Content uploaded by Gao Yifei
Author content
All content in this area was uploaded by Gao Yifei on Nov 22, 2022
Content may be subject to copyright.
Nature Chemical Biology
nature chemical biology
https://doi.org/10.1038/s41589-022-01204-2Perspective
A brief guideline for studies of
phase-separated biomolecular condensates
Yifei Gao1,2,3,5, Xi Li1,4,5, Pilong Li 1,2,3 & Yi Lin 1,2,4
Cells are exquisitely compartmentalized to achieve precise spatiotemporal
regulation of myriad processes and pathways. Phase separation oers
one way to achieve territorial organization in the cellular context, via the
creation of membrane-less organelles (MLOs). MLOs formed through
phase separation are associated with numerous critical biological functions.
Although hundreds of publications on related topics are produced each
year, robust criteria for the determination of biologically meaningful
phase separation are yet to be well established. Here we present some
principles and propose a few guidelines for phase-separation studies
in biology. Specically, we provide an in-depth experiment pipeline for
phase-separation studies, including mechanisms of the molecular
driving forces, ways to correlate in vivo and in vitro observations, and
strategies to relate the phase-separation phenomenon to biological
functions. We also intend to contribute to streamlining the aforementioned
diagnostic criteria by further stressing a few common caveats
in the eld.
Eukaryotic cells are extensively compartmentalized for the efficient
organization of materials, biochemical reactions and information
flow via various membrane-enclosed organelles as well as numer-
ous membrane-less organelles (MLOs)1. The latter, also referred to
as biomolecular condensates, fulfill various cellular functions. It has
recently been revealed that biomolecular condensates are formed in
part through a process called phase separation, predominantly liquid–
liquid phase separation (LLPS), in cells
1,2
. Phase-separated condensates
sometimes transition to solid-like forms (liquid–solid phase separa-
tion) in physiological or pathological scenarios3,4. Phase separation is
a well-understood physiochemical phenomenon in disciplines such as
polymer chemistry and soft-matter physics, and indicates the genera-
tion of multiple distinct phases out of homogeneous solutions. It is now
clear to biologists that phase separation leads to the concentration of
specific biological macromolecules and is a major driving force for the
formation of cellular MLOs.
With a growing appreciation of the importance of phase separation
in biology, there are many situations in which researchers may want
to experimentally study the phase-separation properties of a system.
A clear grasp of phase-separation properties will help biologists to
better understand the functions and mechanisms of action of MLOs.
Sophisticated and systematic experimental evidence, combining both
in vivo and in vitro results, is required. When accumulating the neces-
sary evidence, many principles should be noted and several caveats
should be taken into account. In this Perspective we briefly introduce
aspects that investigators should consider in the design of objective and
comprehensive experiments. A general scheme is shown in Fig. 1. Spe-
cifically, we have established a proper experimental design to validate a
phase-separation phenomenon or characterize a phase-separating bio-
macromolecular system, and we have explained the molecular features
of constituents within condensates. Moreover, we have emphasized that
studies of MLOs have to be closely related to their biological functions.
We also point out some caveats that are often encountered and provide
strategies to avoid them in future studies. Note that we focus mainly
on protein phase separation, as proteins are the major components of
biomolecular condensates. Other biological macromolecules, such as
DNA, RNA or even lipids and glycogen5, are extensively involved in this
process but are beyond the scope of this Perspective.
Received: 20 May 2022
Accepted: 10 October 2022
Published online: xx xx xxxx
Check for updates
1School of Life Sciences, Tsinghua University, Beijing, China. 2Tsinghua-Peking Joint Center for Life Sciences, Beijing, China. 3Beijing Advanced Innovation
Center for Structural Biology, Beijing Frontier Research Center for Biological Structure, Beijing, China. 4IDG/McGovern Institute for Brain Research at
Tsinghua University, Beijing, China. 5These authors contributed equally: Yifei Gao, Xi Li. e-mail: pilongli@mail.tsinghua.edu.cn; linyi@mail.tsinghua.edu.cn
Nature Chemical Biology
Perspective https://doi.org/10.1038/s41589-022-01204-2
determined by the threshold concentration for phase separation
and their absolute concentration within condensates. For clients,
inclusion is mainly determined by the effective affinities of direct or
indirect interactions with the scaffolds. As an example, Rab3 inter-
acting molecule (RIM) and RIM-binding proteins (BPs) are essential
for mediating the formation of condensates at presynaptic active
zones via LLPS. Simultaneous knockout of RIM and RIM-BP abolishes
the dense structure in the active zone
8
. Voltage-gated Ca
2+
channels,
which are unable to phase-separate on their own, can be enriched in
condensates by direct interaction with RIM and RIM-BP
9
. Owing to
Distinguish scaffolds and clients
Biomolecular condensates typically contain hundreds or more molec-
ular components, but only a small number of them are essential for
structural integrity
6
. These molecules are called scaffolds, form con-
densates through multivalent interactions, and are indispensable for
driving phase separation. Other components favorably partition into
condensates via direct or indirect association with scaffolds. These
components are referred to as clients7.
Both scaffolds and clients are enriched within condensates versus
the surroundings. For scaffolds, the degree of enrichment is largely
MLOs
Beginning of a phase-
separation-related project
Proteins of interest
Detailed imaging of
the cellular MLOs
Predict phase-
separation potential
Phase diagrams
in vitro and in vivo
Scaolds Clients
Driving force for phase separation
Dissect valency
dependence
Phase separation
in vivo
Do not phase
separate
YesNo
Connect phase separation to biological functions
Genetic manipulation of
key components in MLOs
Disruption of condensates
Rescue by grafting
Optogranule
Multivalency from
LCD sequences
Multivalency from
tandem structured
domains
Cooperation of
structured domains
and LCDs
1,6-HD Light
WT
LCD A
LCD BB
LCD B
AA
1
2
3
4
Components
Identify components of MLOs
Distinguish the scaolds
and clients of MLOs
In vitro reconstitution
Functions of MLOs
Constitutive
Signal-controlled
Stress-induced
Properties
Molecular diusion and exchange
(fusion, FRAP...)
Photobleaching
Fusion
Spatiotemporal distribution
Size and morphology
1
2
3
1
2
3
1
2
3
Mutant
A general guideline for studies of phase-separated biomolecular condensates
Fig. 1 | A general guideline for studies of phase-separated molecular
condensates. There are two major scenarios for the initiation of a
phase-separation-related project. The first is hypothesis-driven: researchers
perceive it as biologically meaningful if their proteins of interest (POIs) form
condensates in cells. The second is observation-based: an MLO is observed, often
during the imaging of fluorescently tagged proteins in cells. After studying the
functions of an MLO and testing its dynamic properties, we identify the major
components of the MLO and tentatively distinguish scaffolds from clients.
From this effort we obtain a list of potential scaffolds as our POIs. Subsequently,
both scenarios converge on a common step—to identify the multivalency of
the POIs. For POIs without LCDs, the multivalency may come from interactions
between the structural domains of several POIs. For POIs with LCDs, their
multivalency may come from interactions at the amino-acid level, which can
be largely clarified by sequence analysis. Next, we can purify POIs to perform
various phase-separation experiments in vitro. Phase diagram analysis is of
great value, and the dynamic properties of phase-separated condensates should
be measured using fusion assays and FRAP. Some associated biochemical
experiments are required, such as perturbation of the phase-separation process
using reagents and mutant proteins, assessing client recruitment and monitoring
other biochemical activities. In parallel, we should observe the endogenous
morphology and localization of MLOs in cells. Manipulation of condensates
by light or genetics can help to determine whether phase separation is critical
for a specific biological function. Rescue by grafting, that is, replacing the
phase-separating domains with ones that are evolutionarily distant but with
similar phase-separation capacity, is highly encouraged.
Nature Chemical Biology
Perspective https://doi.org/10.1038/s41589-022-01204-2
their distinct architectural roles and hence modes of regulation, it is
important to determine whether a component within a condensate is
a scaffold or a client.
It is worth noting that although clients may be dispensable for the
structural integrity of biomolecular condensates, some clients may
function in regulating the condensates. The client valency determines
client recruitment into condensates, and thus the droplet proper-
ties
10
. Accordingly, the function and contribution of clients cannot
be ignored.
