Chapter 1 - ban

Mechanical regulation of chromatin

and transcription

Mechanical force

A force that is caused by

contact with another object

and that induces a change

in state of rest or motion.

1Department of Molecular

Medicine, University of Padua

Medical School, Padua, Italy.

2Stem Cells and Metabolism

Research Program, Faculty

of Medicine, University of

Helsinki, Helsinki, Finland.

3Helsinki Institute of Life

Science, Biomedicum

Helsinki, University of

Helsinki, Helsinki, Finland.

4Wihuri Research Institute,

Biomedicum Helsinki,

University of Helsinki,

Helsinki, Finland.

5Department of Cell

and Tissue Dynamics,

Max Planck Institute for

Molecular Biomedicine,

Muenster, Germany.

✉e- mail: sirio.dupont@

unipd.it; sara.wickstrom@

helsinki.fi

s41576-022-00493-6

Sirio Dupont

1 ✉ and Sara A. Wickström

2,3,4,5 ✉

Abstract | Cells and tissues generate and are exposed to various mechanical forces that act across

a range of scales, from tissues to cells to organelles. Forces provide crucial signals to inform cell

behaviour during development and adult tissue homeostasis, and alterations in forces and in their

downstream mechanotransduction pathways can influence disease progression. Recent advances

have been made in our understanding of the mechanisms by which forces regulate chromatin

organization and state, and of the mechanosensitive transcription factors that respond to the

physical properties of the cell microenvironment to coordinate gene expression, cell states and

behaviours. These insights highlight the relevance of mechanosensitive transcriptional regulation

to physiology, disease and emerging therapies.

The generation, maintenance and repair of functional

tissues requires intricate, coordinated regulation of cell

fate, behaviour and position, and the ability of cells to

respond and adapt to dynamic changes in their microenvironment.

The close, highly reproducible correlation

between cell­ type­ specific morphology, function

and identity points to constant, bidirectional feedback

between microenvironmental signals, the contractile

cytoskeleton that responds to these signals and determines

cell shape, and the transcriptional circuitry that

establishes and maintains cell identity1–3

.

One central microenvironmental signal modality is

mechanical force, which acts at the level of tissues, cells,

organelles and molecules, and which has fundamental

effects on cell behaviour through regulation of gene

transcription. These mechanical forces are highly tissue

specific and include compression, shear, tensile stress

and hydrostatic pressure (Fig. 1). To generate and sustain

their distinct force environments, tissues display specific

mechanical properties, including elasticity, viscosity and

friction. Importantly, the manner in which cells interact

with and respond to these dynamic forces is determined

by the physical properties of the cells and extracellular

matrix (ECM) (for recent reviews see for example4–7).

As many diseases lead to alterations in tissue function

and architecture, they almost invariably also change

the mechanical properties and forces within tissues, for

example, through fibrotic reactions8. Further, common

diseases such as atherosclerosis, arthritis, osteoporosis

and cardiomyopathies, as well as several developmental

disorders, including Hutchinson–Gilford progeria and

Duchenne muscular dystrophy, entail abnormal physiological

responses to mechanical forces9,10. This highlights

the importance of understanding biological forces and

their effects on cell state and behaviour.

The coupling of extrinsic forces to the intracellular

force­ sensing machineries is referred to as mechanotransmission.

As in any 3D structure, forces within

cells are transmitted across structures that are physically

interconnected — the ECM, adhesions, cytoskeleton and

the nucleus (Fig. 1). Thus, all cells contain structures that

respond to local forces in a multistep process that involves

initial sensing of forces through specialized proteins or

protein complexes (referred to as mechanosensing)

and subsequent activation of signalling molecules and

signal propagation (mechanotransduction), ultimately

resulting in changes in cell behaviour. The outcomes of

mechanotransduction encompass virtually all biologically

relevant aspects of cell behaviour, including cell

proliferation, survival, metabolism, fate determination,

migratory properties and morphological features.

One central mechanism by which mechanical forces

affect cell behaviour is by altering gene expression. This

Review discusses the multiple ways by which mechanical

signals are propagated and relayed into the nucleus

to regulate chromatin state and transcription, and the

specific biologically relevant outcomes of this regulation.

In particular, we focus on the crucial roles of the actin

cytoskeleton dynamics and various transcription factors

(TFs) that respond specifically to mechanical signals in

regulating transcriptional outcomes of mechanotransmission.

Finally, as much of the work in this rapidly

evolving field is cell biological, we critically evaluate the

current evidence for the (patho)physiological relevance

of this regulation.

Mechanisms of force sensing

The ability of cells to respond to extrinsic mechanical

stimuli depends on their ability to sense extrinsic forces

and to transmit them to intracellular force­ sensing and

624 | October 2022 | volume 23

www.nature.com/nrg

0123456789();:

Shear

Occurs when a fluid applies

a tangential force that pushes

one part of the cell in one

direction, and the rest of the

cell is dragged in the opposite

direction by adhesion to the

extracellular matrix or to

other cells.

Tensile stress

The action–reaction forces

acting at the cell–extracellular

matrix or cell–cell contact sites

to stretch cells and to resist

deformation. Tensile stress is

the opposite of compression.

Hydrostatic pressure

A pressure exerted by a fluid

onto a contact surface as

a result of gravity.

Elasticity

The ability of a material to

return to its original shape after

a deformation- inducing force

has been removed.

force­ generating machineries to be converted into biochemical

signals. Thus, force sensing is often associated

with force­ generating structures. Multiple cellular compartments

can be involved in mechanotransmission and

mechanotransduction, including the plasma membrane,

the nucleus and other organelles (such as mitoch­ ondria,

Golgi complex and cilia) through specific but also

partially overlapping mechanisms.

The plasma membrane is a central location for

mechanotransmission. Transmembrane receptor complexes,

such as integrin­ based cell–matrix adhesions

and cadherin­ based cell–cell adhesions, connect with

the contractile actomyosin cytoskeleton to both exert

forces on their surroundings and sense mechanical properties

or dynamic deformation of the ECM substrate or

neighbouring cells11–13 (Fig. 1). Mechanical forces applied

on these multi­ protein adhesion complexes can be converted

into biochemical signals, for example, through

mechanical unfolding of individual proteins and/or

complexes found at cell–substrate and cell–cell adhesion

sites14,15 (Fig. 2a,b). Another central mechanism of force

sensing at the plasma membrane is the stretch­ induced

activation of ion channels, such as Piezo1, Piezo2 or

TRPV (Fig. 2c). These channels are activated in response

Blood flow

Pressure

Intrinsic forces

Extrinsic forces

Glycocalyx

Cortex

Tetraspanins Tetraspanins

Shear stress

Nuclear

deformation

Actomyosin tension

Cadherins

Cell–cell

adhesions

Integrins

ECM Cell–ECM adhesions

ECM Cell-ECM adhesions

Stretch

Viscoelastic resistance

Fig. 1 | Cellular structures involved in mechanotransmission. Diagram of an epithelial

or endothelial cell subjected to extracellular (or extrinsic) forces and developing oppos-

ing intrinsic tension in response to these by contraction of the actomyosin cytoskeleton.

Multiple adhesion complexes can mediate the transmission of forces between the cell

and the surrounding extracellular matrix (ECM) or other cells, including integrin- based

focal adhesions, cadherin- based adherens junctions and tight junctions (containing the

tetraspanin proteins occludins and claudins). These receptors are connected to F- actin

by junctional adaptor proteins (in green). The nucleus is exposed to compressive and

tensile stresses resulting in deformation that triggers mechanosignalling. In endothelial

cells, shear stress results from blood flowing on the cell surface, which is transmitted

through the glycocalyx and actin cortex, and from opposing resistance provided by cell

attachment.

NAture RevIeWS | GeNeTiCs

0123456789();:

R e v i e w s

to stretch, compression, shear and possibly even hydrostatic

pressure to trigger ion­ dependent intracellular

mechanosignalling16–18. Plasma membrane structures

such as caveolae and membrane tension itself are also

involved in mechanotransmission19–21 (Fig. 2d). Activation

of these various mechanosensors initiates 'classical' intracellular

signalling cascades based on phosphorylation,

ion fluxes and other second messenger activation.

Recent evidence suggests that the nucleus is a mechanosensor

that senses its own deformation to trigger

mechanosignalling (Fig. 1). Nuclear deformation can

occur downstream of extrinsic forces as a result of force

transmission from adhesion complexes, via the cytoskeleton

and the LINC complex, to the nuclear envelope,

nuclear lamina and chromatin10,22,23 (Fig. 2d). The nuclear

lamina is composed of the intermediate filament proteins

lamin A/C and B­ type lamins in mammals and

is one of the key elements that determine the stiffness

of the nucleus, but it also participates in the regulation of

mechanosensitive TFs such as NF­ κB through mechanisms

that are poorly understood24–28. Lamins also interact

directly, and indirectly through chromatin­ binding

adaptor proteins, with specific genomic regions called

lamina­ associated domains. These are cell­ type­ specific

regions that are positioned at the nuclear periphery and

are rich in silenced heterochromatin29,30. Thus, lamins are

involved in regulating cell­ type­ specific chromatin

organization and gene expression31–33. In addition, levels

of lamin A protein scale with increasing ECM stiffness,

providing a mechanism that couples tissue stiffness

with nuclear stiffness and amplifies mechanosensitive

transcription24,28,34,35. Despite its high stiffness (in the

range 1–10 kPa36,37) compared with the rest of the cell

(for example, the cell cortex is typically 0.1 kPa38), the

nucleus also undergoes direct deformation in response

to cell compression or stretch. It is interesting to note that

while the mechanical properties of lamins govern the

elastic deformation of nuclei under large defor­ mations,

the mechanical resistance of chromatin governs elastic

deformations of the nucleus under small (<3 μm)

extensions, suggesting that both chromatin and the

nuclear lamina are relevant mechanical components

of the nucleus37. Downstream of extrinsic mechanical

force, nuclear deformation triggers nuclear membrane

stretching and subsequent Ca2+ and phospholipase signalling,

which activates migratory escape responses and

mechanoprotective changes in nuclear and chromatin

stiffness in conditions of genotoxic compression36,39–42

(Fig. 2d). Another response to sensing nuclear deformation

might involve regulation of nuclear pore distribution

and permeability to control intranuclear levels of

TFs43–45. This possibility is particularly interesting considering

the recent observation that the diameter of the

pore opening is gated by membrane tension46 (Fig. 2d).

Finally, emerging evidence suggests that other organelles,

such as the Golgi and mitochondria, display

mechanosensitive properties and respond to mechanical

forces and actomyosin contraction by changing

their structure and function, resulting in propagation

of downstream biochemical signals that ultimately

affect the activity of TFs (see Emerging concepts)47–51

.

Primary cilia have also been shown to participate in

volume 23 | October 2022 | 625

R e v i e w s

a b

Cytoplasm

F-actin

NMII

Talin

Force

Tension-dependent

protein unfolding

Vinculin

Tension-

induced

protein–

protein

interactions

Cytoplasm

α-Catenin Vinculin

β-Catenin

Force

Tension-dependent

protein unfolding

Tension-induced

protein–protein

interactions

Membrane

Integrin

Soft ECM

Soft ECM Stiff ECM

Stiff ECM

Cadherin

Neighbouring cell

Intrinsic forces

Extrinsic forces

c

d

Cytoplasm

Force

Mechanically

activated

ion currents

Nucleus

Lamina

NPC LINC

Ions

TAN lines

Cytoplasm

ECM

Piezo

Ions

Membrane

Membrane

tension

tension

Fig. 2 | Mechanotransduction of forces into biochemical information.

Cells are exposed to external forces that can be sensed and transmitted via

adhesions and the contractile cytoskeleton, which is capable of generating

intrinsic forces transmitted to the extracellular environment. a | Cells

attach to the extracellular matrix (ECM) through integrin receptor

complexes, which are connected to the contractile actomyosin

cytoskeleton via several adaptor proteins, here depicted for simplicity by

the talin protein. On soft ECM, in the absence of resisting forces and of

opposing cytoskeletal tension, talin remains in a closed conformation,

limiting the maturation of focal adhesions (left). On stiff ECM, higher

contractile forces that depend on non- muscle myosin II (NMII) activity lead

to talin unfolding and recruitment of additional proteins to focal adhesions,

depicted here by vinculin (right). Recruitment of these proteins in response

to contractile force initiates signalling within the cell. b | Interactions

between cells occur through cell–cell adhesion complexes. In the example

shown, neighbouring cells interact via cadherin receptors, whose

cytoplasmic domains connect with F- actin via catenin adaptor proteins

(left). In the presence of low cell–cell forces these adhesion complexes do

not mature further. Analogous to talin, higher forces across cell–cell

adhesions lead to unfolding of α-

catenin and recruitment of vinculin and

Ca2+

cPLA2

Arachidonic

acid

Deformation

Indirect force

transmission

to the nucleus

Direct force

transmission

to the nucleus

Caveola

Adhesion

receptor

Extracellular

tension

possibly other signalling molecules to reinforce the adhesion complexes

and enable resistance by cell- generated tension (right). c | When membrane

tension is low, the piezo channel protein remains in a closed conformation.

When membrane tension becomes elevated by direct deformation of the

lipid bilayer, or indirectly by application of forces on the ECM and/or

cytoskeleton, opening of the piezo channel protein allows inward ion

currents. d | Forces can be transmitted to the nucleus either directly

(deformation) or indirectly through the actin cytoskeleton, which is

tethered to the nuclear envelope and to the nuclear lamins (lamina) by

LINC complexes. In response, nuclear compression promotes the inward

permeability of nuclear pore complexes (NPCs) associated with

transmembrane nucleus- associated actin (TAN) lines. Nuclear membrane

tension can trigger the release of calcium ions into the cytoplasm and from

intracellular reservoirs, which synergistically induce the recruitment and

activation of the phospholipase cPLA2 at the inner nuclear membrane; the

resulting cytoplasmic increase in arachidonic acid stimulates cytoplasmic

actin dynamics. Endocytic caveolae also respond to membrane tension,

acting as a membrane reservoir to accommodate stretching and regulate

signalling pathways. Adhesion receptor indicates both cell–ECM and

cell–cell adhesions.

mechanosensing, particularly in flow sensing within the

vasculature and kidney tubules52

.

Collectively, current evidence indicates that mechanical

stress results in strain (deformation) of a broad range

of mechanosensitive structures and organelles, triggering

downstream signalling. What determines which

mechanosensory organelle and/or structure is activated

is unclear, but most likely depends on the type and magnitude

of the biological force as well as the properties

of the cell itself, including the contractile state of the

cytoskeleton and the tension of the plasma and nuclear

membranes. It could also be envisioned that multiple

626 | October 2022 | volume 23

0123456789();:

www.nature.com/nrg

a

Nucleus

G-actin

Lamina

NPC

Remodelling

complex

RNAPII

RNAPII

LINC

NMII

Transcriptional repression

H3K27me3

Chromatin compaction

Me

H3K9me3

Me

Ca2+

Ca2+

Ion channel

F-actin

Cytoplasm

Extracellular

tension

Adhesion

receptor

b

XPO6

MRTF

Nucleus

NPC

Polymeric

F-actin

NMII

IPO9

MRTF

Monomeric

G-actin

Cytoplasm

• Actomyosin

• Cell–ECM adhesion

• Reinforcement

Focal adhesions Focal adhesion

• Stiff ECM

• Stiff ECM

• Cell geometry

• Cell geometry

• Stretch

• Stretch

c

Y/T

ARID1A

ARID1A

Stress fibres

RAP2

Deformation

Intrinsic forces

Extrinsic forces

MRTF

SRF

NPC

Nucleus

NMII

NPC

Y/T

TEAD

Y/T

LATS1/2

P

P P

Y/T

Cytoplasm

Focal adhesions Focal adhesion

• Stiff ECM

• Stiff ECM

• Cell geometry

• Cell geometry

• Stretch/compression

• Stretch/compression

• Shear stress

• Shear stress

NAture RevIeWS | GeNeTiCs

0123456789();:

R e v i e w s

Fig. 3 | General mechanisms by which mechanical forces

can regulate gene expression. a | Forces can influence

chromatin structural organization, epigenetic modifica-

tions and gene transcription. Peripheral chromatin is physi-

cally connected, via the nuclear lamins (lamina) and LINC

complexes, to the cytoplasmic actomyosin cytoskeleton.

Several chromatin and DNA- associated factors interact

with, and are regulated by, nuclear globular actin (G- actin)

and filamentous actin (F- actin). Nuclear and cytoplasmic

actin dynamics are connected by the rate- limiting trans-

port of G- actin across the nuclear pore complex (NPC),

and increased cytoplasmic actin polymerization in response

to mechanical strain can decrease nuclear G- actin levels to

repress transcription, which subsequently permits histone

H3 trimethylation at lysine 27 (H3K27me3) to promote

gene silencing. Tension- dependent release of calcium

within the cell can also influence such processes, increas-

ing or decreasing H3K9me3 and chromatin compaction,

depending on the cellular context. b | Binding of mono-

meric globular actin to the MRTF coactivator protein

precludes the interaction of MRTF with the DNA- binding

transcription factor SRF. When cells are subjected to forces

that induce cytoskeletal tension and F- actin polymerization,

the decrease in G- actin results in increased MRTF nuclear

entry and activation of SRF–MRTF- dependent gene

expression. G- actin is dynamically shuttled into and out

of the nucleus by importin 9 (IPO9) and exportin 6 (XPO6)

transporters, respectively, and can therefore act on MRTF

both in the cytoplasm and in the nucleus. c | Actomyosin

tension regulates the nuclear localization of the YAP or

TAZ (Y/T) coactivator and its association with the TEAD

family of transcription factors via multiple parallel mecha-

nisms. Force applied through focal adhesions leads to

inhibition of the RAP2 small GTPase, which relieves the

inhibitory action of the LATS1 and LATS2 kinases on cyto-

plasmic YAP/TAZ. Deformation of the nucleus in response

to adhesion forces facilitates the nuclear translocation of

YAP/TAZ through the NPC. Binding of the ARID1A protein

to nuclear F- actin titrates ARID1A away from YAP/TAZ,

facilitating the interaction between YAP/TAZ and

TEAD transcription factors. ECM, extracellular matrix;

NMII, non- muscle myosin II; RNAPII, RNA polymerase II.

mechanosensors can be activated simultaneously and

their signals integrated further downstream. In the

following sections we discuss how mechanical signals

are relayed into the nucleus, what is the nature of these

signals and how they regulate transcription to produce

physiologically meaningful responses.

Chromatin responses to nuclear forces

The organization of chromatin represents an important

layer of transcriptional regulation. Consequently,

force­ induced changes in gene expression might be

expected to be mediated by chromatin changes, including

to topology, accessibility and post­ translational

modifications (PTMs) of histones. Although our understanding

remains rudimentary, emerging data indicate

that the effect of force on chromatin is time, magnitude

and cell type dependent.

Remodelling chromatin through force application. The

association of chromatin with the nuclear lamina provides

it with a direct mechanical link all the way from

the extracellular environment through adhesion complexes,

the cytoskeleton and the LINC complex (Fig. 3a).

volume 23 | October 2022 | 627

R e v i e w s

Viscosity

The ability of a fluid to resist

gradual deformation. Most

biological materials are

considered viscoelastic, that is,

they display properties of both

elastic and viscous materials.

Viscoelastic materials have a

strain (deformation) rate that

is dependent on time and

they dissipate energy when a

load is applied and removed

whereas a purely elastic

material does not.

Friction

A resistance that a surface

encounters when it is moving

over another surface, for

example, in joints, tendons,

eye or skin.

LINC complex

A multi- protein complex that

crosses the nuclear envelope

and provides a physical link

between cytoplasmic and

nuclear cytoskeletal structures.

Nuclear lamina

Structure between the

inner nuclear membrane

and peripheral chromatin,

composed mainly of

intermediate filament

proteins, the lamins, and

lamin- associated proteins.

Epidermal differentiation

complex

(EDC). A gene complex of

>50 genes that encode proteins

involved in terminal differentia-

tion and cornification of skin

epidermal keratinocytes.

Indeed, mechanical force transmitted from cell adhesions

to the nucleus has been shown to result in immediate

'stretching' of chromatin within seconds of force

application, which correlates with activation of mechanosensitive

gene expression when this force is applied

continuously for 1 h53,54. How this specificity of forcemediated

activation of only certain genes is achieved

remains a key open question. One possibility is that it

depends on the positioning of the genes. For example,

the epidermal differentiation complex (EDC) gene cluster

is inactive in undifferentiated epidermal stem cells,

in which the locus resides in close proximity to the

lamina. Upon stem cell differentiation, the locus has been

shown to relocate more centrally within the nucleus, correlating

with the activation of EDC gene expression55

.

Application of tension on the lamina through the LINC

complex is crucial for maintaining the compaction of

chromatin and gene silencing at this locus, preventing

expression of differentiation genes. Consequently, reduction

in tension through genetic deletion of the LINC

complex in mice leads to precocious terminal differentiation

of epidermal stem cells56. On longer time scales

of minutes, both cell­ extrinsic (dynamic tensile stretch;

10% increase in the substrate area) and cell­ intrinsic

(elevation of cellular contractility) mechanical stimulation

triggers active, biochemical signalling­ dependent

condensation of euchromatin into heterochromatin

in cultured human mesenchymal stem cells, which is

associated with increased expression of differentiation

genes such as TGFB. Whether the chromatin remodelling

directly contributes to the alteration of gene expression

remains to be demonstrated57,58. By contrast, within

30 min of applying 5–40% cyclic uniaxial tensile stretch

to cultured human epidermal stem cells, decreased levels

of histone H3 trimethylation at lysine 9 (H3K9me3)

were observed, but, although this PTM is associated

with condensed heterochromatin and gene repression,

no substantial effects on expression of protein­ coding

genes was detected. Instead, the decrease in stiff, laminaproximal

heterochromatin contributes to nuclear softening,

which is required to dissipate mechanical energy

to prevent DNA damage36. If this mechanical stretch

stimulus persists for several hours, cells align perpendicular

to the direction of stretch, thereby minimizing

strain on the nucleus and allowing the cells to restore

steady­ state chromatin architecture36. Interestingly, if the

stretch is biaxial and thus cells are not able to realign to

avoid strain, the regions that have lost H3K9me3 will,

in the scale of days, gain histone H3 trimethylation at

lysine 27 (H3K27me3), another silencing mark, most

likely as a compensatory mechanism to ensure proper

silencing of these regions59. The application of longterm

biaxial stretch at similar amplitudes to those that

alter H3K27me3 and H3K9me3 will also deplete free

nuclear G­ actin to trigger transcriptional repression

via accumulation of H3K27me3 (Fig. 3a). Although the

transcriptional repression affects a broad range of genes,

the accumulation of H3K27me3 seems to occur particularly

at promoters of genes that have been implicated as

specific targets of the Polycomb repressive complex 2,

the methyltransferase complex responsible for catalysing

H3K27me3, which is expressed at low levels in the

628 | October 2022 | volume 23

0123456789();:

stem cells59,60. As this group of genes includes epidermal

differentiation genes, tensile stretch thus prevents epidermal

stem cell differentiation59, consistent with the

observed differentiation­ suppressive effects of nuclear

envelope tension mediated by the LINC complex56. Thus,

the potential of nuclear strain to trigger heterochromatin

changes seems to be cell­ type and cell­ state specific

and at least partly determined by the steady­ state nuclear

stiffness36,57,61. For example, human cancer cell lines

with low lamin A — and thus low nuclear stiffness and

membrane tension — are refractory to force­ induced

depletion of H3K9me3 (and in some cases even showed

increases in H3K9me3 in response to force), but could be

rendered mechanosensitive by overexpressing lamin A36

.

Cell and tissue geometry influence nuclear architec-

ture. The notion that histone modifications respond

to changes in mechanical force is further supported by

studies using extrinsic constraints on cell morphology,

which directly correlates with distribution of tension

within cells. Forcing cultured human mammary epithelial

cells into rounded shapes using adhesive micropatterns

results in global histone deacetylation, chromatin

condensation and overall reduction in gene expression62

.

Similarly, forcing human mesenchymal stem cells into

elongated shapes leads to increased histone deacetylase

activity and subsequent globally decreased histone

acetylation, as determined by immunofluorescence63

,

whereas increasing the cell spread area of fibroblasts

triggers increased histone acetylation and changes

in expression of cytoskeletal genes through effects on

actomyosin contractility64. Moreover, studies in murine

melanoma cells show that tissue curvature, most likely

through its effect on increasing mechanical stress,

promotes increased deposition of histone H4 methylation

at lysine 4 (H3K4me2) and histone H3 acetylation

at lysine 9 (H3K9ac) to increase expression of proon­

cogenic genes65. Intriguingly, chromatin architecture

can feed back to regulate the nuclear mechanical state as

it has been shown that haploinsufficiency of the chromatin

modifier MLL4–COMPASS complex in Kabuki

syndrome triggers increased levels of H3K27me3 and

Polycomb complex clustering, leading to increased

nuclear stiffness66

.

The mechanisms by which stretch, compression and

changes in cell and nuclear shape drive epigenetic

and transcriptional changes are still unclear but some

mechanistic insights are beginning to emerge. As changes

in cell and nuclear shape trigger stretch­ induced ion

channels, intracellular Ca2+ signalling has been implicated

in heterochromatin regulation downstream of

mechanical deformation36,37,61 (Fig. 3a). Interestingly, formation

of the perinuclear actin ring that mediates nuclear

actin levels is also driven by elevation of intracellular

Ca2+, linking nuclear strain with actin­ driven effects on

transcription36,59,67,68. In addition, actomyosin contractility

and subsequent local tension on the nuclear envelope is

likely to play a part in tension­ mediated effects on chromatin

through modulation of stress transmission and

thus nuclear deformability, as illustrated by the effects of

manipulating LINC complex or lamin A on chromatin

and gene expression in response to force28,56,69. Whether

www.nature.com/nrg

these effects of local tension are driven by nuclear strain,

thus converging with the above­ described mechanisms

involving nuclear deformation, remains an intriguing

open question. Another alternative or parallel layer of

regulation could involve more local nuclear deformations,

driven by the perinuclear cytoskeleton and leading

to highly localized nuclear invaginations and intranuclear

polarization70. The amount of perinuclear actin­ rich

invagination correlates with the degree of dedifferentiation

in various cell types71. Changes in gene expression

during early differentiation of human haematopoietic

stem cells have recently been shown to be caused by

compressive forces from microtubules that trigger lobular

nuclear shapes and local loss of H3K9me3 heterochromatin

from within these nuclear envelope invaginations72

.

Similarly, cytomegalovirus has been shown to use positioning

of the microtubule organizing centre to regulate

the nuclear lamina and intranuclear polarization to spatially

segregate viral DNA from compact chromatin of the

host DNA, thus maximizing virus replication73

.

Collectively, these studies emphasize the intimate,

functionally relevant connection between cell shape and

nuclear shape changes and 3D chromatin organization

and epigenetic state in regulating gene expression. The

key question is how these changes interface with biochemical

signals to provide specificity while preventing

uncontrolled changes in activity that could result from

the constant exposure of cells to mechanical force. This

crosstalk between mechanical and biochemical signals

and their physiological relevance will be the focus of the

rest of this Review.

Actin dynamics in transcription

Most of the mechanisms for sensing and responding to

mechanical forces described above converge on actomyosin

dynamics, either through direct regulation of

the cytoskeleton or by impacting signalling pathways

that influence actin dynamics. In addition to the role

of the contractile actomyosin cytoskeleton in regulating

nuclear deformation as discussed above, there are three

well­ described mechanisms by which actin can regulate

transcription downstream of mechanical signals:

regulation of the core transcriptional machinery and

chromatin modifiers by nuclear actin; regulation of the

SRF–MRTF signalling pathway; and regulation of YAP/

TAZ mechanosensitive transcriptional coactivators that

respond to changes in actomyosin.

Nuclear actin in mechanical regulation of core tran-

scription. After initial controversy, it has become well

accepted that actin is also found in the nucleus, both as

monomeric G­ actin and as filamentous F­ actin74. Actin is

transported through the nuclear pore as a monomer,

and the availability of monomers is rate­ limiting for the

transport in both directions75. Thus, any mechanochemical

signalling process that affects actin dynamics and thus

the ratio of free G­ actin to bound filamentous F­ actin in

the cytoplasm, including processes such as cell spreading,

is likely to influence nuclear actin levels76 (Fig. 3a).

Nuclear actin has multiple roles in regulation of

transcription regulation and initiation, chromatin reorganization

and DNA repair (for a recent comprehensive

NAture RevIeWS | GeNeTiCs

0123456789();:

R e v i e w s

review see, for example, rEF.

74). Nuclear G­ actin levels

positively correlate with global transcription rates59,77–81

.

Although the precise molecular mechanism or mechanisms

still need to be worked out, interactome studies

have revealed association of actin with several proteins

involved in transcription, such as components

of the pre­ initiation complex, pre­ mRNA splicing

and processing factors, and transcription elongation

factors78,80,82 (Fig. 3a). For example, in Drosophila mela-

nogaster oocytes, actin associates with RNA polymerase

II (RNAPII) on gene bodies of actively transcribed

genes81, but a specific association between actin and

chromatin in mammalian cells remains to be demonstrated.

Nuclear actin levels are dynamically regulated by

force, with decreased nuclear actin observed in response

to substrate stretching at 10% amplitude. Here, consistent

with the transcription­ enhancing function of actin,

the reduction in nuclear G­ actin as it is used for formation

of a tight perinuclear F­ actin ring in stretched

epidermal stem cells leads to a decrease in RNAPII transcriptional

elongation, with subsequent accumulation of

H3K27me3 at promoters silencing epidermal differentiation

genes59. Signals from a specific ECM substrate,

such as laminin 111 (trimer composed of LAMA1,

LAMB1 and LAMC1) can also reduce nuclear actin levels,

suppressing transcription and promoting quiescence

in mouse mammary epithelial cells83. This signalling

axis is disrupted in human breast cancer cells, resulting

in continuous proliferation77. Whether this laminin

111­ dependent effect is of mechanical or biochemical

nature remains to be investigated.

G­ actin has also been identified as a structural

component and allosteric regulator of several chromatin

remodelling complexes, including INO80, SWI/

SNF and TIP60/NuA4 (rEFS84,85) (Fig. 3a). Chromatin

remodelling complexes control chromatin accessibility

for replication, transcription and DNA damage

repair. Functionally, the G­ actin­ containing structural

modu­ les participate in allosteric control of the motor

subunit of the complexes. Indeed, the ATPase subunit

of the BAF complex has reduced chromatin association

and activity in mouse embryonic fibroblasts lacking

functional β­ actin, which also display defects in gene

expression86–88. Curiously, BAF complexes also contain

an additional actin­ like subunit, ACTL6A (also known

as BAF53), that has no known cytoskeletal function,

suggesting the evolution of a dedicated factor that

might have retained only the transcription­ associated

functions of actin in the context of BAF chromatin

remodellers. Whether these functions of actin and/or of

ACTL6A are affected by dynamic regulation of nuclear

actin, and thus could potentially be mechanosensitive,

remains an important open question.

Besides monomeric G­ actin, central regulators of

F­ actin dynamics and polymerization such as the wave

regulatory complex (WRC) and ARP2/3 complexes, as

well as certain myosin motor proteins, have also been

linked to transcription or transcription­ related processes

through their ability to associate with and regulate

RNAPII, indicating a role for nuclear F­ actin in

transcriptional regulation89–94. Such F­ actin­ mediated

transcriptional processes include reactivation of the

volume 23 | October 2022 | 629

R e v i e w s

Mechanoadaptation

The process by which a cell

subjected to mechanical forces

reinforces its force- bearing

structures.

otherwise silenced Oct4 pluripotency gene during

oocyte­ mediated nuclear reprogramming of somatic

nuclei95–97

.

Taken together, these findings indicate that dynamic

communication between the cytoplasmic and nuclear

actin pools can transmit information from the extracellular

environment into the nucleus (Fig. 3a). Further

studies are needed to establish the biochemical basis of

actin­ mediated transcriptional regulation and the role

of actin dynamics in functional regulation of chromatin

remodelling complexes. Advances in this field require

development of tools to better dissect the causative

relationships between co­ occurring regulation of the

transcriptional machinery, TF activity and chromatin

remodelling in response to changes in actomyosin

dynamics.

SRF–MRTF transcriptional feedback on adhesion

and actomyosin. The second main mechanism by

which actin regulates gene expression is by controlling

the assembly of SRF–MRTF TF complexes (Fig. 3b;

TabLE 1), which mediate gene activation by facilitating

either RNAPII recruitment or promoter escape of preloaded

RNAPII, depending on the genomic context98–103

.

G­ actin directly binds to MRTFs and maintains them in

an inhibited state in the cytoplasm, so that upon RHO

activation and cytoplasmic F­ actin polymerization

(and a subsequent drop in the levels of free G­ actin)

MRTFs are released to translocate to the nucleus and

to bind SRF104. Evidence also supports a specific role

for the nuclear F­ actin pool in this regulation76,105,106

.

Upon higher demand for F­ actin (for example, owing

to increased motility or in response to mechanical

stress), the SRF–MRTF system is activated and promotes

expression of ECM, adhesion and cytoskeletal

genes to facilitate a feedback mechanoadaptation107–113

.

This central role of SRF in regulating transcription

of cytoskeletal and ECM components and mechanoadaptation

is also important for epidermal homeostasis

and skin barrier function114–118, and in endothelial

cells to promote vessel growth and maturation119–121

.

Finally, SRF can also associate with TCFs in response

to ERK signalling, representing an additional mechanism

by which mechanical control of cell proliferation

and differentiation occurs122–125. TCFs themselves may

also have a role in the response to mechanical forces, as

indicated by the mechanosensitive induction of the classical

immediate­ early gene EGR1, which is a target of

TCF126–128. The SRF–MRTF system thus relays changes

in extracellular forces and cytoskeletal actin dynamics

into the nucleus (Fig. 3b).

The YAP/TAZ coactivators as a mechanics- to- biology

transduction module. Changes in the actomyosin

cytoskeleton also modulate the activity of the ubiquitously

expressed paralogous factors YAP and TAZ (also

known as WWTR1), which have a central role in regulating

transcription downstream of mechanical force generated

by cell geometry, ECM stiffness, stretching and

shear stress, among others129,130 (Fig. 3c). YAP/TAZ (used

here to indicate either YAP1 or TAZ) dynamically shuttle

between the cytoplasm and the nucleus in response

to multiple inputs including mechanical force and the

Hippo pathway, a kinase cascade that controls tissue

and organ size across animals. In the nucleus, YAP and

TAZ act as transcriptional coactivators and bind the

TEAD family of TFs (TabLE 1). Moreover, YAP–TEAD

complexes can interact with other TF complexes that

bind DNA in the vicinity of TEADs; complexes including

AP-1, SRF–MRTF, E2F and MYC cooperate with

YAP–TEAD, whereas TRPS1 dampens YAP–TEADregulated

chromatin remodelling and transcription131

.

Table 1 | The main transcription factors regulated in response to mechanical stimuli

Transcription factor DNA binding motifa Mechanical input

AP-1 (Jun–Fos dimers) 5′

-

TGA G/C TCA-3′ β-

Catenin–LEF/TCF 5′

-

AGATCAAAGG-3′ Tissue compression

MEF2C (driving expression of KLF2) NF- κB (homo- and heterodimers

of the NF- κB and Rel proteins)

Shear stress (disturbed flow); ECM stiffness

5′-(C/T)TA(A/T)4TA(G/A)-3′ Shear stress (laminar flow)

5′

-

GGGRNYYCC-3′ (where R is

a purine, Y is a pyrimidine, N any

nucleotide)

Shear stress (disturbed flow); ECM stiffness

NICD–RBPJ 5′

-

GTGGGGAA-3′ Cell–cell tugging forces; shear stress

(laminar flow)

NRF2 5′

-

TGACtcaGCa-3′ antioxidant

response elements

Shear stress (laminar flow); ECM stiffness

SMAD2–SMAD3–SMAD4 5′

-

GTCTAGAC-3′ (SMAD binding

element motif), 5′

-

CCAGACA-3′

(CAGA motif)

Traction force- mediated unfastening

of TGFβ from 'ECM traps'; shear stresses

(laminar flow)

SMAD1–SMAD5–SMAD4 5′

-

GGC/AGCC-3′ (GC- rich SBE)

in the vicinity of a 5′

-

AGAC-3′ (SBE)

Shear stress (laminar flow)

SRF–MRTF 5′

-

CC(A/T)6GG-3′ ECM stiffness; cell geometry; cell stretching

TEAD–YAP and TEAD–TAZ 5′-(G/A)CATTCC(A/T)-3′ ECM stiffness; cell geometry; cell crowding;

cell stretching; cell compression; shear

stresses (disturbed flow)

ECM, extracellular matrix; SBE, SMAD binding element. aDNA binding motifs are based on a consensus from the literature and do not

take into account alternative binding sites identified through chromatin- immunoprecipitation experiments.

630 | October 2022 | volume 23

www.nature.com/nrg

0123456789();:

Mediator complex

A multisubunit protein

complex that bridges

transcription factors and the

basal RNA polymerase ii

transcriptional machinery.

Contact inhibition of growth

The process by which

cell crowding inhibits cell

proliferation through the

establishment of cell–cell

contacts and the reduction

of cell size.

Cell competition

The active elimination of a

viable but undesirable cell

population by competitive

interactions within a tissue.

YAP–TEAD complexes are found preferentially at distal

superenhancer elements, and regulate transcription

by recruiting the Mediator complex and BRD4 epigenetic

coactivators to drive RNAPII activity at proximal

promoters131

.

The 'mechanosensitive' control of YAP/TAZ is

an integral part of the mechanisms that coordinate

local cell growth with global tissue size, such as contact

inhibition of growth132–134

, cell competition within epithelia135–142,

the recently proposed 'lateral inhibition'

or 'leader selection' process143,144, and compensatory

tissue growth in response to mechanical expansion114

.

YAP/TAZ respond to F­ actin, rather than to G­ actin,

in a manner that is largely independent of the type

of adhesive ECM ligands and integrins involved, but

dependent on the mechanosensitive Talin 1 and Talin 2

proteins129,130,145,146. Various mechanisms that control

YAP/TAZ activity in response to mechanical forces

have been proposed, and they likely act in parallel131

.

A decrease in forces leads to activation of the RAP2A,

RAP2B and RAP2C small GTPases, which activate the

LATS1 and LATS2 kinases and thus promote YAP and

TAZ phosphorylation and inactivation147. Decreased

forces also set free the ARID1A protein from nuclear

F­ actin, so that ARID1A can bind YAP and TAZ in the

nucleus and prevent their interaction with TEADs148

.

Finally, decreased forces favour the acquisition of a more

spherical nuclear shape, which inhibits general nuclear

pore permeability, including YAP and TAZ nuclear

import43–45,149 (Fig. 3c). Collectively these studies indicate

that YAP/TAZ are substantially nuclear when ECM

elasticity is above a Young's elastic modulus of 10 kPa,

and largely cytoplasmic when this falls below 1 kPa. It

must be noted that this range should not be considered

absolute for any mechanoresponse, as, for example,

skeletal muscle progenitors self­ renew and differentiate

optimally on 12 kPa substrates, resembling the physiological

elastic modulus of the muscle, but not at higher or

lower stiffness150–153. Moreover, ECM is not purely elastic

but also displays a viscous component, whose effect on

YAP/TAZ is only starting to be understood154–156. The

definition of an absolute stiffness threshold for mechanoresponses

is also not meaningful as cells not only passively

respond to extracellular forces, but can actively

tune the cytoskeletal response to extracellular forces and

the ensuing YAP/TAZ activation. Key factors involved in

this tuning process are the F­ actin capping and severing

proteins CAPZA, CAPZB, CFL1, CFL2 and GSN, the

focal adhesion components KRIT1 and PDCD10 (also

known as cerebral cavernous malformation 1 and 3),

SHARPIN and the F­ actin bundling protein FSCN1

(rEFS132,157–162). Finally, it must also be considered that

most cells actively remodel the ECM, which may change

the temporal dynamics of local stiffness sensing7

.

Similarly to SRF–MRTF, YAP/TAZ signalling also

feeds back into the expression of cytoskeletal genes,

facilitating mechanoadaptation163–166. However, YAP/

TAZ differ from the SRF–MRTF system as they translate

mechanical inputs into broader biological responses.

Indeed, YAP/TAZ communicate the degree of extracellular

forces, and the corresponding degree of intracellular

tension, to the nucleus to regulate proliferation,

NAture RevIeWS | GeNeTiCs

0123456789();:

R e v i e w s

apoptosis, differentiation and their associated metabolic

pathways129,167. Thus, the relevance of the mechanically

regulated YAP/TAZ transcriptional response seems

to be widespread and spans many organ systems in

physiological and pathological scenarios.

In addition to SRF–MRTF and YAP/TAZ, several

other TFs have been shown in certain circumstances

to display mechanosensitivity (TabLE 1). However, the

physiological relevance of these pathways seems to be

more specific or remains to be demonstrated. Below, we

discuss the most prominent examples of physiologically

relevant mechanosensitive transcriptional pathways for

which in vivo genetic evidence is available, or which

operate in tissues or organs that have been shown to

display physiologically relevant mechanoresponses.

Mechanosensitive TFs in development

During development, cell fate needs to be tightly coordinated

with cell morphology and position. In addition,

morphogenetic movements generate dynamic stretching

and compression forces as well as changes in tissue

curvature1–3. Thus, it has long been speculated that

changes in forces and cell morphologies could impact

TF activity to provide feedback control of development.

Although the direct link between mechanical forces and

the transcriptional response downstream of the TFs still

needs to be demonstrated in most cases, gene knockout

studies of mechanosensitive TFs have revealed essential

roles for these pathways in development.

Tissue deformation, cell–cell tension and the regulation

of β- catenin. Several in vivo studies demonstrate the

role of mechanical regulation of the WNT/β­ catenin

pathway in coupling morphogenetic movements and

transcriptional regulation. In D. melanogaster embryos,

morphogenetic movements that elongate the germband

tissue compress the neighbouring cells of the developing

foregut, leading to higher expression of the Twist differentiation

gene in these cells168 (Fig. 4a). Build­ up of tissue

tension in response to deformation activates Src, leading

to phosphorylation of β­ catenin. This phosphorylation

releases β­ catenin from cell–cell adhesions, where it has

a structural role, enabling it to enter the nucleus, where it

acts as a transcriptional coactivator of TCF/LEF to drive

Twist expression169,170. Given that the pool of β­ catenin

that is not engaged in cell–cell adhesions is known to be

immediately degraded by the APC destruction complex,

the observed nuclear entry suggests that compression

releases enough β­ catenin to overcome degradation

in these cells, or that phosphorylation protects it from

degradation. Indeed, nuclear β­ catenin accumulation

generally occurs in response to inhibition of APC by

the WNT signalling pathway171. The coactivator function

of β­ catenin depends on recruitment of multiple

transcriptional and chromatin regulatory factors172,173

.

Tension­ mediated regulation of β­ catenin also occurs in

vertebrate embryos during mesoderm induction as well

as in the mouse intestinal epithelium174,175, and has been

observed in the skin of mice in which cell contractility

has been artificially elevated by overexpression of active

Rho kinase (Rock)176. In human pluripotent stem cells, a

soft ECM promotes the formation of cell–cell junctions

volume 23 | October 2022 | 631

R e v i e w s

a

Stomodeum

compression

that function as a 'reservoir' for WNT­ induced β­ catenin

activation, so that upon localized high cell–cell tension

β­ catenin can be locally activated to drive mesoderm

Drosophila

embryo

Germband extension

b

Intrinsic forces

Extrinsic forces

Adherens

junction

specification177,178. The precise mechanical input that

drives such activation remains elusive, not least because

of conflicting studies that show that, instead of being

Granulosa

Compression

Expansion

TAZ

Chorion

Oocyte

Micropylar

cell

O

S

H H

N

HO

Latrunculin B

O

iPSC

O

O

Endoderm

PDX1+

pancreatic

progenitor

+ Latrunculin B

P

βCAT

TCF

βCAT

SRC

TWIST

• Morphogenetic

movements

• Tissue folding

Cell fate

c d

Proliferation

Retinoic acid

Y/T

Y/T

Y/T

Y/T

Y/T

NGN3

Expansion

Y/T

Contraction

Y/T

NGN3

– Latrunculin B

Y/T

β-Cell

Lung bud

epithelium

Stiff TC plastics Stiff TC plastics

SMC

Pancreas Cell replacement therapy to treat diabetes

Fig. 4 | Mechanosensitive transcription factors in development and dis-

ease. a | Deformation of Drosophila melanogaster embryonic tissues, which

occurs as a result of morphogenetic movements (germband extension), rein-

forces expression of the Twist transcription factor in mesoderm cells (such as

the stomodeum, indicated by blue nuclei), coordinating cell fate specifica-

tion with morphogenesis. Cell compression induces phosphorylation (P) of

β-

catenin (βCAT) by the SRC kinase, which favours the liberation of β-

catenin

from the junctional pool and its accumulation in the nucleus, where it binds

to TCF transcription factors to promote Twist expression. b | Mechanical

competition for limited space in the zebrafish egg granulosa cells limits the

activation of TAZ to a single cell at the animal pole of the oocyte. TAZ pro-

motes differentiation of the micropylar cell, which forms the micropyle chan-

nel in the chorion through which the sperm enters to fertilize the egg.

c | Budding morphogenesis of the lung primordia is the result of two oppos-

ing forces: the expansion of the growing lung epithelium; and the localized

contraction of the bud by smooth muscle cells (SMCs). These two forces are

coordinated by YAP/TAZ (Y/T), which promotes epithelial proliferation (cre-

ating the expansion forces) and the release of retinoic acid, which promotes

the localized differentiation of SMCs (leading to contraction). d | The in vitro

production of functional pancreatic endocrine β-

cells is a key objective for

cell- based replacement therapies. Terminal differentiation of human induced

pluripotent stem cells (iPSCs) into insulin- secreting β-

cells is hampered by

culture on stiff tissue culture (TC) plastics. Disabling YAP/TAZ mechanotrans-

duction in pancreatic progenitors with the F- actin inhibitor latrunculin B

enables expression of neurogenin 3 (NGN3) and efficient terminal

differentiation. In part d, Y/T indicates YAP–TEAD or TAZ–TEAD complexes.

632 | October 2022 | volume 23

0123456789();:

www.nature.com/nrg

Trophectoderm

The tissue of the pre-

implantation mammalian

embryo that will contribute

to formation of the placenta.

Inner cell mass

The tissue of the pre-

implantation mammalian

embryo that will contribute

to formation of the tissues

of the fetus.

Granulosa

The somatic cells of the female

gonad that surround oocytes

and support their growth.

Branching morphogenesis

The developmental process

by which a growing epithelium

buds branches in the

surrounding mesenchyme

to form a tree- like structure.

Naive pluripotent state

Pluripotent stem cells

equivalent to those found in

the early inner cell mass, with

largely unmethylated genome,

and still able to differentiate

into germ cells.

Regenerative medicine

The clinical use of stem cells

to stimulate repair mechanisms

and restore function in

damaged body tissues or

organs.

Laminar flow

When particles in a moving

liquid follow linear paths

without lateral mixing.

released in response to tension, β­ catenin is instead

recruited to stabilize the junctions179. Conversely,

mechanical tension at endothelial cell–cell contacts

can unmask a phosphorylation site on VE­ cadherin

that is normally masked by β­ catenin binding, enabling

increased turnover of cell–cell adhesions179. Thus, a

possible reconciliation of these different findings is that

the ability of cell–cell tension to release and activate

β­ catenin depends on cell­ type­ specific cell–cell

adhesion dynamics180: in cells with more dynamic

and unstable adhesions, such as endothelial and perhaps

embryonic cells, tension leads to adhesion disassembly

and β­ catenin release181, whereas in cells

with stable adhesions such as epithelial cells, tension

retains β­ catenin at junctions179, resulting in opposing

transcriptional outcomes.

YAP/TAZ mechanotransduction in development. YAP/

TAZ regulation by cell or tissue mechanics is relevant

in several embryonic contexts. One example is the

mouse blastocyst, in which activation of YAP/TAZ in

response to Rock activity and tissue tension promotes

differentiation of the trophectoderm or trophoblast

stem cell fate, whereas their activity is inhibited in the

pluripotent inner cell mass by the Hippo pathway182–187

.

Indeed, whereas development of Yap1­ deficient mouse

embryos arrests around embryonic day 8.5 owing to

severe embryonic and extra­ embryonic defects188, and

Taz­ deficient embryos are viable but develop multicystic

kidney disease during development and rarely

survive to adulthood189–191, embryos that lack both

proteins die at the morula stage, indicating that YAP

and TAZ are redundant but collectively essential for

very early embryogenesis185. Another example of YAP/

TAZ­ mediated mechanotransduction occurs in the

zebrafish granulosa cells surrounding the oocyte, in

which mechanical competition enables a single cell to

maintain TAZ activity and to differentiate into the single

micropylar cell, which is key for fertilization144,192

(Fig. 4b). Finally, recent findings indicate that lung

branching morphogenesis depends on the mechanical

equilibrium between transmural pressure in the growing

buds and localized contraction by surrounding smooth

muscle cells, which is reliant on YAP/TAZ activation in

epithelial cells and their retinoic acid­ mediated crosstalk

with differentiating mesenchymal cells193,194 (Fig. 4c).

Exploiting mechanoresponsive transcription for regener-

ative medicine. In addition to crucial roles in morphogenesis,

the mechanics of the cell microenvironment

have profound effects on ex vivo stem cell amplification

and in vivo engraftment152,153,195,196. In the case of intestinal

organoids or intestinal injury in mice, a stiff ECM

promotes stem cell expansion and regenerative stem

cell reprogramming through YAP/TAZ197,198. Another

example is the mechanochemical regulation of mesenchymal

stem cell differentiation on stiff substrata via

YAP/TAZ, and other mechanoregulated TFs such as

RARG, and SREBP1/SREBP2 (rEFS28,35,130,151,162,199–201).

This signalling axis is likely of physiological relevance,

as modulation of ECM rigidity or actomyosin contractility

alter mesenchymal stem cell differentiation

NAture RevIeWS | GeNeTiCs

0123456789();:

R e v i e w s

trajectories also in vivo202,203. Other studies in mice have

indicated that YAP/TAZ are important in the context of

'emergency' stem cell activation for tissue regeneration

after damage, rather than for adult homeostatic stem

cell renewal204–206. Perhaps reflecting this idea, YAP

activation can favour reprogramming of differentiated

cells to the naive pluripotent state, but its requirement

for human pluripotent stem cell self­ renewal is rather

limited165,207,208. These findings have triggered interest

in exploring the druggability of the ECM–YAP/TAZ

transcription axis, and some studies have used actomyosin

inhibitory drugs to shut down unwanted YAP/TAZ

activity to enable the in vitro differentiation of pancreatic

β­ cells209–212 (Fig. 4d). Conversely, the availability of

small­ molecule compounds to activate YAP/TAZ213–215

may enable a more efficient expansion of stem cells from

patients affected by genetic deficiencies in cell–ECM

adhesion receptors216–218. Thus, the regulation of ECM

mechanotransduction and the associated transcriptional

responses bears great potential in the context of

regenerative medicine, although the ability to target such

a pleiotropic mechanism to drug specific mechanotransduction

processes while avoiding unwanted side effects

remains a key challenge in the field.

Mechanotranscription in the vasculature

The physiological relevance of force­ mediated transcriptional

control is probably best understood in the

cardiovascular system, which thus serves as an excellent

paradigm to highlight the in vivo evidence for mechanosignalling.

Tangential shear forces associated with

blood flow and sensed by endothelial cells are major

determinants of vascular morphogenesis during development,

and of vascular remodelling during adult life.

Moreover, variations in the magnitude and pattern of

blood flow can contribute to inflammatory responses

and to disease. Flow acting on the apical surface of

endothelial cells is transmitted through the membrane

and cortical cytoskeleton to cell–cell junctions, where

tension is sensed by a complex between VE­ cadherin,

PECAM1 and VEGFR2, as well as to the basal surface,

where tension is sensed at integrin attachment sites219,220

.

As with other mechanical inputs, these responses entail

both morphological and cytoskeletal rearrangements

and the modulation of TF activity9,221 (Fig. 5).

Homeostatic transcriptional responses to laminar flow.

Experimental measurements have shown that the magnitude

of fluid shear stress induced by blood flow in

the human vasculature ranges from 1 to 6 dyn cm−2

in the venous system and from 10 to 70 dyn cm−2 in

arteries, and in the mouse, for example, these values are

even higher222. High laminar flow guides angiogenesis

during development, stabilizes vessels, promotes the

alignment of endothelial cells in the direction of flow,

decreases endothelial cell turnover, suppresses inflammation

and activates antioxidant pathways, preventing

the formation of atherosclerotic plaques. At the level of

intracellular signalling, high laminar flow (in the range

20–40 dyn cm−2) promotes the activation of several TFs.

NRF2 is a TF that regulates transcription by forming

heterodimers with members of the sMaf protein family.

volume 23 | October 2022 | 633

R e v i e w s

a

PECAM1

VEGFR2

High LSS

Extrinsic forces

Notch

Apical

cortex

Basal

membrane ERK5

P

MEF2

NICD

NRF2

KLF2

P P

SM2/3

CDKs P P

P P

SM1/5

NRP1

ALK5

ENG

ALK1

• EC alignment

• Quiescence

• Antioxidants

b

Apical

cortex

Low LSS

KLF2

P P

SM2/3

NRP1

ALK5

Inward

remodelling

c Apical

cortex

NF-κB

AP-1

Y/T

Fig. 5 | Mechanosensitive transcription factors in vascular physiology

and pathological remodelling. a | Under normal conditions, high- intensity

laminar shear stress (LSS) caused by blood flowing at the surface of an

endothelial cell (EC) is transmitted via the cell cortex to cell–cell (cadherins

and PECAM1) and cell–extracellular matrix (ECM) adhesion sites. LSS

activates multiple pathways and transcription factors and modulates

activation of membrane receptors and coreceptors for the TGFβ (ALK5,

neuropilin 1 (NRP1)), BMP (ALK1, endoglin (ENG)) and Notch pathways,

which are key regulators of vascular development and homeostasis.

DSS

• Proliferation

• Inflammation

• Atherosclerosis

b | Under conditions of lowered LSS, TGFβ pathway activity is sustained,

while KLF2 expression is low, enabling activation of SMAD2–SMAD3–SMAD4

(SM2/3) transcription complexes to promote inward vessel remodelling. The

reduction in size of the vessel lumen re- establishes high LSS, which activates

KLF2 expression to shut down SM2/3. c | Disturbed shear stress (DSS)

activates a distinct set of pro- inflammatory and pro- atherosclerotic

transcription factors. Thus, branching points and arches of the vascular tree,

where the blood flow is turbulent, are predisposed to develop

atherosclerotic plaques. Y/T, YAP/TAZ.

NRF2 is known to be activated by oxidative stress and

its main transcriptional targets are protective antioxidant

genes223,224. However, NRF2 is also activated by

blood flow not only to empower endothelial cell antioxidant

metabolism, but also to suppress inflammation

(Fig. 5a). This anti­ inflammatory response depends on

the transmission of shearing forces by the glycocalyx,

entails the activation of PI3K/AKT and the production

of mitochondrial reactive oxygen species (ROS), and

is modulated by COX2 activity, which catalyses the

formation of prostaglandins225–231

.

Another TF complex activated by high laminar flow

is SMAD1–SMAD5–SMAD4 (SM1/5 in Fig. 5a). Flow

promotes the activation of the SMAD1 and SMAD5

factors downstream of the BMP type I receptor ALK1

and of its co­ receptor endoglin (ENG)232–234, and this

634 | October 2022 | volume 23

www.nature.com/nrg

0123456789();:

Sprouting angiogenesis

The growth of new capillary

vessels out of pre- existing

blood vessels.

Fibrotic response

Tissue remodelling character-

ized by the deposition of colla-

genous extracellular matrix,

which can have a physiological

function during wound healing

(scarring) or a pathological

function that can interfere with

or totally inhibit the normal

architecture and function of

the underlying organ or tissue.

Cancer- associated

fibroblasts

A population of cells, likely

deriving from the fibroblast

lineages, that are found

in tumours and have an

elongated morphology,

are negative for epithelial,

endothelial and leukocyte

markers, and lack the

mutations found in

cancer cells.

activation can be further refined by cilia­ dependent

signalling235,236. This pathway is relevant for the stabilization

of vessels under flow, which explains why mutations

in the ALK1 and ENG genes cause vessel instability and

the arteriovenous malformations observed in patients

with haemorrhagic telangiectasia232,235,236. At the same

time, activation of KLF2 prevents excessive BMP signalling,

which accounts for the protective effects of high

laminar shear stress against vascular calcifications237

.

Fluid shear stresses can also regulate gene expression

through Notch (Fig. 5a). Tension across the Delta–

Notch signalling complex at cell–cell contacts facilitates

its proteolytic cleavage and the release of the Notch

intracellular domain (NICD), which acts as a nuclear

coactivator for the RBPJ DNA­ binding factor238–243

.

This mechanism is relevant for the regional specification

of the endothelium during development, for the

crosstalk with vascular smooth muscle cells and for

the formation of the heart structures244–250. Interestingly,

activation of Notch depends on the extent of tugging

force at cell–cell contacts in epithelial cells251–254, suggesting

the interesting possibility that forces between

adjacent cells in a monolayer can regulate Notch also

in other tissues.

Also, the regulation of YAP/TAZ by mechanical

cues is important for adjusting endothelial cell

proliferation130,255,256 during the maintenance of tissue

stiffness homeostasis257 and during developmental

sprouting angiogenesis, in which YAP/TAZ are required

for the maturation of the vascular barrier and for direct

angiogenesis along a tissue stiffness gradient258–262

.

In this context, YAP/TAZ cooperate with the two antagonistic

TFs GTF2I (also known as general transcription

factor II­ I) and GATA2 to regulate the expression of the

key angiogenic VEGFR2 gene263

.

The magnitude of laminar flow is subjected to homeostatic

regulation that keeps it within a narrow range.

When arteries are subjected to a laminar flow of lower

magnitude, a homeostatic mechanism reduces lumen

diameter by inward remodelling to restore the optimal

shear stress (Fig. 5b). This mechanism is based on the

crosstalk between two pathways: at low shear stress

intensity, only the SMAD2–SMAD3–SMAD4 transcriptional

complex is activated downstream of the TGFβ

type I receptor ALK5 and of its co­ receptor neuropilin 1

(NRP1) to induce inward remodelling. As the lumen

diameter decreases and shear stress levels become elevated,

the ERK5–KLF2 system is activated as described

above, leading to inhibition of SMAD2–SMAD3 to

terminate the remodelling process264

.

Taken together, several transcriptional regulators are

activated by laminar flow to regulate vascular development

and homeostasis. A crucial next step for the field

is to understand the specific versus overlapping roles of

these pathways as well as the mechanisms of crosstalk

and cooperativity by which transcriptional specificity

in response to specific flow magnitudes is achieved.

Pathological transcriptional responses to disturbed blood

flow. Not only the magnitude but also the type (that is

disturbed flow versus laminar flow) of shear stress is a

fundamental regulator of endothelial homeostasis265,266

.

NAture RevIeWS | GeNeTiCs

0123456789();:

R e v i e w s

Disturbances in fluid flow dynamics associated with

branching or turning points of the arterial tree facilitate

the emergence of pathological phenotypes including

misalignment and proliferation of endothelial cells and

low­ level chronic inflammation. In response to high

levels of tension, several signalling mediators are activated,

including PI3K, SRC, PLC/PKC, the small GTPase

RAC1 and NOX. These in turn promote the activation of

the NF­ κB, AP-1 and YAP/TAZ TF complexes, resulting

in the promotion of pro­ inflammatory gene expression

and enhancing proliferation267–272. Over long periods and

in cooperation with other risk factors, these phenotypes

predispose such regions to formation of atherosclerotic

plaques (Fig. 5c). These observations indicate that YAP/

TAZ inhibition could be a potential therapy against

atherosclerosis, or to normalize vascular malformations

in hereditary haemorrhagic telangiectasia273,274

.

The interplay between shear stress and cell–ECM

adhesions in regulating the response to flow is complex,

because specific ECM ligands that engage their

specific integrin receptors can have opposite effects on

NF­ κB275. This may also account for the opposite regulation

of NF­ κB in response to ECM stiffness observed

in different cell types276,277. Moreover, transmission of

tension from cell–ECM attachment sites to the nucleus

can regulate NF­ κB in response to cell stretching27,278,279

.

Mechanosensitive TFs in disease

In addition to the role of disturbed flow in regulating

transcriptional responses relevant to vascular disease,

several other common diseases involve altered force

environments and mechanosensitive transcription.

A loop of mechanosensitive transcription in fibrosis. A

key paradigm of pathological mechanotransduction is

the fibrotic response. During fibrosis, a self­ sustaining

loop between fibroblast activation and proliferation and

ECM secretion and contraction is key to pathological

tissue remodelling and stiffening8. This loop was key

to discovering how cells 'read' ECM stiffness by developing

active tension to resist extracellular rigidity12,280

.

Multiple mechanosensitive TFs participate in this loop

and establish an integrated and multi­ tiered feedforward

system (Fig. 6a).

Several data indicate a role for YAP/TAZ in fibrotic

reactions in mice and humans49,163,281,282 and that disabling

YAP/TAZ mechanotransduction may be used as

a strategy to prevent scarring and to promote wound

healing283–285. This YAP/TAZ­ driven mechanical feedback

loop also maintains the activated myofibroblast

phenotype during the contraction model of alveolar

formation and lung regeneration and mediates the

ECM­ mediated crosstalk between cancer cells and

cancer-

associated fibroblasts162,163,286,287. The mechanoadaptive

function of SRF in promoting transcription

of cytoskeletal and ECM genes also has a key role during

fibrosis and is amplified in fibrotic disease. Thus,

small­ molecule modulators of SRF–MRTF can be

used to prevent tissue fibrosis or, conversely, to promote

wound healing288–301. During fibrosis, SRF–MRTF

cooperates with YAP–TEAD transcription complexes

at adjacent promoter elements108,109,302. This crosstalk

volume 23 | October 2022 | 635

R e v i e w s

a

αvβ integrin

ALK5

TGFβRII

Stiff ECM

Stiff ECM

c

is also hijacked in cancer cells to sustain malignant

growth303–307

.

Another example of the role of mechanosensitive

TFs in the fibrotic response is the activation of SMAD2–

SMAD3–SMAD4 TF complexes308: active pulling of cells

on the ECM leads to integrin complex­ mediated liberation

of TGFβ signalling molecules from 'ECM traps'309–316

and activation of SMAD2–SMAD3–SMAD4 TF complexes,

which cooperate with mechanically regulated

TGFβ LAP–LTBP

Intrinsic forces

Extrinsic forces

ECM

P

P

SM2/3

Y/T

MRTF

ZNF416

SNAIL1

• Fibroblast activation

• ECM remodelling

• Tissue fibrosis

• Scarring

Liver

• Proliferation

• Transdifferentiation

• Malignant progression

• Organ homeostasis

• Liver cell fate

• Systemic metabolism

SRF–MRTF110,111. Finally, the SNAIL1 and ZNF1416

TFs can also be activated in the context of fibrosis and

contribute to this self­ amplification loop317,318 (Fig. 6a).

YAP/TAZ in mechanical control of breast cancer pro-

gression. During breast cancer development, cancer

cells actively remodel the normal soft tissue ECM

into a stiffer microenvironment. ECM stiffness in

turn supports cancer cell malignant phenotypes,

b

Mammary

duct

Blood

vessel

Metabolic

cooperation

Y/T

MRTF

Stiff

ECM

CAF

• Proliferation

• EMT

• Invasion

Y/T

MRTF

TWIST1

Cancer cell

d

Soft ECM

Soft ECM Blood vessel

Metastatic site

Metastatic site

CAPZ

Y/T

Bundled

F-actin

AKT/NICD

oncogenes

Golgi

apparatus

Mitochondria

Retrograde

trafficking

Fission

mtROS

FSCN1

SREBP

NRF2

Antioxidants

Lipids

Tissue softness

Soft ECM

Resistance to

chemotherapy

636 | October 2022 | volume 23

www.nature.com/nrg

0123456789();:

◀

Epithelial to mesenchymal

transition

(EMT). The differentiation

process by which cells lose

epithelial identity and the

ability to form stable cell–cell

adhesions, and gain expression

of mesenchymal markers

associated with increased

migratory ability.

Cholangiocellular

transdifferentiation

The differentiation of

hepatocytes into cells that

express markers typically

found in bile duct cells and in

bipotent liver cell progenitors.

often dominating the genetic and oncogenetic makeup

of the cell to promote the loss of epithelial polarity

and epithelial to mesenchymal transition (EMT), the

acquisition of migratory and invasive behaviour, and

proliferation319–321. YAP/TAZ can promote several of

these phenotypes in cancer cells322–330 and they are relevant

in the crosstalk between ECM stiffness and oncogenes

in promoting pro­ tumorigenic phenotypes331,332

(Fig. 6b). Moreover, YAP/TAZ lie at the centre of a multilayered

feedforward loop that involves high mammographic

density and/or ECM stiffness, EMT and

microRNAs323,333–336. In line with these data, YAP/TAZ

promote breast cancer progression in mice, although

stronger in vivo genetic evidence is needed to support

a role of ECM mechanotransduction components in

mammary tumorigenesis and for their effects on YAP/

TAZ and other mechanosensitive TFs325,337–341

.

Mechanical control of liver homeostasis and cancer

through YAP/TAZ. Hepatocytes were among the first

cell types used to study mechanoresponses to ECM

stiffness342, and the liver tissue is a model system for

organ­ size regulation and tumour suppression by the

Hippo pathway343. Recent data indicate that control

of YAP/TAZ by the soft tissue mechanical properties of

the liver are key to maintaining hepatocyte proliferative,

metabolic and cell fate homeostasis, and to controlling

liver organ size159. This observation could relate to the

recent finding that homeostasis of liver sinusoids is regulated

by mechanical pressure from the blood vessels,

which is important during tissue regeneration344. Liver

tissue softness also represents a tumour­ suppressive

mechanism that can be bypassed by cell­ autonomous

and oncogene­ driven regulation of the fascin 1 actin

bundling protein, which supports YAP/TAZ mechanotransduction,

cholangiocellular transdifferentiation and

the formation of cholangiocarcinomas160 (Fig. 6c).

Fig. 6 | Mechanosensitive transcription factors in fibrosis and cancer. a | Transcription

factors (TFs) and signalling networks that control tissue fibrosis. Deposition and remodel-

ling of a collagenous extracellular matrix (ECM) activates multiple mechanosensitive TFs

that cooperate to maintain fibroblast activation and to further amplify ECM remodelling.

Increased ECM stiffness and cell contractility also facilitates the liberation of extracellular

TGFβ1, TGFβ2 and TGFβ3 ligands from ECM- bound 'traps' composed of the latency asso-

ciated peptide (LAP) and latent TGFβ binding protein (LTBP), by pulling via αvβ integrins.

Liberated TGFβ ligands can engage their receptors (ALK5) leading to activation of

pro- fibrotic SMAD2–SMAD3–SMAD4 (SM2/3) TFs. b | During breast cancer development,

cancer cells and cancer- associated fibroblasts (CAFs) cooperate to remodel and stiffen

the ECM. This maintains the activated state of CAFs, enables metabolic crosstalk between

CAFs and cancer cells and promotes malignant cancer cell behaviour by regulating mech-

anosensitive TFs. c | The mechanical properties of the liver prevent YAP/TAZ activation,

which maintains organ homeostasis. In the normal liver, binding of the CAPZ capping

protein to F- actin limits actomyosin tension and enables efficient inhibition of YAP/TAZ.

However, cell- autonomous activation of the fascin 1 (FSCN1) F- actin bundling protein by

oncogenes is able to overcome this mechanical tumour- suppressive tissue microenviron-

ment, facilitating YAP/TAZ activation and transdifferentiation of hepatocytes into chol-

angiocellular cells, and promoting cholangiocarcinoma progression. d | Metastatic breast

cancer cells often disseminate to organs with a soft ECM microenvironment. Reduced

actomyosin tension regulates peri- organelle F- actin pools, leading to decreased rigidity

of the Golgi apparatus, and to increased peri- mitochondrial F- actin. The resulting altera-

tions of organelle dynamics activate the SREBP1/SREBP2 and NRF2 TFs, which mediate

metabolic reprogramming and influence chemotherapy resistance. EMT, epithelial to

mesenchyme transition; MRTF indicates SRF–MRTF complexes; mtROS, mitochondrial

reactive oxygen species; Y/T indicates YAP–TEAD or TAZ–TEAD complexes.

NAture RevIeWS | GeNeTiCs

0123456789();:

R e v i e w s

Overall, experimental evidence obtained across several

fields of research has led to the identification of multiple

molecular mechanisms by which tissue mechanics can

influence genome organization, chromatin epigenetics

and gene expression, and genetic ablation experiments in

model organisms point to the importance of these mechanisms

in the context of normal and diseased cells. As most

of the factors highlighted in this section can also be activated

by purely biochemical signals, the challenge in the

field is now to more precisely dissect the precise, direct

role of mechanical forces in modulating these pathways

in physiological and pathological scenarios in vivo.

Emerging concepts

The direct transmission of forces through the actin

cytoskeleton, and its structural remodelling in response

to forces, is a powerful mechanism by which mechanical

information can be communicated inside the cell.

This mechanism is particularly interesting in light of the

observation that several other organelles in addition to

the nucleus are surrounded by actin filaments, which

are important for their subcellular localization, transport

and dynamics.

As discussed above (in 'Mechanisms of force sensing'

and 'Chromatin responses to nuclear forces') evidence

indicates that calcium is released from the nuclear envelope

in response to transmission of force to the nucleus

or deformation of the ER proximal to the nucleus36,42–45

.

This mechanism could perhaps affect the activity of TFs

that originate as ER­ resident transmembrane proteins345

.

Another example of organelle mechanosensitivity is the

Golgi apparatus, whose rheology mirrors cell tension in

response to extracellular mechanical cues47,49. This tensional

adaptation links extracellular mechanical cues

with activation of the SREBP1/SREBP2 TFs, by regulating

trafficking of the SREBP1/SREBP2 transmembrane

precursors between the Golgi apparatus and the ER49,199

(Fig. 6d). This mechanism accounts for lipid accumulation

in cells on a soft ECM, and couples the promotion

of mesenchymal stem cell differentiation into adipocytes

with activation of the corresponding lipogenic metabolic

programme.

In other recent studies, mitochondrial morphology

has been shown to be regulated by the interplay between

peri­ mitochondrial actin and mechanical cues from the

ECM50,346–352 (Fig. 6d). Mechanical forces can even bypass

the requirement for actin for mitochondrial fission, by

directly inducing mitochondrial deformation48. Cells

on a soft ECM display enhanced recruitment of the

DRP1 fission factor at mitochondria and subsequently

enhanced mitochondrial fission, which likely depends on

cooperation between multiple mechanisms, and which

mediates activation of the NRF2 TF to empower antioxidant

metabolism, ultimately making cells on a soft

ECM better able to resist oxidative stress50,353. The finding

that a similar pathway is relevant for neural stem cell commitment

in the mouse brain, among the softest tissues in

most species, suggests that this link may have multiple

physiological roles beyond redox homeostasis354,355

.

These studies collectively support the emerging

concept that forces can regulate peri­ organelle

actin pools and that this regulation may be important

volume 23 | October 2022 | 637

R e v i e w s

Box 1 | Outstanding questions in mechanical regulation of gene expression

• Mechanical transmission of forces to the nuclear lamina and to chromatin has been

predicted to regulate chromatin dynamics. What specific chromatin processes are

regulated by this force transmission and are they physiologically relevant? Are there

genes whose activity directly depends on force transmission to DNA across the

nuclear lamina?

• What is the functional significance of the interaction of F- actin and of F- actin

regulatory proteins with the basal transcription machinery? Does nuclear F- actin

also regulate other protein complexes (such as chromatin modifiers, chromosome

structure, splicing factors) and are they regulated by the mechanosensitive dynamics

of G- actin and F- actin? How do these generic mechanisms translate into specific

phenotypic effects elicited by forces?

• Several transcription factors can be regulated by various mechanical stimuli,

and the same stimulus often regulates several transcription factors in parallel. Is there

a common principle by which each transcription factor is regulated in response to a

broad spectrum of inputs? Are there even more general mechanisms that apply

to a large number of transcription factors?

• Mechanical forces widely regulate cell metabolism. Conversely, cell metabolism

provides key precursors for epigenetic modifications (including acetylation and

methylation). Could mechanical regulation of metabolism directly regulate

epigenetic modification and transcriptional memory?

for mediating mechanoresponsive signalling and

downstream transcriptional responses.

Concluding remarks

Taken together, recent advances in the field of mechanical

regulation of transcription indicate that multiple

organelles, including the plasma membrane, nucleus,

mitochondria and ER, display mechanosensitivity and

are capable of activating downstream signalling to regulate

transcription. Going forward it will be crucial

to determine under which scenarios and in response to

what kind of forces these various organelles are triggered

to respond and how they cooperate to determine

coherent and integrated transcriptional outputs (bOx 1).

Moreover, the molecular mechanosensors that are

responsible for direct sensing of forces within the specific

mechanosensitive organelles and thus the mechanisms

of initial conversion of mechanical force into a biochemical

signal remain elusive. One key obstacle to progress

is the absence of tools that allow direct visualization

and quantification of tension and stress in intact tissues

in vivo. Such tools, in combination with single­ molecule

imaging of transcriptional output, will allow quantitative

relationships between specific mechanical forces and

their transcriptional outputs to be established, which,

in turn, would enable the force­ mediated effects of

mechanosensitive TF systems to be differentiated from

their non­ mechanosensitive functions (bOx 1). Finally,

how TFs cooperate with the pleiotropic effects of force

on chromatin and transcription remains an important

open question (bOx 1). Unravelling the molecular details

of these processes will help to reveal whether the specific

mechanical properties of a tissue could function as

a layer of 'epigenetic' regulation to maintain stable gene

transcription and cell identity, acting as a 'template' to

ensure appropriate transcriptional responses during

regeneration and repair.

Published online 23 May