HUMAN BIOLOGY: EPIGENETICS, GENETICS AND EVOLUTION

BIO346, SUMMER 2019

Instructor: Dr. Steve Phelps

Email: sphelps@mail.utexas.edu
Cell phone: To be updated onsite
Meeting times: M-Th, 11:45-1:45pm
Office hours: M, W 2-3pm
This is a fun but challenging course that surveys molecular and evolutionary underpinnings of
human biology. The course is intended as a broad summary and synthesis of diverse areas of
biology with an emphasis on its unifying themes. Exams emphasize integration and application
of concepts, and are designed to mimic the range of questions students can expect to find on
pre-professional exams, like the Medical College Admissions Test. Our major learning
objectives are to:
 Understand the molecular basis of interactions between genomes and environments
 Understand how societal factors (poverty, stress, trauma) influence genome function
 Understand genetic contributions to health and disease
 Describe how human movements have shaped genetic variation
 Detect evidence of natural selection within the genome
 Understand how selection and mutation shape health and disease
 Identify the relationship between the human body and its homologs in other species
 Understand the continuity of biological processes across timescales
The material of the course is organized into three broad themes: 1. Epigenetics, genetics and
society. 2. Human diversity, and 3. Becoming human. The class will use principles of biology
that are derived from work with many species and apply them to understanding human form and
function. The first theme, Epigenetics, genetics and society, discusses how our experience in
the world interacts with our genome. Human diversity explores concepts of geographic variation,
population evolution and recent human history. The last theme, Becoming human, moves
progressively farther back in time to understand the origins of the structure of our bodies and
the cells that comprise them. The class assumes students having successfully completed
the pre-requisite course BIO325, or its equivalent (or ANT 301 for students enrolled in ANT
348K). To evaluate your understanding, there will be three in-class exams, three quizzes, and a
variety of active learning activities related to excursions. Grades will be weighted as follows.
Exam 1, Thursday, June 6 100pts
Exam 2, Monday, June 20 100pts
Exam 3, Thursday, July 4 100pts
In-class activities and quizzes 100pts
Attendance, participation, in-class work 50pts
Excursion-related activities, 50pts
Total 400pts
Scores on tests are curved. We will calculate the curve by averaging the top two scores in the
class, subtracting the average from 100, and adding the difference to everyone’s exam scores.
The total percentage of correct items will be converted into letter grades using the familiar
standards: A = 90.0-100, B = 80.0-89.9, C = 70.0-79.9, and D = 60.0-69.9.
The course reading is a mixture of scientific and popular press articles distributed as pdf files.
These pdfs are available in the course Canvas site. Underlined readings are original scientific
papers (so allow more time to get through them). Items in blue refer to excursions.

Date Theme 1: Epigenetics, genetics & society Student reading

T, 5/28 Arrival, dinner Taladrid 2019, Wilfird 2012

W, 5/29 Orientation, Santander excursion Mukherjee 2016, Bonetta 2008,
Th, 5/30 Introduction: Epigenetics, genetics & evolution Barker 1990, Sapolsky 2005
F, 6/1 Excursion: The caves of Cantabria

M, 6/3 Transgenerational epigenetics Rando 2015, Heijmans et al 2008

T, 6/4 Heritability, mutation and association Wray 2008, Kruglyak 2008
W, 6/5 GWAS, “missing heritability” & PRS Maher 2008
Th, 6/6 Exam 1

Theme 2: Human diversity

M, 6/10 Holiday
Cavalii-Sforza 1969
T, 6/11 Ancient migrations: The peopling of Europe Barbujani and Colonna 2010
W, 6/12 Evolution of agriculture and diet Curry 2013
Th, 6/13 Two million years of culture: the genus Homo Pontzer 2012

M, 6/17 Hominin skeletons

T, 6/18 Neanderthals: genomes, epigenomes, culture Mooallem 2017
W, 6/19 Review
Th, 6/20 Exam 2
F, 6/21 Excursion: Atapuerca & Museum of Human Evol.

Theme 3: Becoming human

M, 6/24 Big brains and phylogeny: Homo and beyond Pollard 2009, Boyd Silver 2015
T, 6/25 Breath, voice & language
W, 6/26 Tetrapod limbs to opposable thumbs Riddle & Tabin 1999; Prabhakar Noonan 2008
Th, 6/27 Limbs continued
F-Su, Overnight Excursion: Madrid, National Museum
6/28-6/30 of Archaeology, National Museum of Natural
Sciences

M, 7/1 Faces and jaws Sheehan Nachmann 2014

T, 7/2 Tissues, cells & nuclei Mattiroli Luger 2017
W, 7/3 Review
Th, 7/4 Exam 3
 Pre-departure reading

Because this is a summer class that will go at a fast pace, I recommend you do t he following several
readings in advance. The good news is that they are short and 3 of 4 are pretty fun too.
Spain’s Open Wounds
Decades after Franco’s regime, the country’s citizens continue
to unearth the crimes of the past.
By Stephania Taladrid, The New Yorker, January 2019
January 10, 2019

The Valley of the Fallen, one of the largest mass graves in Europe, has a discomforting
air of normalcy and represents the public’s still muddled opinions of the Franco regime.
Photograph by Miquel Gonzalez / Laif / Redux

In the course of thirty-seven years, María Martín López sent more than a
hundred handwritten letters to the Spanish authorities. She wrote to King
Juan Carlos I, and to his successor; to half a dozen Prime Ministers; to
judges of the Spanish Supreme Court; and to all the “bigwigs” she could
think of. The letters, written in cursive and punctuated by misspellings,
made a single request: the right to exhume her mother’s remains, which
had lain in a mass grave since 1936. Her father, Mariano Martín de la Cruz,
had until his last days sought to give his wife a dignified burial. “You’ll take
her to the cemetery when pigs fly,” Francoist rebels told him. Invariably,
Martín López signed off her letters as “the woman who is still waiting for
pigs to fly.”

Martín de la Cruz, a reaper, met Faustina López in Pedro Bernardo, a town

in the center-west province of Ávila. Flanked by a valley and a mountain
range, Pedro Bernardo is known for its vast, picturesque views over the
Tiétar River. In September of 1936, barely two months after the military
coup that marked the beginning of the Spanish Civil War, Francoist troops
occupied the town. Martín López’s parents were not especially political, but
they had wed, in 1921, in a discreet civil ceremony in France, and they upset
Catholic sensibilities by refusing to remarry before the Church in Spain. On
September 20th, while Martín de la Cruz hid in a town north of Pedro
Bernardo, Francoist vigilantes detained his wife and two other women in
the local girls’ school, shaved their heads, and paraded them naked around
the town.

At the time, Martín López was six years old. The last memory she has of her
mother’s abduction is of her older sister trying to prevent it and being
pushed aside by a member of the Civil Guard with the stock of his rifle. A
day later, the remains of two women and four men were found on a
roadside a few miles away from her home, by a tributary that flows into the
river. As the two sisters waited for their father’s return, in the care of an
aunt, they were forced to expiate the sins of their deceased mother. For
several years after, vigilantes often disciplined Martín López with a dose of
six chili peppers and half a liter of castor oil, a noxious laxative. Every now
and then, she found herself running from a throat-slitting gesture or a gun-
wielding guard.

During those years, Martín López refused to disclose any details about her
suffering, or about her mother’s murderers, many of whose children
attended school with her. Only silence would guard her family from another
loss, she thought. When the war came to an end, in the spring of 1939, and
Francoist troops emerged victorious, the murder of Martín López’s mother
was relegated to a buried chapter in history. Francisco Ferrándiz, a cultural
anthropologist working with the Spanish National Research Council, has
described the treatment of Franco’s victims as a “funerary apartheid,”
manifest in the more than two thousand mass graves scattered across
Spain.

Martín López’s story—and those of many victims of Franco’s regime—have

been captured in the documentary “The Silence of Others,” which was
produced by Pedro Almodóvar and shortlisted for an Academy Award
nomination. The idea for the film emerged when its directors, Almudena
Carracedo and Robert Bahar, watched the scandal over thousands of
children who were abducted during Franco’s dictatorship unfold. The topic
resonated strongly with Carracedo and Bahar, who were first-time parents,
and they moved to Madrid to work on the project. “We wanted to spark a
conversation anew, because the common view is that these are trite subjects
and that we need to move on,” Carracedo told me. “But so many people are
suffering because they can’t forget, and they cannot be forced to forget.”

The exhumation, in Villamayor de los Montes, of forty-five political prisoners who

were executed and buried in a mass grave in 1936 by Franco’s rebels. Photograph by Clemente
Bernad / Contrasto / Redux

The hope to restore historical memory is, in many ways, a response to the
Franco regime’s appropriation of the past. For decades after the Civil War,
children were taught that the 1936 coup was justified; later on, the notion
that both sides were equally to blame for the war’s atrocities gained ground.
(The conflict claimed an estimated five hundred thousand lives.) Even
today, historical reckoning is often seen as an undesirable subject, inciting
resentment and sorrow. For the most part, Spanish governments have
remained on the sidelines of these debates, although Pedro Sánchez, the
leader of the Spanish Socialist Workers’ Party, who became Prime Minister
last June, has pledged to redeem the memory of the vanquished. Until he
does, testimonies such as those presented in “The Silence of Others”
constitute a plea against inaction. The film raises “a human-rights
question,” Bahar told me. “What is the reason, in a democracy with forty
years of trajectory behind it, that María Martín cannot exhume her
mother’s remains?”

After Franco’s death, in 1975, Spaniards bid farewell to four decades of

authoritarianism. In the transition period that followed, the country
embraced the notion that only a clean slate could prevent the recurrence of
conflict. The result was the 1977 Amnesty Law, which was negotiated by the
regime’s most moderate factions and a handful of opposition parties, and
which stipulated that no one would be brought to account for crimes
committed during the Civil War or the dictatorship. In the decades after,
amnesty lawsbecame common in countries emerging from military
regimes. Many of those laws would be overturned, but Spanish authorities
have continued to defend their version as a pillar of democracy.

Martín López wrote her first letter to a government official around the time
that the amnesty was enacted. She was living with her husband and their
three teen-age children in a town a few miles away from Pedro Bernardo,
on the opposite side of the Tiétar River. It seemed like the right time to
request her mother’s exhumation: forty years had passed since Faustina
López’s murder, and, after Franco’s death, relatives of Republican victims
had dug up mass graves across the country. Before long, though, it became
clear that Spanish authorities would evade responsibility for the
exhumations. Martín López continued to write two or three letters a year in
hopes of getting a response.

In 2004, Martín López attended a mass-grave exhumation in the nearby

town of Casavieja and met Emilio Silva, a co-founder of the Association for
the Recovery of Historical Memory. Silva, an indefatigable man of fifty-
three, belongs to a generation known for its fight against Spain’s so-called
Pact of Forgetting; his own grandfather was killed by a Francoist death
squad. A few years after they met, Silva encouraged Martín López to join
the A.R.H.M., alongside thirteen other associations of victims’ families, in
filing a petition before the National Court to look into the killings of a
hundred and fourteen thousand people between 1936 and 1951.
In 2008, the criminal investigation—the first of its kind at the national
level—was taken up by Judge Baltasar Garzón, who had gained notoriety in
the late nineties for attempting to prosecute the Chilean former dictator
Augusto Pinochet. Garzón now denounced his own country’s past, asserting
that “impunity has been the rule in dealing with events that could qualify as
crimes against humanity.” The investigation provoked a public outcry, and
prosecutors argued that retroactive justice could not be applied to the
crimes in question. Within a month, Garzón was forced to drop the case,
and he found himself on trial years later, when two far-right associations
accused him of violating the Amnesty Law. The case against Garzón,
opened in 2009, was the only instance in which a top Spanish court has
heard the testimonies of victims of the Civil War and the dictatorship.

Martín López, along with dozens of other petitioners, testified before

Spain’s Supreme Court in the winter of 2012. Then eighty-two, she wore a
black robe and tied her frizzy white hair in a loose bun. She seemed
overwhelmed by the circumstances—her voice faltered, her breath was
unsteady, and she often palmed her eyes. Several minutes into her
testimony, she was cut short by the presiding judge and returned to her
seat, wheeling her walker slowly. She had been determined to testify in
court because Garzón had written an unusually empathic response to one of
her letters. “We cannot look to the future, unless we do so on the basis of a
known and healed past,” he wrote. “At the very least, we cannot do so with
dignity. Dignity like that which you represent, and so many of us lack.”

At the time of the trial, Martín López also began to correspond with Ana
Messutti, an Argentine lawyer. Messutti represented several plaintiffs who
filed a lawsuit on Franco-era crimes before a criminal court in Buenos
Aires. To Martín López, the Argentine lawsuit represented a last resort. The
two women met at Martín López’s house to talk about her attempts to
recover her mother’s remains. The closest she had come was in 1999, when
the road next to the grave was being renovated, and she had joined the son
of another victim in requesting an exhumation. The man’s brother sent a
letter to a local judge, urging her to scuttle their petition; he had married
the daughter of one of his father’s murderers, and preferred not to unearth
the past. The judge abstained from initiating proceedings and the
construction moved forward. Martín López got in the habit of tucking a
bunch of flowers into the guardrail of the road.
María Martín López sits by the road which covers the mass grave containing her mother’s
remains.
Photograph by Almudena Carracedo

Martín López had planned to meet Messutti again in the summer of 2014,
but she died that July. Messutti, who mourned the news with a sense of
helplessness, thinks of the Argentine lawsuit as a “large, transatlantic ship
slowly making its way across the ocean.” The lawsuit now represents three
hundred and twenty plaintiffs and eight different types of victims, including
many who were repressed during the dictatorship. Chato Galante, a rangy
man of seventy, was imprisoned in his twenties for opposing Franco’s
regime. He spent six years in jail and was routinely tortured by a policeman
known as Billy the Kid, who now lives peacefully in Madrid. “We’ve been
told that we shouldn’t reopen old wounds,” Galante told me. “But do you
think someone has bothered to ask if they’ve actually closed?”

An Argentine judge, invoking the same principle of universal jurisdiction

that Garzón used against Pinochet, has issued several arrest warrants,
which Spanish courts have refused to act upon. But Messutti remains
hopeful. She believes that the lawsuit is raising awareness of Francoist
crimes outside of Spain, and, above all, offering victims an alternative
avenue for justice. When Martín López died, her daughter, María Ángeles
Martín, went back to a white, crocheted bag that held certified mail
receipts of her mother’s letters and hand-drawn maps to the grave. She
transcribed a sample of letters and took them to the Argentine consulate in
Madrid. “There are victims on both sides, and sorrow on both sides, too,”
she told me. “But those on the winning side of the war were able to take the
remains of their people home, while the others weren’t even allowed to pick
them up. I’m not asking for anything the other side didn’t receive.”

On the third floor of a mustard-colored house, barely a thousand feet from

the Bernabéu stadium, in Madrid, lies the headquarters of the National
Francisco Franco Foundation. Nothing on the outside of the complex
suggests a Francoist stronghold. On the inside, however, the dictator’s
presence is ubiquitous—in gilt-framed portraits and sculpted busts; in
black-and-white photographs and ornate tapestries; and in military insignia
and embossed poems. Juan Chicharro, a stout man with a scruffy gray
beard, took over the foundation last March. (“In the midst of the battle,” as
he put it.) Since Pedro Sánchez was sworn in as Prime Minister, in June, he
has insisted that there can be no place for Francoism in Spain’s democracy.
He has pledged to amend the Historical Memory Law, a 2007 bill that
many victims’ groups saw as insufficient; to transfer Franco’s remains from
a state-funded mausoleum; and is also studying a ban of the Franco
Foundation.

Over the years, critics have condemned the foundation’s glorification of

Franco and its links to conservative governments. Chicharro, a former
military general and assistant to King Juan Carlos I, insists that the
foundation has no other mission than to defend the figure of Franco, as well
as Spain’s “historical truth.” At the heart of his argument is the notion that
both the victors and the vanquished have their victims. “My grandmother
was a young widow, my mother was orphaned, and my father lost three
brothers and a sister. But we had forgotten it—I would even say we had
forgiven it!” he told me. There was a brief pause. “Well, I don’t know
whether we had forgiven it. But we had certainly forgotten it.”

In September, the Spanish Parliament approved plans to exhume Franco’s

remains, which lie in the Valley of the Fallen, a monument that sits an hour
outside Madrid. But the government has struggled to reach an agreement
with the Franco family on where the next burial site will be. Franco’s
relatives have appealed Sánchez’s decision before the National Court and
argued that the only other place where Franco can rest, outside of the
Valley of the Fallen, is in their family crypt at Almudena Cathedral, in
Madrid.

The administration is also looking into ways of redefining the mausoleum.

“The Valley of the Fallen cannot be considered national patrimony, because
a section of the population does not recognize it as their own,” José Guirao,
the Minister of Culture, told me. Thousands of Republican prisoners
constructed the monument, which was ordered by Franco in 1940 and took
nineteen years to complete. The project was conceived as a monument to
war victims—it houses the remains of at least thirty-three thousand
people—but they are stacked into chambers, inaccessible to visitors, while
the tombs of Franco and José Antonio Primo de Rivera, the founder of the
Falange, are marked. This did not seem to disturb Chicharro. “What better
symbol of reconciliation can there be?” he said. “The remains of people on
both sides lay there, hugging, under a cross. They are intertwined.”

The monument’s appeal suggests the public’s still muddled opinions of the
Franco regime. On a recent Sunday, a long line of cars waited in front of a
wrought-iron fence to pay the admission fee. Past the entrance, a forested
road wound through a former concentration camp to the monument, where
a five-hundred-foot stone cross towered over the front esplanade. The site—
one of the largest mass graves in Europe—has a discomforting air of
normalcy. Hundreds of people had swarmed to celebrate Mass and gather
for a drink or a pincho at the cafeteria. Lottery sellers strolled the grounds,
and kids played around ranks of pines. The valley is managed by
Benedictine monks, who have pledged to maintain “the cult with
splendour” and who recently declared that the government would be barred
from entering the monument’s grounds until it reached an agreement with
Franco’s relatives.

Inside, in a basilica crowned by a dome of golden tesserae, is a large figure

of Christ mounted on a juniper cross, which was handpicked by Franco. The
dictator’s remains lie on the east side of the altar. Dozens of people were
gathered around his tombstone; some kneeled to place a fresh bouquet of
flowers, while others exchanged thoughts in murmurs. A group of teen-
agers giggled; one of them raised a middle finger at the tomb and captured
the gesture with his smartphone. A few minutes later, a man in his sixties
asked a stranger for a picture and stood in front of the tombstone. His
parents, who were from the countryside, had moved to a government-
subsidized house in Madrid during the dictatorship. “His exhumation
should be decided in a referendum,” the man whispered. “But only those
ages fifty and above should be allowed to vote, not the eighteen-year-olds.”

Adjacent to Franco’s tombstone were two small chapels, which held a pair
of unadorned brass crosses that read “Fallen for God and for Spain 1936-
1939 RIP.” No other mentions of the victims could be found. An official
guide, published by the office of National Patrimony, supplied little
historical context on the monument, focussing instead on its aesthetics,
which are “due to Francisco Franco, the Spanish Head of State from the
time of the Spanish Civil War (1936-1939) until his death.” In front of a gift
store, a representative of the office of National Patrimony explained that
the number of visitors had increased significantly since June, when Sánchez
announced his intention to exhume Franco’s remains. Like everyone else,
she seemed unsure about the timing. “One day, we’ll come in to work, and
they will have taken him,” she said.

This year, Martín López’s daughter, María Ángeles Martín, is touring

Europe with Carracedo, the documentary director, to present “The Silence
of Others.” Her mother’s passing became the wellspring of Martín’s
activism. At fifty-four, she wears a serious, attentive expression, and she
has a penchant for floral-patterned clothes. Sitting at a sun-drenched
terrace in Madrid, she recalled that it was the apologetic tone of her
mother’s letters that moved her to action. “They all end with, ‘I am very
sorry for the inconvenience’; ‘Please forgive me if I have offended you in
any way’; ‘I am not seeking revenge,’ ” she told me.

It pains her to think that her mother could have seen herself as a nuisance.
At home, Martín López was often silent, a discipline that marked her
childhood in Pedro Bernardo. When she began to collaborate with victims’
groups, she slowly opened up. “She felt like talking about it whenever she
asked us to drive her to an exhumation ceremony, or to join her in placing
fresh flowers on the site of her mother’s grave,” Martín said. “In those small
moments, she would tell us her story.” Martín López passed on to her
daughter the same burden that she inherited from her father. It is one that
generations in Spain have handed down while government after
government fails to provide closure.

Martín refuses to write any more letters. “If the royal family, if our
government, and our Prime Minister are not responsible, or cannot be
made responsible, where should we go?” Since joining the Argentine
lawsuit, she has written one letter, addressed to Martín López on the
anniversary of her death. With a shaky voice, Martín recalled the words she
read at the cemetery in front of her family. “Because you wrote so many
letters, I’ll address one to you today. I wish I could say that I recovered the
remains, and that you are now resting side by side with my grandmother,
but it is, sadly, not the case.” She went on, “We’re still here, though. Let’s
see what I’ll have to say next year.”

 Stephania Taladrid is a member of The New Yorker’s editorial staff.

4/30/2019 With Science, New Portrait of the Cave Artist - The New York Times

https://nyti.ms/Lctupx

SCIENCE

With Science, New Portrait of the Cave Artist

By JOHN NOBLE WILFORD JUNE 14, 2012
Stone Age artists were painting red disks, handprints, clublike symbols and geometric patterns on European cave
walls long before previously thought, in some cases more than 40,000 years ago, scientists reported on Thursday,
after completing more reliable dating tests that raised a possibility that Neanderthals were the artists.

A more likely situation, the researchers said, is that the art — 50 samples from 11 caves in northwestern Spain —
was created by anatomically modern humans fairly soon after their arrival in Europe.

The findings seem to put an exclamation point to a run of recent discoveries: direct evidence from fossils that
Homo sapiens populations were living in England 41,500 to 44,200 years ago and in Italy 43,000 to 45,000 years
ago, and that they were making flutes in German caves about 42,000 years ago. Then there is the new genetic
evidence of modern human-Neanderthal interbreeding, suggesting a closer relationship than had been generally
thought.

The successful application of a newly refined uranium-thorium dating technique is also expected to send other
scientists to other caves to see if they can reclaim prehistoric bragging rights.

In the new research, an international team led by Alistair W. G. Pike of the University of Bristol in England
determined that the red disk in the cave known as El Castillo was part of the earliest known wall decorations, at a
minimum of 40,800 years old. That makes it the earliest cave art found so far in Europe, perhaps 4,000 years older
than the paintings at Grotte Chauvet in France.

The handprints common at several of the Spanish caves were stencils, probably made by blowing pigment on a hand
placed against the cave wall. The oldest example, at El Castillo, proved to be at least 37,300 years old, which the
scientists said “considerably increases the antiquity of this motif and implies that depictions of the human hand were
among the oldest art known in Europe.”

At Altamira, the researchers obtained a date of at least 35,600 years for a red club-shaped symbol.
Archaeologists said this indicated that Altamira’s artistic tradition started about 10,000 years earlier than once
estimated, and the cave appeared to have been revisited and painted many times over a span of 20,000 years.

In a report published online in the journal Science, Dr. Pike and his colleagues noted that the oldest dated art is
“nonfigurative and monochrome (red), supporting the notion that the earliest expression of art in Western Europe
was less concerned with animal depiction and characterized by red dots, disks, line and hand stencils.” The more
stunning murals of bison and horses came gradually, later.

Although the early dates coincide with recent evidence of a Homo sapiens presence in Europe, the scientists
wrote that because 40,800 is only a minimum age, “it cannot be ruled out that the earliest paintings were symbolic
expressions of the Neanderthals,” who were living in that part of Spain until at least 42,000 years ago.

https://www.nytimes.com/2012/06/15/science/new-dating-puts-cave-art-in-the-age-of-neanderthals.html 1/3
4/30/2019 With Science, New Portrait of the Cave Artist - The New York Times

These close relatives of modern humans had lived in Europe and parts of Asia since at least 250,000 years ago,
becoming extinct about 30,000 years ago.

In another article for the journal, John Hellstrom of the University of Melbourne in Australia, an authority on
dating prehistoric artifacts, praised the research. “The scope of their study has allowed them to unambiguously
identify a number of examples that challenge and overturn the previous understanding of that art’s origin,” he wrote.

Dr. Hellstrom said that “3 of the 50 examples dated show art to have been created in Spain at around (indeed
possibly before) the time of the arrival of modern humans, bringing current ideas of the prehistory of human art in
southern Europe into question.”

In a teleconference for reporters on Wednesday, Dr. Pike said the older dates suggested three possible
interpretations. One: Homo sapiens entered Europe with the tradition of cave art already part of the culture. There is
increasing evidence that the African ancestors of Homo sapiens had for thousands of years developed expressions of
symbolic thinking in the form of perforated beads, engraved eggshells and decorative pigments. Such has been the
standard hypothesis.

Another possibility is that this artistic culture arose shortly after modern humans reached Europe. “It might have
been the result of competition for resources with Neanderthals,” Dr. Pike said. “The rate of cultural innovation was
accelerating, and this was a byproduct.”

The third possibility, which the scientists said they had not anticipated at the start of their project, is that some of
these earliest works of cave art might be attributed to Neanderthals. Until recently, archaeologists usually saw
Neanderthals as incapable of creating artistic works much beyond simple abstract markings and personal
ornamentation.

Other scientists were expected to be skeptical, pending more evidence of even earlier dates for cave art or of
painting associated with Neanderthal tools or fossils. Eric Delson, a paleoanthropologist at Lehman College of the
City University of New York, said, “There is no need to hypothesize that Neanderthals created these paintings, as we
have evidence of artistic Homo sapiens already in Western Europe.”

But João Zilhão, a prehistorian and Neanderthal specialist at the University of Barcelona and a member of the
research team, made a forceful defense of the hypothesis in the teleconference.

“We have sufficient evidence to the effect that Neanderthals possessed a symbolic culture,” Dr. Zilhão said. “They
are close enough to modern humans to have interbred with us. This is sufficient to think about Neanderthals as
fundamentally human beings with perhaps racial differences.”

Saying that Neanderthals were “more advanced than they have been given credit” for, Dr. Zilhão conceded that
their identification with any of the cave art “cannot be proven at this time. It’s just my gut feeling, and needs dates
older than 42-, 43-, 44,000 years to sort it out.”

The research is “most important,” Dr. Delson said, because it introduces a significant advance in techniques for
more reliable, precise and older dating of antiquities, especially cave art that in most cases does not lend itself to
reliable dating by radiocarbon methods. Dr. Hellstrom, in his article, recommended a wider application of the
improved uranium-thorium dating method.

It is actually a 50-year-old technique, but a vastly improved one. Cave art is typically found in limestone terrain.
Water seeping into caves leaves deposits of calcium carbonate, or calcite, as stalactites and stalagmites or simpler

https://www.nytimes.com/2012/06/15/science/new-dating-puts-cave-art-in-the-age-of-neanderthals.html 2/3
4/30/2019 With Science, New Portrait of the Cave Artist - The New York Times

crusts cover cave surfaces. To date a painting under such a crust, researchers remove a piece of the calcite, dissolve
the sample and extract the traces of uranium and thorium atoms. Over time, the uranium in the crust decays into
thorium. A measure of the ratio of uranium to thorium gives the minimum age of the art beneath the crust.

This is an improvement over radiocarbon dating, which becomes less reliable at ages over 30,000 years and is
not usable in dating art unless the pigment contained carbon. The uranium-thorium method has been made more
sensitive, so that calcite samples about as small as a grain of rice can do the job.

Asked how the Neanderthal question could be resolved, Dr. Pike said, “Simply go back and date more of these
samples and find something that predates modern humans in Europe.”

A version of this article appears in print on June 15, 2012, on Page A1 of the New York edition with the headline: With Science, New Portrait of the
Cave Artist.

© 2019 The New York Times Company

https://www.nytimes.com/2012/06/15/science/new-dating-puts-cave-art-in-the-age-of-neanderthals.html 3/3
Annals of Science MAY 2, 2016 ISSUE
Same but Different
How epigenetics can blur the line between
nature and nurture.
By Siddhartha Mukherjee
On October 6, 1942, my mother was born twice
in Delhi. Bulu, her identical twin, came first,
placid and beautiful. My mother, Tulu, emerged
several minutes later, squirming and squalling.
The midwife must have known enough about
infants to recognize that the beautiful are often
the damned: the quiet twin, on the edge of
listlessness, was severely undernourished and
had to be swaddled in blankets and revived.
The first few days of my aunt’s life were the
most tenuous. She could not suckle at the
breast, the story runs, and there were no infant
bottles to be found in Delhi in the forties, so she
was fed through a cotton wick dipped in milk,
and then from a cowrie shell shaped like a
spoon. When the breast milk began to run dry,
at seven months, my mother was quickly
weaned so that her sister could have the last
remnants.
Tulu and Bulu grew up looking strikingly similar:
they had the same freckled skin, almond-
shaped face, and high cheekbones, unusual
among Bengalis, and a slight downward tilt of
the outer edge of the eye, something that
Italian painters used to make Madonnas exude
a mysterious empathy. They shared an inner
language, as so often happens with twins; they
had jokes that only the other twin understood.
They even smelled the same: when I was four or
five and Bulu came to visit us, my mother, in a
bait-and-switch trick that amused her endlessly,
would send her sister to put me to bed;
eventually, searching in the half-light for
identity and difference—for the precise map of
freckles on her face—I would realize that I had
been fooled.
But the differences were striking, too. My
mother was boisterous. She had a mercurial
temper that rose fast and died suddenly, like a
gust of wind in a tunnel. Bulu was physically
timid yet intellectually more adventurous. Her
mind was more agile, her tongue sharper, her
wit more lancing. Tulu was gregarious. She
made friends easily. She was impervious to
insults. Bulu was reserved, quieter, and more
brittle. Tulu liked theatre and dancing. Bulu was
a poet, a writer, a dreamer.
Over the years, the sisters drifted apart. Tulu
married my father in 1965 (he had moved to
Delhi three years earlier). It was an arranged
marriage, but also a risky one. My father was a
penniless immigrant in a new city, saddled with
a domineering mother and a half-mad brother
who lived at home. To my mother’s genteel
West Bengali relatives, my father’s family was
the embodiment of East Bengali hickdom: when
his brothers sat down to lunch, they would pile
their rice in a mound and punch a volcanic
crater in it for gravy, as if marking the insatiable
hunger of their village days. By comparison,
Bulu’s marriage, also arranged, seemed a vastly
safer prospect. In 1967, she married a young
lawyer, the eldest son of a well-established clan
in Calcutta, and moved to his family’s sprawling,
if somewhat decrepit, mansion.
***
By the time I was born, in 1970, the sisters’
fortunes had started to move in unexpected
directions. Calcutta had begun its spiral into
hell. Its economy was fraying, its infrastructure
crumbling. Internecine political movements
broke out frequently, closing streets and
businesses for weeks. Between the city’s cycles
of violence and apathy, Bulu’s husband kept up
the pretense of a job, leaving home every
morning with the requisite briefcase and tiffin
box, but who needed a lawyer in a city without
laws? Eventually, the family sold the mildewing
house, with its grand veranda and inner
courtyard, and moved into a three-room flat.
My father’s fate mirrored that of his adoptive
city. Delhi, the capital, was India’s
overnourished child, fattened by subsidies,
grants, and the nation’s aspirations to build a
mega-metropolis. Our neighborhood, once
girded by forests of thornbushes and overrun
with wild dogs and goats, was soon transformed
into one of the city’s most affluent pockets of
real estate. My family vacationed in Europe. We
learned to eat with chopsticks, twisted our
tongues around the word “croissant,” and
swam in hotel pools. When the monsoons hit
Calcutta, the mounds of garbage on the streets
clogged the drains and turned the city into a
vast, infested swamp. A stagnant pond,
festering with mosquitoes, collected each year
outside Bulu’s house. She called it her own
“swimming pool.”
Why are identical twins alike? In the late
nineteen-seventies, a team of scientists in
Minnesota set out to determine how much
these similarities arose from genes, rather than
environments—from “nature,” rather than
“nurture.” Scouring thousands of adoption
records and news clips, the researchers gleaned
a rare cohort of fifty-six identical twins who had
been separated at birth. Reared in different
families and different cities, often in vastly
dissimilar circumstances, these twins shared
only their genomes. Yet on tests designed to
measure personality, attitudes, temperaments,
and anxieties, they converged astonishingly.
Social and political attitudes were powerfully
correlated: liberals clustered with liberals, and
orthodoxy was twinned with orthodoxy. The
same went for religiosity (or its absence), even
for the ability to be transported by an aesthetic
experience. Two brothers, separated by
geographic and economic continents, might be
brought to tears by the same Chopin nocturne,
as if responding to some subtle, common chord
struck by their genomes.
One pair of twins both suffered crippling
migraines, owned dogs that they had named
Toy, married women named Linda, and had
sons named James Allan (although one spelled
the middle name with a single “l”). Another
pair—one brought up Jewish, in Trinidad, and
the other Catholic, in Nazi Germany, where he
joined the Hitler Youth—wore blue shirts with
epaulets and four pockets, and shared peculiar
obsessive behaviors, such as flushing the toilet
before using it. Both had invented fake sneezes
to diffuse tense moments. Two sisters—
separated long before the development of
language—had invented the same word to
describe the way they scrunched up their noses:
“squidging.” Another pair confessed that they
had been haunted by nightmares of being
suffocated by various metallic objects—
doorknobs, fishhooks, and the like.
The Minnesota twin study raised questions
about the depth and pervasiveness of qualities
specified by genes: Where in the genome,
exactly, might one find the locus of recurrent
nightmares or of fake sneezes? Yet it provoked
an equally puzzling converse question: Why are
identical twins different? Because, you might
answer, fate impinges differently on their
bodies. One twin falls down the crumbling stairs
of her Calcutta house and breaks her ankle; the
other scalds her thigh on a tipped cup of coffee
in a European station. Each acquires the
wounds, calluses, and memories of chance and
fate. But how are these changes recorded, so
that they persist over the years? We know that
the genome can manufacture identity; the
trickier question is how it gives rise to
difference.
David Allis, who has been studying the
genome’s face for identity and difference for
three decades, runs a laboratory at Rockefeller
University, in New York. For a scientist who has
won virtually all of science’s most important
prizes except the Nobel (and that has been
predicted for years), Allis is ruthlessly self-
effacing—the kind of person who offers to leave
his name on a chit at the faculty lunchroom
because he has forgotten his wallet in the
office. (“We know who you are,” the woman at
the cash register says, laughing.)
As a child, Allis grew up in the leeward shadow
of his sister, a fraternal twin, in Cincinnati, Ohio.
She was the studious one, the straight-A
student; he was the popular kid, the high-school
fraternity president casual about his
schoolwork. “We were similar but different,”
Allis said. At some point in college, though,
Allis’s studies became a calling rather than a
chore. In 1978, having obtained a Ph.D. in
biology at Indiana University, Allis began to
tackle a problem that had long troubled
geneticists and cell biologists: if all the cells in
the body have the same genome, how does one
become a nerve cell, say, and another a blood
cell, which looks and functions very differently?
In the nineteen-forties, Conrad Waddington, an
English embryologist, had proposed an
ingenious answer: cells acquired their identities
just as humans do—by letting nurture
(environmental signals) modify nature (genes).
For that to happen, Waddington concluded, an
additional layer of information must exist within
a cell—a layer that hovered, ghostlike, above
the genome. This layer would carry the
“memory” of the cell, recording its past and
establishing its future, marking its identity and
its destiny but permitting that identity to be
changed, if needed. He termed the
phenomenon “epigenetics”—“above genetics.”
Waddington, ardently anti-Nazi and fervently
Marxist, may have had more than a biological
stake in this theory. The Nazis had turned a
belief in absolute genetic immutability (“a Jew is
a Jew”) into a state-mandated program of
sterilization and mass murder. By affirming the
plasticity of nature (“everyone can be anyone”),
a Marxist could hope to eradicate such innate
distinctions and achieve a radical collective
good.
Waddington’s hypothesis was perhaps a little
too inspired. No one had visualized a gene in
the nineteen-forties, and the notion of a layer
of information levitating above the genome was
an abstraction built atop an abstraction,
impossible to test experimentally. “By the time I
began graduate school, it had largely been
forgotten,” Allis said.
Had Allis started his experiments in the
nineteen-eighties trying to pin down words like
“identity” and “memory,” he might have found
himself lost in a maze of metaphysics. But part
of his scientific genius lies in radical
simplification: he has a knack for boiling
problems down to their tar. What allows a cell
to maintain its specialized identity? A neuron in
the brain is a neuron (and not a lymphocyte)
because a specific set of genes is turned “on”
and another set of genes is turned “off.” The
genome is not a passive blueprint: the selective
activation or repression of genes allows an
individual cell to acquire its identity and to
perform its function. When one twin breaks an
ankle and acquires a gash in the skin, wound-
healing and bone-repairing genes are turned
on, thereby recording a scar in one body but
not the other.
***
But what turns those genes on and off, and
keeps them turned on or off? Why doesn’t a
liver cell wake up one morning and find itself
transformed into a neuron? Allis unpacked the
problem further: suppose he could find an
organism with two distinct sets of genes—an
active set and an inactive set—between which it
regularly toggled. If he could identify the
molecular switches that maintain one state, or
toggle between the two states, he might be
able to identify the mechanism responsible for
cellular memory. “What I really needed, then,
was a cell with these properties,” he recalled
when we spoke at his office a few weeks ago.
“Two sets of genes, turned ‘on’ or ‘off’ by some
signal.”
Allis soon found his ideal subject: a bizarre
single-celled microbe called Tetrahymena. Blob-
shaped cells surrounded by dozens of tiny,
whiskery projections called cilia, Tetrahymena
are improbable-looking—each a hairy
Barbapapa, or a Mr. Potato Head who fell into a
vat of Rogaine. “Perhaps the strangest thing
about this strange organism is that it carries
two very distinct collections of genes,” he told
me. “One is completely shut off during its
normal life cycle and another is completely
turned on. It’s really black-and-white.” Then,
during reproduction, an entirely different
nucleus wakes up and goes into action. “So we
could now ask, What signal, or mechanism,
allows Tetrahymena to regulate one set of
genes versus the next?”
By the mid-nineteen-nineties, Allis had found an
important clue. Genes are typically carried in
long, continuous chains of DNA: one such chain
can carry hundreds of thousands of genes. But a
chain of DNA does not typically sit naked in
animal cells; it is wrapped tightly around a core
of proteins called histones. To demonstrate,
Allis stood up from his desk, navigated his way
through stacks of books and papers, and
pointed at a model. A long plastic tube,
cerulean blue, twisted sinuously around a series
of white disks, like a python coiled around a
skewer of marshmallows.
“Histones had been known as part of the inner
scaffold for DNA for decades,” Allis went on.
“But most biologists thought of these proteins
merely as packaging, or stuffing, for genes.”
When Allis gave scientific seminars in the early
nineties, he recalled, skeptics asked him why he
was so obsessed with the packing material, the
stuff in between the DNA. His protozoan studies
supplied an answer. “In Tetrahymena, the
histones did not seem passive at all,” he said.
“The genes that were turned ‘on’ were
invariably associated with one form of histone,
while the genes that were turned ‘off’ were
invariably associated with a different form of
histone.” A skein of silk tangled into a ball has
very different properties from that same skein
extended; might the coiling or uncoiling of DNA
change the activity of genes?
In 1996, Allis and his research group deepened
this theory with a seminal discovery. “We
became interested in the process of histone
modification,” he said. “What is the signal that
changes the structure of the histone so that
DNA can be packed into such radically different
states? We finally found a protein that makes a
specific chemical change in the histone, possibly
forcing the DNA coil to open. And when we
studied the properties of this protein it became
quite clear that it was also changing the activity
of genes.” The coils of DNA seemed to open and
close in response to histone modifications—
inhaling, exhaling, inhaling, like life.
Allis walked me to his lab, a fluorescent-lit
space overlooking the East River, divided by
wide, polished-stone benches. A mechanical
stirrer, whirring in a corner, clinked on the edge
of a glass beaker. “Two features of histone
modifications are notable,” Allis said. “First,
changing histones can change the activity of a
gene without affecting the sequence of the
DNA.” It is, in short, formally epi-genetic, just as
Waddington had imagined. “And, second, the
histone modifications are passed from a parent
cell to its daughter cells when cells divide. A cell
can thus record ‘memory,’ and not just for itself
but for all its daughter cells.”
***
By 2000, Allis and his colleagues around the
world had identified a gamut of proteins that
could modify histones, and so modulate the
activity of genes. Other systems, too, that could
scratch different kinds of code on the genome
were identified (some of these discoveries
predating the identification of histone
modifications). One involved the addition of a
chemical side chain, called a methyl group, to
DNA. The methyl groups hang off the DNA
string like Christmas ornaments, and specific
proteins add and remove the ornaments, in
effect “decorating” the genome. The most
heavily methylated parts of the genome tend to
be dampened in their activity.
In the ensuing decade, Allis wrote enormous,
magisterial papers in which a rich cast of
histone-modifying proteins appear and
reappear through various roles, mapping out a
hatchwork of complexity. (His twin, Cathy Allis,
is an ace crossword-puzzle constructor, having
supplied Times readers with nearly a hundred
puzzles—an activity that is similar but
different.) These protein systems, overlaying
information on the genome, interacted with
one another, reinforcing or attenuating their
signals. Together, they generated the
bewildering intricacy necessary for a cell to
build a constellation of other cells out of the
same genes, and for the cells to add
“memories” to their genomes and transmit
these memories to their progeny. “There’s an
epigenetic code, just like there’s a genetic
code,” Allis said. “There are codes to make parts
of the genome more active, and codes to make
them inactive.”
And epigenetics could transform whole animals.
“The idea that cells can acquire profoundly
different properties by manipulating their
epigenome was becoming known,” Danny
Reinberg told me. “But that you could create
different forms of a creature out of the same
genome using epigenetics? That was a real
challenge.”
Reinberg’s lab is at New York University’s
School of Medicine. His office— by the East
River around Thirty-first Street—is like Allis’s:
another nest of books and offprints, a wide
river view, and another model of DNA twisted
around histones, although this room is filled
with Reinberg’s private botanical obsession:
huge, overgrown succulents from other climes
that assert themselves with a defiant
muscularity. Intense, articulate, with a
cultivated stubble, Reinberg resembles an
athlete—a gymnast, or a wrestler—whose skill
depends on compaction and repetition. He
grew up in Santiago, Chile, the child of parents
who ran a jewelry business. He scored an A-
minus in his first biochemistry class in college, in
Valparaiso, but felt that he hadn’t really
mastered the material, so he applied to take the
class again. The professor looked at him as if he
were mad before relenting.
Like Allis, Reinberg became interested in
epigenetics in the nineteen-nineties. He
explored how modified histones were copied
when a cell divides, right down to the molecular
level. Allis described Reinberg’s early work as
“some of the most elegant experiments in the
field.” But Reinberg sought a more advanced
instance of epigenetic instruction—a whole
animal, not just a cell, whose form and identity
could be shifted by shifting the epigenetic code.
“So imagine that you tighten some parts of the
DNA and loosen other parts by changing the
structures of the histones,” Reinberg said. “Can
you change the form or nature of an animal
simply by coiling and uncoiling various parts of
its genome?”
One blistering summer day in 2005, Reinberg
found himself stuck in a van ferrying a group of
scientists to an epigenetics meeting outside
Mexico City. “The traffic was jammed for
miles”—he shrugged, signalling South American
resignation—“and I sat next to another
scientist, Shelley Berger, whose work I had long
admired, and we started talking.” Berger, a
molecular biologist who studies epigenetics at
the University of Pennsylvania, had just
returned from Costa Rica, where she had been
looking at ant colonies.
Ants have a powerful caste system. A colony
typically contains ants that carry out radically
different roles and have markedly different
body structures and behaviors. These roles,
Reinberg learned, are often determined not by
genes but by signals from the physical and
social environment. “Sibling ants, in their larval
stage, become segregated into the different
types based on environmental signals,” he said.
“Their genomes are nearly identical, but the
way the genes are used—turned on or off, and
kept on or off—must determine what an ant
‘becomes.’ It seemed like a perfect system to
study epigenetics. And so Shelley and I caught a
flight to Arizona to see Jürgen Liebig, the ant
biologist, in his lab.”
The collaboration between Reinberg, Berger,
and Liebig has been explosively successful—the
sort of scientific story (“two epigeneticists walk
into a bar and meet an entomologist”) that
works its way into a legend. Carpenter ants, one
of the species studied by the team, have
elaborate social structures, with queens (bullet-
size, fertile, winged), majors (bean-size soldiers
who guard the colony but rarely leave it), and
minors (nimble, grain-size, perpetually moving
foragers). In a recent, revelatory study,
researchers in Berger’s lab injected a single
dose of a histone-altering chemical into the
brains of major ants. Remarkably, their
identities changed; caste was recast. The major
ants wandered away from the colony and began
to forage for food. The guards turned into
scouts. Yet the caste switch could occur only if
the chemical was injected during a vulnerable
period in the ants’ development.
***
Since 2012, Reinberg, continuing his partnership
with Berger and Liebig, has been cultivating ant
colonies in his own lab. One afternoon in April, I
put on sky-blue sterile gloves and an apron, and
accompanied a postdoctoral researcher in
Reinberg’s lab, Hua Yan, to the ant room. It is a
neatly kept, gently lighted space with the
slightly dank smell of sugar and dead maggots—
ant food. In a nightmarish inversion of an
American picnic idyll, the ants live inside
Tupperware containers, and the people watch
from outside.
The most mature colonies in Reinberg’s
collection belong to a species called the
“jumping ant,” a pugnacious social insect from
southern India. Like most ant species, jumping
ants segregate into castes. When the queen is
removed from the colony, the workers, sensing
opportunity, launch a vicious, fight-to-the-death
campaign against one another—stinging, biting,
sparring, lopping off limbs and heads, until a
few workers win and become queenlike. The
behavior of these “pseudo-queens,” as
Reinberg calls them, changes dramatically; their
life spans increase. The pseudo-queen (the
scientific term is “gamergate,” not to be
confused with the vicious, fight-to-the-death
campaign against female video-game-makers)
acquires reproductive fecundity, and dominates
the colony.
I looked through a transparent Tupperware lid
at a teeming colony of jumping ants, and
thought, inevitably, of the city around us. The
workers scurried around the edges of the
container with inexhaustible energy, gathering
food and garbage. The gamergates, in contrast,
moved lazily above their brood in the center of
the container. The workers worked. The
gamergates lounged—waking late, moving
little. When a worker approached a gamergate,
the dominant ant Tasered it with her antennae,
warning the worker to keep off her royal
territory. The worker retreated, its antennae
lowered.
“The remarkable thing about workers and
gamergates,” Yan told me, “is that they are
almost genetically identical.” The gene
sequence before and after the transition is the
same. Yet, as DNA methyl groups or histone
modifications get shifted around those gene
sequences, the worker transforms into a
gamergate, and virtually everything about the
insect’s physiology and behavior changes.
“We’re going to solve how the change can have
such a dramatic effect on longevity,” Reinberg
said. “It’s like one twin that lives three times
longer than the other”—all by virtue of a
change in epigenetic information.
The impact of the histone-altering experiment
sank in as I left Reinberg’s lab and dodged into
the subway. (How could I resist the urge, that
spring afternoon, to categorize the passengers
on the No. 6 train into the three basic New
Yorker archetypes: worker, soldier, queen?) All
of an ant’s possible selves are inscribed in its
genome. Epigenetic signals conceal some of
these selves and reveal others, coiling some,
uncoiling others. The ant chooses a life between
its genes and its epigenes—inhabiting one self
among its incipient selves.
Epigeneticists, once a subcaste of biologist
nudged to the far peripheries of the discipline,
now find themselves firmly at its epicenter.
“Fifteen years ago, a meeting on cell biology
would hold a session on histones or DNA
methylation—and no one would be at that
session,” Allis told me. Now there are meetings
on the epigenetics of human memory, of ants,
of cancer, of mental illness. Part of the reason
for the excitement is that epigenes may be
vastly more tractable than genes. “Gene
therapy was all the rage when I began my
career, but manipulating genes has turned out
to be much harder than envisioned,” Allis said.
Genes, after all, are the permanent repository
of a cell’s information system, and thus more
tamperproof. (If genes are hardware, epigenes
are firmware.) But by altering epigenes—the
manner in which DNA is coiled or uncoiled,
methylated or demethylated—one should be
able to alter which genes are activated.
Medical epigeneticists are most excited about
the implications for cancer. In some cancers,
such as leukemias, malignant cells have
markedly aberrant patterns of DNA methylation
or histone modification. “Clearly, there’s a
signal that epigenetic information is important
for a cancer cell,” Allis said. “But can a drug
safely change the epigenome of a cancer cell
without touching a normal cell?” In my own
leukemia- and lymphoma-focussed clinic,
dozens of epigenetic drugs are on trial. Some
alter methylation, while others perturb the
histone-modification system. One woman with
pre-leukemia had a spectacular remission on a
drug called azacitidine, but, oddly, she began to
have sudden spurts of anxiety. Were these
symptoms related to the drug’s effect on the
epigenomes of brain cells?
***
Other researchers, following Reinberg and his
colleagues, have looked at how epigenetics
might change behaviors—not just cellular
memory and identity but an organism’s
memory and identity. The neuroscientist and
psychiatrist Eric Nestler, who studies addiction,
gave mice repeated injections of cocaine, and
found that the histones were altered in the
reward-recognizing region of the brain. When
the histone modification was chemically
blocked, the mice were less likely to become
addicted. In 2004, a team of researchers at
McGill University noticed that rats raised by
low-nurturing mothers were likely to be notably
stressed as young adults. The memory of
childhood neglect in rats appears to be related
to epigenetic changes: a gene that acts as a set
point for stress—an anxiety rheostat—is
dampened in these poorly nurtured rats,
resulting in higher levels of stress hormones.
McGill researchers went on to study the brains
of human beings who were abused as children
and later committed suicide, and found similar
epigenetic alterations.
The medical impact of epigenetics remains to
be established, but its biological influence has
been evident for nearly a decade. Diffuse,
mysterious observations, inexplicable by
classical genetics, have epigenetic explanations
at their core. When a female horse and a male
donkey mate, they produce a longer-eared,
thin-maned mule; a male horse and a female
donkey typically generate a smaller, shorter-
eared hinny. That a hybrid’s features depend on
the precise configuration of male versus female
parentage is impossible to explain unless the
genes can “remember” whether they came
from the mother or the father—a phenomenon
called “genomic imprinting.” We now know that
epigenetic notations etched in sperm and eggs
underlie imprinted genes.
Perhaps the most startling demonstration of the
power of epigenetics to set cellular memory
and identity arises from an experiment
performed by the Japanese stem-cell biologist
Shinya Yamanaka in 2006. Yamanaka was taken
by the idea that chemical marks attached to
genes in a cell might function as a record of
cellular identity. What if he could erase these
marks? Would the adult cell revert to an
original state and turn into an embryonic cell?
He began his experiments with a normal skin
cell from an adult mouse. After a decades-long
hunt for identity-switching factors, he and his
colleagues figured out a way to erase a cell’s
memory. The process, they found, involved a
cascade of events. Circuits of genes were
activated or repressed. The metabolism of the
cell was reset. Most important, epigenetic
marks were erased and rewritten, resetting the
landscape of active and inactive genes. The cell
changed shape and size. Its wrinkles unmarked,
its stiffening joints made supple, its youth
restored, the cell could now become any cell
type in the body. Yamanaka had reversed not
just cellular memory but the direction of
biological time.
It’s one thing to study epigenetic changes
across the life of a single organism, or down a
line of cells. The more tantalizing question is
whether epigenetic messages can, like genes,
cross from parents to their offspring.
The most suggestive evidence for such
transgenerational transmission may come from
a macabre human experiment. In September,
1944, amid the most vengeful phase of the
Second World War, German troops occupying
the Netherlands banned the export of food and
coal to its northern parts. Acute famine
followed, called the Hongerwinter—the hunger
winter. Tens of thousands of men, women, and
children died of malnourishment; millions
suffered it and survived. Not surprisingly, the
children who endured the Hongerwinter
experienced chronic health issues. In the
nineteen-eighties, however, a curious pattern
emerged: when the children born to women
who were pregnant during the famine grew up,
they had higher rates of morbidity as well—
including obesity, diabetes, and mental illness.
(Malnourishment in utero can cause the body
to sequester higher amounts of fat in order to
protect itself from caloric loss.) Methylation
alterations were also seen in regions of their
DNA associated with growth and development.
But the oddest result didn’t emerge for another
generation. A decade ago, when the
grandchildren of men and women exposed to
the famine were studied, they, too, were
reported to have had higher rates of illness.
(These findings have been challenged, and
research into this cohort continues.) “Genes
cannot change in an entire population in just
two generations,” Allis said. “But some memory
of metabolic stress could have become
heritable.”
Both Allis and Reinberg understand the
implications of transgenerational epigenetic
transmission: it would overturn fundamental
principles of biology, including our
understanding of evolution. Conceptually, a key
element of classical Darwinian evolution is that
genes do not retain an organism’s experiences
in a permanently heritable manner. Jean-
Baptiste Lamarck, in the early nineteenth
century, had supposed that when an antelope
strained its neck to reach a tree its efforts were
somehow passed down and its progeny evolved
into giraffes. Darwin discredited that model.
Giraffes, he proposed, arose through heritable
variation and natural selection—a tall-necked
specimen appears in an ancestral tree-grazing
animal, and, perhaps during a period of famine,
this mutant survives and is naturally selected.
But, if epigenetic information can be
transmitted through sperm and eggs, an
organism would seem to have a direct conduit
to the heritable features of its progeny. Such a
system would act as a wormhole for
evolution—a shortcut through the glum cycles
of mutation and natural selection.
***
My visit with Allis had ended on a cautionary
note. “Much about the transmission of
epigenetic information across generations is
unknown, and we should be careful before
making up theories about the kind of
information or memory that is transmitted,” he
told me. By bypassing the traditional logic of
genetics and evolution, epigenetics can arouse
fantasies about warp-speeding heredity: you
can make your children taller by straining your
neck harder. Such myths abound and
proliferate, often dangerously. A child’s autism,
the result of genetic mutation, gets attributed
to the emotional trauma of his great-
grandparents. Mothers are being asked to
minimize anxiety during their pregnancy, lest
they taint their descendants with anxiety-ridden
genes. Lamarck is being rehabilitated into the
new Darwin.
These fantasies should invite skepticism.
Environmental information can certainly be
etched on the genome. But such epigenetic
scratch marks are rarely, if ever, carried forward
across generations. A man who loses a leg in an
accident bears the imprint of that accident in
his cells, wounds, and scars, but he does not
bear children with shortened legs. A hundred
and forty generations of circumcision have not
made the procedure any shorter. Nor has the
serially uprooted life of my family burdened me,
or my children, with any wrenching sense of
estrangement.
In the fall of 2013, Bulu travelled to the United
States. I had not seen her for nearly a decade,
and I drove out to Robbinsville, New Jersey,
with my family to visit her. It was October 6th,
the birthday that she shared with my mother.
She had cooked my favorite meal—shrimp
curry, a signature Tulu dish, tangy with just a
hint of bitterness from lime rind—and the
house smelled of the heady mixture of boiled
shellfish, lime, and the floral brand of hair oil
that both sisters preferred, my private
madeleine. Bulu’s face was leaner and more
angular than I remembered it, but when she
smiled the angles rearranged themselves and
softened into a distant evocation of my
mother’s.
We made our way to the park outside the
house, while the kids played in the garden. The
October light was oblique and sepulchral, a
halo-endowing, New World light that does not
exist in Delhi or Calcutta. There had been an
uncomfortable irony in that Bulu, who loved
adventure, had spent most of her life in the
same stodgy city, while Tulu, an inveterate
homebody, fussy about mattresses and food,
had been dragged across the globe by my
travel-obsessed father. I asked Bulu about her
encounter with America, the adventure of it all.
“Oh, but I’ve been here so many times,” she
said, laughing. “Every time Tulu took a trip
abroad, I bought a guidebook and travelled,
too.” There was something about the remark
that reminded me of my mother. It was almost
rueful, although without the aftertaste of
bitterness. She shared my mother’s lightness
about fate—an equanimity that borders nobility
but comes with no pride.
As we meandered through the park over fallen
leaves, Bulu reminisced about how the
vicissitudes of their lives had reshaped her and
her sister in different ways, while I couldn’t help
noting how fiercely they had converged. In
calculus, the first derivative of a curve at any
point refers not to the position of the point but
to its propensity to change its position; not
where an object is but how it moves. This
shared quality was the lasting link between my
mother and her twin. Tulu and Bulu were no
longer recognizably identical—but they shared
the first derivative of identity.
It is easy to think of twins as comedies of
nature. The rhyming names, the matching sailor
suits, the tomfoolery of mistaken identities, the
two-places-at-the-same-time movie plot—
genetics for gags. But twins often experience
parts of their lives as tragedies of nature. My
mother and her sister grew up in a walled
garden, imagining each other not as friends or
siblings but as alternate selves. They were
separated not at birth but at marriage, as sisters
often are. Jeta Tulur, sheta Bulur, my
grandfather would say: “What is Tulu’s is also
Bulu’s.” But that wistful phrase, a parent’s
fantasy of perfect parity for his children, was
absurd; how could it possibly last? The grief
that twins experience as they drift apart in life is
unique, but it abuts a general grief: if eternal
sameness will not guarantee eternal closeness,
then what hope is there for siblings, or parents,
or lovers?
Why are twins different? Well, because
idiosyncratic events are recorded through
idiosyncratic marks in their bodies. If you
sequence the genomes of a pair of identical
twins every decade for fifty years, you get the
same sequence over and over. But if you
sequence the epigenomes of a pair of twins you
find substantial differences: the pattern of
epigenetic marks on the genomes of their
various cells, virtually identical at the start of
the experiment, diverges over time.
Chance events—injuries, infections,
infatuations; the haunting trill of that particular
nocturne—impinge on one twin and not on the
other. Genes are turned on and off in response
to these events, as epigenetic marks are
gradually layered above genes, etching the
genome with its own scars, calluses, and
freckles. Prospero, raging against the deformed
Caliban in “The Tempest,” describes him as “a
devil, a born devil, on whose nature/Nurture
can never stick.” Caliban is destined to remain a
genetic automaton, a windup ghoul—vastly
more pathetic than anything human. He
experiences the world, but he has no capacity
to be changed by it; he has a genome that lacks
an epigenome.
It is a testament to the unsettling beauty of the
genome that it can make the real world stick.
Hindu philosophers have long described the
experience of “being” as a web—jaal. Genes
form the threads of the web; the detritus that
adheres to it transforms every web into a
singular being. An organism’s individuality,
then, is suspended between genome and
epigenome. We call the miracle of this
suspension “fate.” We call our responses to it
“choice.” We call one such unique variant of
one such organism a “self.”
A strange thing happened on the way out of
Reinberg’s ant room. One of the ants leaped out
of the Tupperware box onto my shirt. There was
a momentary commotion—“They bite,” Yan
said, matter-of-factly—and then we found the
ant on my shoulder, making a desperate break
for my ear. Yan pulled out a pair of forceps and,
after a few attempts, she was returned to the
colony.
The retrieval had been masterfully delicate, but
the ant was injured: a leg had been bruised, and
she waddled lopsidedly for a while. The wound
would heal, I knew, but a scar would remain.
She had done it: she had made difference out of
similarity. The clone was somehow no longer
quite a clone. I watched her make her way back
to the colony—the One That Almost Got Away,
to be memorialized in song and verse—until she
vanished into the metropolis of soldiers,
workers, and queens. ♦
THE MAIL
Letters from our readers.
THE REGULATORS
Siddhartha Mukherjee’s article about twins and
epigenetics misrepresents the processes by
which genes are regulated and how the
environment influences the genome (“Same but
Different,” May 2nd). Mukherjee centers his
article on the work of David Allis and Danny
Reinberg, who think that “epigenetic”
mechanisms play a causative, instructive role in
gene regulation. But many researchers consider
these mechanisms to be downstream
processes, secondary to the work of proteins
called transcription factors, which turn genes on
or off. Ignoring the vast body of work on gene
regulation from the past half century,
Mukherjee gives the lay reader the impression
that “epigenetics” is providing new answers to
an unsolved problem in biology, when scientists
already have a very good understanding of how
the environment influences the genome. (And,
rather than referring to a process that functions
“above genetics,” the term “epigenetic” was
introduced in the nineteen-thirties as the
adjective form of “epigenesis,” the process by
which structures form de novo in the
developing embryo.) Instead of explaining gene
regulation to readers, Mukherjee introduces
confusion into one of science’s most important
domains, one that touches deeply on who we
are as biological beings.
Florian Maderspacher
Senior Editor, Current Biology
Salt Lake City, Utah
Mukherjee replies:
I agree that the broader context of gene
regulation is critical, and “Same but Different,”
with its focus on recent research into histone
modification and DNA methylation, left out
foundational work by other scientists on
transcriptional activators, repressors, and
regulators, a class I refer to as gene “regulators”
or “master regulators” in my new book, “The
Gene: An Intimate History.” These regulatory
factors (and regulatory RNA) are indeed the
primary mediators of the biological response to
the environment. My book, from which the
article drew, details these fundamentals of gene
regulation—such as Jacques Monod and
François Jacob’s research on gene regulation in
bacteria; Shinya Yamanaka’s ongoing work on
regulatory genes in stem cells; and studies of
transcriptional master regulators, such as SRY,
which lie atop gene-regulatory hierarchies. Such
material would have given readers a fuller, and
sounder, view of gene regulation in response to
various stimuli.
1/17/2017

Epigenomics: The New Tool in Studying Complex Diseases
By: Laura Bonetta, Ph.D. (Science writer based in Garrett Park, MD) © 2008 Nature Education 
Citation: Laura, B. (2008) Epigenomics: The new tool in studying complex
diseases. Nature Education 1(1):178

Identical twins often develop different characteristics, even though they carry the same
sequence of DNA nucleotides. How can this be? The answer lies in epigenomics.

Having one version of a gene rather than another may
render a person more likely to develop cancer or
diabetes, or perhaps to be overweight rather than
slender. However, identical twins, who inherit exactly
the same set of genes from their parents, can end up
with different physical characteristics, behaviors, and
even predispositions to disease. How is this possible?
In short, such differences arise because various
epigenetic factors influence whether certain genes are
active or not, as well as when and where these genes
become active.

DNA methylation and histone modification (Figure 1)
are two primary examples of processes that result in
epigenetic modifications to the genome. Changes linked
to both of these processes accumulate throughout a
person's life and can be passed on from one generation
to the next. Epigenetic mechanisms regulate many
important biological processes, such as those
responsible for cell division and cell differentiation.
These mechanisms are also involved in many diseases,
including cancer and diabetes.
Although a person's complement of genes—in other
words, his or her genome—remains essentially the
same from birth onward (except for the occurrence of
mutations that can change the function of genes),
different environmental exposures during development
and throughout a person's life chemically modify DNA
and the proteins bound to it. Thus, even identical
Figure 1: An illustration of how epigenetic
mechanisms can affect health
Epigenetic mechanisms are affected by several
factors and processes including development in
utero and in childhood, environmental chemicals,
drugs and pharmaceuticals, aging, and diet. DNA
methylation is what occurs when methyl groups,
an epigenetic factor found in some dietary
sources, can tag DNA and activate or repress
genes. Histone modification occurs when the
binding of epigenetic factors to histone “tails”
alters the extent to which DNA is wrapped around
histones and the availability of genes in the DNA
to be activated. All of these factors and processes
may result in cancer, autoimmune disease, mental
disorders, or diabetes, among other illnesses.
National Institutes of Health.

genomes from twins can be subject to distinct Figure Detail
epigenetic changes. For instance, when one group of
researchers analyzed the chromosomes of identical
twins, they found differences in the extent to which the DNA was tagged by methyl groups—a process known as
DNA methylation, which plays a role in regulating the transcription of genes. These researchers also found that
each individual's histones, or the proteins around which DNA winds when it is compacted into chromosomes,
carried different chemical tags. These tags are thought to alter the extent to which DNA is wrapped around the
histones, thereby affecting the availability of genes for activation. The researchers reasoned that these changes
accounted for, at least in part, the various physical and behavioral differences between identical twins (Fraga et
al., 2005).

1/6
1/17/2017

Thus, the field of epigenetics focuses on the study of heritable changes rather than mutations in the primary
DNA sequence. The term "epigenetics" was coined in 1940 by biologist Conrad Waddington, who defined it as
"the interactions of genes with their environment which bring the phenotype into being" (Waddington, 1940).
Over time, the field of epigenetics gave rise to that of epigenomics, which is the study of epigenetic modifications
across an individual's entire genome. Epigenomics has only become possible in recent years because of the
advent of various sequencing tools and technologies, such as DNA microarrays, cheap whole­genome
resequencing, and databases for studying entire genomes (Bonetta, 2008).

Why Is Epigenomics Important?
The human genome contains approximately 25,000 genes that encode all the proteins and other molecules that
make up an individual. The epigenome is important to this vast collection of nucleotides in that it provides
instructions for when some of these proteins are produced, as well as in which cells or tissues production takes
place. In the words of researcher Dana Dolinoy (2007), "A useful analogy is to think of the epigenome as the
software that directs the genomic hardware of a computer." Another way to think of the epigenome is to picture it
as DNA "formatting"—here, if our genome is referred to as the book of life, written in A's, T's, G's, and C's, then
the epigenome represents the spaces and punctuation that format the text. Epigenetic marks are literally made
on the DNA itself, or on the proteins that package it in the nucleus. Epigenetics and epigenomics help explain the
mechanism by which our environment affects our phenotype.

Similar to the recently completed Human Genome Project, which decoded the complete sequence of the human
genome, several large­scale projects and consortia are now underway in an attempt to characterize the many
human epigenomes (Jones et al., 2008). However, such projects will undoubtedly be more challenging than the
Human Genome Project, because the epigenome differs over an individual's lifetime and is affected by several
factors, including environmental chemicals, drugs and pharmaceuticals, aging, and diet. In addition, different
tissues have different epigenetic marks across their genomes. Thus, epigenomics requires the identification of
changes not only among different individuals, but also among different tissues, developmental stages, and
disease types.

Studying DNA Methylation
One of the most studied epigenetic mechanisms is DNA methylation, particularly the process by which a methyl
group is added to a cytosine residue, thereby transforming the cytosine into 5­methylcytosine. Currently,
researchers have many methods for detecting such changes in methylation across the entire genome. Some of
these approaches are described in the following sections.

Methylated DNA Immunoprecipitation

One method employed in the study of methylation involves the use of antibodies to 5­methylcytosine; here, after
cutting up genomic DNA, scientists can use these antibodies to pull out DNA fragments that contain methylated
cytosine. The DNA that is bound to the antibodies is then purified, amplified, labeled with a fluorescent tag, and
applied to the surface of a DNA microarray containing a set of probes from across the genome. When a
fluorescently labeled piece of DNA binds to a particular probe in the microarray, it produces a signal. Scientists
can then use the signal to determine where the probe maps within the genome, thereby revealing where DNA
methylation has occurred.

Weber et al. (2005) were among the first investigators to use this technique—called methylated DNA
immunoprecipitation or MeDIP—to find differences in methylation between normal and cancerous human cell
lines. They determined that the methylation patterns in the two types of cells are similar, but they also noted
defined regions on different chromosomes of differential methylation. Figure 2 shows the methylation of normal
skin cells (specifically, fibroblasts, which are depicted in red) and of SW48 cancer cells (depicted in green). As

2/6
1/17/2017

shown in the figure, one region on the long arm of
chromosome 3 (3q) has less methylation in the cancer
cells. By looking at a genetic database (RefSeq) that
tells where genes are located in the genome, the
scientists determined that the region on chromosome
3q that is hypomethylated in the cancer cell line has few
genes.

Other studies have shown that the promoter regions of
genes become more methylated in cancer cells. In
humans, cytosine often becomes methylated when it is
next to a guanosine—this arrangement is known as a
CpG dinucleotide. There are many CpG dinucleotides
in the genome, and most of them are methylated.
However, groups of unmethylated CpG dinucleotides
often occur in the regulatory regions of genes and are
referred to as CpG islands. Unmethylated CpG islands
Figure 2: Chromosomal profiles of DNA
allow genes to be transcribed when the necessary
methylation in colon cancer cells
transcription factors are present.
Methylation profile in primary fibroblasts (red) and
SW48 cancer cells (green) over a 10­Mb window
on chromosome 3q. The RefSeq gene
annotations below the graph illustrate that the
region of extensive hypomethylation in SW48 cells
coincides with a gene­poor domain.
© 2005 Nature Publishing Group Weber, W. et al.
Chromosome­wide and promoter­specific analyses identify sites
of differential DNA methylation in normal and transformed human
cells. Nature Genetics 37, 856 (2005). All rights reserved. 

Comparative Methylation Hybridization, Bisulfite Treatment, and DNA Sequencing

Another approach for detecting differences in methylation that also uses microarrays is comparative methylation
hybridization. Using this method, genomic DNA is digested with enzymes that are either sensitive or insensitive
to methylation—in other words, they either will or will not cut a particular site if the DNA is methylated—before
the DNA is applied to a microarray.

A third way to find methylated sites across the genome uses bisulfite treatment of DNA, which converts
unmethylated cytosine into uracil. Bisulfite­treated DNA can be amplified using a version of polymerase chain
reaction (PCR) that amplifies sequences containing 5­methylcytosines and not uracils—in other words, it
amplifies those sequences that were not converted by the bisulfite (Clark et al., 1994). Bisulfite­treated DNA can
also be hybridized to microarrays containing two sets of oligonucleotides, one of which is complementary to the
unaltered methylated sequence, and the other which is complementary to the converted unmethylated
sequence. Hybridization to one set of probes but not the other reveals where methylation occurs across the
genome.

Of course, one problem associated with the use of microarrays is that they do not contain every single piece of
DNA within the genome. Thus, to analyze the entire genome, scientists have started using a different approach:
DNA sequencing. For instance, a collaboration between the Wellcome Trust Sanger Institute in the United
Kingdom, the Center National de Genotypage in France, and the German company Epigenomics used this

3/6
1/17/2017

method in their analysis of methylation patterns on several regions of human chromosomes 6, 20, and 22 in 12
different tissues. The investigators found that regions that have been conserved throughout evolution show the
greatest differences in DNA methylation. During their research, these scientists analyzed the sequence of 873
genes and discovered that these genes were differentially methylated in their regulatory regions (Eckhardt et al.,
2006). More recently, Cokus et al. (2008) combined bisulfite treatment with new sequencing technologies to look
at patterns of DNA methylation across the entire genome of the plant Arabidopsis thaliana.

Finding Patterns of Histone Modification
Beyond methylation of individual DNA nucleotides,
another type of epigenetic modification consists of
changes to histone proteins. The most common types
of histone modifications are methylation, acetylation,
and ubiquitination, all of which involve the addition of
different types of chemicals to certain amino acids (in
particular, those in the N­terminal regions, or tails, of
histones). Throughout the past decade, scientists have
used antibodies specific for particular modifications to
develop a catalog of all possible histone modifications
that occur in different cells and under different biological
conditions.

More recently, researchers have also started using the
Figure 3: Mass spectrometry analysis of
technique of mass spectrometry to determine
histone H4 isoforms in human stem cells
modification patterns on different histones. For
When differentiation was induced in the cells,
example, Joshua Coon's group at the University of
levels of unmethylated isoforms fell and
Wisconsin­Madison used mass spectrometry to map all
dimethylated isoforms rose.
the modifications that occur in the 23 residues of the N­
© 2008 Modified with permission from Joshua Coons. All rights
terminal tail of the histone H4 in human embryonic stem
reserved. 
cells (Phanstiel et al., 2008). Figure 3 shows a map of all
74 histone H4 combination codes detected in human Figure Detail
embryonic stem (ES) cells. Each box represents a
potential site of modification, and coloration indicates
the presence or absence of modifications. These mass spectrometry experiments show that the amino acid
lysine at position 20 (K20) is the target of two methylation events.

The front panels in the figure show the effects on methylation once cells were treated with a chemical that
induces differentiation. Specifically, levels of unmethylated histone H4 isoforms decreased in human ES cells
during differentiation, while dimethylated isoforms increased. Note that the control human ES cells were
characterized by a large percentage of unmethylated histone H4 (approximately 20%) and that upon 75 hours of
treatment, this population was virtually eliminated in favor of dimethylation. The red line in the panel shows the
results of this experiment using somatic cells (fibroblasts) exposed to the same differentiation agent. Note that no
significant change was observed in this cell population.

The Epigenome and Disease
Cancer cells exhibit a number of phenotypes different from their healthy cellular counterparts due to the
misregulation of many cellular pathways. Some of those alterations are due to heritable changes at the genetic
level, while others are due to epigenetic changes. Indeed, several studies have found that the global methylation
state of tumors is different than that of their normal, healthy cellular counterparts. Often, overall methylation
increases on a global scale, but at the level of individual genes, both hyper­ and hypomethylation have been
4/6
1/17/2017

found. This gene­specific change in methylation status
often corresponds to areas of the genome encoding
tumor suppressors (which feature increased
methylation) and oncogenes (which feature decreased
methylation) (Jones & Baylin, 2002).

Today, many companies provide microarrays that
contain gene promoters, CpG islands, or whole
genomes so that scientists can look for methylation at
these sites using MeDIP (Bonetta, 2008). Techniques
Figure 4: DNA methylation data can
like this allow scientists to study epigenetic changes
classify different types of cancer
across entire genomes, determining where these
(A) This graph shows the differential methylation
changes occur and under what circumstances. By
of genes between the two types of cancer
combining such studies with those that determine the
explored in this experiment: AML and ALL. Data
binding sites of transcription factors and other
from two ALL and three AML patients are shown.
regulatory proteins across the genome, scientists will
(B) A 'heatmap' shows genes that had a decrease
gain a better understanding of how epigenetic
in expression (blue) and an increase in expression
modifications occur and what their consequences are.
(red). Each row represents expression of a
Moreover, by determining patterns of methylation tags
different gene.
across the whole genome and of histone modifications
Creative Commons Figueroa, M. E., et al. An Integrative
in different types of cells and tissues, scientists will be
Genomic and Epigenomic Approach for the Study of
able to find various "signatures" associated with
Transcriptional Regulation. PLOS ONE 3 (3) 1882 (2008)
particular biological processes or diseases states.
doi:10.1371/journal.pone.0001882.

One example of such a signal exists in acute myeloid
leukemia (AML) and acute lymphocytic leukemia (ALL).
Scientists have compared the global methylation patterns from these two leukemias and found that they could
easily be distinguished (Figure 4).

Epigenomic studies might also play a role in the development of new treatments for disease. Although
epigenomic changes are heritable, drug treatments can potentially reverse them, opening the door for possible
treatments. For instance, in one study using mice that had different coat colors depending on the amount of
methylation in the DNA upstream of a particular gene, researchers found that giving pregnant mothers bisphenol
A (BPA), a chemical widely used in the manufacture of plastic, increased DNA methylation in their offspring. This
increased methylation was associated with higher body weight in the offspring. The scientists were able to
change the effects of BPA on methylation and body weight in the offspring by feeding pregnant mice foods that
contained a lot of methyl groups, such as foods rich in folic acid and soy protein (Dolinoy et al., 2007).

The Future of Epigenomic Research
Currently, several research groups are exploiting whole­genome approaches in an attempt to characterize the
epigenome. Indeed, multiple different consortia aimed at the study of human epigenomics have formed in the
past decade. These groups are working to identify, catalogue, and interpret genome­wide DNA methylation
patterns of all human genes in all major tissues. Such collaborations are multinational and multidisciplinary, and
they are a hallmark of twenty­first­century "­omics" research. Epigenomics is a new field with the potential to
dissect nongenetic components of complex diseases. The hope is that with full genome sequence and
epigenomic maps of the DNA methylation and modified histone landscapes, researchers will be able to tell
exactly which genes are "turned on" and in which tissues. Integration of this massive amount of data promises to
revolutionize our understanding of gene­environment interactions and offer new ways to diagnose and treat
complex diseases.

5/6
1/17/2017

References and Recommended Reading

Bonetta, L. Epigenomics: Detailed analysis. Nature 454, 795–798 (2008) doi:10.1038/454795a ( link to article)

Callihan, P. A., & Feinberg, A. P. The emerging science of epigenomics. Human Molecular Genetics 15, R95–R101 (2006)

Clark, S. J., et al. High sensitivity mapping of methylated cytosines. Nucleic Acids Research 22, 2990–2997 (1994)

Cokus, S. J., et al. Shotgun bisulphite sequencing of the Arabidopsis genome reveals DNA methylation patterning. Nature 452, 215–219 (2008) ( link
to article)

Dennis, C. Epigenetics and disease: Altered states. Nature 421, 686–688 (2007) doi:10.1038/421686a ( link to article)

Dolinoy, D. C. A Tale of Two Mice (video clip). NOVA website, http://www.pbs.org/wgbh/nova/sciencenow/3411/02.html (2007) (link to article)

Dolinoy, D. C., et al. Maternal nutrient supplementation counteracts bisphenol A­induced DNA hypomethylation in early development. Proceedings of
the National Academy of Sciences 104, 13056–13061 (2007)

Eckhardt, F., et al. DNA methylation profiling of human chromosomes 6, 20, and 22. Nature Genetics 38, 1378–1385 (2006) doi:10.1038/ng1909 ( link
to article)

Feinberg, A. P., & Tycko, B. The history of cancer epigenetics. Nature Reviews Cancer 4, 143–153 (2004) doi:10.1038/nrc1279

Fraga, M. F., et al. Epigenetic differences arise during the lifetime of monozygotic twins. Proceedings of the
National Academy of Sciences 102, 10604–10609 (2005)

Jirtle, R. L., & Skinner, M. K. Environmental epigenomics and disease susceptibility. Nature Reviews Genetics 8, 253–262 (2007) doi:10.1038/nrg2045
( link to article)

Jones, P. A., & Baylin, S. B. The fundamental role of epigenetic events in cancer. Nature Reviews Genetics 3, 415–428 (2002) doi:10.1038/nrg816
( link to article)

Jones P. A., et al. Moving AHEAD with an international human epigenome project. Nature 454, 711–715 (2008) doi:10.1038/454711a (link to article)

Phanstiel, D. J., et al., Mass spectrometry identifies and quantifies 74 unique histone H4 isoforms in differentiating human embryonic stem cells.
Proceedings of the National Academy of Sciences 105, 4093–4098 (2008)

Seligson, D. B., et al. Global histone modification patterns predict risk of prostate cancer recurrence. Nature 435, 1262–1266 (2005)

Waddington, C. H. Organisers and Genes (Cambridge, UK, Cambridge University Press, 1940)

Weber, M., et al. Chromosome­wide and promoter­specific analyses identify sites of differential DNA methylation in normal and transformed human
cells. Nature Genetics 37, 853–862 (2005) doi:10.1038/ng1598 ( link to article)

6/6
Epigenetics readings
 3 Rogers DB, Shohat M, Pertersen GM, et al. Familial Mediterranean fever in Armenians: autosomal
recessive inheritance with high gene frequency. Amj Med Genet 1989;34:168-72.
4 Barakat MH, Karnik AM, Majeed HWA, El-Sobki NI, Fenech FF. Familial Mediterranean fever
(recurrent hereditary polyserositis in Arabs-a study of 175 patients and review of the literature.
QjMed 1986;60:837-47.
5 Cook GC. Periodic disease, recurrent polyserositis, familial Mediterranean fever, or simply
"FMF." QJ Med 1986;60:819-23.
6 Moore PJ, Mansour A, McDonald JD, Kemp A, Kamath KR, Dorney SFA. Familial
Mediterranean fever in six Australian children. MedjAust 1989;151:108-10.
7 Gertz MA, Petitt RM, Perrault J, Kyle RA. Autosomal dominant familial Mediterranean fever-like
syndrome with amyloidosis. Mayo Clin Proc 1987;62:1095-100.
8 Melamed A, Cabili S, Zakuth V, Spirer Z. The immune regulation in familial Mediterranean fever
(FMF). J Clin Lab Immunol 1988;26:125-8.
9 Matzner Y, Brzezinski A. C5a-inhibitor deficiency in peritoneal fluids from patients with familial
Mediterranean fever. N Englj Med 1984;311:287-90.
10 Shohat M, Korenberg JR, Schwabe AD, Rotter JI. Hypothesis: familial Mediterranean fever-A
genetic disorder of the lipocortin family? Am j Med Genet 1989;34:163-7.
11 Aisen PS, Haines KA, Given W, et al. Circulating hydroxy fatty acids in familial Mediterranean
fever. Proc Natl Acad Sci USA 1985;82:1232-6.
12 Anton PA, Targan SR, Vigna SR, Durham M, Schwabe AD, Shanahan F. Enhanced neutrophil
chemiluminescence in familial Mediterranean fever. J Clin Immunol 1988;8:148-56.
13 Barakat MH, El-Khawad AO, Gumaa KA, El-Sobki NI, Fenech FF. Metaraminol provocative
test: a specific diagnostic test for familial Mediterranean fever. Lancet 1984;i:656-7.
14 Barakat MH, Gumaa KA, Malhas LN, El-Sobki NI, Moussa MA, Fenech FF. Plasma dopamine
beta-hydroxylase: rapid diagnostic test for recurrent hereditary polyserositis. Lancet 1988;ii:
1280-3.
15 Hawkins PN, Lavender JP, Pepys MB. Evaluation of systemic amyloidosis by scintigraphy with
I-labelled serum amyloid P component. N Englj Med 1990;323:508-13.
16 Dinarello CA, Wolff SM, Goldfinger SE, Dale DC, Alling DW. Colchicine therapy for familial
Mediterranean fever: a double-blind trial. N Englj Med 1974;291:934-7.
17 Zemer D, Revach M, Pras M, et al. A controlled trial of colchicine in preventing attacks of familial
Mediterranean fever. N Englj Med 1974;291:932-4.
18 Zemer D, Pras M, Sohar E, Modan M, Cabili S, Gafni J. Colchicine in the prevention and treatment
of the amyloidosis of familial Mediterranean fever. N Englj Med 1986;314:1001-5.
19 Anonymous. Colchicine in amyloidosis [Editorial]. Lancet 1986;ii:724-5.
The fetal and infant origins
of adult disease
The womb may be more important than the home
A hundred years ago, when tuberculosis and rheumatic heart
disease were common, the proposition that the childhood
environment affects adult health would have been selfevident.
This proposition may still hold, even though infective disease
has given place to degenerative disease.
Studies in Norway, Finland, Britain, and the United States
have shown that death rates from cardiovascular disease are
inversely related to adult height, and geographical differences
in cardiovascular mortality are related to past differences in
infant mortality.'7 These findings have been interpreted as
evidence that adverse living conditions during childhood,
such as poor housing and diet, increase the risk of ischaemic
heart disease.4 Case-control studies have generally supported
this8-'2: patients with myocardial infarction have higher infant
death rates among their siblings,89 are more likely to come
from larger families, and are more likely to have fathers who
were unemployed.'0 Now studies in Finland show that men
with ischaemic heart disease had worse socioeconomic con-
ditions in childhood (p 112 1)'2-an observation also made in
Britain.9
The completeness of infant mortality records in England
and Wales from 1911 onwards has allowed detailed geo-
graphical comparisons of the relation between infant
mortality 70 years ago and mortality from cardiovascular
disease today. Differences in the death rates from cardio-
vascular disease among the 212 local authority areas of
England and Wales are closely related to past differences in
neonatal mortality.6"' Most neonatal deaths were associated
with low birth weight, and rates were high in areas where
mothers had poor health and high death rates during
childbirth. '4 1' These findings suggested that research should
be redirected towards the intrauterine environment rather
than the environment in later childhood-housing, family
income, diet, and other influences. The Medical Research
Council employed a historian to search for old records of birth
BMJ VOLUME 301 17 NOVEMBER 1990
and infancy. In Hertfordshire health visitors recorded the
birth ,weight of all babies born in the county from 1911
onwards and visited their homes periodically throughout
infancy. Follow up studies of the men and women born 60 and
more years ago show that those who weighed more at birth
and, if they were breast fed, at 1 year, had lower death rates
from ischaemic heart disease and stroke. 16 The differences in
death rates were large.
We are beginning to identify processes that link fetal and
infant growth with cardiovascular disease. A recent study of
449 men and women aged 50 years who had been born in one
hospital in Preston, England, showed that their current blood
pressure and risk of hypertension were strongly related to
their placental and birth weight.'7 Pressures were highest
when birth weight had been lower than expected from
placental weight. Discordance between placental and birth
weights may be interpreted as fetal growth failure. Its causes
are unknown, but maternal nutrition is an obvious suspect.
These epidemiological findings point to the importance of
long term programming in early life and parallel findings in
clinical and animal research. For example, the composition of
infant food has been shown to have an important effect on
motor development in preterm babies,'8 and programming
of lipid metabolism by early feeding has been shown in
baboons.'9 Knowledge of the fetal processes that may deter-
mine programming is beginning to emerge.20 A recent sympo-
sium heard evidence that diseases other than cardiovascular
disease may also be determined by the maternal environ-
ment.21 Schizophrenia and obstructive lung disease are two
examples.
The old model of adult degenerative disease was based on
the interaction between genes and an adverse environment in
adult life. The new model that is developing will include
programming by the environment in fetal and infant life.
D J P BARKER
Director,
MRC Environmental Epidemiology Unit,
University of Southampton,
Southampton General Hospital,
Southampton S09 4XY
1 Waaler HT. Height, weight and mortality. The Norwegian experience. Acta Med Scand
1984;679(suppl): 1-56.
2 Notkola V. Living conditions in childhood and coronary heart disease in adulthood. Helsinki: Finnish
Society of Science and Letters, 1985.
3 Marmot MG, Shipley MJ, Rose G. Inequalities in death specific explanations of a general pattern?
Lancet 1984;i: 1003-6.
4 Forsdahl A. Are poor living conditions in childhood and adolescence an important risk factor for
arteriosclerotic heart disease? British Journal ofPreventive and Social Medicine 1977;31:91-5.
5 Barker DJP, Osmond C, Golding J. Height and mortality in the counties of England and Wales.
Ann Hum Biol 1990;17:1-6.
6 Barker DJP, Osmond C. Infant mortality, childhood nutrition and ischaemic heart disease in
England and Wales. Lancet 1986;i: 1077-81.
7 Buck C, Simpson H. Infant diarrhoea and subsequent mortality from heart disease and cancer.
J Epidemiol Community Health 1982;36:27-30.
8 Rose G. Familial pattems in ischaemic heart disease. British Journal of Preventive and Social
Medicine 1964;18:75-80.
9 Coggon DNM, Margetts B, Barker DJP, et al. Childhood risk factors for ischaemic heart disease
and stroke. Paediatric and Perinatal Epidemiology 1990;4:464-70.
10 Burr ML, Sweetnam PM. Family size and patemal unemployment in relation to myocardial
infarction. J Epidemiol Community Health 1980;34:93-5.
11 Hasle H. Association between living conditions in childhood and myocardial infarction. BMJ
1990;300:5 12-3.
12 Kaplan GA, Salonen JT. Socioeconomic conditions in childhood are associated with ischaemic
heart disease during middle age. BMJ7 1990;301:1121-3.
13 Barker DJP, Osmond C, Law C. The intra-uterine and early postnatal origins of cardiovascular
disease and chronic bronchitis. J7 Epidemiol Community Health 1989;43:237-40.
14 Campbell JM, Cameron D, Jones DM. High maternal mortality in certain areas. London: HMSO,
1932. (Ministry of Health reports on public health and medical subjects, No 68).
15 Barker DJP, Osmond C. Death rates from stroke in England and Wales predicted from past
maternal mortality. BMJ 1986;295:83-6.
16 Barker DJP, Winter PD, Osmond C, Margetts B, Simmonds SJ. Weight in infancy and death from
ischaemic heart disease. Lancet 1989;ii:577-80.
17 Barker DJP, Bull AR, Osmond C, Simmonds SJ. Fetal and placental size and risk of hypertension
in adult life. BMJ 1990;301:259-62.
18 Lucas A, Morley R, Cole TJ, et al. Early diet in preterm babies and developmental status at 18
months. Lancet 1990;335:1477-81.
19 Mott GE, Lewis DS and McGill HC, Jr. Programming of cholesterol metabolism by breast or
formula feeding. In: Bock GR, Whelan J, eds. The childhood environment and adult disease.
Chichester: John Wiley and Sons (in press). (Ciba Foundation symposium No 156.)
20 Dawes GS, Zacutti A, Borruto F, Zacutti A, Jr. Fetal autonomy and adaptation. Chichester: John
Wiley and Sons, 1990.
21 Bock GR, Whelan J, eds. The childhood environment and adult disease. Chichester: John Wiley and
Sons (in press). (Ciba Foundation symposium No 156.)
1lll
 New studies
suggest that
the stress of
being poor has
a staggeringly
harmful influence
on health