Understand the molecular mechanisms of phase separation
While studying functional MLOs, a vital step is to understand the molec-
ular driving forces underlying phase separation. In this section we
demonstrate a few major driving forces for scaffolds to phase-separate
spontaneously (Fig. 2).
Multivalent tandem structured domains drive phase separation.
It has long been appreciated that proteins are modular, meaning that
evolutionarily conserved regions of proteins often autonomously fold
into discrete domains, and large proteins are often composed of mul-
tiple domains. Perplexingly, some proteins are composed of multiple
copies of the same type of domain. The topology of repeating domains
within one molecule or one complex is referred to as multivalency.
Often, multivalent domains are involved in protein–ligand interac-
tions, where the ligands can be protein domains, peptide motifs, RNA
motifs, DNA motifs, sugar residues within polysaccharides or lipid
moieties. The ligands themselves are frequently multivalent11. For
example, multivalent RNA-recognition motifs (RRMs) can each bind
multiple short and degenerate RNA motifs within a reasonably long
RNA oligo. Multivalent interactions between binding partners may
increase the overall mutual affinity via additivity. It has been estab-
lished that the additivity of multivalent interactions can achieve an
apparent affinity stronger than that of avidin and biotin12. However,
nature must have designed such multivalent interactions to conduct
more specific missions.
Indeed, multivalent interactions have been shown to lead to phase
separation and the formation of biomolecular condensates in test tubes
and in cells
13
. Theories of multivalency-driven phase separation are well
established in the field of polymer chemistry
14–16
. Mechanistically, the
multivalency of an individual molecule allows its simultaneous dynamic
engagement with multiple binding partners. Multivalency of the latter
further enables the formation of dynamic interacting networks, a pro-
cess sometimes coupled with phase separation (Fig. 2). Interestingly,
modular multivalent proteins often only undergo phase separation in
the presence of their multivalent binding partners. Examples include
N-WASP/Nck
13
, PTBP1/RNA oligos
17
, PSD95/SynGAP
18
, p62/polyUb
19
and SPOP/Daxx20.
Low-complexity domains are another major driving force for
phase separation. Another major driving force for biological phase
separation arises from certain less solidly structured regions of pro-
teins. These proteins or domains are termed ‘intrinsically disordered
regions’ based on their lack of known structure or ‘low-complexity
domains’ (LCDs) based on their biased amino-acid distribution, often
containing repeated segments. The two groups overlap substantially
21
.
Such sequences are quite prevalent, but have long been underrep-
resented among structured proteins in the Protein Data Bank
22
, so
investigations into their functions and molecular mechanisms used
to be quite limited. In recent years, many of these proteins were found
to have a strong tendency to undergo self-assembly, both in vitro and
in vivo
23,24
, and are thus considered a major contributor to LLPS or
liquid–solid phase separation. The phase separation of proteins with
LCDs is related to essential biological functions such as transcription
regulation25, cell division26 and differentiation27.
The mechanism by which LCDs trigger phase separation is con-
troversial. Because of their special sequence identity, it was proposed
that the overall amino-acid composition directs interactions and is
important in driving the assembly process. For example, positively
charged residues and aromatic/hydrophobic residues are prevalent
and typically cluster within LCDs. They form cation–π interactions, the
electrostatic interaction between a cation and a polarizable π-electron
cloud. However, another point of view suggests that there is a defined
structural basis underlying LCD-driven phase separation (Fig. 2).
A few pieces of evidence have been presented to support this theory.
(1) Computational assessments have indicated that a statistically sig-
nificant proportion of LCDs have a high propensity to form secondary
structures
22,28
. (2) Single amino-acid substitutions along the sequence
have different effects on the phase-separation capability, so there could
be structured core and peripheral disordered regions that perform
distinct roles in driving phase separation
3,29
. (3) The structure of LCDs,
although impractical to resolve in their in-cell state, can be solved in
their in vitro configuration. These LCD sequences can form cross-β
structures in vitro while adopting an amyloid fibril state. Mutations
that disrupt the cross-β structures can also inhibit phase separation
30,31
.
(4) Besides side-chain interactions, the main-chain hydrogen bonding
also contributes to labile cross-β interactions, promoting self-assembly
and phase separation of LCDs
32
. (5) Footprinting results show that
the in vitro-solved structure has great overlap and close correlation
with the in vivo state, thereby supporting the structure-driven phase
separation theory29,33.
Further investigations are needed to determine which of the
above theories is closest to reality. New techniques in structural
biology, with higher resolution and superior capability to capture
transient or dynamic states, will help us get to the real state more
confidently. Also, with the assistance of cellular imaging techniques,
such as cryo-electron tomography, the state of cellular proteins will
be described more precisely.
Interplay between tandem structured domains and LCDs. Although
multivalent structured domains and LCDs can each independently
drive phase separation, in many cases the two contribute together.
For example, numerous transcription and splicing factors, including
most heterogeneous nuclear ribonucleoproteins, harbor tandem
π π
π+
+
–
Multivalency from
oligomerization
Multivalency from
low-complexity sequences
Multivalency from the cooperation
of structural domains and LCDs
Multivalency from
tandem repeats
... ...
Fig. 2 | A schematic diagram showing the constituents of phase-separated
condensates. Phase separation is driven by multivalent interactions. The
multivalency can originate from tandem repeats or oligomerization of structural
domains; from charge–charge, dipole–dipole, cation–π or cross-β interactions
of low-complexity sequences; or from the cooperation of structural domains and
low-complexity sequences.
Nature Chemical Biology
Perspective https://doi.org/10.1038/s41589-022-01204-2
RRMs, along with LCDs
34
. The interplay between the tandem folded
domain and the low-complexity region can modify the dependence
of phase behavior35. There is also a strong connection between the
two in the cellular context. It has been demonstrated that scaffolding
proteins form a specific and robust meshwork through multivalent
interactions, supporting the structure of the postsynaptic density
and the presynaptic active zone
9,36
. Meanwhile, abundant proteins
with LCDs are also reported to phase-separate locally in synapses.
Synapsin 1, a synaptic vesicle protein, undergoes phase separation via
its C-terminal proline-rich LCD
37
. In addition, synapsin 1 interacts with
structured SH3-domain-containing scaffolding proteins and generates
valence-dependent droplets. This indicates that tandem structured
domains and low-complexity regions can cooperatively regulate the
clustering of synaptic vesicles at the presynapse. Mechanistically, the
homotypic or heterotypic interactions of LCDs are also multivalent
in nature, similar to the tandemly repeated structured domains. So,
both forces can be intermingled to form the final condensates (Fig. 2).
Predict phase-separation potential
Scientists have developed sophisticated computational tools to
predict the phase-separation potentials of specific proteins21. For
example, SEG partitions globular and non-globular regions and helps
to identify low-complexity regions within target proteins
38
. LARKS
(low-complexity aromatic-rich kinked segments) focuses on the poten-
tial of a sequence to form an amyloid-like structure39. PhaSePred built
machine learning models by integrating multimodal features to screen
self-assembling and partner-dependent phase-separation proteins
separately
40
. Investigators can also refer to databases that curate pro-
tein entries with the potential to undergo phase separation41,42.
As stated above, there are two major types of scaffold: proteins
containing LCDs and proteins harboring multivalent tandem struc-
tured domains. However, most sequence-based prediction tools
tend to recognize LCD-specific features. These algorithms largely
ignore phase separation of proteins with low LCD content, especially
proteins with multivalent folded domains. Although these tools and
resources can help researchers determine whether their systems of
interest potentially undergo phase separation, it is advisable to look
into phase separation after experimental observation of MLOs and/
or scientific reasoning, invoking a phase-separation mechanism that
enables advanced understanding of a biological phenomenon.
Carry out detailed phase-diagram analysis
A phase diagram is used to describe the physical state of a protein or
other molecules under a series of different conditions. The boundary
between the states represents a critical condition, such as compo-
nent concentrations, when two aqueous phases start to separate in a
switch-like manner. The critical concentration for a phase-separating
system varies with environmental properties such as temperature, ion
concentration, pH and pressure. Crowding reagents are used to mimic
cellular environments with macromolecular crowding, but there are
potential pitfalls that should be noted (Box 1). Furthermore, although
most phase diagrams depict the phase-separating behavior of isolated
molecules in vitro, it is very valuable to construct phase diagrams in vivo
by observing the condensates in the presence of a range of environ-
mental conditions (temperature, salt concentrations, pressure, pH
and so on)
43
. The nature of a specific molecule should be considered
as well as its cellular behavior, to obtain a full-scale characterization
from the phase diagram.
Dissect the valency dependence of phase separation
As we explained above, multivalency underlies phase separation.
Therefore, the dependence of phase separation on valency should be
well demonstrated. Decreasing or increasing the number of repeated
elements should substantially impact the phase diagram (Fig. 1). Iden-
tifying mutants that disrupt multivalency and quantifying how (much)
each mutant interferes with the condensation process can reveal basic
information about how and why the protein domains are organized in
the way they are.
There is a misunderstanding in the field that the presence of LCDs
is an essential condition for phase separation. In fact, more than 4,000
human proteins contain LCDs longer than 30 amino acids
44
. Among
them, only a fraction actually possess phase-separation potential.
Sequences with specific amino-acid compositions are more likely to
undergo phase separation than others. As an example, in heat-stressed
cells, the phase-separation property of Pub1 is determined by its folded
RRM domains, and its LCD acts as a modifier45. Therefore, while carrying
out phase-separation studies, it is critical to notice that LCDs are not
essential for phase separation.
Correlate cellular observations with in vitro reconstitution
A phase-separation-related project might start from observation of a
cellular condensate, or identification of a protein with a high potential
to undergo phase separation. However, to eventually arrive at the con-
clusion that phase separation plays a role in related biological functions,
both cellular observations and in vitro reconstitution experiments
should be performed (Fig. 1). The capture of cellular condensates
might suggest that the target molecule is related to phase separation
in some way, but in vitro reconstitution experiments reveal more about
the mechanism of how the condensate is formed. On the other hand,
in vitro results alone are insufficient, as they lack actual biological or
physiological relevance. Phase separation is a highly concentration and
condition-dependent process. Many in vitro experiments are carried
out at concentrations way beyond physiological ones. Therefore, the
actual physiological states should be considered when evaluating a
biologically relevant phase separation. Moreover, prediction software
is mainly sequence-based and does not take physiological concentra-
tions into consideration. Actual experiments must be done to verify
the physiological relevance. For in vivo experiments, numerous cellular
imaging experiments are performed in an overexpression setting.
Therefore, immunostaining of the protein itself or imaging of an endog-
enously tagged protein helps nail down phase-separation phenomena
at the physiological level. Also, the expression levels of endogenous
proteins are dynamically controlled and constantly changing. Conse-
quently, it is important to combine in vitro experiments with in vivo
observations to arrive at concrete conclusions.
Box 1
Potential artifacts due to
crowding reagents
Crowding reagents such as poly(ethylene glycol), dextran and Ficoll
are sometimes used to demonstrate that a system is capable of
undergoing LLPS in vitro. There have been extensive investigations
into the physiochemical principles of the eects of crowding
reagents on phase separation of macromolecules94. The underlying
rationale for using these reagents is to mimic the crowded
environment in cells. Although crowding reagents are highly
eective in promoting LLPS in general, the legitimacy of the in vivo
mimicking eect is less certain, as crowding reagents-induced
LLPS says little about their biological signiicance. The ield of
biomolecular condensates pursues physiologically relevant
phase separation. Accordingly, we should be cautious of rushing
the interpretation of phase-separation phenomena that can only
happen in the presence of crowding reagents.
Nature Chemical Biology
Perspective https://doi.org/10.1038/s41589-022-01204-2
Moreover, although some technologies can be used to study
condensates both in vitro and in cells, the same measurements of
the same components are sometimes inconsistent in these two set-
tings. For example, the recovery rate may differ in fluorescence recov-
ery after bleaching (FRAP) experiments on the same components
in vivo and in vitro. The compositions of intracellular condensates
are much more complex, making the biophysical properties distinct
from those in in vitro-reconstituted droplets. Some cellular conden-
sates are anisotropic in structure. Stress granules (SGs), for example,
have a rigid core structure surrounded by a liquid-like shell, but the
in vitro-reconstituted droplets do not have such subcompartmental-
ized structures46. Also, the in vitro environment in a test tube is quite
static, whereas intracellular molecules are constantly moving or traf-
ficking. The dynamic intracellular environment, with its constant flux
of molecules, undoubtedly modulates the local environment and the
properties of condensates therein.
Be cautious in interpreting imaging results
Optical imaging is often used for characterizing a phase-separating
system, but caution should be exercised to avoid false positives. Puncta
observed in an immunostaining experiment do not directly equate
to phase-separated condensates. First, not all antibodies are of high
enough quality and specificity to produce trustworthy immunofluores-
cence results. Also, many punctum-like objects observed in cells—such
as autophagosomes and lysosomes—are not membrane-less. If a novel
condensate is observed, we suggest first determining its relationship
with known cellular organelles, such as nuclear paraspeckles, cyto-
plasmic P-bodies, autophagosomes, lysosomes and so on. If possible,
electron microscopy should be employed to investigate whether the
condensate is membrane-bound or not. Moreover, diffraction-limited
fluorescence imaging can obscure the true underlying structure
47
.
Note that the critical concentration is not equal to the concentration
at which droplets are detectable. Phase separation occurs far below the
point when droplets are visible
48
. However, due to the resolution limit
of confocal microscopy, the smallest droplets that can be observed
are hundreds of nanometers or larger in diameter. The diffraction of
fluorescent probes themselves often leads to the inaccurate definition
of condensates, especially when they are small in size. The changes to
protein properties by the fusion of fluorescent tags should also be con-
sidered (Box 2). Combinatorial strategies are encouraged to facilitate
imaging results. A turbidity assay at 600 nm can be used to monitor
the kinetics of phase separation. A sedimentation assay can be used
to pellet the droplets and quantitatively examine phase-separating
proteins, along with the accompanying components49.
Analyze molecular diffusion and exchange
In 2009, P-granules in Caenorhabditis elegans were reported to have
liquid-like properties
2
. Many studies have subsequently demonstrated
that LLPS occurs in many pathways through the formation of MLOs
in vivo. Molecules diffuse in and out of MLOs and exchange with
the surrounding environment. Nevertheless, MLOs could be more
than liquid-like. MLOs possess a continuum of dynamicity, ranging
from liquid to solid, or even amyloid-like. Less dynamic biomolecu-
lar condensates have been described, such as centrosomes, nuclear
pores and functional amyloid-like bodies
50
. The pericentriolar mate-
rial (PCM) is a gel-like scaffold in the proximity of the centriole. The
in vitro-reconstituted SPD-5 condensate that mimics the mitotic PCM
is stable in solution and does not fuse or recover upon photobleach-
ing
51
. Velo1, the protein that maintains the structure of Balbiani bodies,
has amyloid-like properties and forms a stable matrix. Its amyloid-like
self-assembly can cluster mitochondria in vitro52. However, the prop-
erties of these stable condensates can alter, and the condensates may
either disassemble or enter liquid-like states. Likewise, liquid-to-solid
phase transition of proteins is not uncommon and could play key roles
in several different biological processes4,53.
Multiple techniques are available to detect the dynamics of con-
densates. The fusion of droplets is a strong indication of liquidity.
The fusion procedure should be monitored whenever possible, and
the fusion speed should be quantified. FRAP and fluorescence loss in
photobleaching are non-invasive techniques that are frequently used
to explore the dynamic properties of condensates, both in vivo and
in vitro54. Atomic force microscopy can be used to measure the elastici-
ties of condensates
55
. The motions of genetically encoded multimeric
nanoparticles are used to measure the viscosity of condensates56. More
new techniques that evaluate material properties need to be applied
in phase-separation studies.
Analyze the size and morphology of MLOs
Cellular phase-separated MLOs have a number of characteristic bio-
physical properties. A detailed description of these properties is
encouraged. Among them, the size and morphology of condensates,
though diverse, are important factors that help to identify and char-
acterize phase-separated systems.
The diameters of MLOs in cells generally range from a few hundred
nanometers to a few micrometers. The size of the condensates deter-
mines the specific surface area, which further affects the molecular
exchange rate and other dynamic properties of the MLOs. Surface ten-
sion largely determines size and morphology, and thus acts as a critical
regulatory factor of liquid-like condensates. For example, the recruit-
ment of MEG-3 and RNAs changes the surface tension and controls the
size distribution of guanyl-specific endoribonuclease granules (PGL
granules). In this regard, MEG-3 and RNAs act as interfacial stabilizers
of LLPS due to a Pickering effect53,57.
Perfect liquid droplets are expected to be spherical. However,
within the complex cellular environment, condensates can be shaped
by surrounding structures and regulatory molecules. For example,
the addition of target mRNA changes the morphology of TIS11B con-
densates from spherical to mesh-like
58
. Moreover, the shape of con-
densates can undergo time-dependent changes. For example, over
time, spherical droplets formed by several RNA-binding proteins go
Box 2
Fluorescent tags might alter
protein phase-separation
properties
Fluorescent tags are stable, sensitive and powerful tools in
phase-separation studies, both in vitro and in vivo. As well as
green luorescent protein and mCherry, other smaller genetically
encoded fusion protein tags can be used, such as the tetracysteine
tag that couples with the FlAsH/ReAsH luorophores. Their
advantages or disadvantages have been extensively discussed95.
It should be remembered that when luorescent tags are fused to
target proteins or RNAs, there is a risk that the tags might inluence
the overall molecular properties (such as solubility, structural
folding and proteolytic degradation) and the cellular location of the
tagged molecule. In phase-separation studies, in particular, protein
solubility inluences the critical concentration, so using a tag that
changes the solubility eventually alters the phase-separation
property. In some cases, the fusion proteins might misfold and
mislocalize in cells. Therefore, when using a tag, it is imperative to
carefully consider the inluence of the tag on the properties of the
target protein.
Nature Chemical Biology
Perspective https://doi.org/10.1038/s41589-022-01204-2
through morphological changes into fiber-like structures
17,59
. Research
is ongoing to uncover more factors that regulate the size and morphol-
ogy of MLOs.
Study the spatiotemporal distribution of condensates
MLOs must have precise geographical locations to carry out their bio-
logical functions (Fig. 3). The spatiotemporal distributions of MLOs
are affected by the location of their components, directed by their
interacting molecules, and limited by their tethered cell structures.
MLOs form when the scaffold proteins are above critical concentra-
tions, and the locations of these MLOs are largely decided by interact-
ing proteins. We can take the protein SMN (survival of motor neuron)
as an example. The binding of its Tudor domain to ligands carrying
dimethylarginine is sufficient for condensate formation. The location
of dimethylarginine determines whether the two nuclear MLOs—gems
and Cajal bodies—are separated or attached to one another60. Similarly,
the formation of cytoplasm-located PGL granules is driven by the inter-
action of PGL, LAF-1 and RNA, whereas their asymmetrical distribution
is regulated by the local concentration of MEG-357. Furthermore, the
crosstalk between MLOs and cellular-membrane systems is of profound
importance for determining the location of MLOs. Certain condensates
emerge proximal to other organelles or on membranes. For example,
phosphorylated nephrin, a transmembrane protein located on cell
membranes, induces phase separation at the cell periphery to form
domains that move through actomyosin contraction
61
. Moreover, MLOs
might travel along the cytoskeleton or tether to other organelles. In
highly polarized cells such as neurons, membrane-less RNP granules
tether to lysosomes during long-distance transport. In this cellular
process, annexin A11 (ANXA11) acts as a molecular adaptor between
SGs and lysosomes62.
Notably, MLOs are highly dynamic and might change their tempo-
ral distribution to accompany different biological events. For example,
during mitosis, the dual-specificity kinase DYRK3 acts as a central
dissolvase to disassemble several MLOs
61
. These MLOs reappear as
mitosis is completed. Such chronological assembly/disassembly of
MLOs could be monitored to dissect the roles of MLOs in the timing of
related biological activities. In summary, the spatiotemporal features
of MLOs are tightly connected with their functions. When evaluating
whether a protein might phase-separate in vivo, it is critical to compare
the location of the observed puncta to when and where the cellular
biological function is actually performed.
Identify the components of MLOs
Phase-separated cellular condensates are dynamic and lack phos-
pholipid membranes, making them hard to isolate with classic bio-
chemical strategies. However, to gain comprehensive insights into
how these condensates are formed and achieve their biological func-
tions, analyzing their compositions is obligatory. Methods to isolate
phase-separated condensates include centrifugation, proximal labe-
ling
63
, sorting
64
and so on. Large-scale genetic screens are also helpful
for understanding key regulators, which are usually also components
of granules
65
. Furthermore, in vitro reconstitution of phase-separated
granules is helpful for understanding their compositional specificity.
Commonly used approaches include liquid-droplet pelleting
66
and
hydrogel trapping
3
. As yet, none of the currently available strategies
perfectly reflect dynamic compositional changes, so new techniques
are needed.
Study biological functions of phase-separated MLOs
MLOs mediate many cellular pathways and function in numerous physi
-
ological and pathological circumstances, which have already been
extensively reviewed
67
. As the studies of MLOs have to reflect their
biological functions, here we provide a distinct categorization of them,
based on the different functions they perform.
Constitutive condensates perform housekeeping functions. Some
MLOs largely exist throughout the entire lifetime of cells. For example,
cytoplasmic P-bodies are responsible for translational repression and
mRNA degradation68, and the nucleolus facilitates ribosome biogenesis
and protein quality control. The nucleolus, the most prominent dense
structure observed in the eukaryotic nucleus, is a multiphase conden-
sate with a fibrillarin ‘core’ surrounded by a nucleophosmin-rich liquid
‘shell’. These components provide substrate specificity (for example,
the RNA-binding domains of fibrillarin) and recruit their partners to
form the two mutually incompatible liquid phases69. The constitutive
MLOs typically play very basic housekeeping roles.
MLOs respond to intracellular or external signals. Many MLOs
are dynamic phase-separated condensates that only form at the
right time and at the right location under the control of cellular sig-
nals. In these processes, localized signals trigger the recruitment of
scaffolds so that their local concentrations are elevated beyond the
critical concentrations required for phase separation. Many cellular
physiological responses depend on signal-induced regulation of a
phase-separation system. One well-characterized signal-controlled
phase-separation system is the Wnt/β-catenin signaling pathway,
Membrane-attached
condensates
Ligand-induced
receptor condensates
Cytoplasmic
condensates
Endomembrane
system-interacting
condensates
Cytoskeleton-
associated
condensates
Condensates
traicking on the
cytoskeleton
Perinuclear
condensates Nuclear
condensates
Chromatin-associated
condensates
Nucleoporin
condensates
Fig. 3 | Schematic overview of the spatial distribution of MLOs. In eukaryotic
cells, MLOs have specific geographical characteristics. They form subcellular
compartments in the nucleus and cytoplasm, and they can be associated
with the membrane or cytoskeletal systems, and so on. Nuclear-localized
condensates and chromatin condensates are shown in blue. Nucleoporin
condensates, located between the nucleus and the cytoplasm, are displayed in
pink. Perinuclear condensates, located around the nucleus, are shown in pink.
Cytoplasmic condensates are shown in beige. Condensates interacting with
the endomembrane system (the endoplasmic reticulum is illustrated here as an
example) are shown in yellow. Condensates associated with and trafficking on
the cytoskeleton are shown in green. Ligand-induced receptor condensates and
membrane-attached condensates, such as tight junctions, are presented
in purple.
