11/2/20

Phage Display
In 1997, George Smith and his co-author wrote a perspective on the
developing area of Phage Display. In it they wrote -

“Imagine, then, the applied chemist, not as designer of molecules with

a particular purpose, but rather as custodian of a highly diverse
population of chemicals evolving in vitro as if they were organisms
subject to natural selection,”

“A chemicals ‘fitness’ in this artificial biosphere would be imposed by the custodian for his or her own ends. For
instance, the population might be culled periodically of individuals who fail to bind tightly to some biological receptor;
the population would then evolve toward specific ligands for that receptor... Progress toward the custodians chosen
goal would in a sense be ‘automatic’: once appropriate selection conditions are devised, no plan for how the system
is to meet the demands of selection need be specified. And if the chemical population is sufficiently diverse,
perhaps this ‘blind’ process will outperform rational design. The custodian may not comprehend, even in retro-
spect, how the products of selection work, just as biologists have only the sketchiest understanding of how a fruit fly
functions.”

Chem. Rev. 1997, 97, 391−410

Angew. Chem. Int. Ed. 2019, 58, 14428 – 14437
68

Phage Display
One-half of the 2018 Nobel Prize in Chemistry was awarded jointly
to George P. Smith and Sir Gregory P. Winter “for the phage
display of peptides and antibodies”.

Ø Bacterophages or phages are viruses that infect bacteria

Ø They are found everywhere in Nature but are harmless to humans

Ø They were used as early at the early 1900s to try to treat bacterial infections but antibiotics took over in the 1940s

https://www.reddit.com/r/natureismetal/comments/7vst0s/t4_bacteriophage_virus_as_seen_through_an/
•https://doi.org/10.1371/journal.pone.0173341

69

1
 11/2/20

Phage Display
Ø Can be produced in large quantities inexpensively and are easily freed from impurities as they are robust

Ø Virus particles are called virions

Ø Virions have an outer shell called a capsid, composed of a geometric array of phage coat proteins

Ø Inside the capsid is the phage chromosome, whose genes encode the phage proteins

Ø The capsid serves to protect the chromosome while also mediating infection; as the virion attaches to an uninfected
bacterium and the phage chromosome is Lectures
Nobel transferred into the cell’s cytoplasm
Angewandte
Chemie

field. Much of my phage work before[3, 4] and during the

Ø Once inside the cell, the phage genes reprogram the cell’s machinery to
sabbatical was focused on coat protein pIII. The 5 copies of

h=ps://www.reddit.com/r/natureismetal/comments/7vst0s/t4_bacteriophage_virus_as_seen_through_an/
•https://doi.org/10.1371/journal.pone.0173341
Figure 1. Electron micrograph of filamentous virions (left) and the
schematic representation that will be used in this essay (right). The
70 black links represent the phage chromosome; the only phage gene
represented, by the yellow segment, is gene III, encoding phage coat
protein pIII. Five copies of pIII cap one tip of the virion, with surface-
exposed portions that are visible as irregular knobs at the tip of one
particle in the electron micrograph (left-hand pointer), and that are
represented by yellow circles in the schematic (right-hand pointer).
ped of its protein capsid) into bacterial cells, a procedure
called transfection. Artificial introduction of the phage
chromosome into the cell via transfection initiates the same
infection program as does natural infection. Transfection is
make progeny virions, which are ultimately released into the cell and go on
this protein were known to include an N-terminal domain that
is exposed on the surface at one tip of the virion, and that is
to infect many other bacterial cells
sometimes visible in electron micrographs as irregular knobs
loosely connected to the bulk of the virion by very thin
(usually invisible) threads;[5, 6] the pointers in Figure 1 point to
such knobs in the micrograph and in the accompanying
schematic diagram.
It occurred to me that it might be possible to genetically
fuse all or part of a foreign protein to the exposed parts of pIII
without greatly impairing pIII s function in the phage
infection cycle. If so, the foreign amino acids would be
displayed at the tip of the virion, where they would be
accessible to macromolecules such as antibodies and recep-
tors. This was hardly a great leap of imagination. Genetically
fusing all or parts of two genes to make a recombinant gene Nobel Lectures
encoding a recombinant “fusion protein” was a commonplace
of recombinant DNA technology at the time (and today);
choosing a phage coat protein as one of the fusion partners
The cells indeed released virions, which turned out to be
did not seem to stray very far from a well-traveled path.
infective—the first validation of my plan. I will call the
Paul Modrich s lab down the hall from Webster s lab had
three resources I could use to pursue my project: a purified modified virions “test phages.”
protein called EcoRI, which he was studying at the time; the According to my scheme, a peptide fragment of the EcoRI
gene encoding the protein; and antibody against the protein.
protein—the amino acids encoded by the insert—should be
Figure 2 shows schematically how I used the EcoRI gene to
construct the first filamentous phage-display construct. First, exposed at one tip of the virion surface; that fragment is
a DNA fragment from the EcoRI gene was inserted into the symbolized by the blue circles in Figure 2. From now on in this
gene-III segment of naked phage DNA in vitro, resulting in essay, I will refer to the surface-exposed amino acids encoded
a modified phage DNA with a recombinant gene III encoding
by a foreign DNA insert generically as a “peptide,” regardless
an EcoRI-pIII fusion protein; foreign DNA fragments
inserted into a chromosome in this way are called inserts. of the number of amino acids, and whether or not they

Phage Display – key concepts

Nobel Lectures
extremely inefficient (transfectivity many orders of magni-
tude less than 1 percent), but that inefficiency can be
compensated by using huge numbers of input DNA mole-
cules. Transfection is a staple of recombinant DNA technol-
Second, the modified DNA was transfected into bacterial
cells, which were then cultured in the hope that they would
release modified virions bearing theAngewandte
fusion protein at one tip.
Chemie
constitute all or part of a natural protein. Figure 3 shows

ogy, which involves manipulating naked DNA molecules in

vitro before introducing them into cells. field. Much of my phage work before [3, 4]
and during the Figure 4. Results of test propos
sabbatical was focused on coat protein pIII. The 5 copies of
this protein were known to include an N-terminal domain that
3. Filamentous Phage Display is Born is exposed on the surface at one tip of the virion, and that is
sometimes visible in electron micrographs as irregular knobs if the antibody was pre-rea
From July 1983 to August 1984, I went on sabbatical to
loosely connected to the bulk of the virion by very thin no longer affected infectivi
Bob Webster s lab in the Department of Biochemistry at threads;[5, 6] the pointers in Figure 1 point to
(usually invisible) wildtype control phages.
Duke University. He was an accomplished investigator such knobs of in the micrograph and in the accompanying
filamentous phage biology, while I was a newcomer to the peptide encoded by the in
schematic diagram.
It occurred to me that it might be possible to genetically outside of the virion, where
fuse all or part of a foreign protein to the exposed parts of pIII
George P. Smith received a Bachelor of Arts
The results summarized
degree in Biology from Haverfordwithout
College greatly
in impairing pIII s function in the phage my first phage-display artic
infection
degree cycle. If so, the foreign amino acids would be
Pennsylvania in 1963, and a Ph.D. in the test phage “filamentou
Bacteriology and Immunologydisplayed
under Edgarat the tip of the virion, where they would be
Haber at Harvard University in accessible
1970. Afterto macromolecules such as antibodies and recep-
a postdoctoral fellowship at the University
tors. This wasof hardly a great leap of imagination. Genetically
Wisconsin with Oliver Smithies (2007 Nobel
fusing all or parts of two genes to make a recombinant gene
laureate in Physiology or Medicine), he 4. Replacing Screening
encoding
came to the Division of Biological Sciences
a recombinant “fusion protein” was a commonplace
of recombinant
at the University of Missouri, where he DNA technology at the time (and today);
taught for 40 years before choosing
retiring in a phage coat protein as one of the fusion partners
2015. When I submitted my
Figure 1. Electron micrograph of filamentous virions (left) and the
schematic representation that will be used in this essay (right). The did not seem to stray very far from a well-traveled path. had a specific application
Paul Modrich Figure
s lab2. down
Construction of from
the hall test phage. See text
Webster forhad
s lab details.
black links represent the phage chromosome; the only phage gene screening of small conven
represented, by the yellow segment, is gene III, encoding phage coat three resources I could use to pursue my project: a purified
Angew.ofChem.
protein pIII. Five copies Int. Ed.
pIII cap one2019, 58,the
tip of 14428 – 14437
virion, with surface- ⌫ 2019 Wiley-VCH
protein Verlag GmbH
called EcoRI, which& Co.
heKGaA, Weinheim at the time; the
was studying 14429 with easy affinity selectio
Ø One of the proteins that caps the end of the virion is called pIII coat
exposed portions that are visible as irregular knobs at the tip of one
particle in the electron micrograph (left-hand pointer), and that are
gene encoding the protein; and antibody against the protein. libraries. Explaining the d
Figure 2 shows schematically how I used the EcoRI gene to selection will serve to illum
protein
represented by yellow circles in the schematic (right-hand pointer).
construct the first filamentous phage-display construct. First,
advance that phage display
a DNA fragment from the EcoRI gene was inserted into the
ped of its protein capsid) into bacterial cells, a procedure gene-III segment of naked phage DNA in vitro, resulting in The signature convent
called transfection. Artificial introduction of the phage a modified phage DNA with a recombinant gene III encoding lgt11,[8] another phage th
Ø Is it possible to genetically fuse a foreign protein to the exposed parts of pIII
chromosome into the cell via transfection initiates the same an EcoRI-pIII fusion protein; foreign DNA fragments chromosome has an extra, n
infection program as does natural infection. Transfection is inserted into a chromosome in this way are called inserts.
DNA can be inserted. Th
extremely inefficient (transfectivity many orders of magni- Second, the modified DNA was transfected into bacterial Figure 3. Proposed demonstration that EcoRI peptide was displayed on
a genetic fusion of the ph
Ø Used a protein called EcoR1 and the corresponding antibody
tude less than 1 percent), but that inefficiency can be
compensated by using huge numbers of input DNA mole-
cells, which were then cultured in the hope that they would
release modified virions bearing the fusion protein at one tip.
surface of test phage. See text for details.
DNA insert. This recomb
cules. Transfection is a staple of recombinant DNA technol- infected cells: that is, the in
ogy, which involves manipulating naked DNA molecules in
vitro before introducing them into cells.
schematically how I used Modrich s anti-EcoRI antibody and protein encoded by the r
EcoRI protein to test this supposition. If the test phages were constructs like the test pha
mixed with a large excess of the anti-EcoRI antibody (purple vectors, but there is a cruc
3. Filamentous Phage Display is Born ovals), the antibody might well bind to the surface-exposed expressed in lgt11-infected
71 From July 1983 to August 1984, I went on sabbatical to
EcoRI peptide (blue circles), and thereby impair test-phage
infectivity; while the antibody should have no such effect on
but does not become part o
When a limited numbe
Bob Webster s lab in the Department of Biochemistry at
Duke University. He was an accomplished investigator of the infectivity of wildtype control phages. However, if the a vast excess of bacteria and
filamentous phage biology, while I was a newcomer to the
antibody was pre-treated with a large excess of the EcoRI nutrient agar in a Petri dish
protein (irregular blue shape) before being mixed with the has been incubated overn
George P. Smith received a Bachelor of Arts phages, the protein would occupy the antibody s binding sites called plating the virions
degree in Biology from Haverford College in
Pennsylvania in 1963, and a Ph.D. degree in (indentations in the purple ovals), thus blocking its ability to a synonym for Petri dish.
Bacteriology and Immunology under Edgar impair test-phage infectivity. Figure 4 graphs the results of circular zones of clearing in
Haber at Harvard University in 1970. After
a postdoctoral fellowship at the University of
Wisconsin with Oliver Smithies (2007 Nobel
this experiment. As anticipated in Figure 3, mixing the
antibody with the test phage caused a dramatic time-
2
growth. Each clear zone st
cell, which releases hundred
laureate in Physiology or Medicine), he
came to the Division of Biological Sciences
dependent decline in infectivity, while the antibody had no or so, the cell itself being
at the University of Missouri, where he effect on the infectivity of wildtype control phages. Moreover, virions immediately go on t
taught for 40 years before retiring in 2015.

