Twinkles in

the Nucleus
Discovering a new type of subnuclear body taught me how
pursuing the unexpected can lead to new insights.
© ISTOCK.COM, COLDSNOWSTORM

BY ARCHA FOX
 I
have a clear memory of presenting my
initial results about a “failed” protein
at a lab meeting with my postdoctoral
advisor Angus Lamond at the University
of Dundee in Scotland and the rest of his
group. It was the summer of 2000, and
for the first few months of my postdoc
I had been fusing green fluorescent pro-
tein with novel proteins that had recently
been identified by mass spectrometry as
residing in the nucleolus. I engineered
HeLa cells to produce copious amounts
of these fusion proteins, and watched
where they went. Most migrated to the
nucleolus, as expected, but one protein
steadfastly refused. Instead, it formed
nuclear dots that were much smaller than
the large and obvious nucleoli.
I was really worried about the messi-
ness of this result, but also intrigued. To
my relief, instead of being disappointed
that the protein was not doing what we
had expected, Angus and my lab mates
encouraged me to explore it further. The
group had access to antibodies against
many cellular structures, so I quickly
established that these nuclear dots were
different from any known nuclear bodies.
But having generated much of my data
with overexpressed protein, it was critical
to make sure that the endogenous form
of the protein also localized to the same
nuclear dots, and that what I had seen
were not simply artifacts of my approach.
We created an antibody against the
protein and incubated it with HeLa cells.
It was an incredibly nerve-wracking
moment looking down the microscope
to see what the antibody had stained.
Thankfully, it worked. I was ecstatic
when I saw that it had picked out the
same small nuclear dots that I had
identified with GFP. In 2002, I pub-
lished a manuscript introducing the
scientific community to paraspeckles—
orbs of protein and nucleic acid 360
nanometers in diameter, squeezed next
to the more famous and larger struc-
tures called nuclear speckles.1
Since then, paraspeckles have become
an established part of cell biology; there
are more than 250 articles on them, and
they have already found their way into
34 T H E SC I EN TIST | the-scientist.com
some textbooks. We now know they are
membraneless organelles seeded by a
long noncoding RNA (lncRNA), formed
through a well-characterized physical
phenomenon known as liquid-liquid
phase separation, and composed of
numerous proteins and RNA molecules.
We also know that they can alter gene
regulation when cells get stressed, an
important mechanism for maintaining
cell homeostasis and one that appears to
be disrupted in many diseases.
In 2006, I left the UK to establish my
own paraspeckle lab in my home coun-
try of Australia. Reflecting on my post-
doc reminds me of just how far we have
come in such a short period of time. In
addition to all we have learned about
paraspeckle biology from in vitro work,
many studies have now established that
these structures appear in cells biopsied
from human patients and from healthy
mouse tissue samples. To have discov-
ered a new cellular structure and have
watched the birth of a research field
focused on understanding that struc-
ture is an honor and a privilege. Look-
ing ahead, I can see some big opportuni-
ties for paraspeckle biology, from using
them as a model to understand lncRNAs
and phase separation in the cell to devel-
oping therapeutics that modulate them
in different diseases.
Paraspeckles
Nucleus
Discovering paraspeckles
When I joined Lamond’s lab in late
1999, I initially chose a project within
my molecular biology expertise, clon-
ing complementary DNAs for a bunch
of proteins identified in the human
nucleolus. But as I dove further into
the problematic protein that wouldn’t
localize to this known nuclear struc-
ture, I was quickly pulled out of my