Nature Chemical Biology
Perspective https://doi.org/10.1038/s41589-022-01204-2
a b
c d
Wnt signaling inactive Wnt signaling active
Wnt
LRP5/6
Frizzled
β-catenin
β-catenin
β-catenin
pβ-catenin
GSK3β
APC
Axin
Axin
Dishevelled
Degradosome
Signalosome
Enhancesome
Degradation
Cytosol
Plasma
membrane
Transcription
TFs
Nucleus
p62
p62 filaments
p62 body
Phase separation
Poly-Ub
Misfolded
proteins
Autophagy
Soluble TMF
SAM
H2O2
Reduction
Oxidation
TMF phase-
separated
condensate
Disulfide bond
AN
Presynapse
AZ
PSD
Cleft
Clustered
receptors
Cytoskeleton
Postsynapse
Fig. 4 | Signal-controlled dynamic condensates. a, Phase separation-
dependent signaling in the Wnt/β-catenin pathway. Left: in the absence of
Wnt ligands, the degradosome is formed through LLPS driven by multivalent
interactions between the scaffold proteins axin and APC (adenomatous
polyposis coli protein). The degradosome enhances the phosphorylation of
β-catenin and triggers its proteasomal degradation. Right: when Wnt binds
the transmembrane Frizzled receptor and the LRP5/6 coreceptor, Dishevelled
and axin together condense into the signalosome at the plasma membrane.
Unphosphorylated β-catenin is then released and translocates into the
nucleus, where it induces target gene expression. TFs, transcription factors.
b, LLPS in p62-dependent autophagic degradation. Misfolded proteins are
ubiquitinated and recognized by the polymerized form of p62 to generate a p62
body. The p62 body concentrates the cargo–receptor complex for autophagic
uptake. c, Phase separation at the synapse. At the presynaptic bouton, phase
separation of synapsin maintains the reserve pool of synaptic vesicles (SVs).
Similarly, the postsynaptic density (PSD) is organized mainly by LLPS of PSD95,
SynGAP and other scaffold proteins. The function of the PSD is to concentrate
neurotransmitter receptors and downstream signaling molecules. In addition,
the synaptic cleft also assembles distinct nanodomains. AZ, active zone; PSD,
postsynaptic density; SV, synaptic vesicle. d, Regulation of phase separation by
cysteine oxidation. In the tomato shoot apical meristem (SAM, boxed region),
TMF is specifically expressed at the boundary where H2O2 is locally enriched. The
cysteine residues of the TMF transcription factor can form intramolecular and
intermolecular disulfide bonds to concatenate TMF molecules and drive LLPS
upon developmental redox signaling. This oxidation-mediated TMF condensate
binds and sequesters the promoter of the ANANTHA gene (AN in the figure),
repressing its expression and delaying flower differentiation.
Nature Chemical Biology
Perspective https://doi.org/10.1038/s41589-022-01204-2
in which Wnt triggers the LLPS of Dishevelled into the signalosome to
induce translocation of β-catenin and the expression of downstream
genes70,71 (Fig. 4a). Other examples include the assembly of p62 bod-
ies in autophagy activation
19
(Fig. 4b) and downstream signaling of
the T-cell and B-cell receptors in immunology72, as well as the for-
mation of pre- and postsynaptic condensates in neuronal synaptic
sites9,18 (Fig. 4c).
Environmental cues can also change the phase-separation status.
One way is by altering the post-translational modification state of
condensate component proteins. For example, upon developmental
H2O2 oxidation, the cysteine residues of the terminating flower (TMF)
transcription factor can form disulfide bonds to increase multivalent
interactions between LCDs and between TMF and its target promoters.
The oxidation-mediated TMF condensate represses ANANTHA expres-
sion for precise regulation of flowering time73 (Fig. 4d).
Cells adapt to stress conditions through phase separation. Certain
cellular condensates act as an adaptative mechanism for respond-
ing to increased duration under unfavorable conditions. The SGs
quickly form and dissociate in response to cellular stresses such
as altered temperature, cellular redox, pH and osmotic status. For
example, polyadenylate binding protein (Pab1), a key component of
SGs, is a temperature-sensitive protein whose thermally responsive
self-assembly rate is 160-fold higher than typical biological processes,
making it a thermal sensor
74,75
(Fig. 5a). Similarly, upon cold shock,
cold-inducible RNA-binding protein (CIRBP) is upregulated, methyl-
ated and then translocated to the cytoplasmic SGs
76
(Fig. 5b). H
2
O
2
and
arsenite induce hyperoxidation. The methionine-rich LCDs of yeast
Pbp1 and TDP43 could be oxidized to inhibit self-association
33,77
. Melt-
ing the Pbp1 LLPS droplets activates TORC1 and mediates autophagy,
clearing the oxidatively damaged cellular components
78,79
(Fig. 5c).
ab
c d e
f
Soluble
CIRBP
mRNA
mRNA
Released
mRNA
CIRBP condensates
YAP condensate
Pab1 condensates
Pab1
Osmotic
stress
Chromatin
Enhancer
Stress
Heat
Cold Osmotic
Oxidation pH
Hypoxia
Pbp1
condensates
Soluble
Pbp1
Pbp1
ROS
TORC1
TORC1
Autophagy
Met-rich
Met-oxidized
Reduction
Oxidation
Glycolytic
enzymes
Glycolytic
metabolites
Glycolytic body (G body)
Other
proteins
Glucose
ATP
Chaperone
Snf1p –O2
Macromolecular
assemblies
Fluid
cytoplasm
Solid-like
cytoplasm
Entry into
dormancy
Fig. 5 | Schematic diagrams of condensates regulated in response to
environmental stresses. a,b, Temperature stress-regulated phase separation.
In a, heat-stress-triggered phase separation of Pab1. Under physiological thermal
conditions, Pab1 binds to mRNA. Upon heat stress, Pab1 proteins undergo phase
separation, and the bound mRNAs are released. In b, CIRBP localizes in the
nucleus. In response to cold stress, it translocates to the cytoplasm and acts as
an RNA chaperone to stabilize and protect mRNAs. c, Oxidative stress regulates
Pbp1 phase separation. The C-terminal methionine-rich LCD of Pbp1 undergoes
phase separation. Intracellular reactive oxygen species (ROS) oxidize methionine
to methionine sulfoxide and impair the phase separation of Pbp1. TORC1 proteins
sequestrated in Pbp1 condensates are then released, which triggers TORC1-
mediated autophagy to clear the oxidatively damaged cellular components.
d, Formation of the hypoxia-induced glycolytic body. During hypoxic stress,
glycolysis enzymes are concentrated into glycolytic bodies to enhance the rate
of glucose consumption. e, pH-triggered phase transition in the cytoplasm. In
budding yeast, when cells are deprived of energy during dormancy, the lower
intracellular pH results in widespread packing of proteins into macromolecular
assemblies. The cytoplasm transits from a fluid state to a protective solid state.
f, Osmotic stress-induced condensation of YAP. YAP is a transcriptional co-
activator that normally binds to enhancers of target genes in the nucleus. Upon
hyperosmotic shock, YAP forms condensates at super-enhancer regions and
sequestrates the transcription factors TEAD1, TAZ and other co-activators to
activate downstream gene expression.
Nature Chemical Biology
Perspective https://doi.org/10.1038/s41589-022-01204-2
On the other hand, acute hypoxia causes redox imbalance and the con-
densation of glycolytic enzymes that switches the metabolic pathway
to efficiently produce energy via glycolysis
80
(Fig. 5d). In budding yeast,
a drop in cytosolic pH promotes an overall transition of the cytoplasm
from a liquid to a glass-like state. This shifts the cell into a dormant state
of cell-cycle arrest and reduced metabolic activity81 (Fig. 5e). Also, an
acute change of the extracellular ion/salt concentration causes osmotic
stress, which disrupts the intracellular electrostatic balance and leads
to cell swelling or shrinkage. For example, RAD23B can undergo phase
separation with ubiquitin chains in vitro via a multivalent interac-
tion, and can form hyperosmotic stress-responsive puncta in cells
82
.
Moreover, the osmotic stress-responsive multivalent interactions of
LLPS-driving proteins
83
, including the mechanosensing transcription
factors YAP/TAZ
84
and the apoptosis signal-regulating kinase ASK3
(ref. 85), can compensate for cell-volume changes86 (Fig. 5f). Notably,
although many environmental stresses can lead to phase separation
and the formation of cellular condensates, these SGs may be drastically
different in their components, properties and so on.
Connect phase separation to biological functions
How do we prove that phase separation is sufficient and necessary
for fulfilling certain biological functions? This is a frequently asked
question, and indeed it is a critical one for the whole field. To make
such a connection for specific proteins, mutants are employed that
destroy the phase-separation ability of a component while theoreti-
cally maintaining its other native properties. Small molecules, such
as 1,6-hexanediol, are frequently used to melt phase-separated con-
densates, but related experiments have to be designed carefully
(Box 3). Disturbance of the original biological function may thus origi-
nate from a phase-separation defect. A molecular grafting experiment
can demonstrate this point. In other words, to test whether a critical
function of a protein is due to its phase-separation capability, we can
replace its phase-separation domain with an orthogonal one, but with
similar phase-separation properties, and see whether the function
is maintained
87
. Sequences from different species are often used to
avoid local interactions88,89. For example, the prion-like domain in FCA
(flowering control locus A) from Arabidopsis thaliana can be used to
replace potential phase-separating sequences in mammalian studies
90
.
Besides grafting, an alternate approach is to use a light-inducible con-
densate system, termed ‘Optogranule’. Target proteins are fused with
light-sensitive proteins, such as Cry2, iLID and SspB, the oligomeriza-
tion of which can be induced by light at a specific wavelength. Such
engineered condensates can be used to control phase separation and
evaluate their performance in specified biological functions91,92.
Concluding remarks
In this Perspective we have aimed to set up a few rules to help biologists
with the understanding and subsequent study of phase separation.
However, there are far more mysteries behind phase separation and
the formation of MLOs. ‘Phase-separated condensates’ and ‘aggrega-
tion’ are a perplexing pair of concepts. Molecules in phase-separated
condensates are in a more homogeneous and reversible state, whereas
aggregates represent a more amorphous, heterogeneous and less
reversible state. In addition, phase separation is not the only mechanism
that drives the formation of MLOs, and the role of phase separation in
relation to various biological functions remains highly controversial.
For example, the role that LLPS plays in transcription activation is under
extensive debate47,93 and warrants further investigation.
In summary, we propose that phase separation represents a large
group of phenomena for clustering of macromolecules in cells, but
is definitely not universal or exclusive. More mechanistic investiga-
tions are urgently required to explain experimental results and pre-
cisely define concepts in this field. Moreover, future studies of phase
separation should proceed from cells to the tissue or organism level.
Understanding the role of phase separation in higher and more com-
plex physiological functions, including in vivo studies in multicellular
organisms, is a challenging but urgent future direction. A more pano-
ramic view might be required to develop cures for diseases originating
from abnormal phase separation, such as neurodegenerative diseases,
cancer and so on.
References
1. Banani, S. F., Lee, H. O., Hyman, A. A. & Rosen, M. K. Biomolecular
condensates: organizers of cellular biochemistry. Nat. Rev. Mol.
Cell Biol. 18, 285–298 (2017).
This is a comprehensive review of phase-separated cellular
membrane-less organelles.
2. Brangwynne, C. P. et al. Germline P granules are liquid droplets
that localize by controlled dissolution/condensation. Science
324, 1729–1732 (2009).
This paper describes that P-granules exhibit liquid-like
behaviors.
3. Kato, M. et al. Cell-free formation of RNA granules: low
complexity sequence domains form dynamic ibers within
hydrogels. Cell 149, 753–767 (2012).
This paper reconstituted RNA granules in vitro and revealed the
role of LCDs in driving liquid-to-solid phase separation.
4. Bose, M., Lampe, M., Mahamid, J. & Ephrussi, A. Liquid-to-solid
phase transition of oskar ribonucleoprotein granules is essential
for their function in Drosophila embryonic development. Cell 185,
1308–1324 (2022).
5. Liu, Q. et al. Glycogen accumulation and phase separation drives
liver tumor initiation. Cell 184, 5559–5576 (2021).
6. Fong, K.-w et al. Whole-genome screening identiies proteins
localized to distinct nuclear bodies. J. Cell Biol. 203, 149–164
(2013).
Box 3
Experiments using
1,6-hexanediol must be
designed carefully
1,6-Hexanediol (1,6-HD), an aliphatic alcohol, is widely used to
study the physical properties of membrane-less organelles.
1,6-HD was originally noticed for its ability to melt the nuclear
pore and later RNA granules96,97. Since then, 1,6-HD has been
widely used as an inhibitor and indicator of phase separation, even
though the mechanism by which it disrupts condensates is not
thoroughly understood. A few aliphatic alcohols, including 1,6-HD,
2,5-hexanediol and 1,5-pentanediol, were tested for their ability to
melt MLOs, LCD ibers and liquid-like droplets. 1,6-HD turned out to
be the most potent one. It is suggested that 1,6-HD interferes with
the weak hydrophobic protein–protein or protein–RNA interactions
that are required for these condensates to form98.
Here we raise a signiicant concern about the limitations and
cellular eects of 1,6-HD when it is applied to phase-separation
studies. Improper treatment with 1,6-HD may trigger apoptosis or
non-apoptotic cell death99,100. We recommend suicient controls
and careful usage of 1,6-HD in cellular phase-separation research.
The cell state should be closely monitored during the experiments.
Other aliphatic alcohols, such as 2,5-hexanediol and/or
1,5-pentanediol, should be included as controls, to avoid cell death
due to an overly high concentration or extra long-term treatment.
Nature Chemical Biology
Perspective https://doi.org/10.1038/s41589-022-01204-2
7. Ditlev, J. A., Case, L. B. & Rosen, M. K. Who’s in and who’s out—
compositional control of biomolecular condensates. J. Mol. Biol.
430, 4666–4684 (2018).
8. Acuna, C., Liu, X. & Südhof, T. C. How to make an active zone:
unexpected universal functional redundancy between RIMs and
RIM-BPs. Neuron 91, 792–807 (2016).
9. Wu, X. et al. RIM and RIM-BP form presynaptic active-zone-like
condensates via phase separation. Mol. Cell 73, 971–984 (2019).
10. Banani, S. F. et al. Compositional control of phase-separated
cellular bodies. Cell 166, 651–663 (2016).
11. Letunic, I., Khedkar, S. & Bork, P. SMART: recent updates, new
developments and status in 2020. Nucleic Acids Res. 49,
D458–D460 (2021).
12. Rao, J., Lahiri, J., Isaacs, L., Weis Robert, M. & Whitesides George,
M. A trivalent system from vancomycin·D-Ala-D-Ala with higher
ainity than avidin·biotin. Science 280, 708–711 (1998).
13. Li, P. et al. Phase transitions in the assembly of multivalent
signalling proteins. Nature 483, 336–340 (2012).
This is the leading paper that reconstituted liquid–liquid phase
separation in vitro and demonstrated that LLPS is driven by
multivalency.
14. Flory, P. J. Principles of Polymer Chemistry (Cornell Univ. Press,
1953).
15. Cohen, R. J. & Benedek, G. B. Equilibrium and kinetic theory of
polymerization and the sol-gel transition. J. Phys. Chem. 86,
3696–3714 (1982).
16. Lehn, J.-M. Supramolecular polymer chemistry—scope and
perspectives. Polym. Int. 51, 825–839 (2002).
17. Lin, Y. et al. Formation and maturation of phase-separated liquid
droplets by RNA-binding proteins. Mol. Cell 60, 208–219
(2015).
18. Zeng, M. et al. Phase transition in postsynaptic densities underlies
formation of synaptic complexes and synaptic plasticity. Cell 166,
1163–1175 (2016).
This paper reported a model that phase separation from the
multivalency of SynGAP/PSD-95 drives the formation of
post-synaptic densities.
19. Sun, D., Wu, R., Zheng, J., Li, P. & Yu, L. Polyubiquitin
chain-induced p62 phase separation drives autophagic cargo
segregation. Cell Res. 28, 405–415 (2018).
20. Bouchard, J. J. et al. Cancer mutations of the tumor suppressor
SPOP disrupt the formation of active, phase-separated
compartments. Mol. Cell 72, 19–36 (2018).
21. Shen, B. et al. Computational screening of phase-separating
proteins. Genomics Proteom. Bioinformatics 19, 13–24 (2021).
22. Cascarina, S. M., Elder, M. R. & Ross, E. D. Atypical structural
tendencies among low-complexity domains in the Protein Data
Bank proteome. PLoS Comput. Biol. 16, e1007487 (2020).
23. Frey, S., Richter Ralf, P. & Görlich, D. FG-rich repeats of nuclear
pore proteins form a three-dimensional meshwork with
hydrogel-like properties. Science 314, 815–817 (2006).
This paper describes that LCD in nucleoporin forms reversible
hydrogel, which is required for yeast viability.
24. Kato, M. & McKnight, S. L. A solid-state conceptualization of
information transfer from gene to message to protein. Annu. Rev.
Biochem. 87, 351–390 (2018).
25. Kwon, I. et al. Phosphorylation-regulated binding of RNA
polymerase II to ibrous polymers of low-complexity domains.
Cell 155, 1049–1060 (2013).
26. Yamazaki, H., Takagi, M., Kosako, H., Hirano, T. & Yoshimura, S. H.
Cell cycle-speciic phase separation regulated by protein charge
blockiness. Nat. Cell Biol. 24, 625–632 (2022).
27. Wang, J. et al. Phase separation of OCT4 controls TAD
reorganization to promote cell fate transitions. Cell Stem Cell 28,
1868–1883.e1811 (2021).
28. Mier, P. et al. Disentangling the complexity of low complexity
proteins. Brief. Bioinformatics 21, 458–472 (2020).
29. Xiang, S. et al. The LC Domain of hnRNPA2 adopts similar
conformations in hydrogel polymers, liquid-like droplets, and
nuclei. Cell 163, 829–839 (2015).
30. Murray, D. T. et al. Structure of FUS protein ibrils and its relevance
to self-assembly and phase separation of low-complexity
domains. Cell 171, 615–627 (2017).
31. Cao, Q., Boyer, D. R., Sawaya, M. R., Ge, P. & Eisenberg, D. S.
Cryo-EM structures of four polymorphic TDP-43 amyloid cores.
Nat. Struct. Mol. Biol. 26, 619–627 (2019).
32. Zhou, X. et al. Mutations linked to neurological disease enhance
self-association of low-complexity protein sequences. Science
377, eabn5582 (2022).
33. Lin, Y. et al. Redox-mediated regulation of an evolutionarily
conserved cross-β structure formed by the TDP43 low
complexity domain. Proc. Natl Acad. Sci. USA 117, 28727–28734
(2020).
34. Dreyfuss, G., Matunis, M. J., Pinol-Roma, S. & Burd, C. G. hnRNP
proteins and the biogenesis of mRNA. Annu. Rev. Biochem. 62,
289–321 (1993).
35. Martin, E. W. et al. Interplay of folded domains and the disordered
low-complexity domain in mediating hnRNPA1 phase separation.
Nucleic Acids Res. 49, 2931–2945 (2021).
This paper discusses how the interplay between low-complexity
domains and folded domains drives phase separation.
36. Zeng, M. et al. Reconstituted postsynaptic density as a molecular
platform for understanding synapse formation and plasticity. Cell
174, 1172–1187 (2018).
37. Milovanovic, D., Wu, Y., Bian, X. & Camilli, P. D. A liquid phase of
synapsin and lipid vesicles. Science 361, 604–607 (2018).
38. Wootton, J. C. Non-globular domains in protein sequences:
automated segmentation using complexity measures. Comput.
Chem. 18, 269–285 (1994).
39. Hughes, M. P. et al. Atomic structures of low-complexity protein
segments reveal kinked β sheets that assemble networks. Science
359, 698–701 (2018).
40. Chen, Z. et al. Screening membraneless organelle participants
with machine-learning models that integrate multimodal
features. Proc. Natl Acad. Sci. USA 119, e2115369119 (2022).
41. Li, Q. et al. Protein databases related to liquid-liquid phase
separation. Int. J. Mol. Sci. 21, 6796 (2020).
42. Pancsa, R., Vranken, W. & Mészáros, B. Computational resources
for identifying and describing proteins driving liquid-liquid phase
separation. Brief. Bioinformatics https://doi.org/10.1093/bib/
bbaa408 (2021).
43. Fritsch Anatol, W. et al. Local thermodynamics govern formation
and dissolution of Caenorhabditis elegans P granule condensates.
Proc. Natl Acad. Sci. USA 118, e2102772118 (2021).
44. Cascarina, S. M. & Ross, E. D. Proteome-scale relationships
between local amino acid composition and protein fates and
functions. PLoS Comput. Biol. 14, e1006256 (2018).
45. Kroschwald, S. et al. Dierent material states of Pub1 condensates
deine distinct modes of stress adaptation and recovery. Cell Rep.
23, 3327–3339 (2018).
46. Cirillo, L. et al. UBAP2L forms distinct cores that act in nucleating
stress granules upstream of G3BP1. Curr. Biol. 30, 698–707
(2020).
47. McSwiggen, D. T., Mir, M., Darzacq, X. & Tjian, R. Evaluating
phase separation in live cells: diagnosis, caveats, and functional
consequences. Genes Dev. 33, 1619–1634 (2019).
This review summarizes current strategies to characterize phase
separation in vivo.
48. Peng, S. et al. Phase separation at the nanoscale quantiied by
dcFCCS. Proc. Natl Acad. Sci. USA 117, 27124–27131 (2020).
Nature Chemical Biology
Perspective https://doi.org/10.1038/s41589-022-01204-2
49. Van Lindt, J. et al. A generic approach to study the kinetics of
liquid-liquid phase separation under near-native conditions.
Commun. Biol. 4, 77 (2021).
50. Woodru, J. B., Hyman, A. A. & Boke, E. Organization and function
of non-dynamic biomolecular condensates. Trends Biochem. Sci.
43, 81–94 (2018).
51. Woodru, J. B. et al. The centrosome is a selective condensate
that nucleates microtubules by concentrating tubulin. Cell 169,
1066–1077 (2017).
52. Boke, E. et al. Amyloid-like self-assembly of a cellular
compartment. Cell 166, 637–650 (2016).
53. Zhang, G., Wang, Z., Du, Z. & Zhang, H. mTOR regulates phase
separation of PGL granules to modulate their autophagic
degradation. Cell 174, 1492–1506 (2018).
54. Mitrea, D. M. et al. Methods for physical characterization of
phase-separated bodies and membrane-less organelles.
J. Mol. Biol. 430, 4773–4805 (2018).
This review summarizes biophysical strategies to characterize
phase-separated MLOs.
55. Dazzi, A. et al. AFM–IR: combining atomic force microscopy and
infrared spectroscopy for nanoscale chemical characterization.
Appl. Spectrosc. 66, 1365–1384 (2012).
56. Delarue, M. et al. mTORC1 controls phase separation and the
biophysical properties of the cytoplasm by tuning crowding. Cell
174, 338–349 (2018).
57. Folkmann, A. W., Putnam, A., Lee, C. F. & Seydoux, G. Regulation
of biomolecular condensates by interfacial protein clusters.
Science 373, 1218–1224 (2021).
58. Ma, W., Zheng, G., Xie, W. & Mayr, C. In vivo reconstitution
inds multivalent RNA-RNA interactions as drivers of mesh-like
condensates. eLife 10, e64252 (2021).
This paper shows that RNA can modulate the morphology of
liquid-like condensates.
59. Patel, A. et al. A liquid-to-solid phase transition of the ALS
protein FUS accelerated by disease mutation. Cell 162, 1066–1077
(2015).
60. Courchaine, E. M. et al. DMA-tudor interaction modules control
the speciicity of in vivo condensates. Cell 184, 3612–3625 (2021).
61. Rai, A. K., Chen, J.-X., Selbach, M. & Pelkmans, L.
Kinase-controlled phase transition of membraneless organelles in
mitosis. Nature 559, 211–216 (2018).
62. Liao, Y.-C. et al. RNA granules hitchhike on lysosomes for
long-distance transport, using annexin A11 as a molecular tether.
Cell 179, 147–164 (2019).
