Stem cell mechanobiology: diverselessons from bone marrowIrena L. Ivanovska, Jae-Won Shin, Joe Swift, and Dennis E. Discher

Molecular and Cell Biophysics Laboratory, University of Pennsylvania, Philadelphia, PA 19104, USA

Review

Glossary

Elasticity: a property of a material that causes it to be restored to its original

state after deformation. Stiffness describes how resistant an object is to

deformation.

Monotonic function: a function that, within a given interval, either only

increases or decreases with the argument.

Strain: the deformation of an object in response to stress. When an object is

deformed, distances between points within the object change. Strain describes

the change in distance between points relative to the separation of the points

before deformation. It is a tensor because it is a function of the orientation on

the faces of a cube of material within the object.

Stress: a physical quantity describing the forces acting through a given cross-

sectional area in an object subjected to strain. Stress is also a tensor.

Storage and loss moduli: in viscoelastic materials, they are measures of the

stored energy (elastic component) and the energy dissipated as heat (viscous

component).

Traction forces: the forces that cells exert on their surroundings.

A stem cell niche is defined by various chemical andphysical features that influence whether a stem cellremains quiescent, divides, or differentiates. We reviewmechanical determinants that affect cell fate throughactomyosin forces, nucleoskeleton remodeling, andmechanosensitive translocation of transcription factors.Current methods for physical characterization of tissuemicroenvironments are summarized together withefforts to recapitulate niche mechanics in culture. Wefocus on mesenchymal stem cells, particularly in osteo-genesis and adipogenesis, and on blood stem cells –both of which reside in mechanically diverse marrowmicroenvironments. Given the explosion of efforts withpluripotent stem cells, the evident mechanosensitivityof clinically relevant, multipotent marrow cells under-scores an increasing need to examine and understand in

vivo and in vitro physical properties on length scales thatcells sense.

In vitro approaches to mimic and characterize theextracellular matrix (ECM) in the stem cell nicheStem cells are plastic in that they have the potential todifferentiate into multiple lineages. The control of stem cellfate has been classically attributed to growth factors thatregulate transcription factors. However, because stem cellsdo not exist in isolation in vivo, additional environmentalfactors are now being recognized as likely regulators ofstem cell fate. Indeed, multipotent stem cells or progeni-tors are present in many adult tissues [1] and reside in anenvironment known as a niche that helps to maintain thecells in a naı̈ve state until prompted to differentiate. Theniche combines tissue-specific matrix and nearby differen-tiated cells with key soluble molecules, all of which thestem cell probes and interprets when deciding to undergodifferentiation [2]. Mechanotransduction is also now beingstudied as a differentiation mechanism in which cellsphysically probe their surroundings with mechanicalforces that alter protein organization and ultimately geneexpression.

A role for ECM mechanics in determining cellularfunction – including stem cell differentiation – has beenextensively studied, in part by using reductionistin vitro approaches that simplify the complex mechanical

0962-8924/

� 2015 Elsevier Ltd. All rights reserved. http://dx.doi.org/10.1016/j.tcb.2015.04.003

Corresponding author: Discher, D.E. (discher@seas.upenn.edu).Keywords: extracellular matrix; actomyosin; nucleus; mechanotransduction;osteogenesis; hematopoiesis.

properties of tissue. For example, collagens are the mostabundant proteins in metazoans, but they display com-plex mechanics; collagen fibrils are semi-flexible biopoly-mers with non-linear elastic behavior and, whencrosslinked, form strain-stiffening networks [3]. The de-velopment of biomimetic culture systems depends onmethods to measure the mechanical properties of bothbiological and synthetic systems with high spatial reso-lution, such as rheology, micropipette aspiration, andatomic force microscopy (AFM), as described in Box 1.We discuss here how these techniques provide insightinto the roles of ECM, actomyosin contractility, nuclearmechanics, and mechanosensitive pathways in determin-ing stem cell commitment to specific lineages. We de-scribe some of the mechanical properties of tissues thatincreasingly motivate the characterization and controlof biomimetic platforms at nanoscales to understandthe role of the ECM and mechanotransduction in stemcell biology, with a particular focus on bone marrow stemand progenitor cells.

Influence of matrix mechanics on differentiation of bonemarrow cellsMesenchymal stem cells (MSCs) contribute to an osteo-progenitor population of cells, which differentiate intoosteoblasts that produce the osteoid matrix at the interfacebetween bone marrow and calcified collagen (Figure 1A)

Viscoelasticity: a property of a material that displays both viscous and elastic

characteristics when deformed.

Viscosity: a measure of the resistance of a fluid to gradual deformation by

‘shearing’ stresses (i.e., when layers within a fluid are moved laterally with

respect to each other).

Young’s modulus: the ratio of stress to strain for simple extensional

deformation; a measure of elasticity.

Trends in Cell Biology, September 2015, Vol. 25, No. 9 523

Box 1. Common techniques for measuring mechanical properties of ECM, cells, and the nucleus

Rheological methods

The material properties of natural or synthetic gels can be character-

ized using rheological methods. Measurements can be made of the

complex modulus G* = G’ + iG’’ under shear stress, where the storage

modulus G’ describes the elastic component and the loss modulus G’’

describes the viscous contribution. Extension of any substance, such

as a nucleus aspirated into a micropipette, is characterized in terms

of E*, and a convenient metric of stiffness is approximated as the

root-mean-square of E* on a given time scale for deformation. Gels

formed from different cytoskeletal and extracellular proteins exhibit

strain-stiffening for small to intermediate strains, measured with a

cone and plate rheometer [3] (Figure IA). The deformations of tissues,

cells, or nuclei can be measured on micron scales as they are drawn

into a micropipette under negative pressure. Optical microscopy is

used to image the deformations over time and often the proteins of

interest are fluorescently labeled. The nucleus in Figure IB is from a

cell expressing GFP-tagged lamin A protein (GFP-LamA). This method

was used to show that nuclei stiffen during differentiation [83],

embryonic heart tissues stiffen during development [84], and lamina

composition determines the viscoelastic response of nuclei [52–54].