Figure 2. Construction of test phage. See text for details. 14430 www.angewandte.org ⌫ 2019 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim A

Angew. Chem. Int. Ed. 2019, 58, 14428 – 14437 ⌫ 2019 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim 14429
 Nobel Lectures
essay, I will refer to the surface-exposed amino acids encoded
Chemie
by a foreign DNA insert generically as a “peptide,” regardless
The cells indeed released virions, which turned out to be of the number of amino acids, and whether or not they
infective—the first validation of my plan. I will call the constitute all or part of a natural protein. Figure 3 shows
modified virions “test phages.” 11/2/20
According to my scheme, a peptide fragment of the EcoRI
protein—the amino acids encoded by the insert—should be
exposed at one tip of the virion surface; that fragment is
Figure 4. Results of
symbolized by the blue circles in Figure 2. From now on in this
essay, I will refer to the surface-exposed amino acids encoded
by a foreign DNA insert generically as a “peptide,” regardless
of the number of amino acids, and whether or not they if the antibody w
Phage Display – key concepts
constitute all or part of a natural protein. Figure 3 shows no longer affecte
wildtype contro
peptide encoded
Figure 4. Results of test proposed in Figure 3. See text for details. outside of the vir
The results s
my first phage-di
if the antibody was pre-reacted with excess EcoRI protein, it the test phage “f
no longer affected infectivity of either the test phages or the
wildtype control phages. These results showed that the
peptide encoded by the insert was indeed exposed on the
outside of the virion, where it was accessible to the antibody. 4. Replacing S
The results summarized in this section were reported in
my first phage-display article, in which I called constructs like When I subm
the test phage “filamentous fusion phage.”[7] had a specific
screening of sm
with easy affini
4. Replacing Screening with Affinity Selection
libraries. Explain
When I submitted my first phage-display publication, I selection will ser
had a specific application in mind: to replace laborious advance that pha
screening of small conventional phage expression libraries The signatur
with easy affinity selection from enormous phage-display lgt11,[8] another
libraries. Explaining the difference between screening and
chromosome has
selection will serve to illuminate the essential technological
advance DNA can be in
Figure that phage display
3. Proposed represents.
demonstration that EcoRI peptide was displayed on
72 The signature conventional phage expression vector is a genetic fusion
surface of test phage. See text for details.
[8]
lgt11, another phage that infects E. coli bacteria. Its DNA insert. Th
chromosome has an extra, non-phage gene into which foreign infected cells: th
DNA can be inserted.
schematically how IThe usedresulting
Modrich recombinant
s anti-EcoRI geneantibody
is and protein encoded
Figure 3. Proposed demonstration that EcoRI peptide was displayed on
a genetic fusion of the phage s extra gene and the foreign
surface of test phage. See text for details. EcoRI protein to test this supposition. If the test phages were constructs like th
DNA insert. This recombinant gene is expressed in lgt11-
Phage Display
schematically how I used Modrich s anti-EcoRI antibody and
mixedcells:
infected withthat
ovals),encoded
protein
a large
is, theexcess
the antibody
infectedof cells
the anti-EcoRI
might well bind
by the recombinant
synthesize theantibody
gene. to
fusion
the surface-exposed
Phage-display
(purple vectors, but ther
expressed in lgt
EcoRI protein to test this supposition. If the test phages were EcoRI like
constructs peptide (blue
the test phage circles),
in Figure and
2 are thereby impair test-phage
also expression but does not bec
mixed withOne of excess
a large the proteins that capsantibody
of the anti-EcoRI the end(purple
of the virion is called
vectors, but there
infectivity; pIIIiscoat
whilea the protein.
crucial This pIII
difference:
antibody should theprotein
fusion
have no can
proteinsometimes be
such effect on When a limit
ovals), the seen
antibodyin an