comfort zone. I had to learn new tech-
niques to figure out why the protein was
appearing in mass spectrometry analy-
ses of nucleoli, but not in the nucleoli of
my GFP-fusion–expressing cells.
I ended up adopting a technique
that involved laser-bleaching the flu-
orescence of the wayward GFP-fusion
protein. I then took many microscopic
images over time to track where the
bleached protein, which I named para-
speckle protein 1 (PSP1, subsequently
renamed PSPC1), went in the cell. This
method showed that it was travelling in
and out of nucleoli under steady-state
conditions, even though it was not sub-
stantially enriched within them. Mass
spec was sensitive enough to pick up
this trace of PSPC1 in the nucleoli that
we could not see under the microscope.
I later found that two other proteins—
originally termed P54nrb and PSF, now
called NONO and SFPQ, which are in the
Cytoplasm
GLOWING GREEN:
Twenty years ago,
GFP-labeled proteins in
HeLa cells revealed the
presence of the nuclear
bodies now known as
paraspeckles. (Human
breast cancer cell from
the MDA-MB-231 cell
line shown here; scale
bar is 5 μm.)
ARCHA FOX
same family as PSPC1—were also enriched
within paraspeckles.2 While PSPC1 was an
unstudied protein at the time I named it,
scientists knew that NONO and SFPQ
played roles in gene regulation and RNA
processing. It soon became obvious that
RNA was a major component of para-
speckles, and various RNAs were inte-
gral to their formation. Treating cells with
RNases made PSPC1, NONO, and SFPQ
disappear from paraspeckles, showing it
was RNA holding the proteins there. The
picture emerged that paraspeckles were
conglomerations of protein and RNA that
were formed and maintained by what we
now know as phase separation dynamics,
a physical process that made them part of
a group of cellular structures known as
membraneless organelles. (See “Cellular
Droplets,” The Scientist, December 2018.)
I often saw paraspeckles in clusters on
one side of the nucleus, and I could also see
that only some of the HeLa cells’ micronu-
clei—small structures that carry fragments
of chromosomes—had paraspeckles. Some
nuclei must contain paraspeckle-forming
DNA, while others did not, Angus and I
thought; a particular part of the genome,
maybe even a specific gene, might be caus-
ing paraspeckles to form. But the identity
of that DNA was a mystery.
The period following my initial dis-
covery in the early 2000s was a lonely
time in paraspeckles research. Besides
our group, there was only one other lab
interested in paraspeckles—that of David
Spector at Cold Spring Harbor Labora-
tory. A few years after our initial para-
speckle publication, independent work
from his group found that NONO-bound
RNA was retained in the nucleus, unable
to enter the cytoplasm, where it could be
translated into a protein.3 Then, in 2005,
I received an email from Gérard Pierron
at France’s National Center for Scientific
Research (CNRS) near Paris.
Pierron wondered if paraspeckles might
correspond to structures that his group had
first observed with electron microscopy in
the early 1990s. At the time, the researchers
called the structures inter-chromatin
granule associated zones. These were mor-
phologically distinct from the rest of the
nuclear material, or nucleoplasm, but did
not have a molecular marker to distinguish
them. I sent Pierron some antibody to
PSPC1, which he tested against HeLa cells
in his lab. Sure enough, he could see enrich-
ment in the zones identified by electron
microscopy, suggesting that these granules
were indeed paraspeckles.
So actually, the French group had been
the first to see and publish on what would
later become known as paraspeckles—
and suddenly I was not so alone.
A NEAT link
In 2007, after I left Angus’s lab to start
my own group at the University of West-
ern Australia, I got another email about
my paraspeckle work. Jeanne Lawrence,
an epigenetics researcher based at the