Atomic force microscopy (AFM)

AFM is a widely used instrument to measure a variety of forces

between a sample and a nano-sized probe [85]. The working principal

behind the method is to raster-scan a surface with a small probe at the

end of a flexible cantilever. Interactions with the sample cause the

cantilever to bend and its deflection is detected by measuring the

position of a laser beam reflected from the back of the lever (Figure IC).

AFM can be used for force spectroscopy or ‘force mode.’ With this

application, the tip approaches the sample surface vertically, and is

then retracted. When the tip indents the sample, a force indentation

curve is recorded that can be used to obtain the properties of the

material under compression. When the probe is retracted, material

properties that are under stretching can be measured, or proteins

that are unfolding under tension can be examined. Using the Hertz

model for contact mechanics of elastic solids and its modifications

for different geometries, one can extract Young’s modulus E from

force–indentation curves (Figure ID). Another application of AFM is

imaging structures at high resolution such as the organization and

assembly of matrix proteins. The AFM image in Figure IE shows

the topography of nano-fibrils in a thin molecular crosslinked collagen

film. Moreover, AFM can be used for ‘force mapping’ (Figure IF). With

this method, force curves are recorded at an array of points across

the sample. Elasticity maps of ECM, cells, or tissues on stiff or

flexible substrates can be generated. The image in Figure IF shows

E of an MSC on a flexible, molecularly thin collagen film.

(A) Rheometer

(C) Atomic force microscopy (D) Hydrogel

GFP-LamA

(B) Micropipe�e aspira�on

Photodiode Laser

Can�leverTip

Forc

e

Indenta�on

nm

3

2

1

0

(E) Topography of nanofibrils of thin molecularly crosslinked collagen film

(F) Elas�city map of cell on thin collagen film

30

E, kPa

20

10500 nm5 μm

TRENDS in Cell Biology

Figure I. Common techniques for measuring the mechanical properties of ECM,

cells, and nuclei.

Review Trends in Cell Biology September 2015, Vol. 25, No. 9

[4]. Osteoid contains fibrillar collagen, non-collagenousproteins, and proteoglycans, all of which are crosslinkedby enzymes secreted by osteoblasts. With time, thematrix thickens and mineralization is initiated throughthe deposition of apatite (calcium phosphate mineral)crystals [5]. The nanoscale composition and topology ofthe bone matrix (Figure 1B) defines how cells experiencestresses at the subcellular level. Matrix fibrillar and non-fibrillar proteins, and apatite crystals, have nanoscaledimensions and their structure is likely to affect matrixnanomechanics. Calcium phosphate crystals grow be-tween fibrils but can also be found embedded withinthe fibrils themselves [6]. Interestingly, the orientationof the apatite crystals, rather than the density of themineral, correlates most strongly with the stiffness of thisnascent bone [7].

Isolated from marrow, MSCs are an adherent cell typethat spreads on tissue-culture plastic or glass. In culturewith the proper soluble factors, MSCs are multipotent; in

524

addition to their osteogenic and chondrogenic potential,they certainly possess an adipogenic potential [8,9]. Adipo-cytes are common in human bone marrow [10] and, basedon findings reviewed below, it is tempting to speculate thatthe soft marrow cavity is conducive to adipogenesis where-as the stiffer surface of bone influences osteogenesis. Thelatter is likely an important function of MSCs, and thereis some evidence that marrow adipocytes – perhaps alsodifferentiated from marrow MSCs – contribute to theregulation of hematopoiesis [11,12].

Hematopoietic stem and progenitor cells (HSPCs) isolat-ed from bone marrow are non-adherent compared to MSCs,and HSPCs are also capable of growing in suspension invitro. Nonetheless, interactions in vivo with the surroundingmarrow cells and the ECM seem unavoidable. Indeed,HSPCs express surface adhesion receptors, including integ-rins and cytoskeletal components such as actomyosin that,in principle, enable the sensing of physical forces andECM stiffness from the microenvironment. Historically,

TRENDS in Cell Biology

Mineralized bone

OsteocyteOsteoid

Pre-osteoblastOsteoblast

Adipocyte

Osteoclast

Bone marrowHSC and MSC

Nu

rER

Os

Min. bone

(A) (B)

Figure 1. Matrix material properties are important at both the cellular and nanometer scales. (A) Schematic representation of the spatial organization of bone marrow cells at

their interface with mineralized bone. (B) Transmission electron microscopy (TEM) image of part of an osteoblast [the nucleus (NU) and rough endoplasmic reticulum (rER)

are visible inside] and its extracellular environment with two distinct layers: the fibrillar osteoid (Os) and the mineralized bone (Min. Bone). Adapted from F. Bronner and M.C.

Farach-Carson [4]; scale bar, 1 micron. Abbreviations: HSC, hematopoietic stem cell; MSC, mesenchymal stem cell.

Review Trends in Cell Biology September 2015, Vol. 25, No. 9

HSPCs have been cultured in viscous methylcellulosemedia in combination with soluble factors to controlthe growth and differentiation of progenitors. Unlikesuspended cells, that diffuse away and disperse freelyas they divide, the viscous environment restricts cellmotions, leading to colonies that have allowed investiga-tion of how individual progenitor cells give rise tospecific lineages. While the methylcellulose system mightbe viewed as recapitulating a viscous marrow environ-ment, it is possible that cells behave differently in thissystem than in their natural environment. Recent studieshave used better defined hydrogel systems that allowstiffness to be tuned to study how matrix mechanicscan regulate MSCs and HSPCs either in isolation or incombination with other intrinsic and extrinsic factorsand forces.