mightelectron
well bind microscope image of theexpressed
to the surface-exposed virion asinirregular
the
knobs,
lgt11-infected
infectivity of wildtype
connected
cells accumulates
control
to inthe
phages.
virion
those cells
However,
by very thin
if the a vast excess of b
EcoRI peptidethreads.
(blue Itcircles),
occurredand to Smith,impair
thereby that ittest-phage
may be possible
but does tonot
genetically
become part fuse a foreign
of the progenyprotein
virions to the
they exposed parts of
release.
antibody was pre-treated with a large excess of the EcoRI nutrient agar in a
infectivity; pIII,
whilewithout greatly
the antibody impairing
should have no pIII’s
suchfunction.
effect on At thisWhentime,agenetically
limited number fusing proteins
of lgt11 virionstogether
are mixedwaswith not a new idea
or technique
the infectivity of wildtypewas. Thephages.
control idea was to display
However, if theforeign protein
a vast excess(irregular
amino acids blue
or peptides
of bacteria and spreadshape) thebefore
onevenly surface being
over the mixed
ofsurface
the of with
virion theto
(fused has been incuba
antibody was pIII). He chose
pre-treated a protein
with calledofEcoRI.
a large excess the EcoRI phages,
He hadnutrient
access the
to
agar protein
the
in would
genedish,
a Petri thatthey occupy
encoded the
the antibody
form plaques protein
after theand s binding
dish sites
the antibody called plating th
againstblue
protein (irregular the shape)
proteinbefore
(this was
beingsimply
mixed awith protein
the that
has (indentations
would recognize
been incubated in theand purple
overnight bindattoovals),
EcoRI).
37 thus
8C. This blocking
Firstly,
procedurethe DNAits
is ability to
fragment a synonym for P
phages, thefrom
proteinthe EcoRI
would gene
occupy thewas inserted
antibody intosites
s binding the gene-III
called
impair segment
plating the of
test-phage the
virions naked phage
(or phages),
infectivity. DNA
“plate”
Figure in
here
4 graphs vitro,
being
theresulting
results inof a circular zones of
(indentations in the purple
modified phageovals),
DNAthus withblocking its ability gene
a recombinant to aIIIthis
synonym
encoding foranPetri
experiment. dish.
As The
EcoRI-pIII plaques
fusion
anticipated areinvisible
protein. Foreign
Figureas small
3,DNA fragments
mixing the growth. Each cle
impair test-phage
inserted infectivity. Figure 4 graphs
into chromosomes arethe resultsasof inserts.
known circular zones of clearing
Secondly, in an opaque
DNAlawn oftransfected
thick bacterial
antibody withthethe modified
test phage wascaused a dramatic into bacterial
time- cell, which releas
this experiment. As anticipated
cells, which were thenincultured
Figure 3,somixing
that theythe would
growth. Each modified
release clear zonevirions
stems frombearinga single
the phage-infected
fusion protein athad
oneno tip.
antibody with dependent decline in infectivity, while the antibody or so, the cell its
This the
wastest phage
the first ‘testcaused
phage’ a and
dramatic time- proof
an excellent cell, which releasesexperiment.
of concept hundreds of progeny virions after an hour
dependent Whendecline the
in infectivity,
test phages whilewere
the antibody
mixed with had no oreffect
a large so, theon
excess
the
cell infectivity
ofitself being lysed
anti-EcoRI
of wildtype
in the process;
antibody
control phages.
(purple the progeny
ovals).
Moreover,
Binding of the
virions immediat
effect on the infectivity of wildtype control phages. Moreover, virions immediately go on to infect
antibody to the test phage had a significant effect on the infectivity of the test phage – they did not infect that hundreds of bacterial cells
bacteria very efficiently, because the key14430 proteinswww.angewandte.org
on the surface were blocked by the ⌫ antibody.
2019 Wiley-VCHOn theVerlag GmbH &ifCo. KGaA, Weinheim
country,
14430 www.angewandte.org ⌫ 2019 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim Angew. Chem. Int. Ed. 2019, 58, 14428 – 14437
the antibody was premixed with a large excess of EcoRI peptide, the antibody was no longer available to bind to
the EcoRI peptide that was fused to the pIII protein. In this case, infection was significantly more efficient.

73

3
 the infectious virion that encodes it that is the essential
advance of phage display. It is that linkage that permits a vast
increase in the number of phage clones that can be effectively
surveyed. It is not necessary to prepare clones as separate (if
very crowded) plaques on Petri dishes—a requirement that
severely limits the number that can be screened. A probe
immobilized on a solid support, such as the surface of a small
empty plastic Petri dish, can readily be exposed to, say, ten
trillion virions representing 100 billion clones. Even if only
a handful of those 100 billion starting clones display a target
fusion protein, a few rounds of affinity selection should suffice
to clone them. Moreover, each round of affinity selection is
far easier and less technically demanding than the screening
procedure outlined above.
affinity selection. Step 1: Selector is immobilized by attach-
ment to an immobilizing substrate, such as the surface of
a polystyrene Petri dish or microplate well. Stringency can be
adjusted by changing the density of immobilized selector on
the substrate surface, as explained later in this section. Step 2:
Input virions are added, so that virions whose displayed
peptides bind the receptor are captured on the substrate
11/2/20
surface. The input to the first round of affinity selection is the
starting unselected library; this input might consist of up to
10–100 trillion virions representing up to 100 billion clones.
The input virions for each subsequent round of selection are
the amplified output virions (Step 5 below) from the previous
round of selection. Step 3: The substrate surface is thoroughly

Affinity Selection Although the physical linkage explained in the previous

paragraph continues to be understood as the essence of phage
display, my vision of the technology s field of application now

‘Bio-panning’ seems very parochial indeed. Not until 1988 were my horizons
greatly expanded, as I will detail in Section 6.

5. Affinity Selection

As I returned to Missouri at the end of my sabbatical in

August 1984, I resolved to refocus my research program on
the phage-display idea. Three practical issues needed to be
addressed: development of an efficient phage-display
vector—one that would make it easy to insert foreign
DNAs into the coat protein gene without unduly impairing
phage function; development of an effective procedure for
affinity selection; and development of methods for creating
Like panning for gold!libraries, so that the potential of affinity
huge phage-display
selection could be fully exploited. The first two tasks were the
dissertation project taken on by Steve Parmley, a graduate
Often multiple rounds
student who of selection
came to myare lab necessary
at that time; due to low
his results were
selectivity and to ensurethree
published onlyyears
the later.
peptides
[10]
The with
fUSEthe highest
vector system he
affinity are subject to further
introduced analysis
continues to be used today, especially vector
fUSE5 (GeneBank Accesssion AF218364). The subject of this
section is the affinity selection process introduced in my 1985
Stringency (high-affinity binders over low-affinity)
article and greatly improved in Parmley s dissertation project.
is key and can be increased by reducing
Affinity selection was closely the density
modeled of selectors
on “panning,” [11]

on the immobilizing
a staplesubstrate. However,research
of cellular immunology stringency is Ionly
of which increased
was well
aware
following the initial at the of
round time (I didn tselection.
affinity think to acknowledge
Can you thisthink
debt inwhy?
any of my previous publications). In immunological panning, Figure 6. Schematic diagram of affinity selection. Steps will be detailed
it is cells rather than virions that are captured by immobilized in the text.
74
14432 www.angewandte.org ⌫ 2019 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim Angew. Chem. Int. Ed. 2019, 58, 14428 – 14437

Affinity Selection
’ The following text is taken directly from Angew. Chem. Int. Ed. 2019, 58, 14428 – 14437:

Step 1: Selector is immobilized by a=achment to an immobilizing substrate, such as the surface of a polystyrene Petri
dish or microplate well. Stringency can be adjusted by changing the density of immobilized selector on the substrate
surface, as explained later in this secVon. Step 2: Input virions are added, so that virions whose displayed pepVdes
bind the receptor are captured on the substrate surface. The input to the first round of affinity selecVon is the starVng
unselected library; this input might consist of up to 10–100 trillion virions represenVng up to 100 billion clones. The
input virions for each subsequent round of selecVon are the amplified output virions (Step 5 below) from the previous
round of selecVon. Step 3: The substrate surface is thoroughly washed to remove virions that are not captured by
immobilized selector. Step 4: Captured virions that remain a]er washing are released by weakening the bond between
the selector and the displayed pepVde ; alternaVvely (not shown ; see next paragraph), they may be released by
cleaving the segment of amino acids linking the displayed pepVde to the rest of the virion. The released virions are the
unamplified output of selecVon. Step 5: The unamplified output virions are “amplified” by infecVng fresh bacterial
cells and shaking the cells overnight in culture medium. Each virion in an unamplified output gives rise to millions or
billions of idenVcal progeny virions in the amplified output. Amplified output virions are the input virions for the next
round of affinity selecVon. Virions from either unamplified or amplified eluates can be cloned so that individual
displayed pepVdes can be sequenced (by sequencing the DNA inserts that encode them) and analyzed in countless
other ways.

75

4
 11/2/20

Phage Display in Drug Discovery and

Development
Next…….

76