63. Padrón, A., Iwasaki, S. & Ingolia, N. T. Proximity RNA labeling
by APEX-Seq reveals the organization of translation initiation
complexes and repressive RNA granules. Mol. Cell 75, 875–887
(2019).
64. Hubstenberger, A. et al. P-body puriication reveals the
condensation of repressed mRNA regulons. Mol. Cell 68, 144–157
(2017).
65. Wheeler, E. C. et al. Pooled CRISPR screens with imaging on
microraft arrays reveals stress granule-regulatory factors. Nat.
Methods 17, 636–642 (2020).
66. Roy, R., Das, G., Kuttanda, I. A., Bhatter, N. & Rajyaguru, P. I. Low
complexity RGG-motif sequence is required for Processing body
(P-body) disassembly. Nat. Commun. 13, 2077 (2022).
67. Tsang, B., Pritišanac, I., Scherer, S. W., Moses, A. M. & Forman-Kay,
J. D. Phase separation as a missing mechanism for interpretation
of disease mutations. Cell 183, 1742–1756 (2020).
68. Parker, R. & Sheth, U. P bodies and the control of mRNA
translation and degradation. Mol. Cell 25, 635–646 (2007).
69. Lafontaine, D. L. J., Riback, J. A., Bascetin, R. & Brangwynne, C. P.
The nucleolus as a multiphase liquid condensate. Nat. Rev. Mol.
Cell Biol. 22, 165–182 (2021).
70. Bienz, M. Head-to-tail polymerization in the assembly of
biomolecular condensates. Cell 182, 799–811 (2020).
71. Zamudio, A. V. et al. Mediator condensates localize signaling
factors to key cell identity genes. Mol. Cell 76, 753–766
(2019).
72. Xiao, Q., McAtee, C. K. & Su, X. Phase separation in immune
signalling. Nat. Rev. Immunol. 22, 188–199 (2022).
73. Huang, X. et al. ROS regulated reversible protein phase
separation synchronizes plant lowering. Nat. Chem. Biol. 17,
549–557 (2021).
74. Riback, J. A. et al. Stress-triggered phase separation is an adaptive,
evolutionarily tuned response. Cell 168, 1028–1040 (2017).
75. Weitzel, G., Pilatus, U. & Rensing, L. Similar dose response of heat
shock protein synthesis and intracellular pH change in yeast. Exp.
Cell. Res. 159, 252–256 (1985).
76. Lenard, A. J. et al. Phosphorylation regulates CIRBP arginine
methylation, transportin-1 binding and liquid-liquid phase
separation. Front. Mol. Biosci. https://doi.org/10.3389/
fmolb.2021.689687 (2021).
77. Kato, M., Tu, B. P. & McKnight, S. L. Redox-mediated regulation of
low complexity domain self-association. Curr. Opin. Genet. Dev.
67, 111–118 (2021).
78. Kato, M. et al. Redox state controls phase separation of the yeast
ataxin-2 protein via reversible oxidation of its methionine-rich
low-complexity domain. Cell 177, 711–721 (2019).
79. Yang, Y.-S. et al. Yeast ataxin-2 forms an intracellular condensate
required for the inhibition of TORC1 signaling during respiratory
growth. Cell 177, 697–710 (2019).
80. Jin, M. et al. Glycolytic enzymes coalesce in G bodies under
hypoxic stress. Cell Rep. 20, 895–908 (2017).
81. Munder, M. C. et al. A pH-driven transition of the cytoplasm from
a luid- to a solid-like state promotes entry into dormancy. eLife 5,
e09347 (2016).
82. Yasuda, S. et al. Stress- and ubiquitylation-dependent phase
separation of the proteasome. Nature 578, 296–300 (2020).
83. Lau, Y., Oamen, H. P. & Caudron, F. Protein phase separation
during stress adaptation and cellular memory. Cells 9, 1302
(2020).
84. Lu, Y. et al. Phase separation of TAZ compartmentalizes the
transcription machinery to promote gene expression. Nat. Cell
Biol. 22, 453–464 (2020).
85. Watanabe, K. et al. Cells recognize osmotic stress through liquid–
liquid phase separation lubricated with poly(ADP-ribose). Nat.
Commun. 12, 1353 (2021).
86. Perez-Gonzalez, N. A. et al. YAP and TAZ regulate cell volume. J.
Cell Biol. 218, 3472–3488 (2019).
87. Rawat, P. et al. Stress-induced nuclear condensation of NELF
drives transcriptional downregulation. Mol. Cell 81, 1013–1026
(2021).
88. Xie, D. et al. Phase separation of SERRATE drives dicing body
assembly and promotes miRNA processing in Arabidopsis. Nat.
Cell Biol. 23, 32–39 (2021).
89. McDonald, N. A., Fetter, R. D. & Shen, K. Assembly of synaptic
active zones requires phase separation of scaold molecules.
Nature 588, 454–458 (2020).
90. Fang, X. et al. Arabidopsis FLL2 promotes liquid–liquid phase
separation of polyadenylation complexes. Nature 569, 265–269
(2019).
91. Shin, Y. et al. Spatiotemporal control of intracellular phase
transitions using light-activated optoDroplets. Cell 168, 159–171
(2017).
92. Schneider, N. et al. Liquid-liquid phase separation of
light-inducible transcription factors increases transcription
activation in mammalian cells and mice. Sci. Adv. 7, eabd3568
(2021).
Nature Chemical Biology
Perspective https://doi.org/10.1038/s41589-022-01204-2
93. Trojanowski, J. et al. Transcription activation is enhanced by
multivalent interactions independent of phase separation.
Mol. Cell https://doi.org/10.1016/j.molcel.2022.04.017 (2022).
94. Annunziata, O. et al. Eect of polyethylene glycol on the
liquid-liquid phase transition in aqueous protein solutions.
Proc. Natl Acad. Sci. USA 99, 14165–14170 (2002).
95. Homann, C. et al. Fluorescent labeling of tetracysteine-tagged
proteins in intact cells. Nat. Protoc. 5, 1666–1677 (2010).
96. Shulga, N. & Goldfarb David, S. Binding dynamics of structural
nucleoporins govern nuclear pore complex permeability and may
mediate channel gating. Mol. Cell. Biol. 23, 534–542 (2003).
97. Kroschwald, S. et al. Promiscuous interactions and protein
disaggregases determine the material state of stress-inducible
RNP granules. eLife https://doi.org/10.7554/eLife.06807 (2015).
98. Lin, Y. et al. Toxic PR poly-dipeptides encoded by the C9orf72 repeat
expansion target LC domain polymers. Cell 167, 789–802 (2016).
99. Kroschwald, S., Maharana, S. & Alberti, S. Hexanediol: a
chemical probe to investigate the material properties of
membrane-less compartments. Matters https://doi.org/10.19185/
matters.201702000010 (2017).
100. Ming, Y. et al. Targeting liquid-liquid phase separation in
pancreatic cancer. Transl. Cancer Res. 8, 96–103 (2019).
Acknowledgements
This work was supported by grants from the Ministry of Science and
Technology of the People’s Republic of China (2022ZD0213900
and 2022ZD0204900 to Y.L.), the National Key R&D Program
(2019YFA0508403 to P.L.) and the National Natural Science Foundation
of China (32170684 to Y.L.; 32150023, 32125010 and 31871443 to P.L.).
Author contributions
X.L. and Y.G. created the igures. Y.G., X.L., P.L. and Y.L. conceived and
wrote this Perspective.
Competing interests
The authors declare no competing interests.
Additional information
Correspondence should be addressed to Pilong Li or Yi Lin.
Peer review information Nature Chemical Biology thanks the
anonymous reviewers for their contribution to the peer review of
this work.
Reprints and permissions information is available at
www.nature.com/reprints.
Publisher’s note Springer Nature remains neutral with regard
to jurisdictional claims in published maps and institutional
ailiations.
Springer Nature or its licensor (e.g. a society or other partner)
holds exclusive rights to this article under a publishing agreement
with the author(s) or other rightsholder(s); author self-archiving
of the accepted manuscript version of this article is solely
governed by the terms of such publishing agreement and
applicable law.
© Springer Nature America, Inc. 2022