Importantly, MSCs have been differentiated from plu-ripotent stem cells, including iPSCs [13,14], and suchcells exhibit the expected multi-lineage potential whengiven standard soluble factors. It is a long-term vision ofthe field to use easily-accessible cells, such as skin cells(as opposed to marrow biopsies), to generate patient-specific iPSCs that could then be used to make well-defined lineages for tissue-engineering applications.HSCPs have also been derived from pluripotent stemcells [14], but the efficiency is much lower and may reflectthe fact that adult HSCPs seem to derive from a type ofendothelial cell that is exposed to early blood flow[15,16]. The mechanotransduction capacities of iPSC-derived MSCs and HSCPs have yet to be tested, thusfurther motivating this review.

Tissue mechanics: the case for nanoscalecharacterizationSolid tissues in general are flexible or elastic on themacroscale, and small deformations can be well de-scribed by Young’s modulus, E (modulus of elasticity,see Glossary). Tissues span a wide range of elasticities:with brain being the most flexible (<1 kPa) [17] andmineralized bone being the stiffest (GPa) [18]. Bonemarrow is a softer tissue, with an apparent elasticitymeasured in situ at 0.3 kPa [19,20]. However, it is

important to note that mechanical properties can varydepending on the resolution with which they are mea-sured. The macroscale elasticity of a given tissue reflectsits structure at a larger scale, and this could differ fromthe elasticity that the cell encounters at micron scales orsmaller.

Many ECM proteins beyond fibrillar collagens canassemble into gels with measurable elasticity. Matrigel,for example, is a commercially extracted mixture of base-ment-membrane proteins secreted from mouse sarcomacells that also contains many growth factors [21]. Althoughsuch gels can be useful substrates for studies of cellbehavior, modifications to these gels to control stiffness(e.g., density changes or crosslinking) generally changeligand density. The complex composition of naturallyderived substances such as matrigel adds to the difficultyof separating mechanical effects from biochemical effects[21]. Synthetic hydrogels can simplify and clarify thebiology. Polyacrylamide (PA) gels coated with non-fibril-lar collagen ligand of a chosen density are a commonsystem introduced by Pelham and Wang [22]; flexibilityis easily controlled by varying the crosslinking and/ordensity of acrylamide. Such systems and others haveincreasingly been used to mimic tissue microenviron-ments and to study cell responses.

Matrix mechanics and actomyosin contractility indifferentiationMost tissue cell types are anchorage-dependent andmust adhere to the ECM to survive and function properly.Cell–matrix interactions are mediated by cell adhesionreceptors such as integrins that mediate transmembraneconnections to the cytoskeleton, signaling across themembrane. Information provided by specific extracellularligand binding sites to the outer integrin domains istransmitted to intercellular domains, a process depen-dent on long-range conformational changes that areallosterically regulated [23]. Cells respond to the spatialorganization of the ECM through the formation of focaladhesions (FAs) – complex dynamic structures that linkintegrins and scaffold proteins to the actin cytoskeleton[24].

525

Review Trends in Cell Biology September 2015, Vol. 25, No. 9

Once cells adhere, they are able to ‘feel’ the mechanicalproperties of their environment by pulling on their sur-roundings with the contractile forces generated by actin–myosin protein networks [25]. Non-muscle myosin-II is afilament-forming, actin-crosslinking protein with force-generating contractile properties that are regulated bythe phosphorylation of its light and heavy chains [26].Forces that myosin-II exerts can cause the matrix todeform, but resistance in the matrix can feed back tocause stress inside the cell where the assembly and orga-nization of stress fibers – cytoskeletal features composedof bundled actin filaments – can be modified. The sameprocesses are found in many contractile cell types, includ-ing stem cells.

Cell fate can be influenced by several mechanical prop-erties, including elasticity, geometry, and adhesion, whichultimately impact on cytoskeletal tension (Figure 2A).Indeed, MSCs cultured on PA gels coated with collagenrespond to matrix elasticity by specifying to a lineage. Onmatrix with osteoid-like stiffness (40 kPa), MSCs becameosteogenic, whereas on softer matrix (10 kPa), MSCs be-came myogenic [27]. The same study showed that pharma-cological inhibition of myosin-II contractility blockedlineage specification. Matrix stiffness favored actomyosinassembly and contractile forces, evidenced by formationof stress fibers, as predicted by theory [28]. In particular,stress fiber alignment within cells was a non-monotonicfunction of matrix elasticity, and there was an optimalratio of matrix to cell elasticity that governed cell polari-zation and stress fiber anisotropy.

In some of the original determinations of traction stres-ses (t) exerted by cells on gels, the mean traction stresses

Cont

rac�

lity

Matrix elas�

Lam

in-A

,C

HSC

Fat

So� matrix S�ff m

MSC

Lamin-

A,C OE

Lam

in-A,C KD

(A)

Figure 2. From outside-in and out again – the relationship between matrix stiffness, cell

cell contractility through regulation of non-muscle myosin-II levels. Nuclear stiffness,

cultured mesenchymal stem cells (MSCs) is higher on a stiffer matrix and increases with

with unknown stiffness and have high lamin-A,C levels. Lamin-A,C knockdown (KD) on

stiffer matrix enhances osteogenesis. (B) Myosin-II proteins polarize in hematopoietic s

By contrast, soft matrix suppresses myosin polarization, and cells divide symmetrically

526

increased with gel elastic modulus E such that the ratio oft/E was �3–4% over a range of physiological elasticity(reviewed in [25]). Cells can only exert a finite stress t

(perhaps in the kPa range), but the elastic modulus can befar higher (e.g., GPa for glass or plastic), such that t/Eapproaches 0%. Tissues also have a non-zero viscosity,especially very soft tissues such as bone marrow, and itis therefore interesting to consider the contribution of theloss modulus to cell behavior. Recent work shows that,when cells were cultured on soft substrates that exhibitsstress relaxation, their spreading was increased comparedwith elastic substrates with the same elastic modulus[29]. In another study, where the loss modulus was variedat fixed E, MSCs appeared sensitive to substrate creep andshowed enhanced differentiation potential towards thelineages commensurate with elasticity [30].

Actomyosin forces can also be modulated by constrain-ing cell shape: geometric features that increase cell con-tractility at a constant spread area favor osteogenesisover adipogenesis, independently of soluble factors [31].Cell fate is also influenced by shaped clusters of ‘nano-pits’– 120 nm in diameter and 100 nm deep – etched intopolymethylmethacrylate substrates by electron beamlithography [32]. Osteogenesis of MSCs is maximal onnano-pit arrays with 50 nm spacing, where long FAs areobserved and prolonged culture results in mineralization.A subsequent study showed that MSCs cultured on squarearrays maintain multipotency [33]. By culturing cells onelastic pillars that allow traction forces to be measured bytracking the bending of pillars, a strong correlation wasobserved between FA distribution, traction forces, and cellspreading [34]. When MSCs were cultured in bipotential

city

atrix

BoneMyosin polarizability

Prolifera�on

Sym

met

ric d

ivisio

n

Asym

met

ric d

ivisio

n

Differen�a�on

HSC

So� S�ff

(B)

TRENDS in Cell Biology

contractility, nuclear composition, and differentiation. (A) Matrix stiffness increases

set by the level of lamin-A,C protein, scales with tissue microelasticity, E, and in

cellular tension. MSCs in culture and expanded on plastic deposit their own matrix

soft matrix enhances adipogenesis; lamin-A overexpression (OE) combined with

tem cells (HSCs) on stiff matrix, leading to asymmetric division and differentiation.

and proliferate.

Review Trends in Cell Biology September 2015, Vol. 25, No. 9

osteogenic–adipogenic media, the spring constant of thepillars was found to be a deciding factor for cell fate, withhigh stiffness favoring osteogenesis and low stiffness fa-voring adipogenesis. The study suggested that tractionforces, and by implication cellular tension, can predictwhich cells will preferentially differentiate upon biochemi-cal stimuli, and moreover hypothesized that cell tensionor stiffness can be used as an early marker for predictingcell fate decisions in a heterogeneous MSC population.

Many ECM proteins contain adhesive sequences suchas Arg-Gly-Asp (RGD), and over 20 integrins are knownto recognize this sequence [35]. Peptides immobilizedon monolayers have generally proven useful for studyingcell–matrix adhesion [36–39]. Substrates presentinghigh-affinity peptides promote osteogenesis, those withlow-affinity peptides at high density promote myogenesis,whereas those with low-affinity peptides at low densitypromote neurogenesis [40]. These lineage specificationsseem in rough agreement with matrix stiffness cues [27],suggesting a unifying physicochemical mechanism: theresistance of substrate engagement to cellular tractionforces feeds back into cytoskeletal tension. However, itremains to be seen whether adhesive ligands are limitingin vivo because tissue stiffness varies in vivo.

Beyond MSCs, hematopoietic differentiation to variouslineages is also regulated by mechanical stresses. Forexample, platelets are generated from megakaryocytes(MKs) that undergo maturation by accumulating DNA con-tent without division. This process, termed polyploidizationor endomitosis, occurs because weak adhesion to externalinterfaces limits cytokinesis. Interestingly, studies withnon-mammalian cells have shown that adhesion to matrixprovides traction forces to pull cells apart [41]. Consistentwith this notion, MK polyploidization is inhibited on stiffmatrix where stronger adhesion increases traction forcesto drive cell division [42]. By contrast, soft matrix (typical ofsoft marrow [19]) inhibits myosin-II forces and maximizesMK maturation. High-ploidy MKs have motile projectionsthat migrate into the nearby bloodstream where shearcauses fragmentation to pro-platelets [43] that then shear-activate myosin-II to promote its division into small,familiar platelets (except when myosin-IIA is mutated) [44].

Very early hematopoietic differentiation is also regulat-ed by ECM mechanics in combination with other physicalfactors. The HSC:progenitor (P) ratio is maintained orincreased on highly-flexible tropoelastin substrate com-pared to stiff crosslinked tropoelastin [45]. As with MSCs,inhibition of myosin II eliminates mechanosensing of HSCsand progenitors. Using mass spectrometry-calibrated in-tracellular flow cytometry, the composition of myosin-IIisoforms was observed to switch from the polarizable Bisoform to the more homogeneous A isoform during he-matopoietic differentiation [19]. External stresses suchas fluid shear and stiff matrix induced polarization ofmyosin-IIB, resulting in the switching of lamin isoforms.By contrast, soft matrix prevented myosin-IIB polariza-tion. When polarization occurred during cell division, my-osin-IIB was asymmetrically segregated to a daughter cellthat maintained its stemness, while the cell with lowmyosin-IIB differentiated (Figure 2B). When myosin-IIBwas downregulated in differentiated cells, myosin-IIA was

activated by dephosphorylation, inducing actin polymeri-zation. Soft matrix also maximized myosin-IIA phosphor-ylation. Together, these findings suggest that soft matrix(away from rigid bone) likely suppresses myosin-II isoformswitching, while stiff matrix (near bone) drives asymmetricdivision (Figure 2B).

By understanding the matrix constraints that regulatecell differentiation, one can begin to rationalize how cellsare organized within a tissue. In bone marrow, cells exist inan environment with a gradient in stiffness over length-scales several times greater than cellular dimensions.Osteoblasts localize to the osteoid interface, which is rela-tively stiff but is adjacent to cells such as pre-osteoblaststhat are surrounded by softer matrix. MSCs reside in theinterior of the bone marrow, but it is not clear how theyreposition during commitment to lineages such as adipo-cytes or pre-osteoblasts. MSCs not only polarize towardand migrate into regions of greater stiffness [46] but alsotend to differentiate according to ECM elasticity cues[47,48]. It is notable that this ‘durotaxis’ process dependson myosin-II, and is evident in 3D matrix systems. Non-muscle myosin II-B (NMII-B) is unpolarized on soft sub-strates but polarizes in the rear of the cell when the cellmigrates from a soft to a stiff substrate. Moreover, stiffersubstrates promote the assembly of NMII-A into stressfibers, which are necessary for NMII-B polarization [46].

Lamins: regulating cell fate with nuclear structureWithin a cell, the nucleus is caged by the cytoskeletonsuch that stresses in the cytoskeleton propagate to thenuclear envelope (Figure 3) [49] and possibly influencethe global organization of chromatin [50]. Key to maintain-ing the structure and physical properties of the nucleus isthe nuclear lamina, a network of intermediate filamentproteins of A-type lamins (lamins A and C are products ofalternative splicing of the LMNA gene) and B-type lamins(lamins B1 and B2 are encoded by LMNB1 and LMNB2genes, respectively). Lamins are located at the interfacebetween the inner nuclear membrane and chromatin[51–54]. The ‘linker of nucleo- and cytoskeleton’ (LINC)complex is a protein complex that might contribute tomechanical transduction of signals to the nucleus[55,56]. Specifically, nesprins – a family of nuclear mem-brane proteins – connect to actin filaments in the cytoplasmthrough their N-terminal domains, while their C-terminaldomains are anchored to SUN (from Sad1p, UNC-84)domain proteins, providing a direct physical link betweenthe cytoskeleton and nucleoskeleton by spanning the innernuclear membrane and in turn binding to the lamina.

Tissue analyses have shown that lamin-A,C levels scalewith the stiffness of a tissue, with stiffer tissues such asbone having higher levels of lamin-A,C than do softertissues such as fat (Figure 2A) [52]. While the expressionof lamin-A,C increased 30-fold across tissues, the constitu-tively expressed B-type lamins varied by less than three-fold. Although ‘the truth is in the tissues’, culture studiesare likely to hint at possible mechanisms. For at leastseveral types of cells in culture, lamin-A,C levels were alsoaffected by cellular tension in a process regulated bymechanosensitive phosphorylation (Figure 3B,E) [57].Studies of MSCs cultured on plastic had shown that

527

Low myosin contrac�lity

Thin ac�nstress fibers

Thick ac�nstress fibersHigh SRFFocal

adhesions

High myosin contrac�lity

S�ffECM

Focaladhesions

ECM So�

Cytoplasmic localiza�on

Cytoplasm

Proteasomal degrada�on

Low LMNA

PP

P

Kinase

Phosphoryla�on Solubiliza�on Degrada�on

Low tension

RARγ

YAP/TAZTEAD

Nucleus

RARE LMNA

RAagonist

Low tension

Adipogenesis

AdipogenesisNeurogenesis

Osteogenesis

High LMNA High tension

NucleusCytoplasm

TEADYAP/TAZYAP/TAZ

RARE LMNA

FA forma�on

Rho

Rock

MLC - P

Myosin ac�vitycontrac�lity

High tension

RAantagonist

Osteogenesis

Osteogenesis

Nuclear transloca�onTranscrip�onal ac�va�on

AdipogenesisChondronesis

RAR-γ

(A)

(C)

(D)

(F) (G)

(E)

(B)

RAR-γ

X

TRENDS in Cell Biology

Figure 3. Stem cell mechanobiology. (A) Mesenchymal stem cells (MSCs) cultured on soft substrates exhibit a low-contractility phenotype: low spread area, poorly

developed focal adhesions (FAs) and stress fibers, and lower levels of lamin-A,C in the nuclear lamina. (B) Lamin-A,C level is regulated by tension-dependent kinase activity.

Low tension allows lamin to be phosphorylated (P), leading to increased solubility and turnover. The filamentous lamina is thus maintained in a state appropriate to the

mechanical environment of the nucleus. (C) On soft substrates, transcription factor (TF) RAR-g remains in the cytoplasm, reducing transcription of the lamin-A,C gene

(LMNA) and favoring adipogenesis, an effect enhanced by agonists of the retinoic acid (RA) pathway. TF YAP/TAZ also remains cytoplasmic, favoring adipose and neuronal

lineages over osteogenesis. (D) By contrast, MSCs cultured on stiff substrates have a high-contractility phenotype: high spread area, well-developed FAs and stress fibers,

high levels of TF serum response factor (SRF), and a nuclear lamina rich in lamin-A,C. (E) High tension on the nuclear lamina inhibits phosphorylation and turnover of lamin-

A,C. (F) The contractility of the actomyosin cytoskeleton is regulated in response to substrate stiffness by the Rho/ROCK signaling cascade. (G) On stiff substrates, TF RAR-g

translocates to the nucleus, increasing transcription of LMNA and driving osteogenesis, an effect enhanced by the addition of antagonists to the RA pathway. YAP/TAZ

levels in the nucleus are also increased, favoring osteogenesis over adipogenic or chondrogenic cell fates. Further abbreviations: MLC, myosin light chain; RAR-g, retinoic

acid receptor-g; Rho, Ras homolog; RARE, retinoic acid response element; ROCK, Rho-associated protein kinase; SRF, serum response factor; TAZ, transcriptional

coactivator with PDZ-binding motif; TEAD, TEA domain protein; YAP, Yes-associated protein.

Review Trends in Cell Biology September 2015, Vol. 25, No. 9

528

Review Trends in Cell Biology September 2015, Vol. 25, No. 9

lamin-A,C knockdown favors adipogenesis [58] whereaslamin-A,C overexpression favors osteogenesis [59]. Byusing PA gels with stiffness tuned to mimic either softfat tissue or stiff osteoid, lamin-A,C was further shown tobe a mechanosensitive enhancer of matrix elasticity-direct-ed lineage specification in vitro (Figure 2A) [52].

As with myosin-II isoform switching in hematopoiesis,the composition of the nuclear lamina not only changeswith differentiation to blood lineages but knockdown andoverexpression also demonstrate a functional influence[53]. Although the direct roles of matrix mechanics innuclear lamin turnover of hematopoietic cells remain un-studied, a stiffer lamina impedes migration and – in prin-ciple – exit from the marrow. Furthermore, lamin-A,Cchanges less than B-type lamins across different hemato-poietic lineages [53]. Because marrow matrix might be toosoft to impact on lamin-A,C, relatively unique marrowfactors likely regulate B-type lamins in this essentialcompartment, which generates millions of new blood cellsper second in humans. The interplay between matrix,cytoskeletal, and nuclear mechanics during hematopoieticlineage specification certainly requires further investiga-tion, but studies thus far with hematopoietic lineagesprovide initial insight into how adhesion, matrix elasticity,and external shear forces can couple to cytoskeletal andnuclear mechanics in processes central to stem cell self-renewal, fate decisions, and even migration.

Mechanical regulation of transcription factorsAn understanding of how mechanical signals are transmit-ted to the nucleus to activate different gene expressionprograms is slowly emerging. Nucleocytoplasmic shuttlingof transcription factors appears to be an essential step inseveral mechanosensitive pathways implicated in differen-tiation across different cell types (Figure 3). For example,SRF (serum response factor) signaling is known to play akey role in myogenic differentiation in diverse muscle cells(e.g., [60]) and depends on RhoA and actin polymerizationvia interactions with cofactors including MRTF (myocardin-related transcription factor) A and B (also known as MAL).MRTFs A and B bind to monomeric G-actin in the cytoplasmbut are released upon actin polymerization and are trans-ferred into the nucleus to bind to and activate SRF[61,62]. Matrix constraints can be a sufficient cue for tran-scription factors to change their nucleocytoplasmic localiza-tion and thereby regulate gene expression. Restricted ECM–cell contacts have been shown to trigger human epidermalstem cells to initiate terminal differentiation as mediatedby actin cytoskeleton and SRF pathway regulation [63].Cytoskeletal components including non-muscle myosin-IIA (MYH9) regulate SRF, as do substrate stiffness as wellas lamin-A,C and emerin levels [52,57,63,64].

Retinoic acid (RA), a metabolite of vitamin A, is widelyknown to regulate development, differentiation, prolifera-tion, and apoptosis [65,66]. Retinoic acid receptors (RAR),RAR-a, RAR-b and RAR-g, encoded respectively by theRARA, RARB, and RARG genes, bind to retinoid X receptorsin the nucleus and regulate the transcription of manygenes, including LMNA. Lamin-A,C expression is regulatedby RARs (Figure 3C,G), but the nuclear-to-cytoplasmic ratioof RARG also increases with matrix elasticity and lamin-A,

C overexpression while decreasing with lamin-A,C knock-down. With MSCs RAR antagonists increase lamin-A,Clevels and enhance osteogenesis on rigid substrates, butneither Retinoic Acid or antagonist affects lamin-A,C in cellson soft substrates [52]. The idea that a purely soluble factordepends for its effects on the elasticity of the matrix thatunderlies a cell underscores the need for deeper insight intothe native mechanics of any niche. Furthermore, given themany pharmacological variants of RA agonists and antago-nists, testing of any such drug should consider matrix.

YAP (Yes-associated protein) and TAZ (transcriptionalcoactivator with PDZ-binding motif) transcriptional coacti-vators are classically part of the Hippo pathway that reg-ulates organ growth and development but might also have arole in mechanotransduction [67,68]. Two kinases MST(mitogen-activated protein kinase kinase kinase 10, alsoknown as MAP3K10) and LATS (large tumor suppressorkinase) canonically control the activity of the transcriptionfactors YAP/TAZ [69]. MST phosphorylates and activatesLATS kinase, and LATS phosphorylation of YAP/TAZ inhi-bits its activity and traps the transcription factor in thecytoplasm. If the LATS kinase is inhibited then YAP andTAZ enter the nucleus and bind to TEAD (TEA domain)transcription factors that drive the transcriptional regula-tion of proliferation (Figure 3C,G) [70,71].

YAP/TAZ subcellular localization depends on matrixstiffness and the actin cytoskeleton in a mechanism distinctfrom the Hippo pathway and independent of MST/LATS,perhaps involving other unknown kinases [72]. Nuclearlocalization of YAP/TAZ in MSCs increases on stiff matrixsimilarly to osteoid (40 kPa) (Figure 3G), and requires RhoGTPase activity and actomyosin contractility (Figure 3F). Ahigh nuclear:cytoplasmic ratio of YAP/TAZ correlates withincreased osteogenic potential in MSCs in culture (Figure 3G)[73], although tissue profiling of YAP does not reveal a cleartrend with tissue stiffness [52]. YAP/TAZ nuclear accumula-tion driven by prolonged passive mechanical stimulationmight increasingly restrict MSC fate choices as a ‘mechanicalmemory’ [74]. Inhibition of nuclear localization of YAP inhPSCs (human pluripotent stem cells) through compliantsubstrates that mimic brain stiffness (0.7 kPa) or by inhibi-tion of F-actin polymerization has been found to promotehighly efficient differentiation of hPSCs into neurons even inthe absence of soluble stimuli (Figure 3C) [75].

A next step in dimensionality and greater complexityIn vivo cells are often surrounded by other cells or matrix,and generally these 3D environments constrain cell shapeand dynamics, restrict access to soluble factors such asoxygen, and impact on how secreted matrix is deposited,among other effects. Given the complexities of the locationsin which stem cells reside in vivo, it remains difficult torecapitulate these environments in vitro and to pinpointthe precise mechanical cues that influence cell fate. Amongthe new cell culture approaches being developed, onehydrogel system could be patterned in 3D and remodeledin situ with a focused laser to control cell location and tospatially define the incorporation of RGDS (Arg-Gly-Asp-Ser) peptides for cell tethering [76]. MSCs encapsulated insuch systems mimic several features of in vivo chondrogen-esis: cells remodel the matrix through downregulation and

529

Review Trends in Cell Biology September 2015, Vol. 25, No. 9

degradation of fibronectin, with subsequent upregulationof glycosaminoglycans (GAGs) and type-II collagen.

MSCs have also been encapsulated within a hydrogel-based matrix where stiffness and ligand density canbe manually controlled [77]. Osteogenesis occurred inhigh-stiffness matrices (11–30 kPa), while adipogenesisoccurred in the softest substrates, which is consistent withfindings on simpler 2D hydrogels. In addition, cells rear-ranged their integrin adhesions with the number of bondsper cell depending non-monotonically on matrix elasticityand peaking at �20 kPa. However, another finding arguesthat hydrogels with breakable crosslinks differ from cova-lently crosslinked hydrogels [78]: MSCs encased in non-degradable, covalently crosslinked hyaluronic acid (HA)hydrogels with RGD functionalization favored adipogen-esis, even in nominally stiff gels. Through adaptation ofa 3D version of traction force microscopy [79], roundedcells were seen to generate forces similar to those observedin cells with pharmacologically inhibited contractility.However, the introduction of proteolytically cleavablecrosslinks made the gels susceptible to cell-mediated deg-radation, which allowed the cells to break symmetry, pulland spread within the matrix, and undergo osteogenesis.Access to soluble factors and the impact on secreted matrixdeposition remain to be understood in such systems. How-ever, when symmetry is forcibly broken by using thin,non-degradable HA gels that are layered on top of MSCspre-cultured on a soft or stiff gel, the cells appear to exhibitcytoskeletal characteristics and polarization similar tocells cultured on a stiffer gel [80]. For MSCs in contactwith the stiff surface of bone and with soft hematopoieticcells above it (Figure 1A), the stiffness of bone will domi-nate the niche signal and tend to induce osteogenesis.

Soluble factors are major drivers of differentiation.Lineage-specific cocktails are often added or even mixedin deconstruction studies of substrate and ECM effects.Sometimes the cocktails include growth factors such astransforming growth factor b (TGF-b) that bind to theECM and are released depending on matrix stiffnessand cell tension [81]. Glucocorticoids such as dexametha-sone are also commonly used in vitro in differential doses toimpact on differentiation, but whether soft or stiff matrixmodulates such important transcription factors as theglucocorticoid receptors has not been examined in vitroor in vivo. Perhaps the insights gained into matrix- andlamina-regulated RAR shuttling will prove relevant. Re-gardless, the relative contributions of each soluble and‘insoluble’ factor, and the extent to which the pathwaysare coupled in series or in parallel, need to be explored indetail with modern methods such as proteomics, transcrip-tomics, and certainly epigenetics. Accounting for cell-cellvariation from single cell measurements can be equallyimportant to clarify the sensitivity to niche signals.

Concluding remarksAlthough insights into the molecules that contribute to themany types of stem cells and niches have accumulated overthe past several decades, measurements of niche mechan-ics and the forces that can influence stem cells have onlyemerged within the past 10 years. Bone marrow stem cellsthat make blood, fat, and bone among other lineages have

530

been the most deeply studied, and considerable evidencefrom tunable hydrogels demonstrates that matrix elastici-ty and perhaps viscosity can affect such cell fates. Path-ways involve actomyosin contractile forces, which cells useto probe the surrounding matrix, and the nucleoskeletalprotein lamin-A, which adjusts nuclear mechanics to themicroenvironment and enhances differentiation programs.In these contexts, mechanical regulation of gene expres-sion at least sometimes takes place via mechanosensitivetranslocation of key transcription factors. Some of thesepathways have also been described for cells growing onsubstrates that are patterned [82], although the physicalrelationships to niches in tissues are not always clear. 3Dhydrogels with controlled properties and patterns are alsoemerging, but the added complexity always needs to becarefully considered for its effects on cell confinement,permeation of factors, and the relevance to 2D nature ofsome tissue formation processes (e.g., osteogenesis).

The physical properties of intact and fresh tissue clearlyneed to be studied in greater detail in normal and diseasedstates to deepen our understanding of stem cell microen-vironments. Such information is already foundational todeveloping and employing optimal cell culture models.Functional studies and characterizations of the epigeneticchanges in stem cells in response to mechanical cues fromtheir environment have provided important mechanisticinsights into how cells choose their fate. Further measure-ments of tissues (and proposed culture models) will enablemany more types of molecular biology perturbations,super-resolution microscopy analyses of key structures,and detailed evaluation of the cells and their nuclei duringdifferentiation. ‘Kinetics reveal mechanism’ in terms ofthe order in which processes occur, and increasing atten-tion to timing will therefore be essential in understandingand controlling the maturation of the many types of plu-ripotent and multipotent stem cell systems that arerapidly emerging.

AcknowledgmentsThis work was supported by the National Institutes of Health (P01-DK032094; R01-HL124106; R01-HL062352; R01-EB007049; P30-DK090969; NCATS-8UL1TR000003), the Human Frontier ScienceProgram (I.L.I. and D.E.D.), the National Science Foundation (CMMI1200834), the Nano Science and Engineering Center-Nano Bio InterfaceCenter, and the American Heart Association (14GRNT20490285). J.W.S.is currently a Postdoctoral Fellow at the School of Engineering andApplied Sciences, Harvard. J.S. is currently at University of Manchester(UK) and acknowledges a David Phillips Fellowship from the Biotechnol-ogy and Biological Sciences Research Council (BBSRC).

References1 Zipori, D. (2009) Biology of Stem Cells and the Molecular Basis of the

Stem State, Humana Press2 Schofield, R. (1978) Relationship between spleen colony-forming cell

and hematopoietic stem-cell – hypothesis. Blood Cells 4, 7–253 Storm, C. et al. (2005) Nonlinear elasticity in biological gels. Nature

435, 191–1944 Bukka, P. et al. (2004) Molecular regulation of osteoblast

differentiation. In Bone Formation (Bronner, F. and Farach-Carson,M.C., eds), pp. 1–17, Springer

5 Coelfen, H. (2010) Biomineralization: a crystal-clear view. Nat. Mater.9, 960–961

6 McKee, M.D. et al. (2005) Hierarchies of extracellular matrix andmineral organization in bone of the craniofacial complex andskeleton. Cells Tissues Organs 181, 176–188

Review Trends in Cell Biology September 2015, Vol. 25, No. 9

7 Ishimoto, T. et al. (2013) Degree of biological apatite c-axis orientationrather than bone mineral density controls mechanical function inbone regenerated using recombinant bone morphogenetic protein-2.J. Bone Miner. Res. 28, 1170–1179

8 Pittenger, M.F. et al. (1999) Multilineage potential of adult humanmesenchymal stem cells. Science 284, 143–147

9 Kern, S. et al. (2006) Comparative analysis of mesenchymal stem cellsfrom bone marrow, umbilical cord blood, or adipose tissue. Stem Cells24, 1294–1301

10 Rosen, C.J. et al. (2009) Marrow fat and the bone microenvironment:developmental, functional, and pathological implications. Crit. Rev.Eukaryot. Gene Expr. 19, 109–124

11 DiMascio, L. et al. (2007) Identification of adiponectin as a novelhemopoietic stem cell growth factor. J. Immunol. 178, 3511–3520

12 Naveiras, O. et al. (2009) Bone-marrow adipocytes as negativeregulators of the haematopoietic microenvironment. Nature 460,259–263

13 Zou, L. et al. (2013) A simple method for deriving functional MSCs andapplied for osteogenesis in 3D scaffolds. Sci. Rep. 3, 2243

14 Hynes, K. et al. (2014) Generation of functional mesenchymal stemcells from different induced pluripotent stem cell lines. Stem Cells Dev.23, 1084–1096

15 Cumano, A. and Godin, I. (2007) Ontogeny of the hematopoietic system.Annu. Rev. Immunol. 25, 745–785

16 Zovein, A.C. et al. (2008) Fate tracing reveals the endothelial origin ofhematopoietic stem cells. Cell Stem Cell 3, 625–636

17 Georges, P.C. et al. (2006) Matrices with compliance comparable to thatof brain tissue select neuronal over glial growth in mixed corticalcultures. Biophys. J. 90, 3012–3018

18 Keller, T.S. et al. (1990) Young modulus, bending strength, and tissuephysical-properties of human compact-bone. J. Orthop. Res. 8, 592–603

19 Shin, J.W. et al. (2014) Contractile forces sustain and polarizehematopoiesis from stem and progenitor cells. Cell Stem Cell 14, 81–93

20 Winer, J.P. et al. (2009) Bone marrow-derived human mesenchymalstem cells become quiescent on soft substrates but remain responsiveto chemical or mechanical stimuli. Tissue Eng. 15, 147–154

21 Vukicevic, S. et al. (1992) Identification of multiple active growth-factors in basement-membrane Matrigel suggests caution ininterpretation of cellular-activity related to extracellular-matrixcomponents. Exp. Cell Res. 202, 1–8

22 Pelham, R.J. and Wang, Y.L. (1997) Cell locomotion and focaladhesions are regulated by substrate flexibility. Proc. Natl. Acad.Sci. U.S.A. 94, 13661–13665

23 Hynes, R.O. (2002) Integrins: bidirectional, allosteric signalingmachines. Cell 110, 673–687

24 Geiger, B. et al. (2009) Environmental sensing through focal adhesions.Nat. Rev. Mol. Cell Biol. 10, 21–33

25 Discher, D.E. et al. (2005) Tissue cells feel and respond to the stiffnessof their substrate. Science 310, 1139–1143

26 Vicente-Manzanares, M. et al. (2009) Non-muscle myosin II takescentre stage in cell adhesion and migration. Nat. Rev. Mol. CellBiol. 10, 778–790

27 Engler, A.J. et al. (2006) Matrix elasticity directs stem cell lineagespecification. Cell 126, 677–689

28 Zemel, A. et al. (2010) Optimal matrix rigidity for stress-fibrepolarization in stem cells. Nat. Phys. 6, 468–473

29 Chaudhuri, O. et al. (2015) Substrate stress relaxation regulates cellspreading. Nat. Commun. 6, 7

30 Cameron, A.R. et al. (2011) The influence of substrate creep onmesenchymal stem cell behaviour and phenotype. Biomaterials 32,5979–5993

31 Kilian, K.A. et al. (2010) Geometric cues for directing thedifferentiation of mesenchymal stem cells. Proc. Natl. Acad. Sci.U.S.A. 107, 4872–4877

32 Dalby, M.J. et al. (2007) The control of human mesenchymal celldifferentiation using nanoscale symmetry and disorder. Nat. Mater.6, 997–1003

33 McMurray, R.J. et al. (2011) Nanoscale surfaces for the long-termmaintenance of mesenchymal stem cell phenotype and multipotency.Nat. Mater. 10, 637–644

34 Fu, J. et al. (2010) Mechanical regulation of cell function withgeometrically modulated elastomeric substrates. Nat. Methods 7,733–736

35 Ruoslahti, E. (1996) RGD and other recognition sequences forintegrins. Annu. Rev. Cell Dev. Biol. 12, 697–715

36 Derda, R. et al. (2010) High-throughput discovery of synthetic surfacesthat support proliferation of pluripotent cells. J. Am. Chem. Soc. 132,1289–1295

37 Derda, R. et al. (2007) Defined substrates for human embryonic stemcell growth identified from surface arrays. Acs Chem. Biol. 2, 347–355

38 Hudalla, G.A. and Murphy, W.L. (2010) Immobilization of peptideswith distinct biological activities onto stem cell culture substratesusing orthogonal chemistries. