Macromolecular condensation buffers intracellular water potential

842

| Nature | Vol 623 | 23 November 2023

Article

Macromolecular condensation buffers

intracellular water potential

Joseph L. Watson

, Estere

Seinkmane

1,1 2

, Christine T. Styles

2,1 2

, Andrei

Mihut

, Lara K. Krüger

Kerrie E. McNally

, Vicente Jose Planelles-Herrero

, Michal

Dudek

, Patrick M. McCall

4,5,6

Silvia Barbiero

, Michael

Vanden Oever

, Sew Yeu Peak-Chew

, Benjamin T. Porebski

Aiwei Zeng

, Nina M. Rzechorzek

, David C. S. Wong

, Andrew D. Beale

Alessandra Stangherlin

1,1 1

, Margot

Riggi

, Janet

Iwasa

, Jörg

Morf

, Christos

Miliotis

Alina Guna

, Alison J. Inglis

, Jan

Brugués

4,5,6

, Rebecca M. Voorhees

, Joseph E. Chambers

Qing-Jun Meng

, John S. O’Neill

✉

, Rachel S. Edgar

✉

& Emmanuel

Derivery

✉

Optimum protein function and biochemical activity critically depends on water

availability because solvent thermodynamics drive protein folding and

macromolecular interactions

. Reciprocally, macromolecules restrict the movement

of ‘structured’ water molecules within their hydration layers, reducing the available

‘free’ bulk solvent and therefore the total thermodynamic potential energy of water,

or water potential. Here, within concentrated macromolecular solutions such as the

cytosol, we found that modest changes in temperature greatly affect the water

potential, and are counteracted by opposing changes in osmotic strength. This

duality of temperature and osmotic strength enables simple manipulations of

solvent thermodynamics to prevent cell death after extreme cold or heat shock.

Physiologically, cells must sustain their activity against fluctuating temperature,

pressure and osmotic strength, which impact water availability within seconds. Yet,

established mechanisms of water homeostasis act over much slower timescales

;

we therefore postulated the existence of a rapid compensatory response. We find that

this function is performed by water potential-driven changes in macromolecular

assembly, particularly biomolecular condensation of intrinsically disordered

proteins. The formation and dissolution of biomolecular condensates liberates and

captures free water, respectively, quickly counteracting thermal or osmotic

perturbations of water potential, which is consequently robustly buffered in the

cytoplasm. Our results indicate that biomolecular condensation constitutes an

intrinsic biophysical feedback response that rapidly compensates for intracellular

osmotic and thermal fluctuations. We suggest that preserving water availability

within the concentrated cytosol is an overlooked evolutionary driver of protein (dis)

order and function.

Water is critical to life, providing a dynamic hydrogen-bonded envi

ronment that supports macromolecule solvation. Far from being a

passive solvent, water drives protein folding and macromolecular

interactions that optimize the network of H

O hydrogen bonds

. Pro

tein structure, supramolecular assembly and activity are therefore

highly sensitive to changes in water thermodynamics

, which must

be tightly regulated to preserve function over multiple timescales.

Reciprocally, macromolecules in solution impose a profound ener

getic cost on neighbouring water molecules within their hydration

layers by lowering their translational and rotational entropy

. In other

words, water is required to hydrate macromolecules and make them

fold properly, but this restricts the movement of water molecules and

thereby diminishes their availability. Thus, cells must maintain water

availability within an optimal range for protein activity, biochemical

efficiency and, ultimately, viability.

Water–macromolecule interactions are integral to every biological

process. Here we refer to hydration-layer water molecules with lower

entropy as structured, in contrast to the free water molecules that form

the bulk solvent. The impact of a macromolecule on water depends on

both the size and chemistry of its solvent-accessible surface area

–

https://doi.org/10.1038/s41586-023-06626-z

Received: 17 November 2022

Accepted: 8 September 2023

Published online: 18 October 2023

Open access

Check for updates

MRC Laboratory of Molecular Biology, Cambridge, UK.

Department of Infectious Disease, Imperial College London, London, UK.

Wellcome Centre for Cell Matrix Research, University of

Manchester, Manchester, UK.

Cluster of Excellence Physics of Life, TU Dresden, Dresden, Germany.

Max Planck Institute of Molecular Cell Biology and Genetics, Dresden, Germany.

Max

Planck Institute for the Physics of Complex Systems, Dresden, Germany.

Department of Biochemistry, University of Utah, Salt Lake City, UT, USA.

Laboratory of Nuclear Dynamics, Babraham

Institute, Cambridge, UK.

California Institute of Technology, Pasadena, CA, USA.

Cambridge Institute for Medical Research, Cambridge, UK.

Present address: Cluster of Excellence Cellular

Stress Responses in Aging-associated Diseases (CECAD), Faculty of Medicine and University Hospital Cologne, University of Cologne, Cologne, Germany.

These authors contributed equally:

Estere Seinkmane, Christine T. Styles.

✉

e-mail:

oneillj@mrc-lmb.cam.ac.uk

;

rachel.edgar@imperial.ac.uk

;

derivery@mrc-lmb.cam.ac.uk

Nature | Vol 623 | 23 November 2023 |

843

as the entropic penalty of hydration can be offset by enthalpically

favourable interactions between water and hydrophilic surfaces or

accentuated at hydrophobic interfaces

The total thermodynamic potential energy of water, or water poten

tial (

), has pressure, gravimetric and other components, but the most

biologically relevant is the solute or osmotic potential (

and −

respectively), defined as zero for pure water. The addition of solutes

lowers the potential energy by reducing the free water available in

the system to perform work. In an ideal solution,

depends solely

on the number of particles in solution, rather than their nature, and

can be measured directly from colligative properties such as vapour

pressure. However, the behaviour of water in the concentrated intra

cellular environment is far from ideal. In vitro, hydrophilic macro

molecules such as glycosylated mucin or tRNA have a much greater

impact on water potential compared with smaller solutes such as

KCl or sucrose (Fig.

and Extended Data Fig. 1a), as their hydra

tion constrains many more water molecules. For these highly polar

or charged molecules,

changes linearly with concentration,

g

mss4

PH–GFP

Low

High

135

0246

0.5

1.0

1.5

Normalized

membrane:cytosol ratio

Time after osmolarity

change (min)

Hypoosmotic shock of –382 mOsm l

−1

Hypoosmotic

shock

32 ºC

WT (

= 30)

mss4

(

= 16)

mss4

(39 ºC, no shock)

Free H

(bulk)

Free:

structured

O ratio

Structured

Heat shock

(39 ºC)

Prediction:

hypoosmotic

shock

Control (32 ºC)

Cortical

PH–GFP

PIP(4)P

PIP(4,5)P

mss4

PIP(4)P

PIP(4,5)P

mss4

PH–GFP

500

1,000

Concentration (mM)

BSA 27 ºC

BSA 37 ºC

tRNA 27 ºC

tRNA 37 ºC

KCl 37 ºC

KCl 27 ºC

Mucin 27 ºC

Mucin 37 ºC

–

/RT (mOsm kg

–1

)

High

Low

Enthalpic compensation

Calcium signalling induced by hyperosmotic shoc

Temperature (ºC)

24 h cold shock

100

Viability (%)

30 min heat shock

O (%)

Temperature (ºC)

100

Viability (%)

350

150

250

350

150

250

Osmolarity (mOsm l

–1

)

f

Control (

= 20)

+100 mOsm l

−1

(

= 20)

–150 mOsm l

−1

(

= 25)

Δ

T = –17 ºC

–150 mOsm l

–1

(

= 27)

Δ

T = –17 ºC (

= 18)

Fluo-4

(a.u.)

Time (min)

Medium

addition

Cell death induced by temperature shocks

Hypothermia

External

hyperosmolarity

Hyperthermia

External

hypoosmolarity

27 °C

BSA

37 °C

BSA

= 5.8 × 10

–11

= 0.0030

< 0.0001

–

Integrated Fluo-4 signal (

)

< 0.0001

Fig. 1 | The duality of thermal and osmotic perturbation on water potential

and cellular function.

, Vapour-pressure osmometry measurements for the

indicated solute concentrations and temperatures (left). Data are mean ± s.e.m.

= 3. BSA exerts a nonlinear effect on solvent thermodynamics that is

accentuated at 27 °C compared with at 37 °C, whereas other macromolecules

exhibit temperature-independent quasi-linear relationships, indicating that

this is not a crowding effect (Extended Data Fig. 1). Right, instead, we propose

that structured water increases as the temperature decreases. Statistical analysis

was performed using two-way analysis of variance (ANOVA).

, The model for

duality of osmotic and thermal perturbations on free:structured water in cells

(Supplementary Video 3).

–

, Hypoosmotic shock phenocopies heat shock for

thermosensitive yeast mutants.

, Using WT

mss4

and thermosensitive

mss4

strains of

S. cerevisiae

expressing PIP(4,5)P

GFP probe (PH–GFP) to monitor

mss4

PIP(4) kinase activity, we found that the cortical GFP signal decreased

when cells shifted from permissive (32 °C) to restrictive (39 °C) temperatures

(Extended Data Fig. 2). We predicted that hypoosmotic shock at 32 °C would

mimic 39 °C heat shock.

, The PH–GFP signal after hypoosmotic shock was

monitored using spinning-disk confocal microscopy (SDCM; single confocal

planes). Scale bar, 5 μm.

, The normalized cortical to cytosol ratio of the PH–GFP

signal. Data are mean ± s.e.m. As predicted, thermosensitive

mss4

mutants

lose the cortical PH–GFP signal after hypoosmotic shock but the WT strains do

not, excluding indirect effects on PH–GFP signal or PIP

levels through membrane

tension or PIP(4,5)P

phosphatases.

, The interaction between osmolarity and

temperature on calcium signalling in primary chondrocytes (Fluo-4

signal).

Data are mean ± s.e.m.

values indicate the number of fields of view analysed.

See also Extended Data Fig. 3. Statistical analysis was performed using one-way

ANOVA followed by Dunnett’s test;

values are indicated.

, Manipulating

water thermodynamics rescued Raji cell viability after cold or heat shock. Data

are mean ± s.e.m.

= 3. Hypoosmotic conditions increased survival at 0 °C.

Similarly, D

O increased survival after extreme heat shock as increased

H-bonding network strength preserves the hydration layer size. Statistical

analysis was performed using two-way ANOVA with Dunnett’s post hoc test;

values are indicated.

844

| Nature | Vol 623 | 23 November 2023

Article

as described by van’t Hoff ’s law −

iCRT

, where

is the van’t Hoff

factor,

is the gas constant and

is the temperature in kelvin. As

expected, a small decrease in temperature from 37 °C (310 K) to

27 °C (300 K) has a minimal impact on the water potential of these

solutions.

By contrast,

deviates significantly from van’t Hoff ’s linearity in

concentrated solutions of macromolecules with exposed surfaces that

are more hydrophobic and less electrostatically favourable, such as

bovine serum albumin (BSA), haemoglobin (Hb) or polyethylene glycol

(PEG)

–

(Fig.

and Extended Data Fig. 1b). Departure from ideality

reflects how much hydrogen bonding (enthalpy) and water movement

(entropy) is perturbed compared with pure water. Owing to its size,

mucin restricts the movement of thousands of water molecules, but

its heavily glycosylated surface provides sufficient enthalpic compen

sation (Fig.

). However, most macromolecular surfaces have both

hydrophilic and hydrophobic regions that differentially alter water

motion and hydrogen bonding networks compared with bulk solvent.

To formally evaluate this effective solute–water interaction, denoted

, we followed the work of Fullerton and colleagues and modelled

our osmometry curves with the empirical equation (

)

(the rationale

is provided in the Supplementary Discussion):

−=

1−

)

πs

where

is the solute concentration and constant

is a function of

solute mass. A smaller

component indicates less deviation from

van’t Hoff ’s linearity (that is,

≪

lim=

), and fitting our data to equa

tion (

) confirmed that proteins such as BSA have very high

values

compared with more hydrophilic macromolecules and small solutes

(Extended Data Fig. 1c). Furthermore, as expected,

scales with chain

size for polymers such as PEG (Extended Data Fig. 1d,e). Notably, the

water potential of concentrated protein solutions becomes sensitive

to physiologically relevant temperature changes. For example, a mod

est temperature decrease from 37 °C (310 K) to 27 °C (300 K) alters the

of a BSA solution by twofold as bulk solvent becomes limiting

(Fig.

). Overall, the departure from van’t Hoff ’s linearity (

) of BSA,

Hb and PEG was strongly increased as temperature decreased (Extended

Data Fig. 1d–i).

Molecular dynamics simulations and studies of protein cold dena

turation have both previously suggested that increased macromo

lecular hydration occurs at lower temperatures and is also consistent

with temperature-dependent changes in linear alcohol hydration

–

The notable effect of temperature decrease on

was observed only

in concentrated colloidal solutions, where the relationship with mac-

romolecule concentration departs from linearity. We infer that this

occurs because more water molecules are recruited to hydration lay

ers as temperature falls, as the increased strength of hydrogen bond-

ing extends the structured water surrounding each macromolecule.

Similar to higher macromolecule concentrations in which there are

more surfaces to hydrate, colder temperatures would increase the

proportion of structured water compared with free water to greatly

increase −

. In both cases, the entropic cost of structured water

increases disproportionately as the bulk solvent becomes limiting

(Fig.

The cellular interior is a concentrated colloidal solution. From our

observations in solution, we predicted that intracellular water poten-

tial would be similarly sensitive to acute changes in macromolecular

hydration elicited by perturbation of temperature and extracellular

osmotic strength, because both affect the ratio of free to structured

water (Fig.

). For example, an abrupt fall in temperature is expected

to decrease the proportion of available free water to structured water,

similar to external hyperosmotic conditions in which free bulk water

immediately leaves the cell to restore the

equilibrium across the

cell membrane.

Duality of thermal and osmotic shocks

In light of our findings, we hypothesized that acute changes in tem

perature could rapidly affect macromolecular structure and enzymatic

activity indirectly by altering water availability and thermodynam

ics, in addition to direct kinetic effects. If true, decreased external

osmotic strength would have an equivalent effect on intracellular

to increased temperature (Fig.

). Initially, we tested this prediction

using temperature-sensitive yeast mutants. Thermosensitive muta

tions are thought to modify protein stability so that a slight elevation in

temperature causes the protein to reversibly unfold

. After transfer to

the restrictive temperature of 39 °C, the well-established thermosensi

tive mutant of the phosphatidylinositol 4,5-bisphosphate (PIP(4,5)P

)

kinase Mms4p in

Saccharomyces cerevisiae

mms4

, becomes inactive

and PIP(4,5)P

is therefore lost from the plasma membrane

(Fig.

and

Extended Data Fig. 2a–c). Notably, an external hypoosmotic shock mim

icked the temperature phenotype and led to a similar loss in PIP(4,5)P

signal at the membrane in the

mms4

mutant, but not in the wild-type

(WT) controls, over comparable timescales (Fig.

1d,e

and Extended Data

Fig. 2d–f for recovery control). We validated this concept in another

thermosensitive mutant in another species. The established

Schizosac

charomyces pombe

thermosensitive mutant of the spindle assembly

kinesin-5 Cut7,

Cut7-24

, induces monopolar spindle formation at the

restrictive temperature

. As we predicted, this phenotype was also

observed at the permissive temperature after external hypoosmotic

shock (Extended Data Fig. 2g–i).

We next investigated the duality of thermal and osmotic pertur

bation in primary mouse chondrocytes, in which Ca

signalling in

response to changes in osmotic strength is well established in vivo

Using Fluo-4 imaging, we confirmed that acute hyperosmotic treatment

evoked a dose-dependent increase in Ca

signalling and validated our

prediction that an acute temperature decrease would evoke a similar

response, whereas hypoosmotic treatment had no effect (Fig.

and

Extended Data Fig. 3a–d). Critically, when hypoosmotic treatment and

temperature decrease were applied simultaneously, the Ca

-signalling

response was completely abolished (Fig.

). This suggests that the

observed Ca

signalling in chondrocytes is regulated by the ratio of

intracellular free:structured water. Given the fast, second-scale kinetics

of this response, we propose that membrane Ca

-channel opening may

directly respond to

, as opposed to indirect modulation by sensors

of solute concentration or membrane tension.

Finally, we tested this duality hypothesis at the global level of the

cell, focusing on the viability of mammalian cells during stress. Cel

lular responses to osmotic and thermal stress are thought to involve

different pathways and mechanisms

, but our previous observa

tions suggested a shared component that senses and responds to

resultant changes in water availability. We reasoned that overwhelm-

ing the cell’s ability to buffer intracellular

could contribute to cell

death when exposed to temperature extremes. Consequently, we

predicted that manipulation of solvent thermodynamics to oppose

temperature-driven changes in the ratio of free:structured water would

attenuate the effect of heat or cold shock. Consistent with this predic

tion, by combining hypoosmotic shock with prolonged exposure to

0 °C, we observed that viability was markedly increased for both sus-

pension and adherent cells (Fig.

(left) and Extended Data Fig. 3e–g).

According to our model, the deleterious increase in the proportional

amount of structured water at low temperatures was counteracted by

increased uptake and availability of bulk free water under hypoosmotic

conditions (Fig.

At supraphysiological temperatures (>43 °C), protein denaturation

arises through a combination of increased kinetic energy and decreased

effective strength of hydrogen bonds. The resulting aggregation of

unfolded proteins leads to cell death

. In dilute solutions, protein

thermal stability can be rescued by

manipulations, such as greatly

increasing osmolarity with small solutes (for example, sucrose

) or

Nature | Vol 623 | 23 November 2023 |

845

by using heavy water (D

O) instead of H

O (ref.

). In cells, the high

extracellular osmolarities that would be needed to increase protein

stability also lead to decreased cell volume that increases the aggrega

tion of thermally denatured proteins and cell death. By contrast, replac

ing hydrogen with deuterium has complex effects but, importantly,

cell volume is unchanged and hydrogen bonds are stronger in D

compared with in H

O (refs.

–

). Consequently, the water potential

of PEG/D

O solutions are less sensitive to changes in macromolecule

concentration and temperature compared with H

O solutions

(Extended Data Fig. 4a,b), that is, the effective interaction between

solute and D

O (

) is lower than for H

O and the temperature depend

ency of

is attenuated for D

O (Extended Data Fig. 4c). We therefore

predicted that substitution of H

O with D

O in macromolecule hydra-

tion layers would mitigate the effect of high temperature and preserve

protein stability. In agreement with this conceptual framework, we

found that D

O substantially rescued cell viability from an otherwise

cytotoxic heat shock (Fig.

(right) and Extended Data Fig. 4d–f ).

Similarly, D

O partially rescued the effects of the restrictive tempera-

ture on monopolar spindle formation in the

S. pombe

thermosensitive

mutant

Cut7-24

(Extended Data Fig. 2h,i).

We conclude that, while temperature and osmotic shock clearly have

many different effects on cell biology, their intersection with regards

to solvent thermodynamics (

) impacts fundamental properties of

life such as protein structure and function and, ultimately, cell survival.

homeostasis involves condensates

Our results highlight the need for cells to maintain

homeostasis over

different challenges and timescales. In the body, cells must tolerate and

adapt to anatomical and temporal variation in temperature and osmotic

strength. The osmolality of plasma is around 290 mOsm l

−1

compared

with 1,200 mOsm l

−1

in parts of the human kidney, for example, while the

temperature of the dermal tissue is around 30 °C whereas that of the

deep brain can exceed 40 °C in healthy individuals

. Yet, cell volume is

robust against physiological fluctuations in osmolarity, temperature,

pressure and intracellular macromolecule concentration

, the effect

of which on intracellular water potential is almost instantaneous. This

suggests that intracellular

is defended over subsecond timeframes

to maintain the optimum balance of free:structured water for protein

function and biochemical activity.

We hypothesized that cells possess a fast-acting compensatory

mechanism to preserve water availability and reasoned that since some

proteins can strongly affect

(Fig.

), this response could be mediated

by proteins themselves. If true, we expected that proteins involved in

water potential homeostasis would show consistent changes in expres-

sion or activity after long-term exposure to either thermal or osmotic

stresses, to defend against any further perturbations from their new

setpoint. Specifically, we sought to identify proteins and phospho

proteins of which the abundance varies not only with external osmotic

strength, but also inversely with temperature, due to their antagonistic

impact on the free:structured water ratio in concentrated colloidal solu

tions (Fig.

). To this end, confluent (quiescent) cultures of primary

mouse fibroblasts were allowed to adapt over 2 weeks to conditions

of lower/higher external osmolarity (±100 mOsm l

−1

) or temperature

(32 °C or 40 °C). The cellular (phospho)proteome composition was then

compared with the controls (37 °C, 350 mOsm l

−1

) using quantitative

mass spectrometry (MS; Fig.

). Validating our approach, we observed

expected changes in the abundance and/or phosphorylation of known

heat-shock proteins (such as HSPA13 and HSPH1), cold-shock proteins

(for example, CIRBP and RBM3) and osmoregulated proteins (such

as HMOX1 and SLC5A3) for each relevant stressor, as reported previ

ously

–

(Extended Data Fig. 5 and Supplementary Tables 1 and 2).

We found a significant over-representation of proteins for which

abundance and phosphorylation was both postively correlated with

temperature and negatively correlated with osmotic strength (Fig.

Methods and Extended Data Fig. 5b–f ). Gene Ontology analysis of these

putatively

-responsive proteins revealed significant enrichment

for localization to membraneless organelles (MLOs), including the

nucleolus (Extended Data Fig. 5c). Furthermore, querying published

databases of phase-separating proteins confirmed significant enrich

ment of proteins known to participate in the formation of biomolecular

condensates (

= 3.6 × 10

−5

; Fig.

). This suggests that protein conden

sation may be involved in the response and adaptation to changes in

MLOs are biomolecular condensates that behave as liquid–liquid

phase-separated compartments and are associated with the presence

of intrinsically disordered regions (IDRs) within their constituent

proteins

–

. IDRs in solution have a greater effect on solvent entropy

compared with soluble globular proteins because their higher ratio of

surface area:volume requires proportionally more hydration water

IDR-containing proteins, such as fused in sarcoma (FUS), can revers

ibly form condensates depending on their local environment and

post-translational modifications

–

. The entropic cost of IDR hydra-

tion can be enthalpically compensated by electrostatic factors—such

as phosphorylation—and, throughout the proteome, most protein

phosphorylation indeed occurs within IDRs

. On the basis of previous

research in yeast

, it therefore seemed plausible that global changes

in IDR phosphorylation might provide one means for modulating the

effect of intracellular proteins on

, thereby buffering

against

applied thermal or osmotic changes. Our analysis of the putative

-responsive phosphoproteome supported this paradigm: the rela-

tive proportion of IDR phosphorylation increased during adaptation to

both hyperosmolarity and lower temperature, and vice versa (Fig.

Notably, temperature had the opposite effects to external osmolarity on

the phosphorylation of OXSR1 kinase, a key osmo-effector, at a known

regulatory site within its IDR

(Fig.

and Extended Data Fig. 5f ). Sali

ently,

-responsive phosphosites were enriched for motifs recognized

by promiscuous kinases with established preference for IDRs (casein

kinase 1, casein kinase 2, glycogen synthase kinase 3; Extended Data

Fig. 5g). Collectively these results show that chronic osmotic or thermal

perturbations elicit similar adaptations in the (phospho)proteome

that implicate MLOs and intrinsically disordered proteins as frontline

defenders of intracellular

Cellular control of condensates by

Macromolecules within condensates are predicted to be less

hydrated compared with in bulk solvent

. Given the involvement of

IDR-containing proteins during osmotic and thermal adaptation

(Fig.

), and that condensation of the intrinsically disordered protein

FUS releases entropically unfavourable hydration water in vitro

, we

hypothesized that changes in biomolecular condensation could buffer

intracellular water potential. For example, under acute hypoosmotic

or hyperthermal challenge, a transient increase in free H

O molecules

available for protein hydration could provide the bioenergetic drive to

liberate IDR-containing proteins from condensates, hydration of which

would proportionally decrease free water and thereby minimize the

impact of the challenge on cellular water potential.

To test this hypothesis, we used FusLC–GFP—a fusion of the FUS

N-terminal IDR with the GFP fluorescent reporter that is an established

model for phase separation

–

. If formation or dissolution of biomo

lecular condensates acts to oppose variations in

, then one would

expect a rapid increase in condensation after acute hyperosmotic chal

lenge, releasing previously structured water into the bulk solvent to

counteract the externally applied change. Our prediction was con-

firmed by experimental observation: a modest, transient increase in

extracellular osmotic strength elicited a rapid (within seconds) and

a reversible increase in FusLC–GFP condensation; by contrast, GFP

alone showed no significant change and remained diffuse through

out (Fig.

3a,b

; see the Methods and Extended Data Figs. 6 and 7 for

automated, deep-learning-based quantification and Extended Data