By Robert Sapolsky

Sickof
POVERTY
R
udolph Virchow, the 19th-century German neuro-
scientist, physician and political activist, came
of age with two dramatic events — a typhoid out-
break in 1847 and the failed revolutions of 1848.
Out of those experiences came two insights for him: fi rst,
that the spread of disease has
much to do with appalling liv-
ing conditions, and second, that It is not a subtle
those in power have enormous
means to subjugate the power- statistical phenomenon.
less. As Virchow summarized
in his famous epigram, “Physi- When you compare the highest
cians are the natural attorneys
of the poor.” versus lowest rungs of the
Physicians (and biomedical
scientists) are advocates of the socioeconomic ladder, the risk
underprivileged because pov-
erty and poor health tend to go of some diseases varies 10-fold.
hand in hand. Poverty means
bad or insufficient food, un-
healthy living conditions and endless other factors that lead
to illness. Yet it is not merely that poor people tend to be un-
healthy while everyone else is well. When you examine socio-
economic status (SES), a composite measure that includes
Overview/Status and Health
■ Researchers have long known that people with low
socioeconomic status (SES) have dramatically higher
disease risks and shorter life spans than do people in
the wealthier strata of society. The conventional
explanations — that the poor have less access to health
care and a greater incidence of harmful lifestyles such
as smoking and obesity — cannot account for the huge
discrepancy in health outcomes.
■ New studies indicate that the psychosocial stresses
associated with poverty may increase the risks of many
illnesses. The chronic stress induced by living in a poor,
violent neighborhood, for example, could increase one’s
susceptibility to cardiovascular disease, depression
and diabetes.
■ Other studies have shown a correlation between income
inequality and poor health in the U.S. Some researchers
believe that the poor feel poorer, and hence suffer
greater stress, in communities with wide gaps between
the highest and lowest incomes.
94 SCIENTIFIC A MERIC A N
income, occupation, education and housing conditions, it be-
comes clear that, starting with the wealthiest stratum of so-
ciety, every step downward in SES correlates with poorer
health.
This “SES gradient” has been documented throughout
Westernized societies for prob-
lems that include respiratory and
cardiovascular diseases, ulcers,
rheumatoid disorders, psychiat-
ric diseases and a number of
cancers. It is not a subtle statis-
tical phenomenon. When you
compare the highest versus the
lowest rungs of the SES ladder,
the risk of some diseases varies
10-fold. Some countries exhibit
a five- to 10-year difference in
life expectancy across the SES
spectrum. Of the Western na-
tions, the U.S. has the steepest
gradient; for example, one study showed that the poorest
white males in America die about a decade earlier than the
richest.
So what causes this correlation between SES and health?
Lower SES may give rise to poorer health, but conversely,
poorer health could also give rise to lower SES. After all,
chronic illness can compromise one’s education and work
productivity, in addition to generating enormous expenses.
Nevertheless, the bulk of the facts suggests that the arrow
goes from economic status to health— that SES at some point
in life predicts health measures later on. Among the many
demonstrations of this point is a remarkable study of elderly
American nuns. All had taken their vows as young adults and
had spent many years thereafter sharing diet, health care and
housing, thereby controlling for those lifestyle factors. Yet in
their old age, patterns of disease, incidence of dementia and
P A U L F U S C O M a g n u m P h o t o s ( p r e c e d i n g p a g e s)
longevity were still significantly predicted by their SES status
from when they became nuns, at least half a century before.
Inadequate Explanations
s o , t o u s e a m a r v e l o u s p h r a s e common to this
field, how does SES get “under the skin” and influence health?
The answers that seem most obvious, it turns out, do not hold
much water. One such explanation, for instance, posits that
for the poor, health care may be less easily accessible and of
lower quality. This possibility is plausible when one considers
that for many of the poor in America, the family physician
DECEMBER 2005
 does not exist, and medical care consists solely of trips to the 2.0
emergency room.
Adjusted for age
But that explanation soon falls by the wayside, for reasons
made clearest in the famed Whitehall studies by Michael G. Adjusted for other
Marmot of University College London over the past three de- risk factors
cades. Marmot and his colleagues have documented an array 1.5
of dramatic SES gradients in a conveniently stratified popula-

from Chronic Heart Disease

tion, namely, the members of the British civil service (ranging

Relative Rate of Deaths

from blue-collar workers to high-powered executives). Office
messengers and porters, for example, have far higher mortal-
ity rates from chronic heart disease than administrators and 1.0
professionals do [see top illustration at right]. Lack of access
to medical attention cannot explain the phenomenon, because
the U.K., unlike the U.S., has universal health care. Similar
SES gradients also occur in other countries with socialized
0.5
medicine, including the health care Edens of Scandinavia, and
the differences remain significant even after researchers factor
in how much the subjects actually use the medical services.
Another telling finding is that SES gradients exist for dis-
eases for which health care access is irrelevant. No amount of 0.0
Administrators Professionals/ Clerical Office Support
medical checkups, blood tests and scans will change the like- Executives Staff Staff
lihood of someone getting type 1 (juvenile-onset) diabetes or WHITEHALL STUDIES show that low-ranked British civil servants (office
rheumatoid arthritis, yet both conditions are more common messengers and other support staff) are almost twice as likely to die
among the poor. from heart disease as administrators of the same age. Differences in risk
“ S U B J E C T I V E S O C I A L S T A T U S : I T S D E T E R M I N A N T S A N D I T S A S S O C I A T I O N W I T H M E A S U R E S O F I L L- H E A LT H I N T H E W H I T E H A L L I I S T U D Y, ”

The next “obvious” explanation centers on unhealthy life- factors — for example, higher smoking rates among the support staff —
styles. As you descend the SES ladder in Westernized societ- account for less than half the gap in mortality rates.
ies, people are more likely to smoke, to drink excessively, to
B Y A . S I N G H - M A N O U X , N . E . A D L E R A N D M . G . M A R M O T, I N S O C I A L S C I E N C E A N D M E D I C I N E , V O L . 5 6 ; 2 0 0 3 ( b o t t o m g r a p h)

be obese, and to live in a violent or polluted or densely popu-

lated neighborhood. Poor people are also less likely to have 50
access to clean water, healthy food and health clubs, not to
mention adequate heat in the winter and air-conditioning in
Men
the summer. Thus, it seems self-evident that lower SES gets
Women
(percentage reporting health as poor or fair)

under the skin by increasing risks and decreasing protective 40

factors. As mordantly stated by Robert G. Evans of the Uni-
versity of British Columbia, “Drinking sewage is probably
E M I LY C O O P E R ; S O U R C E S : M I C H A E L G . M A R M O T U n i v e r s i t y C o l l e g e L o n d o n ( t o p g r a p h) ;

Prevalence of Ill Health

unwise, even for Bill Gates.”

What is surprising, though, is how little of the SES gradi- 30
ent these risk and protective factors explain. In the Whitehall
studies, controlling for factors such as smoking and level of High
1
exercise accounted for only about a third of the gradient. This 20 status
same point is made by studies comparing health and wealth 2
among, rather than within, nations. It is reasonable to assume 3
that the wealthier a country, the more fi nancial resources its
citizens have to buy protection and avoid risk. If so, health 10 4
should improve incrementally as one moves up the wealth 5
gradient among nations, as well as among the citizens within
6
individual nations. But it does not. Instead, among the wealth-
iest quarter of countries on earth, there is no relation between 0 7
1 and 2 3 and 4 5 and 6 7 and 8 9 and 10
a country’s wealth and the health of its people. 8
Subjective Status Rank
Thus, health care access, health care utilization, and ex-
posure to risk and protective factors explain the SES/health PARTICIPANTS in one of the Whitehall studies were asked to 9
show their place in society by marking a rung on a ladder. Low
gradient far less well than one might have guessed. One must 10 status
Far more people at the bottom rungs reported that their
therefore consider whether most of the gradient arises from a health was either poor or fair. The strong correlation
different set of considerations: the psychosocial consequenc- between subjective status and health indicates that
es of SES. “feeling poor” may somehow increase disease risks.

w w w. s c ia m . c o m SCIENTIFIC A MERIC A N 95

CHRONIC STRESS may explain how poverty “gets under the skin” and
exerts a harmful influence on health. The risk of stress-sensitive
diseases increases if individuals lack social support, have no outlets for
their frustration and feel that their circumstances are worsening —
exactly the conditions in many poor communities in the U.S.
Psychosocial Stress
i de a l ly, t h e b ody is in homeostatic balance, a state in
which the vital measures of human function — heart rate,
blood pressure, blood sugar levels and so on — are in their
optimal ranges. A stressor is anything that threatens to dis-
rupt homeostasis. For most organisms, a stressor is an acute
physical challenge — for example, the need for an injured ga-
zelle to sprint for its life or for a hungry predator to chase
down a meal. The body is superbly adapted to dealing with
short-term physical challenges to homeostasis. Stores of en-
ergy, including the sugar glucose, are released, and cardiovas-
cular tone increases to facilitate the delivery of fuel to exercis-
THE AUTHOR
ROBERT SAPOLSKY is professor of biological sciences, neurol-
ogy and neurological sciences at Stanford University and a re-
search associate at the National Museums of Kenya. In his lab-
oratory work, he focuses on how stress can damage the brain
and on gene therapy for the nervous system. In addition, he
studies populations of wild baboons in East Africa, trying to de-
termine the relation between the social rank of a baboon and its
health. His latest book is Monkeyluv and Other Essays on Our
Lives as Animals (Scribner, 2005).
96 SCIENTIFIC A MERIC A N
ing muscle throughout the body. Digestion, growth, tissue
repair, reproduction and other physiological processes not
needed to survive the crisis are suppressed. The immune sys-
tem steps up to thwart opportunistic pathogens. Memory and
the senses transiently sharpen.
But cognitively and socially sophisticated species, such as
we primates, routinely inhabit a different realm of stress. For
us, most stressors concern interactions with our own species,
and few physically disrupt homeostasis. Instead these psycho-
social stressors involve the anticipation (accurate or other-
wise) of an impending challenge. And the striking character-
istic of such psychological and social stress is its chronicity.
For most mammals, a stressor lasts only a few minutes. In
contrast, we humans can worry chronically over a 30-year
mortgage.
Unfortunately, our body’s response, though adaptive for
an acute physical stressor, is pathogenic for prolonged psy-
chosocial stress. Chronic increase in cardiovascular tone
brings stress-induced hypertension. The constant mobiliza-
tion of energy increases the risk or severity of diseases such as
type 2 (adult-onset) diabetes. The prolonged inhibition of di-
gestion, growth, tissue repair and reproduction increases the
risks of various gastrointestinal disorders, impaired growth
in children, failure to ovulate in females and erectile dysfunc-
tion in males. A too-extended immune stress response ulti-
mately suppresses immunity and impairs disease defenses.
And chronic activation of the stress response impairs cogni-
tion, as well as the health, functioning and even survival of
some types of neurons.
An extensive biomedical literature has established that in-
dividuals are more likely to activate a stress response and are
more at risk for a stress-sensitive disease if they (a) feel as if
they have minimal control over stressors, (b) feel as if they
have no predictive information about the duration and inten-
sity of the stressor, (c) have few outlets for the frustration
caused by the stressor, (d) interpret the stressor as evidence of
circumstances worsening, and (e) lack social support for the
duress caused by the stressors.
Psychosocial stressors are not evenly distributed across so-
ciety. Just as the poor have a disproportionate share of physical
stressors (hunger, manual labor, chronic sleep deprivation
with a second job, the bad mattress that can’t be replaced),
they have a disproportionate share of psychosocial ones.
Numbing assembly-line work and an occupational lifetime
spent taking orders erode workers’ sense of control. Unreliable
cars that may not start in the morning and paychecks that may
not last the month infl ict unpredictability. Poverty rarely al-
lows stress-relieving options such as health club memberships,
costly but relaxing hobbies, or sabbaticals for rethinking one’s
priorities. And despite the heartwarming stereotype of the
“poor but loving community,” the working poor typically
have less social support than the middle and upper classes,
JACOB HOLDT
thanks to the extra jobs, the long commutes on public transit,
and other burdens.
Marmot has shown that regardless of SES, the less au-
DECEMBER 2005
 tonomy one has at work, the worse one’s cardiovascular
health. Furthermore, low control in the workplace accounts
for about half the SES gradient in cardiovascular disease in
his Whitehall population.
Feeling Poor
t h r e e l i n e s of r e se a rc h provide more support for the
influence of psychological stress on SES-related health gradi-
ents. Over the past decade Nancy E. Adler of the University of
California, San Francisco, has explored the difference between
objective and subjective SES and the relation of each to health
[see bottom illustration on page 95]. Test subjects were shown
a simple diagram of a ladder with 10 rungs and then asked,
“In society, where on this ladder would you rank yourself in
terms of how well you’re doing?” The very openness of the
question allowed the person to defi ne the comparison group
that felt most emotionally salient.
As Adler has shown, a person’s subjective assessment of
his or her SES takes into account the usual objective measures
(education, income, occupation and residence) as well as mea-
sures of life satisfaction and of anxiety about the future.
Adler’s provocative fi nding is that subjective SES is at least as
good as objective SES at predicting patterns of cardiovascular
function, measures of metabolism, incidences of obesity and

THE GOOD AND BAD EFFECTS OF STRESS

The human body is superb at responding to the acute stress of a physical challenge, such as chasing down prey or escaping
a predator. The circulatory, nervous and immune systems are mobilized while the digestive and reproductive processes are
suppressed. If the stress becomes chronic, though, the continual repetition of these responses can cause major damage.

EFFECTS OF EFFECTS OF
ACUTE STRESS CHRONIC STRESS
Brain Brain
Increased alertness and Impaired memory and
less perception of pain increased risk of depression

Thymus Gland and Thymus Gland and

Other Immune Tissues Other Immune Tissues
Immune system readied for Deteriorated immune
possible injury response

Circulatory System Circulatory System

Heart beats faster, and Elevated blood pressure
blood vessels constrict and higher risk of
to bring more oxygen cardiovascular disease
to muscles

Adrenal Glands Adrenal Glands

Secrete hormones that High hormone levels slow
mobilize energy supplies recovery from acute stress

Reproductive Organs Reproductive Organs

Reproductive functions are Higher risks of infertility
temporarily suppressed and miscarriage
B R YA N C H R I S T I E D E S I G N

w w w. s c ia m . c o m SCIENTIFIC A MERIC A N 97
levels of stress hormones — suggesting that the subjective feel-
ings may help explain the objective results.
This same point emerges from comparisons of the SES/
health gradient among nations. A relatively poor person in the
U.S. may objectively have more fi nancial resources to pur-
chase health care and protective factors than a relatively
wealthy person in a less developed country yet, on average,
will still have a shorter life expectancy. For example, as Ste-
phen Bezruchka of the University of Washington emphasizes,
people in Greece on average earn half the income of Ameri-
cans yet have a longer life expectancy. Once the minimal re-
sources are available to sustain a basic level of health through
adequate food and housing, absolute levels of income are of
remarkably little importance to health. Although Adler’s
work suggests that the objective state of being poor adversely
affects health, at the core of that result is the subjective state
of feeling poor.
Being Made to Feel Poor
a n o t h e r b ody of r e se a rc h arguing that psychoso-
cial factors mediate most of the SES/health gradient comes
from Richard Wilkinson of the University of Nottingham in
England. Over the past 15 years he and his colleagues have
reported that the extent of income inequality in a community
is even more predictive than SES for an array of health mea-
sures. In other words, absolute levels of income aside, greater
disparities in income between the poorest and the wealthiest
in a community predict worse average health. (David H. Ab-
bott of the Wisconsin National Primate Research Center and
I, along with our colleagues, found a roughly equivalent phe-
nomenon in animals: among many
nonhuman primate species, less
egalitarian social structures cor-
The surest way to feel
relate with higher resting levels of
a key stress hormone — an index
poor is to be endlessly
for worse health— among socially
subordinate animals.)
made aware of the haves
Wilkinson’s subtle and critical
finding has generated considerable
when you are a have-not.
controversy. One dispute concerns
its generality. His original work
suggested that income inequality was relevant to health in
many European and North American countries and commu-
nities. It has become clear, however, that this relation holds
only in the developed country with the greatest of income
inequalities, namely, the U.S.
Whether considered at the level of cities or states, income
inequality predicts mortality rates across nearly all ages in the
U.S. [see illustration on opposite page]. Why, though, is this
relation not observed in, say, Canada or Denmark? One pos-
sibility is that these countries have too little income variabil-
ity to tease out the correlation.
Some critics have questioned whether the linkage between
income inequality and worse health is merely a mathematical
quirk. The relation between SES and health follows an
98 SCIENTIFIC A MERIC A N
INCOME INEQUALIT Y appears to exacerbate the stress of poverty. As the
divide between the rich and poor grows wider, social support typically
becomes less available and the frustrations of the poor intensify.
asymptotic curve: dropping from the uppermost rung of so-
ciety’s ladder to the next-to-top step reduces life expectancy
and other measures much less drastically than plunging from
the next-to-bottom rung to the lowest level. Because a com-
munity with high levels of income inequality will have a rela-
tively high number of individuals at the very bottom, where
health prospects are so dismal, the community’s average life
expectancy will inevitably be lower than that of an egalitar-
ian community, for reasons that have nothing to do with psy-
chosocial factors. Wilkinson has shown, however, that de-
creased income inequality predicts better health for both the
poor and the wealthy. This result strongly indicates that the
association between illness and
inequality is more than just a
mathematical artifact.
Wilkinson and others in the
field have long argued that the
more unequal income in a com-
munity is, the more psychosocial
stress there will be for the poor.
Higher income inequality intensi-
fies a community’s hierarchy and
makes social support less avail-
able: truly symmetrical, reciprocal, affiliative support exists
only among equals. Moreover, having your nose rubbed in
your poverty is likely to lessen your sense of control in life, to
aggravate the frustrations of poverty and to intensify the sense
of life worsening.
NORBER T MIL L A UER AFP/Getty Images
If Adler’s work demonstrates the adverse health effects of
feeling poor, Wilkinson’s income inequality work suggests
that the surest way to feel poor is to be made to feel poor— to
be endlessly made aware of the haves when you are a have-not.
And in our global village, we are constantly made aware of the
moguls and celebrities whose resources dwarf ours.
John W. Lynch and George A. Kaplan of the University of
Michigan at Ann Arbor have recently proposed another way
that people are made to feel poor. Their “neomaterialist” in-
DECEMBER 2005
 terpretation of the income inequality phenomenon— which is RATING THE STATES
subtle, reasonable and, ultimately, deeply depressing— runs as
follows: Spending money on public goods (better public tran- 1,020
MS LA
sit, universal health care and so on) is a way to improve the

(deaths per 100,000 population)

960 SC GA AL
quality of life for the average person. But by definition, the NV TN WV
KY

Age-Adjusted Mortality
bigger the income inequality in a society, the greater the finan- 900
DE MD NC IL AR
OH OK NY
cial distance between the average and the wealthy. The bigger PA AK IN
VA MI MO TX
NJ
this distance, the less the wealthy have to gain from expendi- 840 VT ME Best fit for data
NH WY OR MT CA
tures on the public good. Instead they would benefit more WA NE RI MA AZ NM
from keeping their tax money to spend on their private good— 780 WI KS
IA ID CO CT FL
ND
a better chauffeur, a gated community, bottled water, private
UT MN SD
schools, private health insurance. So the more unequal the 720
HI
income is in a community, the more incentive the wealthy will
have to oppose public expenditures benefiting the health of the 660
community. And within the U.S., the more income inequality 27 28 29 30 31 32 33 34 35
Robin Hood Index
there is, the more power will be disproportionately in the
hands of the wealthy to oppose such public expenditures. Ac- ROBIN HOOD INDEX is a measure of income inequality; it is the
percentage of total community income that must be taken from the rich
cording to health economist Evans, this scenario ultimately
(those with above-average incomes) and given to the poor (those below
leads to “private affluence and public squalor.” the average) to achieve an equal distribution. U.S. states (blue dots)
E M I LY C O O P E R ; S O U R C E : “ I N C O M E D I S T R I B U T I O N A N D M O R T A L I T Y: C R O S S S E C T I O N A L E C O L O G I C A L S T U D Y O F T H E R O B I N H O O D I N D E X I N T H E U N I T E D S T A T E S , ”

This “secession of the wealthy” can worsen the SES/health
gradient in two ways: by aggravating the conditions in low-
income communities (which account for at least part of the
increased health risks for the poor) and by adding to the psy-
chosocial stressors. If social and psychological stressors are
entwined with feeling poor, and even more so with feeling
poor while being confronted with the wealthy, they will be
even more stressful when the wealthy are striving to decrease
the goods and services available to the poor.
Social Capital
a t h i r d br a n c h of su p p or t for psychosocial expla-
nations for the relation between income inequality and health
B Y B . P. K E N N E D Y, I . K A W A C H I A N D D . P R O T H R O W - S I T H , I N B R I T I S H M E D I C A L J O U R N A L , V O L . 31 2 ; 1 9 9 6
with a high Robin Hood index also tend to have high mortality rates.
None of this is surprising. As a culture, America has ne-
glected its social safety nets while making it easier for the
most successful to sit atop the pyramids of inequality. More-
over, we have chosen to forgo the social capital that comes
from small, stable communities in exchange for unprecedent-
ed opportunities for mobility and anonymity. As a result, all
measures of social epidemiology are worsening in the U.S. Of
Westernized nations, America has the greatest income in-
equality (40 percent of the wealth is controlled by 1 percent
of the population) and the greatest discrepancy between ex-
penditures on health care (number one in the world) and life

comes from the work of Ichiro Kawachi of Harvard Univer-
sity, based on the concept of “social capital.” Although it is
still being refi ned as a measure, social capital refers to the
broad levels of trust and efficacy in a community. Do people
generally trust one another and help one another out? Do
people feel an incentive to take care of commonly held re-
sources (for example, to clean up graffiti in public parks)?
And do people feel that their organizations — such as unions
or tenant associations — actually have an impact? Most stud-
ies of social capital employ two simple measures, namely, how
many organizations people belong to and how people answer
a question such as, “Do you think most people would try to
take advantage of you if they got a chance?”
What Kawachi and others have shown is that at the levels
of states, provinces, cities and neighborhoods, low social cap-
ital predicts bad health, bad self-reported health and high
mortality rates. Using a complex statistical technique called
path analysis, Kawachi has demonstrated that (once one con-
trols for the effects of absolute income) the strongest route
from income inequality to poor health is through the social
capital measures — to wit, high degrees of income inequality
come with low levels of trust and support, which increases
stress and harms health.
expectancy (as of 2003, number 29).
The importance of psychosocial factors in explaining the
SES/health gradient generates a critical conclusion: when it
comes to health, there is far more to poverty than simply not
having enough money. (As Evans once stated, “Most graduate
students have had the experience of having very little money,
but not of poverty. They are very different things.”) The psy-
chosocial school has occasionally been accused of promulgat-
ing an antiprogressive message: don’t bother with universal
health care, affordable medicines and other salutary measures
because there will still be a robust SES/health gradient after all
the reforms. But the lesson of this research is not to abandon
such societal change. It is that so much more is needed.
MORE TO EXPLORE
Mind the Gap: Hierarchies, Health and Human Evolution. Richard
Wilkinson. Weidenfeld and Nicolson, 2000.
The Health of Nations: Why Inequality Is Harmful to Your Health.
Ichiro Kawachi and Bruce P. Kennedy. New Press, 2002.
The Status Syndrome. Michael Marmot. Henry Holt and Company,
2004.
Why Zebras Don’t Get Ulcers: A Guide to Stress, Stress-Related
Diseases and Coping. Robert Sapolsky. Third edition. Henry Holt and
Company, 2004.

w w w. s c ia m . c o m SCIENTIFIC A MERIC A N 99
 The Scientist »
 December 2015 Issue »
 Features »
 Cover Story

Ghosts in the Genome

How one generation’s experience can affect the next

By Oliver J. Rando | December 1, 2015

I n one of the 20th century’s most disastrous collisions of

political ideology and science, the Russian botanist Trofim
Lysenko steered the USSR’s agricultural research policies
to deemphasize the deterministic concepts of Mendelian
inheritance. Instead, Lysenko was committed to the idea
that, within the space of a single generation, the
environment could alter the phenotype of future
generations, an idea that is now often (imprecisely)
referred to as “Lamarckian” inheritance. In Lysenko’s view,
Mendelian inheritance, along with Darwinian evolution,
emphasizes competition, whereas he believed that biology
was based on cooperation, and that hard work in one
generation should rapidly lead to the betterment of the
species.

Lysenko was among the most infamous purveyors of the

idea that the environment experienced by an organism
could influence the phenotype in future generations, and
he was rightly denounced as a charlatan because he
falsified results in pursuit of his goal. However, the
scientific community has discovered over the past few
decades that the idea that acquired characters can be
inherited may not be completely off the mark. It turns out
that epigenetic marks, information not encoded in the BEYOND DNA: While the haploid DNA of the sperm
genome’s sequence, do respond to environmental and egg contain the blueprint for offspring,
conditions within an organism’s lifetime, and recent numerous molecular processes beyond genomic
evidence suggests that such information may be inherited. inheritance may transmit information from one
generation to the next. Such epigenetic
These findings have helped motivate modern research into mechanisms can convey information about the
the oft-discredited study of transgenerational effects of the parents’ diet, stressful experiences, and
environment. Researchers are now beginning to environmental exposures, influencing offspring
understand the mechanisms of epigenetic inheritance and phenotype in diverse ways.
to generate evidence for the idea that the experiences of See full infographic: WEB | PDF© SCOTT LEIGHTON
an ancestral population can influence future generations.