University of Massachusetts Medical With a better
School, requested some antibody to
PSPC1. I happily shared my reagents, understanding of
and then I did not think much more
about it, until a few months later when why cells make
Lawrence contacted me again to see if I
was interested in a collaboration. paraspeckles, we
Her group was working on a
lncRNA—broadly defined as any RNA could in theory
of more than 200 nucleotides that does
not appear to encode any protein—called therapeutically
NEAT1, and had used my PSPC1 anti-
body to show that the lncRNA colocal- modulate the
ized with paraspeckles. When the lab
knocked down NEAT1, paraspeckles structures.
could no longer form. I thought, here
was the paraspeckle-forming gene that
I had been searching for all these years.
Together, our labs, along with Andrew
Chess’s group at Icahn School of Medicine
at Mount Sinai in New York City, assem-
bled a narrative, published in 2009, of
how NEAT1 seeds paraspeckle formation.4
Around the same time, two other groups—
Spector’s Cold Spring Harbor lab5 and that
of Tetsuro Hirose at Hokkaido University
in Japan6—reported similar results.
We had some inkling that other groups
might be working on the same thing, but
it still came as a surprise. It was unset-
tling to suddenly find myself in a compet-
itive environment after working in rela-
tive obscurity, but the independent studies
greatly strengthened the case for NEAT1
1 2. 201 9 | T H E S C IE N T IST 3 5
 PARASPECKLE
FORMATION
These tiny subnuclear bodies typically measure 360 nanometers in diameter. They are
composed of a long noncoding RNA (lncRNA) molecule called NEAT1, which serves
as the seed. Proteins that bind to NEAT1 accumulate, self-associate, and recruit other
proteins, forming a mature paraspeckle.
360 nm
Because NEAT1
is so long—more
than 23,000
nucleotides—it is
possible to label
different parts of
the RNA and see 5’ end
them appear in NEAT1 IncRNA
different zones
of individual
as a major component of paraspeckles.
paraspeckles. The findings also linked paraspeckles to
the exciting world of lncRNAs at a time
when the notion was just emerging that
lncRNAs are functional, and not simply
transcriptional noise. The debate over
lncRNAs continues today. While there is a
general acceptance that tens of thousands
of lncRNAs are produced by the human
genome, how many of these are functional
is still controversial. NEAT1 has become
an important model lncRNA with a clear
cellular function: forming paraspeckles.
Because NEAT1 is so long—more
than 23,000 nucleotides—it is possible
to label different parts of the RNA and
see them appear in different zones of
individual paraspeckles. A paraspeckle
36 T H E SC I EN TIST | the-scientist.com
PARASPECKLE
PROTEINS (PSPs)
FUS
SFPQ
NONO
Other
PSPs
Core Shell
is roughly spherical, with a shell and
a core, as defined by the distinct pro-
tein and RNA composition of each of
these regions. Using gold labelling and
electron microscopy, Pierron’s group
in Paris, in collaboration with us and
others, showed that the 5' and 3' ends
of NEAT1 are found in the shell, while
the middle sequences of the RNA are
found in the core. 7 (See illustration on
this page.) Another collaborative group
that I was part of, led by Shinichi Nak-
agawa at Hokkaido University, later
confirmed this with super-resolution
© PEG GERRITY
imaging. 8 Hirose’s group, working in
concert with Nakagawa’s team, myself,
and many others, found—by providing
seed sequences for various paraspeckle-
PARASPECKLE FUNCTION
When a cell is stressed, various triggers can cause it to increase the production of the lncRNA NEAT1, leading to the
formation of more paraspeckles. These bodies can grow to up to 2 micrometers in length, changing from a spherical
shape to oblong and sometimes branched structures. They trap various proteins and mRNAs, hindering their function
and thereby affecting the cell’s continued response to stress.

HEALTHY NUCLEUS STRESSED NUCLEUS

NEAT1

As they accumulate proteins and

RNAs, paraspeckles can become
linked together, growing bigger,
oblong, and sometimes branched.

Messenger RNAs are

exported to cytoplasm
for translation. Greater abundance of
NEAT1 leads to more
paraspeckles.

Specific messenger RNAs

become trapped in para-
speckles and cannot
reach the cytoplasm for
translation.

Nuclear proteins
regulate gene
expression.

Paraspeckles act as
a sponge, soaking up
the proteins from the
Nucleolus nucleoplasm.
Paraspeckles trap gene-regulating
proteins, depleting them from
target sites on the genome
and thereby altering
transcription.
 To have discovered
associated proteins to bind—that differ-
a new cellular ent regions of NEAT1 were required for
directing this core-shell organization.9
structure and NEAT1 is now being finely dissected
to understand how this one RNA can
have watched the form a scaffold on which the tiny mem-
braneless organelles can be built.
birth of a research
Stress drives paraspeckles
field focused on Because NEAT1 is essential for para-
speckle formation, deleting NEAT1 in an
understanding animal makes a paraspeckle knockout. In
2011, Nakagawa established the NEAT1
that structure is knockout mouse.10 However, it showed
no obvious phenotype. This was disap-
an honor and a pointing to me, and made it hard to jus-
tify continuing to work on paraspeckles.
privilege. But as the Hokkaido-based team contin-
ued to scrutinize the mutants, it turned
out that there was a phenotype: some
female knockout mice had reduced fer-
tility.11 Nakagawa found that paraspeck-
les are abundant in the corpus luteum
that forms in the ovary and emits pro-
gesterone after the release of an ovum.
Loss of NEAT1 prevented the corpus
luteum from forming in some, but not
all, of the knockout females.
That some animals appeared to be
more reliant on paraspeckles than oth-
ers hinted at the possibility that environ-
mental factors are at play when it comes
to paraspeckle function. Sure enough,
my group and others have since found
that various stressors can trigger the for-
38 T H E SC I EN TIST | the-scientist.com
NUCLEAR SPOTS: A HeLa cell nucleus with
DNA stained blue and paraspeckle markers
in green and red. The large circles within the
nucleus are nucleoli. Scale bar is 5 μm.
mation of abundant, larger-than-normal
paraspeckles that sometimes take on an
oblong shape. Paraspeckles seem to be
part of the cell stress response.
The key is NEAT1 transcription:
more NEAT1 RNA means more para-
speckles. A 2018 genome-wide screen
conducted by Lingling Chen at the
Shanghai Institute of Biochemistry and
Cell Biology identified more than 100
factors that increase NEAT1 transcrip-
tion.12 These include molecular signals
of mitochondrial disturbance, a driver of
many diseases.
As paraspeckle numbers increase
within a cell due to stress, they seques-
ter paraspeckle-associated transcrip-
tion factors such as SFPQ—one of the
first two proteins I discovered to be a
component of paraspeckles, along with
PSPC1—and this changes expression
of the downstream target genes. Large
paraspeckles, which can grow to be up to
2 micrometers long, also sequester spe-
cific mRNAs, stopping them from being
ARCHA FOX
exported from the nucleus and trans-
lated in the cytoplasm. (See illustration
on page 37.) This paraspeckle-driven
 gene regulation is important in immune
responses to many types of infection,
and it helps cells when mitochondria
get stressed by regulating mRNAs that
encode key mitochondrial proteins.13
The story that has emerged is that
paraspeckles act as buffers when cells
are stressed, helping them maintain
homeostasis and avoid apoptosis. This
function is likely useful for long-lived
cells such as neurons, but also may play
a role in some types of tumor forma-
tion. The role of paraspeckles in cancer
is still an active area of study, however,
and thus far there are conflicting data
in different cancer types. While one big
study found that paraspeckles may be
oncogenic, 14 another paper found that
the structures can actually suppress can-
cer.15 Nevertheless, it appears that para-
speckles can be both good and bad when
it comes to disease.
Getting in phase with disorder
Even with all the excitement about
the NEAT1 RNA, I never lost focus on
the paraspeckle proteins that are also
needed for these little structures to
form. In 2012, Hirose and colleagues
published a landmark paper expand-
ing the paraspeckle proteome from the
three I originally found up to a total of
40 different protein types.16
My postdoc Sven Hennig and I
teamed up with Hirose to interrogate
the interactions among these mol-
ecules and found that some parts of
the proteins with no predicted struc-
ture, called low complexity domains,
were very important. In collaboration
with Charlie Bond, also at the Univer-
sity of Western Australia, we showed
that the low complexity domains of the
paraspeckle proteins formed hydrogels,
jelly-like globs that are neither liquid
nor solid, implicating these proteins in
the phenomenon of liquid-liquid phase
separation (LLPS). 17 LLPS is a hot new
area in cell biology, and it may eventu-
© PEG GERRITY
ally explain how cellular structures and
macromolecular complexes form with-
out membranes to compartmentalize
the cytoplasm and nucleus.
In the case of paraspeckles, after bind-
ing to NEAT1, proteins self-associate
and recruit other paraspeckle proteins
that glue RNA-protein particles into a
mature paraspeckle. 18 While the mate-
rial properties of paraspeckles are still
being worked out, I anticipate that our
wealth of knowledge about their com-
position and organization will make
them highly attractive as a model mem-
braneless organelle to help scientists
gain new ground in continuing research
on phase separation.
My hope for paraspeckle biology is
that we will one day have a complete
molecular model of a paraspeckle, down
to the atomic level, with a full suite of
molecular tools to break it down and
build it up. With a better understand-
ing of why cells make paraspeckles, espe-
cially when stressed, we could in theory
use these tools therapeutically to modu-
late the structures in different diseases.
Although we have learned a lot since
the discovery of paraspeckles nearly two
decades ago due to the hard work and
creativity of the members of my and
other research groups, there is undoubt-
edly much more to understand. I look for-
ward to being part of that adventure. g
Archa Fox is an associate professor and
Australian Research Council Future
Fellow at the University of Western Aus-
tralia and an affiliate with the Harry
Perkins Institute of Medical Research.
References
1. A.H. Fox et al., “Paraspeckles: A novel nuclear
domain,” Curr Biol, 12:13–25, 2002.
2. A.H. Fox et al., “P54nrb forms a heterodimer
with PSP1 that localizes to paraspeckles in
an RNA-dependent manner,” Mol Biol Cell,
16:5304–15, 2005.
3. K.V. Prasanth et al., “Regulating gene
expression through RNA nuclear retention,”
Cell, 123:249–63, 2005.
4. C.M. Clemson et al., “An architectural role
for a nuclear noncoding RNA: NEAT1 RNA
is essential for the structure of paraspeckles,”
Mol Cell, 33:717–26, 2009.
5. H. Sunwoo et al., “MEN epsilon/beta nuclear-
retained non-coding RNAs are up-regulated
upon muscle differentiation and are essential
components of paraspeckles,” Genome Res,
19:347–59,
2009.
6. Y.T.F. Sasaki et al.,
“MENepsilon/beta noncoding RNAs are
essential for structural integrity of nuclear
paraspeckles,” PNAS, 106:2525–30, 2009.
7. S. Souquere et al., “Highly ordered spatial
organization of the structural long noncoding
NEAT1 RNAs within paraspeckle nuclear
bodies,” Mol Biol Cell, 21:4020–27, 2010.
8. J.A. West et al., “Structural, super-resolution
microscopy analysis of paraspeckle nuclear body
organization,” J Cell Biol, 214:817–30, 2016.
9. T. Yamazaki et al., “Functional domains
of NEAT1 architectural lncRNA induce
paraspeckle assembly through phase
separation,” Mol Cell, 70:1038–53.E7, 2018.
10. S. Nakagawa et al., “Paraspeckles are
subpopulation-specific nuclear bodies that are not
essential in mice,” J Cell Biol, 193:31–39, 2011.
11. S. Nakagawa et al., “The lncRNA Neat1 is
required for corpus luteum formation and the
establishment of pregnancy in a subpopulation
of mice,” Development, 141:4618–27, 2014.
12. Y. Wang et al., “Genome-wide screening of
NEAT1 regulators reveals cross-regulation
between paraspeckles and mitochondria,” Nat
Cell Biol, 20:1145–58, 2018.
13. T. Hirose et al., “NEAT1 long noncoding
RNA regulates transcription via protein
sequestration within subnuclear bodies,” Mol
Biol Cell, 25:169–83, 2014.
14. C. Adriaens et al., “p53 induces formation
of NEAT1 lncRNA-containing paraspeckles
that modulate replication stress response and
chemosensitivity,” Nat Med, 22:861 -68, 2016.
15. S.S. Mello et al., “Neat1 is a p53-inducible
lincRNA essential for transformation
suppression,” Genes Dev, 31:1095–108, 2017.
16. T. Naganuma et al., “Alternative 3’-end
processing of long noncoding RNA initiates
construction of nuclear paraspeckles,” EMBO
J, 31:4020–34, 2012.
17. S. Hennig et al., “Prion-like domains in RNA
binding proteins are essential for building
subnuclear paraspeckles,” J Cell Biol, 210:529–
39, 2015.
18. M. Lee et al., “The structure of human SFPQ
reveals a coiled-coil mediated polymer essential for
functional aggregation in gene regulation,” Nucleic
Acids Res, 43:3826–40, 2015.
1 2. 2019 | T H E S C IE N T IST 3 9