Mechanisms of biological inheritance

Transmission of traits from parent to child is a universal characteristic of life as we know it, and is
central to the process of evolution. Nevertheless, evolutionary theory has long outpaced our
understanding of the biological mechanisms of inheritance. When Charles Darwin published On the
Origin of Species in 1859, for example, Gregor Mendel was only just beginning to formulate his ideas
of genetic heredity based on his experiments with pea plants. And it would be nearly another century
before researchers elucidated the double-helical structure of DNA, providing us with a clear
understanding of how genetic information is copied and passed on to offspring.

More complicated still is the inheritance of information beyond the sequence of the genome, so-called
epigenetic inheritance. A relatively straightforward example is cell-fate inheritance. Although every cell
in your body carries basically the same genome, when a liver cell divides it always makes two liver
cells, never a skin cell. At one point, researchers hypothesized that differentiation into particular cell
types involved permanent loss of portions of the genome. But beginning in 1958, English
developmental biologist John Gurdon showed that it was possible to generate an intact tadpole by
injecting the nucleus from a differentiated cell into the cytoplasm of a frog ovum, disproving the idea.
Instead, the inheritance of cell type is mediated by the passing on of transcription factors, and of
epigenetic marks on the genome and on the histones around which DNA is wrapped.

Not only is epigenetic information inherited during cellular division, but it can also be passed from one
generation to the next in multicellular organisms, a phenomenon known as transgenerational
epigenetics. This requires that epigenetic information be carried in the gametes—sperm and eggs—
and be maintained throughout the dramatic changes that occur during gamete production,
fertilization, and early development. While researchers once considered this unlikely, recent studies
have begun to demonstrate that parents can and do pass on epigenetic information to their children.

Perhaps the earliest description of transgenerational epigenetic inheritance was the discovery of
paramutation in maize in the 1950s. Plant geneticist R.A. Brink of the University of Wisconsin showed
1

that the repressed allele of a gene involved in pigment synthesis could somehow convert an active
copy of the same gene to an inactive state. When passed on, this inactivated “paramutant” allele
could similarly convert an active copy of the gene to the paramutant state in the plant’s offspring, with
this silenced state being stable for hundreds of generations.

In this case, the gene does not behave as an inviolable particle. It is converted from one state to
another simply by coexisting in the same organism with the “infectious” paramutant variant allele.
Although the precise mechanism underlying paramutation remains the subject of ongoing studies,
researchers have identified a key role for small RNAs expressed from the infectious copy that are
thought to direct chromatin-based silencing of the other gene copy. Indeed, we now know that small
RNA molecules are common carriers of epigenetically inherited information in many systems. For
example, in the process of RNA interference in the worm C. elegans, small RNAs generated from an
injected double-stranded RNA can lead to silencing of genes that are targeted by these RNAs for up to
80 generations. 2

Another interesting example of multigenerational epigenetic inheritance came in the form of the prion
states of budding yeast. A number of yeast proteins form insoluble aggregates at high concentrations.
These aggregates are broken into small chunks that can be transmitted to daughter cells, where they
behave in a dominant, non-Mendelian manner. Crossing a [prion–] stain, in which the protein of
interest remains completely soluble, to a [PRION+] yeast, in which the protein has aggregated into
the prion state, results in [PRION+] offspring in which the aggregated form of the prion protein
converts all soluble protein to the aggregated state.3 While epigenetic inheritance of prion states has
been best studied in yeast, prions such as those responsible for mad cow disease have been found in
many organisms, and it is conceivable that prion-forming proteins could be transmitted by gametes at
fertilization in mammals. (See “The Bright Side of Prions,” The Scientist, January 2014.)

These and other model systems provide evidence that inheritable epigenetic information exists
alongside the inheritable DNA sequence. In addition to RNAs and prion proteins, epigenetic
information carriers include covalent modifications to nucleotides and histones, providing a wide
variety of mechanisms that enable organisms to transmit information extragenomically. (See
illustration above.) And the increasing acceptance of transgenerational epigenetics has, in turn,
spurred renewed interest in the possibility that ancestral environmental conditions might influence the
phenotypes of future generations.

This idea, often referred to as the inheritance of acquired characters, was one aspect of Jean-Baptiste
Lamarck’s early evolutionary theories. But the current use of “Lamarckian inheritance” to refer to
transgenerational epigenetic inheritance is something of a misnomer. In fact, the inheritance of
acquired characters was hardly the defining feature of Lamarck’s beliefs. His evolutionary theory did
not include the basic concept of natural selection, and did not have a place for phenotypic variation
existing prior to environmental challenges. Moreover, both Darwin and Lamarck believed that traits
acquired in one’s lifetime could be passed on. Famously, Darwin even developed a model of
inheritance that invoked “gemmules,” which carried information from all parts of the body to alter the
characteristics of the next generation.

Today, a number of studies document a link between ancestral environmental conditions and changes
in offspring behavior or metabolism, potentially validating some of the thinking of both seminal
evolutionary theorists on this topic. It is now becoming clear that the environments of both the mother
and the father can influence offspring phenotype.

Mama knows best

One of the earliest connections between environmental conditions affecting human parents and
disease rates in their offspring was made in the late 1980s by British epidemiologist David Barker,
who noticed that babies with low birth weights were significantly more likely to suffer from diabetes
and coronary heart disease later in life. He proposed what is now known as the “thrifty phenotype”
4

hypothesis: in response to inadequate nutrition during fetal growth, a person will exhibit a long-lasting
physiological response of aggressively storing calories. This thrifty phenotype then wreaks havoc when
and if times of plenty return, leading to obesity and other diseases related to caloric excess.

It is now becoming clear that the environments of both the

mother and the father can influence offspring phenotype.
The general concept of the thrifty phenotype is supported by studies in multiple human populations,
the most famous being the Dutch survivors of the “Hunger Winter” of World War II. In response to a
strike by railway workers meant to aid the Allied forces, the Nazis occupying the Netherlands imposed
an embargo on food shipments to the western part of the country. The result was a reduction in
government rations to roughly 700 calories per day per person, and a severe famine that lasted from
October 1944 until early 1945. Many people resorted to measures such as eating tulip bulbs to
survive, and at least 18,000 people died of malnutrition. Children of pregnant women exposed to this
famine are more susceptible to obesity, diabetes, cardiovascular disease, and even schizophrenia than
are people who were born just before or conceived after the famine. 5

These human studies have been supplemented by hundreds of experimental investigations using a
variety of mammalian models. In such studies, pregnant female animals are either subjected to a
significant caloric restriction or are surgically treated to restrict the main blood supply to the uterus. In
both scenarios fetal growth is inhibited as a consequence of poor access to nutrients. As seen in
human cohorts, rodents and other animals subjected to in utero growth restriction exhibit a multitude
of changes in metabolism, including poor insulin release and insulin resistance—two factors relevant to
type 2 diabetes in humans. Importantly, even though animals starved during gestation are then
provided with sufficient access to food for the remainder of their lives, the early conditions somehow
establish lifelong metabolic changes. A large field is dedicated to understanding how animals
“remember” conditions present transiently early in their lives.

But the fact that environmental conditions affecting a mother can influence the phenotype of her
offspring is not in itself greatly surprising, as the womb is a baby’s first environment. Mothers who
drink heavily during pregnancy are exposing their babies to a toxin and may give birth to children with
fetal alcohol syndrome. Much more curious than maternal effects on children are cases where the
lifestyle or history of the father has been implicated in disease risk of his children.
Daddy issues
But we now know that sperm carry epigenetic information in addition to their haploid genomic
payload. Moreover, during mating, a male provides his partner with a bolus of seminal fluid, which
carries proteins and other molecules that might have signaling roles. Fathers may also contribute
microbes to their partner and offspring by direct contact or in feces. There is even evidence that in
some species a mother’s investment in offspring is modulated by her judgment of the prospective
father’s adequacy, so the impression a male makes on his mate has the potential to alter his
offspring’s future. Given these contributions, we and others have used lab models to ask how paternal
exposure history may influence offspring behavior, metabolism, or disease risk. 6

Already, a large number of studies in rodents have shown that altering a father’s diet can influence a
number of metabolic traits in his offspring. The typical experimental paradigm used for such studies
involved separating male siblings from one another and providing one brother a control diet and the
other an altered diet, low in protein or with excess fat. At sexual maturity, each brother was mated
with a female who had been fed a control diet. Researchers then examined the offspring for various
phenotypes, most commonly, metabolic features or behavioral traits such as anxiety-related activities.
In multiple studies of this sort, paternal diet was shown to affect glucose control, cholesterol
metabolism, blood pressure, and other cardiovascular measures in offspring. In addition to studies
focused on paternal diet, several studies have shown that males subjected to undernutrition in utero—
the male offspring of mothers subjected to starvation during pregnancy—can sire offspring with
altered glucose and lipid metabolism. These rodent studies are complemented by research in other
7

mammals such as pigs, as well as by human epidemiological studies. This body of work suggests a link
between access to food and metabolic phenotypes in one’s offspring, and possibly in future
generations as well.

But its father’s diet is not the only environmental factor that can affect the biology of a rodent: stress
experienced by fathers can also negatively impact future offspring. A number of studies from Tracy
Bale’s lab at the University of Pennsylvania have shown that prospective mouse fathers subjected to
stressful environments, such as separation from their mothers at a young age, have offspring that
exhibit altered cortisol release in response to stress.8 Similarly, Mount Sinai neurobiologist Eric Nestler
and his colleagues have shown that male mice subjected to social defeat sire offspring with altered
anxiety- and depression-related behaviors, such as decreased time spent in exposed areas. 9

Studies of stress, diet, and additional environmental stimuli such as toxin exposure all strongly
support the idea that a father’s environment can influence the phenotype of his offspring. How exactly
is this information transferred across the generations? In principle, huge amounts of information could
be transmitted to offspring in sperm, as millions of genomic cytosines can each potentially transmit a
“bit” of information in the form of an attached methyl group. Researchers have recently discovered
that the vast majority of cytosine methylation on the paternal genome is erased at fertilization, but
proteins and RNAs in the seminal fluid or in the sperm may also carry information, as could sperm
chromatin packaging. How much detail about a father’s lifestyle sperm really passes on to his offspring
remains unclear, however.

One recent study provides support for the “high bandwidth” hypothesis, in which a very detailed
account of a father’s experiences is transmitted in his sperm, as opposed to a more general quality-of-
life summary. Neurobiologists Brian Dias and Kerry Ressler of Emory University exposed male mice to
different odorants while stressing them with foot shocks, conditioning the mice to startle when they
merely smelled the shock-associated odor. Offspring of these mice showed greater sensitivity to just
the specific odorant that had been paired with the foot shock for their fathers. Smell is mediated by
binding of odorants to a large family of olfactory receptors. Dias and Ressler found that the offspring
of the fear-conditioned males had more neurons expressing the olfactory receptor specific for the
shock-paired odor. This finding raises the exciting and surprising possibility that sperm can transmit
10

thousands of bits of information to offspring. If odorants can induce specific responses in offspring,
why not other factors such as hormones, and
so forth? It will be very exciting to see whether PARSING PATERNAL EFFECTS
other labs have similar results with effects on To determine whether a father’s experiences are
offspring of paternal exposure to odorants and passed down to his offspring via his sperm or
other small molecules. some other mechanism, researchers can breed
mice via in vitro fertilization (IVF), which uses
Although epigenetic marks on sperm DNA seem purified sperm to fertilize an egg. If the effect is
the likeliest carrier of paternal information, still apparent, this points to an epigenetic
fathers transmit much more to mothers and mechanism within the sperm itself. If the effect
children than sperm. One way to disappears, this suggests that the relevant
experimentally probe the question of how a information may be located outside of the sperm,
father transmits information to his offspring is such as in the seminal fluid.
to breed rodents using assisted reproductive
technologies, such as artificial insemination or
in vitro fertilization (IVF), that use purified
sperm to fertilize ova. In the case of Nestler’s
social-defeat paradigm, IVF experiments using
sperm from control or defeated male mice did
not reproduce the effect in offspring, indicating
either that the relevant paternal information is
located outside of sperm (perhaps in the
seminal fluid), or that the disruptive process of
IVF and embryo culture might somehow
prevent accurate transmission of sperm
information to progeny. In contrast, in Diaz
and Ressler’s study pairing odorants with foot
shocks, odor sensitivity was also affected in
offspring generated via IVF using sperm from
exposed versus unexposed male mice.
Male mice exposed to different odorants paired with foot
Unpublished data from our lab shows that a shocks sire offspring with greater sensitivity to those
odorants. Odor sensitivity is affected in offspring
father’s low-protein diet can also alter the
generated via IVF, and the effect is transmitted to those
metabolism of offspring generated via IVF. mice’s offspring as wellsubjected to social defeat spend less
Thus, in some cases, paternal environmental time in exposed areas and exhibit other anxiety-related
information can be transmitted using purified behaviors. When the sperm of these socially defeated fathers is
sperm, strongly pointing to sperm as the used to sire offspring via IVF technology, however, the offspring
are normal.
vehicle for epigenetic inheritance. But rigorous © WESLEY BEDROSIAN
testing of sperm as information carriers in
other species and paradigms is still needed.
And much remains to be done to test whether
such epigenetic changes are necessary and
sufficient to alter metabolism or behavior of
children. One exciting recent advance has
come from studies focused on RNA populations
in sperm, which have provided some evidence
that the effects of paternal stress on offspring
can be mimicked by injecting embryos with
RNAs found in sperm. (See “Sperm RNAs
11,12

Transmit Stress,” The Scientist, October 19,

2015.)

Outlook
While it is clear that the conditions experienced
Offspring of male mice subjected to social defeat spend
by parents can affect their offspring’s less time in exposed areas and exhibit other anxiety-
metabolism and risk for various diseases, it is related behaviors. When the sperm of these socially
important to be cautious in interpreting what defeated fathers is used to sire offspring via IVF
this means for human health. In our lab, we
find that paternal diet explains perhaps 10
technology, however, the offspring are normal.© WESLEY
percent of the overall variance in cholesterol BEDROSIAN
metabolism among inbred mice. In other
words, factors outside of the father’s diet can alter cholesterol metabolism to a far greater extent than
paternal diet, even in genetically identical animals held in carefully controlled conditions. Other
paternal effects are similarly subtle, presumably one of the reasons why paternal environmental
effects have only been uncovered in the past decade or so. The next decade or two should be an
exciting time as we learn more about what, how, and why we tell our children about the world around
us before they’re even born.

Oliver Rando is a professor in the Department of Biochemistry and Molecular Pharmacology at the
University of Massachusetts Medical School. His research interests include structural analysis of the
yeast genome and the mammalian sperm epigenome.

References
1. M.A. Arteaga-Vazquez, V.L. Chandler, “Paramutation in maize: RNA mediated trans-
generational gene silencing,” Curr Opin Genet Dev, 20:156-63, 2010.
2. A. Fire et al., “Potent and specific genetic interference by double-stranded RNA in
Caenorhabditis elegans,” Nature, 391:806-11, 1998.
3. R.B. Wickner, “[URE3] as an altered URE2 protein: Evidence for a prion analog in
Saccharomyces cerevisiae,” Science, 264:566-69, 1994.
4. C.N. Hales, D.J.P. Barker, “Type 2 (non-insulin-dependent) diabetes mellitus: The thrifty
phenotype hypothesis,” Diabetologia, 35:595-601, 1992.
5. L.H. Lumey et al., “Prenatal famine and adult health,” Annu Rev Public Health,
32:10.1146/annurev-publhealth-031210-101230, 2011.
6. O.J. Rando, “Daddy issues: Paternal effects on phenotype,” Cell, 151:702-08, 2012.
7. J.C. Jimenez-Chillaron et al., “Intergenerational transmission of glucose intolerance and
obesity by in utero undernutrition in mice,” Diabetes, 58:460-68, 2009.
8. A.B. Rodgers et al., “Paternal stress exposure alters sperm microRNA content and reprograms
offspring HPA stress axis regulation,” J Neurosci, 33:9003-12, 2013.
9. D.M. Dietz et al., “Paternal transmission of stress-induced pathologies,” Biol Psychiatry,
70:408-14, 2011.
10. B.G. Dias, K.J. Ressler, “Parental olfactory experience influences behavior and neural structure
in subsequent generations,” Nat Neurosci, 17:89-96, 2014.
11. K. Gapp et al., “Implication of sperm RNAs in transgenerational inheritance of the effects of
early trauma in mice,” Nat Neurosci, 17:667-69, 2014.
12. A.B. Rodgers et al., “Transgenerational epigenetic programming via sperm microRNA
recapitulates effects of paternal stress,” PNAS, doi:10.1073/pnas.1508347112, 2015.
Persistent epigenetic differences associated with
prenatal exposure to famine in humans
Bastiaan T. Heijmansa,1,2, Elmar W. Tobia,2, Aryeh D. Steinb, Hein Putterc, Gerard J. Blauwd, Ezra S. Sussere,f,
P. Eline Slagbooma, and L. H. Lumeye,1
Departments of aMolecular Epidemiology, cMedical Statistics, and dGerontology and Geriatrics, Leiden University Medical Center, Leiden, The Netherlands;
bHubert Department of Global Health, Rollins School of Public Health, Emory University Atlanta, GA 30322; eDepartment of Epidemiology, Mailman School
of Public Health, Columbia University, New York, NY 10032; and fNew York State Psychiatric Institute, New York, NY 10032
Edited by Charles R. Cantor, Sequenom Inc., San Diego, CA, and approved September 17, 2008 (received for review July 7, 2008)
Extensive epidemiologic studies have suggested that adult disease
risk is associated with adverse environmental conditions early in
development. Although the mechanisms behind these relation-
ships are unclear, an involvement of epigenetic dysregulation has
been hypothesized. Here we show that individuals who were
prenatally exposed to famine during the Dutch Hunger Winter in
1944 – 45 had, 6 decades later, less DNA methylation of the im-
printed IGF2 gene compared with their unexposed, same-sex
siblings. The association was specific for periconceptional expo-
sure, reinforcing that very early mammalian development is a
crucial period for establishing and maintaining epigenetic marks.
These data are the first to contribute empirical support for the
hypothesis that early-life environmental conditions can cause epi-
genetic changes in humans that persist throughout life.
developmental origins 兩 DNA methylation 兩 insulin-like growth factor II 兩
nutrition 兩 periconception
S uperimposed on the DNA sequence is a layer of epigenetic
information that is heritable, particularly during mitosis, and
controls the potential of a genomic region to be transcribed (1).
Methyl groups coupled to cytosines in cytosine-guanine (CpG)
dinucleotides and modifications of histones that package the
DNA are the two main molecular marks that compose this
information and regulate chromatin structure and DNA acces-
sibility (2).
Animal studies have indicated that certain transient environ-
mental influences can produce persistent changes in epigenetic
marks that have life-long phenotypic consequences (3, 4). Early
embryonic development is of special interest in this respect,
because this is a crucial period for establishing and maintaining
epigenetic marks (5). Indeed, culturing of preimplantation mice
embryos found that epigenetic marks are susceptible to nutri-
tional conditions in the very early stages of mammalian devel-
opment (6, 7).
One of the rare opportunities for studying the relevance of
such findings to humans is presented by individuals who were
prenatally exposed to famine during the Dutch Hunger Winter
(8). This period of famine was the consequence of a German-
imposed food embargo in the western part of The Netherlands
toward the end of World War II in the winter of 1944–45. During
this period, registries and health care remained intact, so that
individuals who were prenatally exposed to this famine can be
traced. Moreover, the period of famine was clearly defined, and
official food rations were documented. These unique features
allow us to assess whether prenatal exposure to famine is
associated with persistent epigenetic differences in humans.
One of the best-characterized epigenetically regulated loci is
insulin-like growth factor II (IGF2). IGF2 is a key factor in
human growth and development and is maternally imprinted (9).
Imprinting is maintained through the IGF2 differentially meth-
ylated region (DMR), the hypomethylation of which leads to
bi-allelic expression of IGF2 (10). We recently studied IGF2
DMR methylation in 372 twins (11). IGF2 DMR methylation is
17046 –17049 兩 PNAS 兩 November 4, 2008 兩 vol. 105 兩 no. 44
a normally distributed quantitative trait that is largely deter-
mined by genetic factors in both adolescence and middle age,
indicating that the methylation mark is stable up to middle age.
Thus, if affected by environmental conditions early in human
development, altered IGF2 DMR methylation may be detected
many years later.
Here we used our ongoing Hunger Winter Families Study (8)
to investigate whether prenatal exposure to famine is associated
with persistent differences in methylation of the IGF2 DMR.
Our primary focus was exposure during periconception, thus
ensuring that the exposure was present during the very early
stages of development that are critical in epigenetic program-
ming. To further investigate the role of timing, we also studied
individuals who were exposed late in gestation.
Results
Periconceptional Exposure. Our primary goal was to test whether
periconceptional exposure to famine was associated with differ-
ences in IGF2 DMR methylation in adulthood. Toward this end,
we selected the 60 individuals from the Hunger Winter Families
Study who were conceived during the famine 6 decades ago. The
exposure period thus included the very early stages of develop-
ment. The exposed individuals were compared with their same-
sex sibling to achieve partial genetic matching. Using a quanti-
tative mass spectrometr y– based method (12, 13), the
methylation of five CpG dinucleotides within the IGF2 DMR
was measured (11). Three CpG sites were measured individually,
and two were measured simultaneously, because they could not
be resolved due to their close proximity. All CpG sites but one
were significantly less methylated among periconceptionally
exposed individuals compared with their siblings (1.5 ⫻ 10⫺4 ⱕ
P ⱕ 8.1 ⫻ 10⫺3; see Table 1). The average methylation fraction
of the IGF2 DMR based on all five CpG sites was 0.488 among
exposed siblings and 0.515 among unexposed siblings. Thus,
periconceptional exposure was associated with a 5.2% lower
methylation (P ⫽ 5.9 ⫻ 10⫺5), corresponding to 0.48 standard
deviations (SDs) of the controls. The association was indepen-
dent of sex (Pinteraction ⫽ 0.20).
Fig. 1A displays the difference in IGF2 DMR methylation
within sibships according to the estimated conception date of the
famine-exposed individual. IGF2 DMR methylation was lowest
in the famine-exposed individual among 72% (43/60) of sibships;
this lower methylation was observed in conceptions across the
famine period. Official daily rations were set weekly during the
Author contributions: B.T.H., A.D.S., E.S.S., P.E.S., and L.H.L. designed research; E.W.T. and
G.J.B. performed research; B.T.H., E.W.T., H.P., and L.H.L. analyzed data; and B.T.H., E.W.T.,
P.E.S., and L.H.L. wrote the paper.
The authors declare no conflict of interest.
This article is a PNAS Direct Submission.
1Towhom correspondence may be addressed. E-mail: b.t.heijmans@lumc.nl or lumey@
columbia.edu.
2B.T.H. and E.W.T. contributed equally to this work.
© 2008 by The National Academy of Sciences of the USA
www.pnas.org兾cgi兾doi兾10.1073兾pnas.0806560105
 Table 1. IGF2 DMR methylation among individuals periconceptionally exposed to famine and their unexposed,
same-sex siblings
Mean methylation fraction (SD)
IGF2 DMR Relative change Difference
methylation Exposed (n ⫽ 60) Controls (n ⫽ 60) exposed in SDs P

Average 0.488 (0.047) 0.515 (0.055) ⫺5.2% ⫺0.48 5.9 ⫻ 10⫺5

CpG 1 0.436 (0.037) 0.470 (0.041) ⫺6.9% ⫺0.78 1.5 ⫻ 10⫺4
CpG 2 and 3 0.451 (0.033) 0.473 (0.055) ⫺4.7% ⫺0.41 8.1 ⫻ 10⫺3
CpG 4 0.577 (0.114) 0.591 (0.112) ⫺2.3% ⫺0.12 .41
CpG 5 0.491 (0.061) 0.529 (0.068) ⫺7.2% ⫺0.56 1.4 ⫻ 10⫺3

P values were obtained using a linear mixed model and adjusted for age.

famine period and were the same for every individual. The
average daily rations were 667 kcal (SD, 151) (Fig. 1 A), and there
was little variation in the percentage of calories from proteins
(⬇12%, of which 4% of animal origin), fat (19%), and carbo-
hydrates (69%) (14).
As a technical validation, IGF2 DMR methylation was remea-
sured in 46 of 60 periconceptionally exposed individuals and
their same-sex siblings, repeating the whole procedure from
bisulfite treatment to quantification. A similarly lower 5.6%
IGF2 DMR methylation was observed (P ⫽ 2.1 ⫻ 10⫺3),
confirming our initial findings.
Late Gestational Exposure. To further investigate the influence of
timing, we selected the 62 individuals who were exposed to
famine late in gestation for at least 10 weeks, so that they were
born in or shortly after the famine. We found no difference in
IGF2 DMR methylation between the exposed individuals and
their unexposed siblings (Table 2; Fig. 1B).
To formally test whether the association with lower IGF2
DMR methylation depended on the timing of exposure, we
analyzed the periconceptional and late exposure groups together
with all 122 controls in a single model (Table 3). Periconcep-
tional exposure was associated with lower methylation (P ⫽
1.5 ⫻ 10⫺5), whereas late exposure was not (P ⫽ 0.69). Further-
more, there was statistically significant evidence for an interac-
tion between timing and exposure (Pinteraction ⫽ 4.7 ⫻ 10⫺3),
indicating that the association was timing-specific.
Birth Weight. The mean birth weight of the 62 individuals exposed
late in gestation was 3126 g (SD, 408), which is 296 g lower (95%
confidence interval [CI], ⫺420 to ⫺170 g) than the mean (3422
g; SD, 464) of 324 reference births in 1943 at the same institu-
tions (P ⫽ 4 ⫻ 10⫺6) (15). The lower birth weight underscores
the impact of the famine during the Hunger Winter notwith-
standing the absence of an association with IGF2 DMR methyl-
ation. The mean birth weight of the 60 individuals who were
exposed periconceptionally was 3612 g (SD, 648), not lower that
that of the reference births (95% CI, ⫹15 to ⫹ 365 g; P ⫽ 0.03).
IGF2 DMR methylation was not associated with birth weight
(P ⫽ 0.39).
Age Association. To put the association of periconceptional
famine exposure with a 5.2% lower IGF2 DMR methylation into
perspective, we assessed the relationship between age and IGF2
DMR methylation in the 122 control individuals. Within the age
range studied (43–70 years), a 10-year-older age was associated
with a 3.6% lower methylation (P ⫽ .015).
Discussion
Here we report that periconceptional exposure to famine during
the Dutch Hunger Winter is associated with lower methylation
of the IGF2 DMR 6 decades later. The hypomethylation that we
observed is highly comparable to that found for the Nr3c1 and
Ppara genes in offspring of female rats fed an isocaloric protein-
deficient diet starting before pregnancy (⫺8.2% and ⫺10.2% vs
⫺5.2% in our human study) (16), although greater effects for the
Agtr1b gene have been found in a similar rat model (17). These
data from animal models are consistent with the interpretation
that famine underlies the IGF2 hypomethylation that we ob-
served and may be related to a deficiency in methyl donors, such
as the amino acid methionine (3). An additional contribution of
other stressors, such as cold and emotional stress (8), cannot be

GENETICS

Fig. 1. Difference in IGF2 DMR methylation between individuals prenatally exposed to famine and their same-sex sibling. (A) Periconceptional exposure:
Difference in methylation according to the mother’s last menstrual period (a common estimate of conception) before conception of the famine-exposed
individual. (B) Exposure late in gestation: Difference in methylation according to the date of birth of the famine-exposed individual. To describe the difference
in methylation according to estimated conception and birth dates, a lowess curve (red or blue) is drawn. The average distributed rations (in kcal/day) between
December 1944 and June 1945 are depicted in green.

Heijmans et al. PNAS 兩 November 4, 2008 兩 vol. 105 兩 no. 44 兩 17047

 Table 2. IGF2 DMR methylation among individuals exposed to famine late in gestation and their unexposed,
same-sex siblings
Mean methylation fraction (SD)
IGF2 DMR Relative change Difference
methylation Exposed (n ⫽ 62) Controls (n ⫽ 62) exposed in SDs P

Average 0.514 0.045 0.519 0.036 ⫺0.9% ⫺0.12 .64

CpG 1 0.460 0.044 0.464 0.048 ⫺0.9% ⫺0.09 .68
CpG 2 and 3 0.462 0.039 0.471 0.039 ⫺1.7% ⫺0.21 .46
CpG 4 0.602 0.085 0.612 0.073 ⫺1.5% ⫺0.12 .30
CpG 5 0.529 0.060 0.531 0.060 ⫺0.3% ⫺0.02 .77

P values were obtained using a linear mixed model and adjusted for age.

ruled out, however. Our study provides the first evidence that
transient environmental conditions early in human gestation can
be recorded as persistent changes in epigenetic information.
In contrast to periconceptional exposure to famine, exposure
late in gestation was not associated with IGF2 DMR methyl-
ation. Epigenetic marks may be particularly vulnerable during
the very early stage of mammalian development, which is a
crucial period for establishing and maintaining epigenetic marks
(5). Experiments in which mouse zygotes were cultured to
blastocysts favor this hypothesis (6, 7). The timing dependence
of the association that we observed also may relate to the timing
of tissue development, however (18). We studied blood and adult
blood cells stem from the hematopoietic system, which is estab-
lished relatively early in mammalian development (eg, day 10.5
in the mouse embryo [cf. weeks 4–6 in human gestation] (19)).
Detailed future studies are needed to establish whether the
susceptibility of epigenetic marks is an intrinsic property of early
mammalian development or a general feature of newly devel-
oping tissues throughout gestation. Our results do not exclude
the occurrence of epigenetic changes later in development (20)
or during aging (21).
The developmental origins hypothesis states that adverse
conditions during development contribute to adult disease risk
(22). Although the mechanisms behind these relationships are
unclear, the involvement of epigenetic dysregulation has been
proposed (22–24). Our findings are a key element in elaborating
this hypothesis. Human studies on the developmental origins of
health and disease often use low birth weight as a proxy for a
compromised prenatal development (22). Our data indicate that
such studies are not necessarily sufficient for testing the involve-
ment of epigenetics and thus extend our previous finding that
birth weight is a poor surrogate for nutritional status during
gestation (15). Epigenetic differences were found among indi-
viduals who were exposed to famine early in gestation and had
a normal birth weight. Exposure to famine late in gestation was
associated with low birth weight, as expected, but not with
epigenetic changes. To monitor the crucial stages of early
development, assessing maternal lifestyle, especially regarding
nutrition (25), and embryo growth using three-dimensional
ultrasonography (26) may be more appropriate than assessing
birth weight.
The current study presents a first example of an association
between a periconceptional exposure and DNA methylation in
humans. It will be of prime interest to investigate whether other
exposures during early development that are more common in
modern societies, including overnutrition (3) and assisted repro-
ductive technologies (27), give rise to similar associations. In
addition, the extent to which epigenetic marks at other genomic
regions are vulnerable to such exposures remains to be estab-
lished. A key area to explore in future studies will be to assess
the phenotypic consequences of changes in epigenetic marks.
Diseases that have been associated with early gestational expo-
sure to famine, such as schizophrenia (28) and coronary heart
disease (29), are of particular interest in this respect. Analogous
to current studies in genetic epidemiology (30), such epigenetic
epidemiologic studies may need to be large and to include
replication. Understanding how epigenetic control depends on
early exposure may shed light on the link between development
and health over the lifespan and ultimately suggest new ways to
prevent human disease.
Methods
Study Population. The design of and recruitment for the Hunger Winter
Families Study were described previously (8). Individuals exposed to famine
prenatally were recruited by identification and follow-up of live singleton
births in 1945 and early 1946 at three institutions in famine-exposed cities (the
midwifery training schools in Amsterdam and Rotterdam and the Leiden
University hospital). As controls, same-sex siblings and unrelated individuals
from the same institutions who were born before or conceived after the
famine period were recruited. Clinical examination, including blood sam-
pling, was completed for 311 exposed individuals, 311 same-sex siblings, and
349 unrelated controls. Birth weight was abstracted from birth records from
the three institutions. No birth weight data are available for the same-sex
siblings who were not born at these institutions.
For the current epigenetic study, we focused on exposed individuals and
their siblings as controls to achieve partial genetic matching in view of the

Table 3. Timing of famine exposure during gestation, IGF2 DMR methylation, and
birth weight
Periconceptional Late gestational
exposure exposure All controls

n 60 62 122
Males, % 46.7 45.2 45.9
Mean age, years 58.1 (SD, 0.35) 58.8 (SD, 0.4) 57.1 (SD, 5.5)
Birth weight, g 3612 (SD, 648) 3126 (SD, 408) —
IGF2 DMR methylation
Average 0.488 (SD, 0.047) 0.514 (SD, 0.045) 0.517 (SD, 0.047)
Pvs all controls 1.5 ⫻ 10⫺5 .69
Pinteraction 4.7 ⫻ 10⫺3

P values were obtained using a linear mixed model and adjusted for age.

17048 兩 www.pnas.org兾cgi兾doi兾10.1073兾pnas.0806560105 Heijmans et al.

high heritability of IGF2 DMR methylation (11). From these, we selected the
sibships with an individual exposed to famine periconceptionally and those
with an individual exposed to famine late in gestation. Periconceptional
exposure was defined as the mother’s last menstrual period before conceiving
the exposed individual between November 28, 1944 and May 15, 1945. This
yielded 60 sibships. Exposure late in gestation was defined as a birth between
January 28 and May 30, 1945, so that the duration of the famine exposure was
at least 10 weeks. This yielded 62 sibships.
DNA Methylation. Methylation of the IGF2 DMR was measured using genomic
DNA from whole blood extracted using the salting-out method. One micro-
gram of genomic DNA was bisulfite-treated using the EZ 96-DNA methylation
kit (Zymo Research). Sibships were bisulfite-treated on the same plate. Three
plates were used to process the 244 samples, each with an equal number of
samples and a similar distribution in periconceptionally and late-exposed
subjects. The region harboring the IGF2 DMR (chr11:2,126,035–2,126,372 in
NCBI build 36.1) was amplified using primers described elsewhere (11). DNA
methylation was measured using a mass spectrometry– based method (Epi-
typer, Sequenom) (12), the quantitative accuracy (R2 duplicate measure-
ments ⱖ 0.98) and concordance with clonal polymerase chain reaction bisul-
fite sequencing of which has been reported previously (13, 31). All
measurements were done in triplicate. CpG dinucleotides whose measure-
ment was confounded by single nucleotide polymorphisms, as we discussed in
a previous report (11), were discarded as part of quality control. The CpG
dinucleotides reported in the current study were located at positions 41, 57
and 60, 202, and 251 bp in the amplicon targeting the IGF2 DMR. Methylation
data were 93% complete. DNA methylation of five CpG dinucleotides could be
measured, three individually and two as a pair because they were directly
adjacent and could not be resolved individually.
1. Jaenisch R, Bird A (2003) Epigenetic regulation of gene expression: How the genome
integrates intrinsic and environmental signals. Nat Genet 33(Suppl):245–254.
2. Bernstein BE, Meissner A, Lander ES (2007) The mammalian epigenome. Cell 128:669 –
681.
3. Waterland RA, Jirtle RL (2003) Transposable elements: Targets for early nutritional
effects on epigenetic gene regulation. Mol Cell Biol 23:5293–5300.
4. Sinclair KD, et al. (2007) DNA methylation, insulin resistance, and blood pressure in
offspring determined by maternal periconceptional B vitamin and methionine status.
Proc Natl Acad Sci U S A 104:19351–19356.
5. Reik W, Dean W, Walter J (2001) Epigenetic reprogramming in mammalian develop-
ment. Science 293:1089 –1093.
6. Doherty AS, Mann MR, Tremblay KD, Bartolomei MS, Schultz RM (2000) Differential
effects of culture on imprinted H19 expression in the preimplantation mouse embryo.
Biol Reprod 62:1526 –1535.
7. Morgan HD, Jin XL, Li A, Whitelaw E, O’Neill C (2008) The culture of zygotes to the
blastocyst stage changes the postnatal expression of an epigentically labile allele,
agouti viable yellow, in mice. Biol Reprod 79:618 – 623.
8. Lumey LH, et al. (2007) Cohort profile: The Dutch Hunger Winter Families Study. Int J
Epidemiol 36:1196 –1204.
9. Smith FM, Garfield AS, Ward A (2006) Regulation of growth and metabolism by
imprinted genes. Cytogenet Genome Res 113:279 –291.
10. Cui H, et al. (2003) Loss of IGF2 imprinting: A potential marker of colorectal cancer risk.
Science 299:1753–1755.
11. Heijmans BT, Kremer D, Tobi EW, Boomsma DI, Slagboom PE (2007) Heritable rather
than age-related environmental and stochastic factors dominate variation in DNA
methylation of the human IGF2/H19 locus. Hum Mol Genet 16:547–554.
12. Ehrich M, et al. (2005) Quantitative high-throughput analysis of DNA methylation
patterns by base-specific cleavage and mass spectrometry. Proc Natl Acad Sci USA
102:15785–15790.
13. Coolen MW, Statham AL, Gardiner-Garden M, Clark SJ (2007) Genomic profiling of CpG
methylation and allelic specificity using quantitative high-throughput mass spectrom-
etry: Critical evaluation and improvements. Nucleic Acids Res 35:e119.
14. Burger GCE, Drummond JC, Sandstead HR (1948) Malnutrition and Starvation in
Western Netherlands, September 1944 –July 1945, Part 1 (General State Printing Office,
The Hague).
15. Stein AD, Zybert PA, van de Bor M, Lumey LH (2004) Intrauterine famine exposure and
body proportions at birth: The Dutch Hunger Winter. Int J Epidemiol 33:831– 836.
Heijmans et al.
Statistical Analysis. The mean methylation fractions of individual CpGs and
their SDs presented in the tables and figures are based on raw data. To obtain
the average methylation of the whole IGF2 DMR presented in the tables and
figures, missing methylation data were first imputed using estimates from
linear mixed models, thereby exploiting the correlations among CpG sites (11).
To test for differences between exposed individuals and their unexposed
siblings, age-adjusted linear mixed models were applied to the raw data
without imputation of missing values. These analyses accounted for age at
examination, family relations, correlated methylation of CpG dinucleotides,
and methylation data missing at random. Exposure status, CpG dinucleotide,
and age were entered as fixed effects, and sibship was entered as a random
effect. The model including both the periconceptional and the late-exposure
groups was extended with a variable indicating timing of the exposure and an
interaction term of exposure status times exposure time. To test for the
association between IGF2 DMR methylation and birth weight, birth weight
was added as a fixed effect. The linear mixed model may be viewed as an
extension of the paired t-test; the model reduces to a paired t-test with
identical outcomes if within-family methylation differences are assessed for a
single CpG nucleotide and if data are complete and age adjustment is omitted.
All P values are two-sided, and all statistical analyses were performed using
SPSS 14.0.
ACKNOWLEDGMENTS. We thank the participants of the Hunger Winter
Families Study, TNO Quality of Life for contact tracing, the staff of the
Gerontology and Geriatrics Study Center at the Leiden University Medical
Center for performing the clinical examinations, Marja Kersbergen and Mar-
got van Schie for extracting genomic DNA, and Dennis Kremer for technical
assistance. This work was supported by grants from the Netherlands Heart
Foundation (2006B083 to B.T.H.), the U.S. National Institutes of Health (RO1-
HL067914 to L.H.L.), the Netherlands Organization for Scientific Research
NWO (911– 03-016 to P.E.S.), and the European Union–funded Network of
Excellence LifeSpan (FP6 036894).
16. Burdge GC, et al. (2007) Dietary protein restriction of pregnant rats in the F0 gener-
ation induces altered methylation of hepatic gene promoters in the adult male
offspring in the F1 and F2 generations. Br J Nutr 97:435– 439.
17. Bogdarina I, Welham S, King PJ, Burns SP, Clark AJ (2007) Epigenetic modification of
the renin-angiotensin system in the fetal programming of hypertension. Circ Res
100:520 –526.
18. Waterland RA, Lin JR, Smith CA, Jirtle RL (2006) Post-weaning diet affects genomic
imprinting at the insulin-like growth factor 2 (Igf2) locus. Hum Mol Genet 15:705–716.
19. Dzierzak E, Speck NA (2008) Of lineage and legacy: The development of mammalian
hematopoietic stem cells. Nat Immunol 9:129 –136.
20. Weaver IC, et al. (2004) Epigenetic programming by maternal behavior. Nat Neurosci
7:847– 854.
21. Bjornsson HT, et al. (2008) Intra-individual change over time in DNA methylation with
familial clustering. JAMA 299:2877–2883.
22. Waterland RA, Michels KB (2007) Epigenetic epidemiology of the developmental
origins hypothesis. Annu Rev Nutr 27:363–388.
23. Cutfield WS, Hofman PL, Mitchell M, Morison IM (2007) Could epigenetics play a role
in the developmental origins of health and disease? Pediatr Res 61:68R–75R.
24. McClellan JM, Susser E, King MC (2006) Maternal famine, de novo mutations, and
schizophrenia. JAMA 296:582–584.
25. Verkleij-Hagoort AC, et al. (2007) Validation of the assessment of folate and vitamin
B12 intake in women of reproductive age: The method of triads. Eur J Clin Nutr
61:610 – 615.
GENETICS
26. Verwoerd-Dikkeboom CM, Koning AH, van der Spek PJ, Exalto N, Steegers EA (2008)
Embryonic staging using a 3D virtual reality system. Hum Reprod 23:1479 –1484.
27. DeBaun MR, Niemitz EL, Feinberg AP (2003) Association of in vitro fertilization with
Beckwith-Wiedemann syndrome and epigenetic alterations of LIT1 and H19. Am J Hum
Genet 72:156 –160.
28. Susser E, et al. (1996) Schizophrenia after prenatal famine: Further evidence. Arch Gen
Psychiatry 53:25–31.
29. Painter RC, et al. (2006) Early onset of coronary artery disease after prenatal exposure
to the Dutch famine. Am J Clin Nutr 84:322–327.
30. Wellcome Trust Case Control Consortium (2007) Genome-wide association study of
14,000 cases of seven common diseases and 3,000 shared controls. Nature 447:661– 678.
31. Ehrich M, et al. (2008) Cytosine methylation profiling of cancer cell lines. Proc Natl Acad
Sci USA 105:4844 – 4849.
PNAS 兩 November 4, 2008 兩 vol. 105 兩 no. 44 兩 17049
Questions for primary literature:
1) What is the hypothesis or hypotheses the authors set out to test?

2) Why was this hypothesis important or interesting to the authors?

3) What are the methods they used? Don’t get too hung up in details here.

4) What are their findings? Can you explain the figures in the paper? What do they show?

5) How did they interpret these findings?

6) Do you feel their conclusions were justified by their data? Why or why not?

7) If this were your work, how would you follow-up on the study?
Weekly self-review:
1) Make a list of terms from this week’s lecture and reading. On another sheet or on note cards,
define them in your own words.

2) What are the most important concepts? Explain them to a friend.

3) What are the terms or concepts you don’t understand? Look them up. Fill details
into your lecture notes based on the reading. Ask a friend. Come to office hours!

4) How do these concepts relate to one another? On another sheet, make a concept map
to link this week’s terms and concepts
Genetics and heritability readings
4/30/2019

Estimating Trait Heritability

By: Naomi R. Wray, Ph.D. (Queensland Institute of Medical Research, Brisbane, Australia) & Peter M.
Visscher (Queensland Institute of Medical Research, Brisbane, Australia) © 2008 Nature Education
Citation: Wray, N. & Visscher, P. (2008) Estimating trait heritability. Nature
Education 1(1):29

Genetic variation in a population can result from a variety of things. What are the ways we can estimate trait heritability?

A central question in biology is whether observed variation in a particular trait is due to environmental or to biological factors, sometimes
popularly expressed as the "nature versus nurture" debate. Heritability is a concept that summarizes how much of the variation in a trait is due to
variation in genetic factors. Often, this term is used in reference to the resemblance between parents and their offspring. In this context, high
heritability implies a strong resemblance between parents and offspring with regard to a specific trait, while low heritability implies a low level of
resemblance.

Quantifying Heritability
Phenotypes that vary between the individuals in a population do so because of both environmental factors and the genes that influence traits, as
well as various interactions between genes and environmental factors. Unless they are genetically identical (e.g., monozygotic twins in humans,
inbred lines in experimental populations, or clones), the individuals in a population tend to vary in the genotypes they have at the loci affecting
particular traits. The combined effect of all loci, including possible allelic interactions within loci (dominance) and between loci (epistasis), is the
genotypic value. This value creates genetic variation in a population when it varies between individuals. In fact, heritability is formally defined as
the proportion of phenotypic variation (VP) that is due to variation in genetic values (VG).

Genotypes or genotypic values are not passed on from parents to progeny; rather, it is the alleles at the loci that influence the traits that are
passed on. Therefore, to predict the average genotypic value of progeny and their predicted average phenotype, investigators need to know the
effect of alleles in the population rather than the effect of a genotype. The effect of a particular allele on a trait depends on the allele's frequency
in the population and the effect of each genotype that includes the allele. This is sometimes termed the average effect of an allele. The additive
genetic value of an individual, called the breeding value, is the sum of the average effects of all the alleles the individual carries (Falconer &
Mackay, 1996). According to the principles of Mendelian segregation, one allele from each locus is present in each gamete, and in this way,
additive genetic values are passed on from parents to progeny. Indeed, because each offspring receives a different set of alleles from its parents,
half of the additive genetic variance in the population occurs within families.

Broad-sense heritability, defined as H2 = VG/VP, captures the proportion of phenotypic variation due to genetic values that may include effects
due to dominance and epistasis. On the other hand, narrow-sense heritability, h2 = VA/VP, captures only that proportion of genetic variation that
is due to additive genetic values (VA). For definitions and decomposition of components of variation, you can read more about phenotypic
variance. Note that often, no distinction is made between broad- and narrow-sense heritability; however, narrow-sense h2 is most important in
animal and plant selection programs, because response to artificial (and natural) selection depends on additive genetic variance. Moreover,
resemblance between relatives is mostly driven by additive genetic variance (Hill et al., 2008).

Given its definition as a ratio of variance components, the value of heritability always lies between 0 and 1. For instance, for height in humans,
narrow-sense heritability is approximately 0.8 (Macgregor et al., 2006). For traits associated with fitness in natural populations, heritability is
typically 0.1–0.2 (Visscher et al., 2008).

Heritability Estimation
Estimation of heritability in populations depends on the partitioning of observed variation
into components that reflect unobserved genetic and environmental factors. In other
words, researchers recognize that genetic and/or environmental variation exists, but they
may not be in a position to assess either directly. However, this does not prevent them
from being able to estimate the relative effects of both genes and environment on
phenotype. Here, heritability can be estimated from empirical data on the observed and
expected resemblance between relatives. The expected resemblance between relatives
depends on assumptions regarding a trait's underlying environmental and genetic
causes. Traditionally, heritability was estimated from simple, often balanced, designs,
such as the correlation of offspring and parental phenotypes, the correlation of full or half Figure 1: Heritability estimation.
siblings, and the difference in the correlation of monozygotic (MZ) and dizygotic (DZ) twin Low (panel a) and high (panel b) heritability can
pairs. Heritability can also be estimated from the ratio of the observed selection response be estimated from the regression (h2) of offspring
(R) to the observed selection differential (S) in artificial selection experiments. This phenotypic values vs. the average of parental
relationship is summarized in the "breeder's equation," R = h2S. phenotypic values.
© 2008 Nature Education All rights reserved.
In Figure 1, examples are given of a scatterplot of progeny phenotypes (y-axis) and the
average of two parental phenotypes (x-axis), for traits with high (0.9) and low (0.1) Figure Detail
heritability. The straight line is the best-fit linear relationship between y and x, obtained
from a statistical technique called linear regression. The slope of the regression line is an
estimate of narrow-sense heritability. For the high heritability of 0.9 (Figure 1b), there still is a lot of variation around the regression line, because

1/3
4/30/2019
the correlation between offspring phenotype and mid-parent value is √(½)h2, which is only 0.64 for h2 = 0.9. Even when the heritability is 1.0 (i.e.,
there is no environmental variation), the phenotypes of offspring and parents are not identical because of random segregation of alleles from
parents to progeny. This explains, for example, why human siblings can vary considerably in height, despite the heritability of height being very
large.

When phenotypic measures are available on individuals with a mixture of relationships, both within and across multiple generations, or when the
design is unbalanced (e.g., there are unequal numbers of observations per family), estimates of additive genetic variance and environmental
components are most efficiently calculated via statistical methods that use all data simultaneously and take account of the exact properties of the
data. Such methods are iterative and computationally more intensive than estimates of heritability that are based upon regression or correlation
coefficients.

Estimating Heritability: Caveats

When estimating heritability from the observed and expected resemblance between relatives, a model is necessary to specify the expected
resemblance in terms of genetic and environmental factors. Sometimes this model is straightforward; for example, it may posit that the observed
resemblance between half-sibling dairy cows on different farms is due solely to additive genetic factors inherited from the common parent. In
other cases, a model's assumptions may be open to questioning. For example, in human twin analysis, it is usually assumed that the
resemblance between monozygotic and dizygotic twin pairs due to shared environment is the same.

Recently, new methods that exploit the use of genetic marker data have been proposed and applied to estimate heritability essentially free of
such assumptions regarding the nature of between-family variation (Visscher et al., 2006). These methods are based upon the correlation
between phenotypic and genetic similarity within families. They exploit the fact that there is variation in identity (defined here as the proportion of
the genome that is shared identical-by-descent) between pairs of individuals that have the same expected value and that this variation can be
measured with genetic markers. Variation in identity arises because of the random segregation of chromosomes during meiosis. For full siblings
in humans, the mean identity is 50%, with a standard deviation of approximately 4%. Hence, some full siblings share only 40% of their genome
by descent, while others share 60%. If those siblings who share more of their genome than average are phenotypically more similar to each other
than those siblings who share less than average, then this similarity is most likely due to genetic factors. This assumption was the basis of a
study by Visscher et al. (2006), who estimated a narrow-sense heritability of height in humans of 0.8 using pairs of full siblings, without making any
assumption about the variation between families.

Heritability Is Not Necessarily Constant

Interestingly, heritabilities are not constant. For example, estimates of heritability for first lactation milk yield in dairy cattle nearly doubled from
approximately 25% in the 1970s to roughly 40% in recent times. Heritability can change over time because the variance in genetic values can
change, the variation due to environmental factors can change, or the correlation between genes and environment can change. Genetic variance
can change if allele frequencies change (e.g., due to selection or inbreeding), if new variants come into the population (e.g., by migration or
mutation), or if existing variants only contribute to genetic variance following a change in genetic background or the environment. The same trait
measured over an individual's lifetime may have different genetic and environmental effects influencing it, such that the variances become a
function of age. For example, variance in weight at birth is influenced by maternal uterine environment, and variance in weight at weaning
depends on maternal milk production, but variance of mature adult weight is unlikely to be influenced by maternal factors, which themselves have
both a genetic and environmental component. Heritabilities may be manipulated by changing the variance contributed by the environment.
Empirical evidence for morphometric traits suggests lower heritabilities in poorer environments, but not for traits more closely related to fitness
(Charmantier & Garant, 2005). Understanding how heritability changes with environmental stressors is important for understanding evolutionary
forces in natural populations (Charmantier & Garant, 2005).

Misconceptions of the Heritability Concept

There are a number of common misconceptions on the exact meaning and interpretation of heritability (Visscher et. al., 2008). Heritability is not
the proportion of a phenotype that is genetic, but rather the proportion of phenotypic variance that is due to genetic factors. Heritability is a
population parameter and, therefore, it depends on population-specific factors, such as allele frequencies, the effects of gene variants, and
variation due to environmental factors. It does not necessarily predict the value of heritability in other populations (or other species).
Nevertheless, it is surprising how constant heritabilities are across populations and species (Visscher et. al., 2008).

Applications of heritability estimation are broad and cross a range of disciplines, from evolutionary biology to agriculture to human medicine. In
humans, estimation of heritability has been applied to diseases and behavioral phenotypes (e.g., IQ), and it has helped establish that a
substantial proportion of variation in risk for many disorders, like schizophrenia, autism, and attention deficit/hyperactivity disorder, is genetic in
origin.

References and Recommended Reading

Charmantier, A., & Garant, D. Environmental quality and evolutionary potential: Lessons from wild populations. Proceedings of the Royal Society, Biological Sciences 272, 1415–1425 (2005)

Falconer, D. S., & Mackay, T. F. C. Introduction to Quantitative Genetics (Harlow, UK, Longman, 1996)

Hill, W. G., et al. Data and theory point to mainly additive genetic variance for complex traits. PLoS Genetics 4, e1000008 (2008)

2/3
4/30/2019
Macgregor, S., et al. Bias, precision and heritability of self-reported and clinically measured height in Australian twins. Human Genetics 120, 571–580 (2006)

Visscher, P. M., et al. Assumption-free estimation of heritability from genome-wide identity-by-descent sharing between full siblings. Public Library of Science Genetics 2, e41 (2006)

———. Heritability in the genomics era—Concepts and misconceptions. Nature Reviews Genetics 9, 255–266 (2008) (link to article)

3/3
Perspectives
Lander and Botstein deemed LD mapping
E S S AY
to be impractical in the general population
owing to the high marker density that
The road to genome-wide would be required, but they proposed
that a map of hundreds of RFLPs might
association studies suffice for LD mapping in recently
founded isolated populations.
The first complete RFLP map of the
Leonid Kruglyak human genome was reported in 1987
(Ref. 11), but human mapping studies really
Abstract | The recent crop of results from genome-wide association studies flourished once microsatellites replaced
might seem like a sudden development. However, this blooming follows a long RFLPs12 (BOX 1). Genome-wide studies
germination period during which the necessary concepts, resources and used family-linkage approaches almost
techniques were developed and assembled. Here, I look back at how the exclusively, with LD being used to refine
necessary pieces fell into place, focusing on the less well-chronicled days before the locations of genes that were mapped
by linkage, as pioneered by Kerem and
the launch of the HapMap project, and speculate about future developments.
colleagues for the cystic fibrosis gene13. In
a groundbreaking and forward-looking
Genome-wide association studies molecular markers6,7. The natural initial study in 1994, Houwen and colleagues
(GWAS) use dense maps of SNPs that applications were to genetically simple reported the first application of LD map-
cover the human genome to look for Mendelian diseases; however, as early as ping in an initial whole-genome search for
allele-frequency differences between 1986, even before the first linkage map was a disease locus14. Following the approach
cases (patients with a specific disease or completed, Lander and Botstein recog- that was envisioned by Lander and
individuals with a certain trait) and con- nized that most human traits and diseases Botstein almost a decade earlier, they used
trols. A significant frequency difference is follow complex modes of inheritance, 256 markers to map the gene responsible
taken to indicate that the corresponding and they discussed several approaches for benign recurrent intrahepatic cholesta-
region of the genome contains functional for studying such complex traits8. One of sis (BRIC) in an isolated fishing com-
DNA-sequence variants that influence the the approaches they proposed was link- munity in the Netherlands. Their success
disease or trait in question. The recent age disequilibrium (LD) mapping, which relied on the rarity of the disease and on
crop of results from GWAS (reviewed recognizes that a mutation that is shared the availability of a population isolate in
in refs 1–4) might seem like a sudden by affected individuals through common which the affected individuals were distant
development. However, this blooming descent will be surrounded by shared relatives. A similar study in a Mennonite
follows a long germination period during alleles at nearby loci, representing the kindred allowed Puffenberger and col-
which the necessary concepts, resources haplotype of the ancestral chromosome on leagues to identify a gene for Hirschsprung
and techniques were developed and which the mutation first occurred (FIG. 1). disease15. However, such studies, which
assembled. Here, I look back at how the The first example of LD between a DNA straddled the border between family
necessary pieces fell into place, starting polymorphism and a disease mutation linkage and LD mapping, remained the
with early ideas and continuing with was provided by an association between exception as linkage approaches domi-
concrete proposals and theoretical and an allele of an RFLP in the β-globin gene nated. Studies of many pairs of relatives
empirical studies that laid the foundation and the sickle-cell form of haemoglobin9. (most commonly, affected sib pairs) were
for The International HapMap Project5. I especially prevalent owing to the ease of
close by contemplating the implications of collecting such samples versus samples
the lessons that were learnt from the initial [Genetic] complexity is from extended families, to their theoreti-
crop of GWAS for future studies of human cal appeal for mapping complex traits16–18
present on multiple levels,
genetic variation. and to the availability of powerful analysis
and might be fruitfully thought tools19. These genome scans were carried
Early milestones of as ‘fractal’. out for many common diseases that show
Genome-wide approaches to human complex inheritance, but they failed to
genetics date back to the proposal in find many reproducible loci. With these
1980 by Botstein and colleagues for the Simple population-genetics arguments findings, the initial belief that a few major
construction of a linkage map of the suggested that LD in the general human genes would explain susceptibility to
human genome, with restriction frag- population would probably be limited to complex diseases gave way to the realiza-
ment length polymorphisms (RFLPs) as distances below 100 kb10. For this reason, tion that the level of complexity was much

314 | april 2008 | volume 9 www.nature.com/reviews/genetics

© 2008 Nature Publishing Group
 Perspectives

higher and that many loci of individually
small effect were involved. Because such
loci are difficult to identify by family link-
age owing to limited power, the search was
on for new approaches.
Modern proposals
In an influential perspective, Risch and
Merikangas argued that association
studies should be more powerful than
family linkage studies for detecting high-
frequency, small-effect polymorphisms20.
Linkage studies rely on allele sharing by
descent among affected relatives, and
their low power to detect such polymor-
phisms is due to two factors. First, when
the increased risk conferred by an allele
is small, some relatives will be affected
because of other causes and will not carry
the risk allele. Second, when an allele is
common, it can enter the family through
multiple founders, erasing clear inheritance
patterns. These effects combine to
decrease sharing by descent to the point
at which an impractically large number
approach of cataloguing all common func-
tional variants and the indirect approach
of relying on a dense map of SNPs for LD
mapping (FIG. 2). A back-of-the-envelope
calculation put the likely size of shared
ancestral haplotypes in the range of
10–100 kb, leading to a proposal to identify
at least 100,000 SNPs. To achieve this goal,
The SNP Consortium, a public–private
partnership, was launched in 1999.
Charting the course
The early proposals for genome-wide stud-
ies were audacious, because the number of
SNPs known at the time was small, and the
approaches to their discovery and geno-
typing were cumbersome. In 1998, Wang
and colleagues performed an important
feasibility study, discovering some 3,000
SNPs and developing an array-based geno-
typing approach that could assay hundreds
of SNPs in parallel23. The SNP consortium
and the HapMap project would eventually
bridge the gap between this early survey
and the much larger number of SNPs
prediction that some 500,000 SNPs would
be required for GWAS, even if relatively
low LD levels between mapped SNPs and
functional variants were deemed accept-
able25. The predicted number of SNPs was
considerably larger than the goals of SNP
discovery projects at the time26, and led to
an increase in the targeted number. Indeed,
less than 2 years later, a map of 1 million
SNPs was reported24. Given the simpli-
fied nature of the model that was used to
calculate the estimate of 500,000 SNPs,
this number has held up remarkably well
— most of the recent successes of GWAS
came when approximately this number of
SNPs could be genotyped within individual
studies, and the current generation of com-
mercial SNP-typing products deploys some
500,000–1,000,000 SNPs.
When the prediction is viewed from the
vantage point of the extensive empirical
data available today (for example, ref. 27),

chromosome
of families must be studied to detect it. required.

Association studies still suffer from the
first effect (which is inherent to searches
for small effects) but not the second, and
therefore they have a higher sensitivity
in detecting common variants with small
effects. The increase in power is such
that even testing large numbers of poly-
morphisms, with the ensuing statistical
costs of multiple testing, does not erase
the advantage of association studies20,21.
Ancestral
How many SNPs are needed? The number
of SNPs that are required for LD mapping
obviously depends on the genomic extent
of LD because genotyped SNPs must be
spaced sufficiently densely to be in LD
with most of the (potentially disease-
associated) variants that are not genotyped.
At the time the proposals for GWAS were
made, few empirical estimates of the extent
Present-day chromosomes

In addition to offering higher power with
the same sample sizes, association studies
also have the practical advantage that large
samples of unrelated cases and controls
can be collected much more easily than
family-based samples.
Risch and Merikangas issued a call for
a catalogue of all variants in human genes,
and set out a challenge “to the molecular
technologists to develop the tools” for
their identification and genotyping20.
This call was echoed by Eric Lander,
who hypothesized that common variants
of modest effect might hold the key to
susceptibility to common diseases21 (this
was subsequently codified as the common
disease–common variant hypothesis).
Lander also noted that the role of noncod-
ing variation might be studied by the use
of LD mapping with a sufficiently dense
polymorphism map.
These proposals were formalized the
following year in a policy forum by Collins,
Guyer and Chakravarti22. They made
explicit the distinction between the direct
of LD were available, and these varied
wildly from observations of LD over hun-
dreds of kb to the breakdown of LD at very
short distances. This range of observations
translated into an uncertainty of up to
three orders of magnitude in the required
number of SNPs — from thousands to mil-
lions. Even several years later, the number
of SNPs required for GWAS was said to be
in the range of 30,000–1,000,000, based
on a survey of empirical studies24. Starting
in 1997, I attempted to reduce this uncer-
tainty by using simple population-genetics
models to calculate the likely extent of
LD. A highly realistic model could not be Figure 1 | Linkage disequilibrium around
Nature Reviews an
| Genetics
constructed at the time because of a lack ancestral mutation. The mutation is indicated
of detailed information regarding both the by a red triangle. Chromosomal stretches that
demographic history of different popula- are derived from the common ancestor of all
mutant chromosomes are shown in light blue,
tions and the variation in recombination
whereas new stretches introduced by recombi-
rate at short distances. Instead, the aim nation are shown in dark blue. Markers that are
was to obtain a reasonable estimate. In the physically close (that is, within the light-blue
model that was designed to approximate regions of present-day chromosomes) tend to
the global human population, moderate remain associated with the ancestral mutation,
levels of LD were confined to regions of even as recombination whittles down the
approximately 6 kb, thus leading to the region of association over time.

nature reviews | genetics volume 9 | april 2008 | 315

© 2008 Nature Publishing Group
Perspectives

Box 1 | A brief history of genetic markers now includes 3 million SNPs, estimated
to cover one-quarter to one-third of all
Human genetic mapping was initially based on restriction fragment length polymorphisms human SNPs with frequencies above 5%.
(RFLPs)6,9,49,50 — fragment length variants generated through the presence and/or absence of Where do we go from here?
restriction enzyme recognition sites6,11,53,54. RFLPs resulted from various sequence changes
The recent crop of discoveries from
including base substitutions, insertions and deletions, and were laboriously assayed by
Southern blots. Southern blots were superseded by PCR-based assays for microsatellite
GWAS is a major advance in our under-
markers (also known as short tandem repeats or simple sequence length repeats)10,51. standing of the genetic basis of common
Microsatellites are di‑, tri‑ or tetranucleotide repeat sequences that are composed of many diseases, as well as normal human varia-
tandem repeats. Many alleles are generally associated with each microsatellite within most tion39,40. Nevertheless, the associated loci
populations, hence their use as markers for carrying out family-based linkage analysis12,55. that have been identified usually have
More recently, SNPs have become the markers of choice; their lower polymorphism is offset small individual effects on phenotype,
by their abundance and ease of genotyping23,56, and their low mutation rates make them and even collectively tend to explain only
especially suitable for linkage disequilibrium mapping. a small fraction of the heritable com-
ponent3. For some diseases studied, no
significant loci have been identified3,41.
it is clear that the actual average extent across the genome5,32–35. Rather, a million This failure to detect loci that explain
of LD is greater than in the model. This or more SNPs would need to be genotyped the bulk of the heritable components of the
is especially true in non-African popula- in substantial numbers of individuals from phenotypes studied could be attributable to
tions, most likely owing to a combination multiple populations in order to select several factors. First, because the detected
of demographic factors and a clustering of sets of several hundred thousand (with loci have small effects, the power to detect
recombination events at hot spots28 (both the precise number depending on the them is low, and more such loci remain
of these effects were anticipated when the population) that would efficiently capture to be discovered as sample sizes increase.
prediction was made25). However, the cal- untyped common variants5,32–34. These Second, association studies can only
culation of the number of SNPs assumed observations gave rise to the HapMap detect the effects that are due to relatively
both a relatively low acceptable level of LD project and a parallel effort by Perlegen common alleles. Rare alleles remain to
and coverage of each region of LD with a Sciences, which eventually joined forces be discovered — both at the loci that are
single SNP. In practice, GWAS have set to produce the SNP panels that are being identified by GWAS because they also have
a higher standard for required LD, and used today29,36,37. These projects also drove common alleles with phenotypic effects,
multiple SNPs are used to ‘tag’ each region the development of rapid and cost-effective and at other loci that do not have such
of high LD. These factors combined lead genotyping technologies, setting the common alleles. The former can be found
to the most current empirical estimates stage for GWAS. These recent develop- by focused resequencing studies of the
of approximately 500,000 SNPs for non- ments are well chronicled elsewhere (for loci identified by GWAS; finding the latter
African and 1,000,000 SNPs for African example, ref. 38). might require resequencing of other genes
populations to ensure adequate coverage in the relevant pathways, of the exons
of the genome in GWAS, even when high The road ahead of all genes42–44 or of the entire genome.
LD levels between mapped SNPs and In the past two decades, GWAS have Third, we might be missing the effects of
functional variants are required29. progressed from visionary proposals, structural variation, of other less well-
made when neither the sequence of the studied types of genome alterations45, and
How should the SNPs be chosen? Initially, human genome nor many variations in of interactions among variants and between
the discussion focused on simply assem- this sequence were known, to routine genetic and environmental factors.
bling a dense collection of SNPs. However, practice of screening 500,000–1,000,000 It is only a matter of time before all
both theoretical considerations and early SNPs in thousands of individuals. The SNPs with appreciable frequencies in the
empirical studies suggested that the physi- recently reported phase 2 of the HapMap29 human population have been discovered.
cal extent and the local patterns of LD
were likely to vary across the genome and
among populations. In commenting on a Direct:
catalogue and test all functional variants for association
one empirical study30 in 1999, I proposed
that LD among a dense collection of
SNPs be measured empirically across the
genome and in different populations in
order to identify the most efficient SNP
panels for association studies (FIG. 3); b Indirect:
such panels would vary in their density use a dense SNP map and test for linkage disequilibrium
by region of the genome and by popula-
tion31. The result of such empirical studies
would constitute an LD map of the human
genome31. As SNP discovery efforts con-
tinued, empirical data confirmed both the Figure 2 | Alternative designs for genome-wide association studies. a | Direct approach of
need for hundreds of thousands of SNPs Nature Reviews | Genetics
testing a catalogue of all common functional variants in the genome. b | Indirect approach
and the fact that these SNPs could not be of testing a dense map of SNPs and relying on linkage disequilibrium to detect associations that
chosen at random or by uniform spacing are due to untested functional variants.

316 | april 2008 | volume 9 www.nature.com/reviews/genetics

© 2008 Nature Publishing Group

Indeed, efforts to discover and genotype
additional SNPs in larger and more
diverse population samples are underway.
Assuming that the genotyping technologies
can keep up the pace, the indirect associa-
tion studies relying on LD will be replaced
by direct association studies that assay all
relatively common SNPs (perhaps the
estimated 11 million SNPs with minor
allele frequencies exceeding 1% in the
population46), although it will probably
still be worthwhile to exclude wholly
redundant SNPs. Thus, LD and haplotype
maps are merely useful but temporary
shortcuts. An interesting finding of
phase 2 of the HapMap is that for approxi-
mately 1% of all SNPs (tens of thousands),
the basic assumption of indirect associa-
tion mapping breaks down — these SNPs
are not in LD with any others, often owing
to their location in hot spots of recombi-
nation, and are thus ‘untaggable’ and must
be assayed directly for phenotypic
associations29. Tailored approaches that are
under development will cover structural
variants47. Studies based on the resequenc-
ing of individual genomes (rather than
genotyping of known variants) will be
needed to begin to comprehensively
address the role of rare variants and
de novo mutations, and will eventually
replace genotyping studies altogether,
although this is likely to take some time. It
is worth noting that resequencing studies
of rare variants have to rely on the rec-
ognition of many different variants, each
of which alters the function of the same
gene or pathway in different individuals.
Whereas recognition of likely functional
variants in coding regions is straightfor-
ward, detecting functional changes in
noncoding DNA poses a major challenge.
This is because regulatory sequences can
be located far from the coding region
and are often difficult to identify, and we
do not have a ready connection between
nucleotide differences and function for
these sequences.
Looking further ahead, we can already
envision the day when the genome
sequences of a significant fraction
of the population are known, at least
in the developed world. Assuming that
the relevant logistical and ethical issues
can be solved, what will we learn from
combining this unprecedented scope of
genetic information with medical records
and other phenotypic data? We are just
beginning to get the first glimpses of the
real underlying genetic complexity of phe-
notypic variation. Complexity is present
nature reviews | genetics volume 9 | april 2008 | 317
Perspectives
Reference SNPs SNPs captured by proxy Uncaptured SNPs
Figure 3 | Schematic of a genomic region to be tested for association Nature
with a Reviews
phenotype. The
| Genetics
four reference SNPs in the mapping panel are indicated by red triangles; these are genotyped
directly. The eight SNPs indicated by yellow triangles are captured through linkage disequilibrium
(by proxy) with the reference SNPs denoted by arrows. The four SNPs indicated by blue triangles
are neither genotyped nor in linkage disequilibrium with the reference SNPs; phenotypic
association that is due to one of these would be missed.
on multiple levels, and might be fruitfully to iteratively refine phenotypic categories
thought of as ‘fractal’. First, many loci are by combining genotypic and phenotypic
involved; we do not yet know how many information. Careful and detailed meas-
but the number could be in the hundreds ures of phenotypes and environmental
for many traits. Second, individual loci exposures will also have an important
can often represent variation in multiple role. Clearly, we have a lot of work to do
linked genes, as has been found in model before an individual genome sequence is
organisms (for example, ref. 48). Third, more phenotypically informative than
each gene is likely to contain multiple it is today49,50. In the meantime, great care
functional variants, including both ‘super- is required in offering genome-based
alleles’ of linked alterations on one haplo- information to individuals51,52.
type and allelic series with a range of allele
frequencies and effect sizes. Non-additive Concluding remarks
interactions can be present at all levels. What is the best future direction for
GWAS detect effects at the locus level, human genetics? There are essentially three
and an important challenge for future avenues to pursue: much larger samples;
studies is identification of the genes, the better assays of genome variation that can
functional variants and the functional capture both common alterations that
mechanisms underlying phenotypic asso- are not in LD with SNP panels and rare
ciations. Currently, such studies require variants; and more detailed phenotyping.
painstaking, low-throughput experiments Undoubtedly, each of these approaches
in cell lines and animal models. has a role, and we do not yet have all the
It is possible that some genetic contri- information needed to decide which will
butions to human phenotypic variation prove most fruitful. Therefore, it is a high
might be too subtle to unravel, even priority to apply a full battery of approaches
when our surveys of the genome become to several model diseases and phenotypes
truly comprehensive and the sample sizes in order to empirically determine the range
approach that of the human population. of outcomes, just as the Wellcome Trust
Aside from the question of how much of Case Control Consortium study of seven
the population variation we will ultimately diseases provided an empirical guide for
be able to explain, we also have to ask GWAS41. In my opinion, the most pressing
how we can piece together individual risk question is the contribution of rare variants,
from many small genetic contributions. both in the genes that harbour common
Will we ultimately be able to classify risk variants and in those that do not. This
individuals into meaningful groups with question is also the most difficult to address
regard to risk of specific common diseases comprehensively with today’s technologies,
or response to drugs, as envisioned in but it seems imperative that we prioritize
personalized medicine? Doubtless this studies to begin to get the answers.
will be (or already is) true in some cases, Leonid Kruglyak is at the Lewis-Sigler Institute for
but it is currently unknown how general Integrative Genomics and the Department of Ecology
such classifications are. We might need and Evolutionary Biology, Princeton University,
Princeton, New Jersey 08544, USA.
to replace some current phenotypic and
disease classifications with ones that bet- e-mail: leonid@genomics.princeton.edu
ter correspond to the underlying genetic doi:10.1038/nrg2316
causes, perhaps by developing methods Published online 19 February 2008
© 2008 Nature Publishing Group
Perspectives

1. Altshuler, D. & Daly, M. Guilt beyond a reasonable 31. Kruglyak, L. Genetic isolates: separate but equal? 48. Sinha, H., Nicholson, B. P., Steinmetz, L. M. &
doubt. Nature Genet. 39, 813–815 (2007). Proc. Natl Acad. Sci. USA 96, 1170–1172 (1999). McCusker, J. H. Complex genetic interactions in a
2. Bowcock, A. M. Genomics: guilt by association. 32. Carlson, C. S. et al. Additional SNPs and linkage- quantitative trait locus. PLoS Genet. 2, e13 (2006).
Nature 447, 645–646 (2007). disequilibrium analyses are necessary for whole- 49. Brenner, S. E. Common sense for our genomes.
3. Gibson, G. & Goldstein, D. B. Human genetics: the genome association studies in humans. Nature Genet. Nature 449, 783–784 (2007).
hidden text of genome-wide associations. Curr. Biol. 33, 518–521 (2003). 50. Levy, S. et al. The diploid genome sequence of an
17, R929–R932 (2007). 33. Gabriel, S. B. et al. The structure of haplotype blocks individual human. PLoS Biol. 5, e254 (2007).
4. Topol, E. J., Murray, S. S. & Frazer, K. A. The genomics in the human genome. Science 296, 2225–2229 51. Anonymous. Risky business. Nature Genet. 39, 1415
gold rush. JAMA 298, 218–221 (2007). (2002). (2007).
5. The International HapMap Consortium. The 34. Reich, D. E., Gabriel, S. B. & Altshuler, D. Quality and 52. McGuire, A. L., Cho, M. K., McGuire, S. E. & Caulfield, T.
International HapMap Project. Nature 426, 789–796 completeness of SNP databases. Nature Genet. 33, Medicine. The future of personal genomics. Science
(2003). 457–458 (2003). 317, 1687 (2007).
6. Botstein, D., White, D. L., Skolnick, M. & Davis, R. W. 35. Daly, M. J., Rioux, J. D., Schaffner, S. F., Hudson, T. J. & 53. Gusella, J. F. et al. A polymorphic DNA marker
Construction of a genetic linkage map in man using Lander, E. S. High-resolution haplotype structure in the genetically linked to Huntington’s disease. Nature
restriction fragment length polymorphisms. human genome. Nature Genet. 29, 229–232 (2001). 306, 234–238 (1983).
Am. J. Hum. Genet. 32, 314–331 (1980). 36. Hinds, D. A. et al. Whole-genome patterns of common 54. Wyman, A. R. & White, R. W. A highly polymorphic
7. Solomon, E. & Bodmer, W. F. Evolution of sickle DNA variation in three human populations. Science locus in human DNA. Proc. Natl. Acad. Sci. USA 77,
variant gene. Lancet 1, 923 (1979). 307, 1072–1079 (2005). 6754–6758 (1980).
8. Lander, E. S. & Botstein, D. Mapping complex genetic 37. The International HapMap Consortium. A haplotype 55. Weber, J. L. & May, P. E. Abundant class of human
traits in humans: new methods using a complete RFLP map of the human genome. Nature 437, 1299–1320 DNA polymorphisms which can be typed using the
linkage map. Cold Spring Harb. Symp. Quant. Biol. (2005). polymerase chain reaction. Am. J. Hum. Genet. 44,
51, 49–62 (1986). 38. Hirschhorn, J. N. & Daly, M. J. Genome-wide 388–396 (1989).
9. Kan, Y. W. & Dozy, A. M. Polymorphism of DNA association studies for common diseases and complex 56. Kruglyak, L. The use of a genetic map of biallelic
sequence adjacent to human beta-globin structural traits. Nature Rev. Genet. 6, 95–108 (2005). markers in linkage studies. Nature Genet. 17, 21–24
gene: relationship to sickle mutation. Proc. Natl Acad. 39. Sulem, P. et al. Genetic determinants of hair, eye and (1997).
Sci. USA 75, 5631–5635 (1978). skin pigmentation in Europeans. Nature Genet. (2007).
10. Bodmer, W. F. Human genetics: the molecular 40. Weedon, M. N. et al. A common variant of HMGA2 is Acknowledgements
challenge. Cold Spring Harb. Symp. Quant. Biol. associated with adult and childhood height in the I thank many colleagues over the years, the participants at
51, 1–13 (1986). general population. Nature Genet. 39, 1245–1250 the 2007 Banbury Center Meeting “From Statistics To Genes:
11. Donis-Keller, H. et al. A genetic linkage map of the (2007). Figuring Out the Molecular Basis of Complex Traits”,
human genome. Cell 51, 319–337 (1987). 41. Wellcome Trust Case Control Consortium. D. Altshuler for discussions, and D. Botstein, A. Chakravarti,
12. Weissenbach, J. et al. A second-generation linkage Genome-wide association study of 14,000 cases of B. Coller, D. Goldstein and L. Rosenberg for comments on the
map of the human genome. Nature 359, 794–801 seven common diseases and 3,000 shared controls. manuscript. I regret that space constraints prevented me
(1992). Nature 447, 661–678 (2007). from citing other important work in the field. Supported by a
13. Kerem, B. -S. et al. Identification of the cystic fibrosis 42. Albert, T. J. et al. Direct selection of human genomic MERIT award from the National Institutes of Health (R37
gene: genetic analysis. Science 245, 1073–1080 loci by microarray hybridization. Nature Methods 4, MH059520) and a James S. McDonnell Centennial Fellowship
(1989). 903–905 (2007). in Human Genetics.
14. Houwen, R. H. J. et al. Genome scanning by searching 43. Hodges, E. et al. Genome-wide in situ exon capture
for shared segments: mapping a gene for benign for selective resequencing. Nature Genet. (2007).
recurrent intrahepatic cholestasis. Nature Genet. 8, 44. Porreca, G. J. et al. Multiplex amplification of large DATABASES
380–386 (1994). sets of human exons. Nature Methods 4, 931–936 OMIM: http://www.ncbi.nlm.nih.gov/entrez/query.
15. Puffenberger, E. G. et al. Identity-by-descent and (2007). fcgi?db=OMIM
association mapping of a recessive gene for 45. Legendre, M., Pochet, N., Pak, T. & Verstrepen, K. J. benign recurrent intrahepatic cholestasis (BRIC) |
Hirschsprung disease on human chromosome 13q22. Sequence-based estimation of minisatellite and Hirschsprung disease
Hum. Mol. Genet. 3, 1217–1225 (1994). microsatellite repeat variability. Genome Res. 17,
16. Risch, N. Linkage strategies for genetically complex 1787–1796 (2007). FURTHER INFORMATION
traits. III. The effect of marker polymorphism on 46. Kruglyak, L. & Nickerson, D. A. Variation is the spice Leonid Kruglyak’s homepage: http://www.molbio2.
analysis of affected relative pairs. Am. J. Hum. Genet. of life. Nature Genet. 27, 234–236 (2001). princeton.edu/index.php?option=content&task=view&id=217
46, 242–253 (1990). 47. Estivill, X. & Armengol, L. Copy number variants and Perlegen Sciences: http://www.perlegen.com
17. Risch, N. Linkage strategies for genetically complex common disorders: filling the gaps and exploring The International HapMap project: http://www.hapmap.org
traits. II. The power of affected relative pairs. Am. J. complexity in genome-wide association studies. All links are active in the online pdf
Hum. Genet. 46, 229–241 (1990). PLoS Genet. 3, 1787–1799 (2007).
18. Risch, N. Linkage strategies for genetically complex
traits. I. Multilocus models. Am. J. Hum. Genet. 46,
222–228 (1990).
19. Kruglyak, L. & Lander, E. S. Complete multipoint
sib-pair analysis of qualitative and quantitative traits.
Am. J. Hum. Genet. 57, 439–454 (1995).
20. Risch, N. & Merikangas, K. The future of genetic
studies of complex human diseases. Science 273,
1516–1517 (1996).
21. Lander, E. S. The new genomics: global views of
biology. Science 274, 536–539 (1996).
22. Collins, F. S., Guyer, M. S. & Chakravarti, A. Variations
on a theme: cataloging human DNA sequence
variation. Science 278, 1580–1581 (1997).
23. Wang, D. G. et al. Large-scale identification, mapping
and genotyping of single-nucleotide polymorphisms in
the human genome. Science 280, 1077–1082 (1998).
24. The International SNP Map Working Group. A map of
human genome sequence variation containing 1
million single nucleotide polymorphisms. Nature 409,
928–933 (2001).
25. Kruglyak, L. Prospects for whole-genome linkage
disequilibrium mapping of common disease genes.
Nature Genet. 22, 139–144 (1999).
26. Collins, F. S. et al. New goals for the US human
genome project: 1998–2003. Science 282, 682–689
(1998).
27. Pe’er, I. et al. Biases and reconciliation in estimates
of linkage disequilibrium in the human genome.
Am. J. Hum. Genet. 78, 588–603 (2006).
28. Reich, D. E. et al. Human genome sequence variation
and the influence of gene history, mutation and
recombination. Nature Genet. 32, 135–142 (2002).
29. The International HapMap Consortium. A second
generation human haplotype map of over 3 million
SNPs. Nature 449, 851–861 (2007).
30. Lonjou, C., Collins, A. & Morton, N. E. Allelic
association between marker loci. Proc. Natl Acad. Sci.
USA 96, 1621–1626 (1999).

318 | april 2008 | volume 9 www.nature.com/reviews/genetics

© 2008 Nature Publishing Group
BIO 346, HERITABILITY AND ASSOCIATION SELF-QUIZ
1. You measure the stature (height) of adults and their offspring in each of two populations, and
generate the following plots. Does the heritability appear to be the same in the two
populations?

Population 1 Population 2

a) Heritability appears higher in population 1, perhaps because of more environmental variation

b) Heritability appears higher in population 2, perhaps because of less environmental variation
c) Heritability must the same in both populations, because heritability is a measure of genetic
contributions to a trait
d) Heritability appears the same in both populations, perhaps because genetic variance is the
same
e) It is not possible to estimate heritability from these data

Scientists examined 535,000 different single nucleotide polymorphisms (SNPs) to identify genetic
markers of early onset high blood pressure. You decide to look closely at a region of chromosome 15
that has previously been implicated in the disorder. You find the following data from individuals with
SNP rs1568727837.

Genotype at # healthy # people with

SNP people high blood
rs1568727837 pressure
T/T 7,000 700
T/G 1,500 150
G/G 500 150

2. What is probability that a random person from this population having high blood pressure?

3. What is the frequency of the T allele?

4. If you are T/T, what is the probability you will have high blood pressure?
5. If you two copies of the G allele, what is the probability you will have high blood pressure?

6. Which of the following is a true statement about the G allele? (Mark all that apply.)
a. Having two copies more than doubles your risk of HBP compared to having no copies
b. The G allele is definitely not associated with HBP
c. This G allele causes HBP
d. The SNP may be in linkage disequilibrium with a causal genetic variant
e. The G allele is a rare variant

7. Create a second table – use the answers from questions 2 and 3 to fill in your null expectations
assuming the locus is not related to high blood pressure. How do they compare?

Genotype at # healthy # people with

SNP people high blood
rs1568727837 pressure
T/T
T/G
G/G
NEWS FEATURE PERSONAL GENOMES
The case of the missing heritability
When scientists opened up the human genome, they expected to find the genetic components of
common traits and diseases. But they were nowhere to be seen. Brendan Maher shines a light on
six places where the missing loot could be stashed away.
f you want to predict
I how tall your children
might one day be, a
good bet would be to
look in the mirror, and at
your mate. Studies going
back almost a century have
estimated that height is 80–90% heritable. So
if 29 centimetres separate the tallest 5% of a
population from the shortest, then genetics
would account for as many as 27 of them1.
This year, three groups of researchers2–4
scoured the genomes of huge populations
(the largest study4 looked at more than 30,000
people) for genetic variants associated with the
height differences. More than 40 turned up.
But there was a problem: the variants had
tiny effects. Altogether, they accounted for
little more than 5% of height’s heritability —
just 6 centimetres by the calculations above.
18
Even though these genome-wide association
studies (GWAS) turned up dozens of variants,
they did “very little of the prediction that you
would do just by asking people how tall their
parents are”, says Joel Hirschhorn at the Broad
Institute in Cambridge, Massachusetts, who
led one of the studies3.
Height isn’t the only trait in which genes
have gone missing, nor is it the most impor-
tant. Studies looking at similarities between
identical and fraternal twins estimate herit-
ability at more than 90% for autism5 and more
than 80% for schizophrenia6. And genetics
makes a major contribution to disorders such
as obesity, diabetes and heart disease. GWAS,
one of the most celebrated techniques of the
past five years, promised to deliver many of
the genes involved (see ‘Where’s the reward?’,
page 20). And to some extent they have, iden-
tifying more than 400 genetic variants that
NATURE|Vol 456|6 November 2008
contribute to a variety of traits and common
diseases. But even when dozens of genes have
been linked to a trait, both the individual
and cumulative effects are disappointingly
small and nowhere near enough to explain
earlier estimates of heritability. “It is the big
ILLUSTRATIONS BY D. PARKINS
topic in the genetics of common disease right
now,” says Francis Collins, former head of the
National Human Genome Research Insti-
tute (NHGRI) in Bethesda, Maryland. The
unexpected results left researchers at a point
“where we all had to scratch our heads and
say, ‘Huh?’”, he says.
Although flummoxed by this missing herit-
ability, geneticists remain optimistic that they
can find more of it. “These are very early days,
and there are things that are doable in the next
year or two that may well explain another size-
able chunk of heritability,” says Hirschhorn. So
where might it be hiding?
NATURE|Vol 456|6 November 2008
Right under everyone’s noses
The inability to find some genes could be
explained by the limitations of GWAS. These
studies have identified numerous one-letter
variations in DNA called single nucleotide
polymorphisms (SNPs) that co-occur with a
disease or other trait in thousands of people.
But a given SNP represents a much bigger
block of genetic material. So, for example, if
two people share one of these variants at a
key location, both may be scored as having
the same version of any height-related gene
in that area, even though one person actu-
ally has a relatively rare mutation that has a
huge effect on height. The association study
might identify a variant responsible for the
height difference, says Teri Manolio, direc-
tor of the Office of Population Genomics at
the NHGRI, but averaging across hundreds
of people could give the appearance that its
effects are pretty wimpy. “It’s going to be
diluted,” she says.
Finding this type of missing heritability is
conceptually easy, because it involves closer
scrutiny of the genes already in hand. “Just
exploring, in a very dense way, genetic vari-
ation at the loci that have been discovered is
probably going to [explain] another incre-
ment of missing heritability,” Hirschhorn says.
PERSONAL GENOMES NEWS FEATURE
Researchers will need to sequence candidate
genes and their surrounding regions in thou-
sands of people if they are to unearth more
associations with the disease.
Helen Hobbs and Jonathan Cohen of the
University of Texas Southwestern Medical
Center in Dallas did this in an attempt to
capture all the variation in ANGPTL4, a gene
their studies had linked to cholesterol and
triglyceride concentrations. They sequenced
the gene in around 3,500 individuals from
the Dallas Heart Study and found that some
previously unknown variants had dramatic
effects on the concentration of these lipids
in the blood7. Mark McCarthy of Britain’s
Oxford Centre for Diabetes, Endocrinology
and Metabolism says that such studies could
reveal much of the missing heritability, but
not a lot of people have had the enthusiasm
to do them. This could change as the cost of
sequencing falls.
Out of sight
Other variants, for which GWAS haven’t
even begun to provide clues, will prove even
harder to find. In the past, conventional
genetic studies for inherited diseases such as
cystic fibrosis identified rare, mutated genes
that have a high penetrance, meaning that the
gene has an effect in almost everyone who
carries it. But it quickly became apparent that
high-penetrance variants would not under-
lie most common diseases because evolution
largely keeps them in check.
What powered the push into
genome-wide association was a
hypothesis that common diseases
would be caused by common,
low-penetrance variants when
enough of them showed up in the
same unlucky person. Now that
hypothesis is being questioned. “A
lot of people are recognizing that
screening for common varia-
tion has delivered less than
we had hoped,” says David
Goldstein, professor of
genetics at Duke University
in Durham, North Carolina.
But between those variants that
stick out like a sore thumb, and those
common enough to be dredged up by
the wide net of GWAS, there is a
potential middle ground of vari-
ants that are moderately pen-
etrant but are rare enough
that they are missed by the
net. There’s also the possi-
bility that there are many
more-frequent variants
that have such a low pen-
etrance that GWAS can’t statistically link them
to a disease.
These very-low-penetrance variants pose
some problems, says Leonid Kruglyak pro-
fessor of ecology and evolutionary biology at
Princeton University in New Jersey. “You’re
talking about thousands of variants that you
would have to invoke to get near 80% or 90%
heritability.” Taken to the extreme, practi-
cally every gene in the genome could have a
variant that affects height, for example. “You
don’t like to think about models like that,”
Kruglyak says.
If rare, moderately penetrant or common,
weakly penetrant variants are the culprits,
then bumping up the number of people in
existing association studies could help find
previously missed genetic associations. Peter
Visscher of the Queensland Institute of Medi-
cal Research in Brisbane, Australia, says that
a meta-analysis of height studies covering
roughly 100,000 people is in the works. Low-
ering the stringency with which an association
is made could drag up more, but confidence in
the hits would drop.
At some point it might make sense to stop
using SNPs, and start sequencing whole
genomes. Collins suggests that the NHGRI’s
1,000 genomes project, which aims to sequence
the genomes of at least 1,000 people from all
over the world, could go a long way towards
finding hidden heritability, and many more
genomes may become possible as the price of
sequencing falls.
Not everyone supports
an all-out sequencing
onslaught. Gold-
stein warns against
19
NEWS FEATURE PERSONAL GENOMES NATURE|Vol 456|6 November 2008

Where’s the reward?

There is more riding on the
case of missing heritability than
academic satisfaction. By finding
variants related to common
disease, genome-wide association
studies promised to deliver
meaningful medical information
and justify the US$3 billion spent
on the human genome and the
multimillion-dollar effort to map
human variation. “The reason for
spending so much money was
that the bulk of the heritability
would be discovered,” says Joseph
Nadeau, a geneticist at Case
Western Reserve University in
Cleveland, Ohio.
The ability to predict someone’s
height from their genes would
be a pretty trivial carnival trick,
but it represents a mastery
over the language of life that
could potentially spill into
most areas of medicine. Aside
from some surprises, though,
such as mutations in immune-
system genes being tied to an
eye disorder called age-related
macular degeneration, many of
the variants found
have only modest effects on
human characteristics. For now,
genetics rarely provides a clearer
predictive answer than a good
family history. And the path to
therapy is not straightforward,
says David Goldstein of Duke
University in Durham, North
Carolina. “This talk about
personalized risk profiles, using
genetics, for most common
diseases, and this talk about a
whole flood of new drug targets. I
think that that’s now pretty clearly
wishful thinking.”
Francis Collins, former head of
the National Human Genome
Research Institute in Bethesda,
Maryland, agrees that the picture
for disease prediction remains
bleak, but is still optimistic about
therapeutic intervention. Most
genetic variants found by genome-
wide
association
“contribute
a relatively modest risk, but
that in no way says the genes
aren’t important,” he says. “The
opportunity for therapy here is
breathtaking.”
Peter Visscher, a geneticist at the
Queensland Institute of Medical
Research in Brisbane, Australia,
agrees. “It would be easy to
knock [genome-wide association
studies] and say everything
was promised and nothing
was delivered. But in terms of
identifying genes and pathways for
disease, it’s been very successful.
I would feel it’s moved the field
forwards tremendously.”
Ultimately, the clinical value of
the variants that
genome-wide association
studies have turned up
may differ from disease
to disease. Still, some say
that the field is too fixated
on clinical application,
be it through prediction,
personalization or identifying
drug targets. Robert Nussbaum of
the University of California, San
Francisco, puts it bluntly: “Human
genetics research always assumes
too quickly that it has to be
translational. They’re doing basic
research.” B.M.

continuing to “turn the crank” without devising
a more rational approach, such as sequencing
the genomes of people who exhibit extreme
manifestations of diseases. “I’m not really sold
on doing the sequencing version of what we did
with [GWAS],” he says. “It’s a big enough, costly
enough job, that I think we want to think a little
bit harder about exactly who gets re-sequenced.”
In the architecture
Some researchers are now homing in on
copy-number variations (CNVs), stretches of
DNA tens or hundreds of base pairs long that
are deleted or duplicated between individu-
als. Variations in these features could begin
to explain missing heritability in disorders
such as schizophrenia and autism, for which
GWAS have turned up almost nothing. Two
recent studies looked at hundreds of CNVs in
normal people and in those with schizophre-
nia, and found strong associations between the
disease and several CNVs8,9. They commonly
arise de novo — in an individual without any
family history of the mutation.
These structural variants might account for
a lot of the genetic variability from person to
person and could account for some of those
rare ‘out-of-sight’ mutations with moderate
penetrance that GWAS can’t pick up. Many
20
CNVs go undetected because they don’t alter
SNP sequences. Duplicated regions can also
be difficult to sequence.
A standard technology for uncovering
CNVs is array comparative genomic hybridi-
zation, in which scientists examine how genetic
material from different individuals hybridizes
to a microarray. If certain spots on an array
pick up more or less DNA, it could indicate
that there’s a CNV. This and several other
techniques are being tested by a consortium
called the Copy Number Variation Project, run
out of the Wellcome Trust Sanger Institute in
Cambridge, UK. The consortium is dedicated
to characterizing as many CNVs as possible so
that associations can be made between them
and diseases. McCarthy says that the role hid-
den CNVs have in heritability “should play out
in the next six months to a year”. But Gold-
stein argues that current technologies will miss
many of the smaller CNVs, from 50 base pairs
down to repeats of just two bases. “All we’ll have
verification of is the big whopping CNVs that
are identifiable, and they clearly do not account
for much of the missing heritability.”
In underground networks
Most genes work together with close partners,
and it is possible that the effects of one on
heritability cannot be found without knowing
the effects of the others. This is an example of
epistasis, in which one gene masks the effect of
another, or where several genes work together.
Two genes may each add a centimetre to height
on their own, for example, but together they
could add five. GWAS don’t cope with epistasis
very well, and efforts to find these interactions
usually require good up-front guesses about
the interacting partners.
Joseph Nadeau, a geneticist at Case West-
ern Reserve University in Cleveland, Ohio,
says that ‘modifier’ genes act even in some
straightforward single-gene diseases. “That’s a
simple kind of epistasis,” he says. Cystic fibro-
sis, for example, is usually caused by muta-
tions in one gene, CFTR, yet can vary greatly
in symptoms and severity. The suspicion has
been that modifier genes are one cause of this
variability.
But despite the years of study, researchers
still struggle to pin down these genes. “People
haven’t modelled truly the effect of epistasis,”
says population geneticist Sarah Tishkoff at the
University of Pennsylvania in Philadelphia.
It’s no surprise that genetics is more com-
plicated than one gene, one phenotype, or
even several genes, one phenotype, but it’s
humbling to realize how much more complex
NATURE|Vol 456|6 November 2008
things are starting to look. In a now classic
study10, Kruglyak and his colleagues found
that expression of most yeast genes is control-
led by several variants, often more than five.
To fill in all the heritability blanks, research-
ers may need better and more varied models
of the entire network of genes and regulatory
sequences, and of how they act together to
produce a phenotype. At some point this
process starts to look more like systems biol-
ogy, and researchers are already applying
systems methods to humans and other organ-
isms (see page 26). “What we’re learning from
these studies is that we need to think about the
more complex of the complex models rather
than the more simple of the complex models,”
Kruglyak says.
The great beyond
What if heritability estimates were wrong
in the first place? Heritability of height was ini-
tially measured by taking the mean height
of parents and comparing that value
to the adult height of their off-
spring. As the average heights of
parents increase, researchers
found, so too does the aver-
age height of their children,
hence the calculated 80–90%
heritability.
Environment, especially factors such as
nutrients or toxins present during important
growth phases, can affect the mean height of
a population considerably — but researchers
have controlled for environment in estimates
of heritability by, for example, comparing
genetically identical twins raised together
with those raised apart. Most
researchers are confident
that the heritability esti-
mates are sound. “I don’t
think anyone’s going to
say that the heritability of
height is 10% and let envi-
ronment get you closer to
the answer,” Kruglyak says. “I
don’t think you can explain it
away.”
But there are lingering doubts
about how precisely environ-
ment has been accounted for
in heritability studies. Adverse
experiences in utero could lead
to lifelong health disparities,
according to David Barker
from the University of South-
ampton, UK, and yet a shared
womb is an aspect of the envi-
ronment that would not be
factored into such studies.
“Heritability estimates are basi-
PERSONAL GENOMES NEWS FEATURE
cally what clusters
in families, and
environment clus-
ters in families,” says
Manolio.
Epigenetics, changes
in gene expression that are
inherited but not caused by
changes in genetic sequence,
confuses things further. Feeding a
mouse a certain diet, for example,
can alter the coat colour not only in
its children, but also in its children’s
children11. Here, the expression of
a coat-colour gene is controlled by
a type of DNA modification called
methylation, but it’s not completely
clear how that methylation pattern is
‘remembered’ by the next generation.
The idea that grandma’s environment
could affect future generations is contro-
versial — and such effects would have been
included in the heritability normally attributed
to genes.
“This complicates everything,” says
Nadeau. “How do we sort out what great-
grandfather and great-grandmother were
exposed to when they were young and hav-
ing children?” Model organisms might help.
Nadeau has investigated testicular germ-
cell tumours in mice that are analogous to
a highly heritable cancer in humans. His
group found that the effects of one weak,
cancer-promoting gene, Dnd1Ter, are greatly
enhanced by several other gene variants,
and the boosted effects are passed on even if
the genes that cause them are not12.
“It’s presumably transmitting
its presence in some epige-
netic way,” says Nadeau.
The mechanisms by which
epigenetic inheritance
might work are still dis-
puted, though; marks
such as methylation that
direct gene expression dur-
ing someone’s life seem to be
wiped clean in a new embryo.
One possible explanation for
Nadeau’s observation, he says, is
that RNA is being inherited alongside
DNA through sperm or eggs.
Collins is not convinced that epi-
genetics will play a big part in missing
heritability in humans. “It just doesn’t
look likely outside of one or two exam-
ples to suggest that this is the case.”
Nadeau disagrees. “It’s hard to imag-
ine that every other organism works
one way and humans are the excep-
tion,” he says.
Lost in diagnosis
There is a nagging worry as
researchers hunt for heritability:
that common diseases might
not, in fact, be common.
Medicine tries hard to lump
together a complex collec-
tion of symptoms and call
it a disease. But if thou-
sands of rare genetic vari-
ants contribute to a single
disease, and the genetic
underpinnings can vary
radically for different
people, how common is
it? Are these, in fact, dif-
ferent diseases?
GWAS could actually
be proving so difficult
because researchers are
seeking shared susceptibility
genes in a group of people who
may share few, if any. And yet with-
out a more refined understanding of genet-
ics, it could be impossible to categorize them
any better. “It may be rare variants, common
disease. And that’s kind of scary to people
because it’s much, much harder to find those,”
says Tishkoff.
There could be scarier and more intractable
reasons for unaccounted-for heritability that
are not even being discussed. “It’s a possibility
that there’s something we just don’t fundamen-
tally understand,” Kruglyak says. “That it’s so
different from what we’re thinking about that
we’re not thinking about it yet.”
Still the mystery continues to draw its
sleuths, for Kruglyak as for many other basic-
research scientists. “You have this clear, tangi-
ble phenomenon in which children resemble
their parents,” he says. “Despite what students
get told in elementary-school science, we just
don’t know how that works.” ■
Brendan Maher is a Features editor for Nature.
1. Visscher, P. M. Nature Genet. 40, 489–490 (2008).
2. Weedon, M. N. et al. Nature Genet. 40, 575–583 (2008).
3. Lettre, G. et al. Nature Genet. 40, 584–591 (2008).
4. Gudbjartsson, D. F. et al. Nature Genet. 40, 609–615
(2008).
5. Sullivan, P. F. PLoS Med. 2, e212 (2005).
6. Freitag, C. M. Mol. Psychiatr. 12, 2–22 (2007).
7. Romeo, S. et al. Nature Genet. 39, 513–516 (2007).
8. Stefansson, H. et al. Nature 455, 232–237 (2008).
9. The International Schizophrenia Consortium Nature 455,
237–241 (2008).
10. Brem, R. B., Yvert, G., Clinton, R. & Kruglyak, L. Science 296,
752–755 (2002).
11. Waterland, R. A. & Jirtle, R. L. Mol. Cell. Biol. 23, 5293–
5300 (2003).
12. Lam, M. Y., Heaney, J. D., Youngren, K. K., Kawasoe, J. H. &
Nadeau, J. H. Hum. Mol. Genet. 16, 2233–2240 (2007).
See Editorial, page 1, and News Feature, page 26.
21
Weekly self-review:
1) Make a list of terms from this week’s lecture and reading. On another sheet or on note cards,
define them in your own words.

2) What are the most important concepts? Explain them to a friend.

3) What are the terms or concepts you don’t understand? Look them up. Fill details
into your lecture notes based on the reading. Ask a friend. Come to office hours!

4) How do these concepts relate to one another? On another sheet, make a concept map
to link this week’s terms and concepts
As you prepare for your exam, here are a few things to keep in mind.

1. Congratulations on doing the reading.

It will be useful. Now how do you make sense of all of it?

I occasionally ask question specifically to see that you’ve done the reading, but they tend not to be
things that are hard to remember if you have done them. Really, you should ask yourself what the most
important concepts are. If it’s an original research article, work through the figures to see if you
understand what they show. Look at your weekly worksheets and primary literature guides to organize
your studying. Fill in your terms and concepts. Which ones were covered in lecture as well? Those are
likely to be on the exam.

2. Will this be on the test?

Every instructor hates to hear this question, so please don’t ask it. My answer is always the same.
Everything in the reading and lecture is fair game. The lecture, however, is my best attempt to
synthesize what is most important about the material, so that stuff tends to be most represented. Don’t
try to memorize details that weren’t in lecture. The reading is here to deepen your understanding. The
smart thing is to do the reading in advance, or as you go. Then focus on your notes when it comes time
to study. If you skipped the reading or left it to the last minute, make peace with the fact that you won’t
get an A. If you study the notes and came to lecture, you might still get a B. (Do better next time.)

My test writing strategy is to take a few questions from each lecture, mixing simple recall of terms with
application of key concepts. I usually have some interesting scenario (often from the news) and write a
series of questions that ask you to interpret data, integrate and apply concepts to new examples. The
practice exam included here is a multiple choice exam from the most recent offering of the class.

3. The test will be challenging, but you will not have trouble finishing it, and there
will not be any big surprises.
Readings, notes, and in-class activities are intended to prepare you for the exam.
NAME (please print)_______________________________

EID____________________________________________

Signature_______________________________________

EXAM ONE
BIO 346, HUMAN BIOLOGY
PROFESSOR PHELPS, FALL 2018

The test below contains 40 multiple-choice questions on 7 pages. Answer the questions both on the test
and on the scantron. Don’t be afraid to ask for clarifications. You have 70 minutes to complete the
exam. Good luck!

Use terms below to answer the questions 1-5 below. Terms may be used more than once or not at all.
a. enhancer b. SNP c. histone modification d. locus e. transcription factor

1. A protein that binds a specific DNA sequence to influence transcription

2. A chemical tag on a nucleosome that helps define functions of different parts of the genome

3. Alternative alleles that differ at a single base

4. A regulatory sequence that is very far from the gene that it regulates

5. A DNA sequence that can be identified by acetylation of histone 3, lysine 27

6. Epigenetics refers to
a. Stable variation in the expression of a phenotype that does not depend on genetic variation
b. Short-lived and reversible changes in phenotype mediated by the activation or deactivation of
enzymes
c. Variation in phenotype driven entirely by genetic variation
d. All of the above
e. None of the above

7. Neuroscientists sometimes use the term epigenetics to describe the molecular processes that
underlie memory. A molecular biologist might argue this is not epigenetics at all. Which of the
following capture their different perspectives?
a. It is epigenetics, because a memory is an enduring property of an individual not specified by DNA
sequence
b. It is not epigenetics, because changes in gene expression are not passed across cell divisions
c. It is not epigenetics, because it does not involve chromatin function or gene expression
d. All of the above
e. A and B but not C
8. Which of the following mutations is incorrectly paired with its molecular origin?
a. DNA polymerase errors, copy number variants
b. Uneven crossing over, single nucleotide polymorphisms
c. DNA repair errors, indels
d. All of the above are correctly paired
e. None of the above are correctly paired

9. An enhancer sequence has no effect on gene A, which is just 10kb (10,000 bp) away from the
enhancer, but it does influence the expression of gene B, which is ~200kb away. This could be
because
a. Genes A and B are in different topologically associated domains
b. Genes A and B are separated by an “insulator”
c. This is not possible, because enhancers modulate their nearest gene
d. A and B but not C
e. None of the above

10. Which of the following accurately describes the relationship between DNA sequence variation and
phenotypic variation?
a. DNA variation can alter transcriptional regulation by changing DNA binding sites
b. Sequence variation can change how the genome interacts with the environment
c. Sequence variation can alter the structure of proteins critical to a phenotype
d. All of the above are accurate
e. A and B but not C

11. The addition of a methyl group to DNA occurs

a. at CpG sites
b. at inactive promoters
c. in active coding sequences
d. all of the above
e. A and B but not C

12. Which of following correctly pairs the histone modification with the DNA sequence it identifies?
a. A single methyl group on lysine 4 of histone 3 (H3K4me1), active promoter
b. A single acetyl group on lysine 27 of histone 3 (H3K27ac), inactive enhancer
c. A triple methyl group on lysine 4 of histone 3 (H3K4me3), inactive enhancer
d. A and B but not C
e. None of the above

13. Which of the following enzymes is correctly paired with its function?
a. DNMT, methylation of DNA
b. TET, demethylation of DNA
c. HDAC, histone deacetylation
d. All of the above are correct
e. None of the above is correct
14. Which of the following hormones are released by the adrenal gland?
a. Adrenaline
b. Cortisol
c. Aldosterone
d. All of the above
e. None of the above

15. Stress-related diseases include

a. Diabetes
b. Cardiovascular disease
c. Depression
d. All of the above
e. None of the above

16. What is the best predictor of the frequency of stress-related disease among developed countries?
a. The prevalence of poverty
b. The mean income
c. The disparity between richest and poorest within a society
d. All of the above are equal predictors
e. none of the above

17. High levels of parental care alters the adult stress reactivity of offspring by
a. Increasing the abundance of the glucocorticoid receptor in the hippocampus
b. Increasing negative feedback on the hypothalamic-pituitary-adrenal stress system
c. Reducing methylation of the glucocorticoid receptor promoter in the hippocampus
d. All of the above
e. none of the above

18. Which of the following make life events more stressful?

a. Very predictable stressors
b. Stressors over which you have no control
c. The presence of an outlet for frustration
d. The presence of social support
e. None of the above contribute to the stressfulness of an event

19. Based on studies in animal models, what childhood intervention would you expect to have the
greatest impact on SES-related disparities in cardiovascular disease?
a. Increased affectionate touch between children and caregivers
b. Training in conflict management
c. Providing children with more books and other cognitively enriching materials
d. All of the above would be equally effective
e. All of the above would be ineffective
20. 14. An animal’s sex can be determined by
a. Its genotype
b. The temperature of its environment
c. Its social surroundings
d. All of the above
e. None of the above

21. Which of the following describes the order of human sex determination?
i. Chromosomal ii. Hormonal iii. Morphological iv. Gonadal v. Behavioral
a. i, ii, iii, iv, v
b. ii, i, iii, v, iv
c. v, iv, iii, ii, i
d. I, iv, ii, iii, v
e. None of the above

22. Organization refers to ______, while activation refers to ______.

a. transient effects of circulating hormones, the lasting changes of hormones during development
b. lasting changes produced by hormones during development, transient effects of current hormones
c. gonadal sex, hormonal sex
d. hormonal sex, gonadal sex
e. None of the above

23. Congenital adrenal hyperplasia is characterized by

a. Excessive negative feedback on ACTH release by the pituitary
b. Loss of the enzyme that cleaves a hydroxyl group from 21-carbon steroids
c. An abnormal SRY gene
d. Excessive production of glucocorticoids by the adrenal
e. Loss of the androgen receptor

24. If you expose a pregnant mom to an intervention, how many generations do you need to examine to
conclude that there is a stable germline transmission across generations?
a. First generation (F1) offspring
b. Second generation offspring (F2)
c. Third generation offspring (F3)
d. Fourth generation offspring (F4)
e. Any of the above is sufficient

25. The idea that early developmental environments shape adult risk for disease is attributed to the
scientist
a. Waddington
b. Barker
c. Darwin
d. Fisher
e. None of the above
26. In studies of the Dutch hunger winter, researchers found that the differentially methylated region of
of insulin-like growth factor 2 (IGF-2 DMR) differed based on when in development people were
exposed to the famine. They found that the effects of famine on methylation were
a. Strongest at conception, despite a lack of differences in birthweight
b. Strongest if they occurred later in development, coincident with birthweight differences
c. Equally prevalent regardless of when in gestation the embryos were exposed
d. A and B but not C
e. None of the above
Scientists examined 1 million single nucleotide polymorphisms (SNPs) in 10,000 individuals to identify
genetic markers of congenital heart disease. You decide to look closely at a region of chromosome
11 that has previously been implicated in the disorder. You find the following data from individuals
with SNP rs116827837.

Genotype at SNP # healthy # people with

rs116827837 people heart
disease
A/A 8020 80
A/C 1781 19
C/C 99 1

27. What is the probability that a random person from this population will have congenital heart disease
a. 0.00 b. 0.01 c. 0.10 d. 0.50 e. 1.00

28. What is the frequency of the C allele?

a. 0.00 b. 0.01 c. 0.10 d. 0.50 e. 1.00

29. If you are C/C, what is the probability you will have congenital heart disease?

a. 0.00 b. 0.01 c. 0.10 d. 0.50 e. 1.00

30. Which of the following is a true statement about the SNP at rs116827837?
a. Having two copies of the C allele doubles your risk compared to having no copies
b. The SNP causes congenital heart disease
c. The SNP does not seem to be associated with congenital heart disease
d. The SNP is in linkage disequilibrium with a causal genetic variant
e. None of the above is true

Integrative questions -- Twin studies and other methods estimate that sexual orientation has a
heritability of approximately 40%. In a paper currently in review at Science, researchers examined
genetic contributions to sexual orientation in a sample of 480,000 volunteers. As part of the UK Biobank
program, survey respondents were asked whether they had ever had sex with someone of the same sex.
They looked for SNPs that predicted responses to this question in men, in women, or in both sexes
combined. Results of this genome-wide survey of SNPs are shown in the image below. The red dashed
line indicated the criterion for statistical significance after correcting for multiple statistical tests.
31. Based on the Manhattan plot above, how many SNPs independently predicted this measure of sexual
orientation?
a. 0
b. 1
c. 4
d. Too many to count in this graph
e. None of the above
32. Each peak in the Manhattan plot is accompanied by many other neighboring sites that exceed the
criterion for significance. This is likely to due to
a. The physical linkage between peak SNP and neighboring SNPs
b. The fact that the neighboring SNPs are likely to be inherited together
c. The most significant SNP is clearly causing the phenotype
d. A and B but not C
e. None of the above
33. The researchers constructed a polygenic score based on the results of the GWAS. Based on what
you know about GWAS and polygenic scores, how do you think the score would perform?
a. It would not predict orientation across the 480,000 people better than chance
b. The polygenic score could be reliably used to predict orientation
c. It would predict orientation significantly in the population, but not in individuals
d. A and B but not C
e. None of the above
34. Twin studies and other methods estimate that sexual orientation has heritability of approximately
40%. If the contributions of all SNPs are put together into a polygenic score, the resulting score explains
a very small percentage of heritable variation in orientation. Such disparities are commonly known as
a. Missing heritability
b. Non-additive genetic variation
c. Phenotypic variance
d. A and B but not C
e. None of the above
35. The fact that specific SNPs explain only a small fraction of heritable variation could be because
a. The trait is governed by very many loci, each of which contribute a tiny amount
b. There are some alleles that have a big effect, but they are rare and hard to detect
c. The SNPs examined are poorly linked to many of the causal variants
d. All of the above
e. None of the above

36. Example of rare variants that are known to have a large effect on sexual differentiation include
a. Mutations to 21-B-hydroxlase
b. Loss of androgen receptor function
c. Duplication of SRY
d. A and B but not C
e. None of the above

37. Examination of SNPs associated with sexual orientation reveals that many of the neighboring genes
are expressed in the hypothalamus. This is interesting because
a. It demonstrates that the SNPs themselves were causing differing sexual orientations
b. In one study, gay men were found to have a “feminized” nucleus in the hypothalamus
c. The hypothalamus is critical to the stress response
d. A and B but not C
e. None of the above

38. In rats, sex-typical sexual behavior is influenced by

a. Organizational influences of androgens
b. Activational influences of androgens
c. Activational effects of estrogen
d. All of the above
e. None of the above

39. One SNP the researchers was associated with sexual orientation in men was previously found to be
associated with male pattern baldness. Treatments for male pattern baldness include finasteride, an
inhibitor of 5a-reductase. This suggests that sexual orientation in men might be mediated in part by
effects of
a. The androgen receptor
b. The progesterone receptor
c. 21-b-hydroxylase
d. A and B but not C
e. None of the above

40. One researcher was interested in what the fitness consequences were for sexual orientation. The
team found that homosexual orientation was associated with significantly lower evolutionary fitness.
How could variation in this trait persist?
a. New mutations arise faster than selection can clear them
b. Mutations that are harmful in one sex might be beneficial in another
c. Orientation may be influenced by non-heritable factors, like cultural forces or hormonal exposure
d. All of the above are plausible
e. None of the above are plausible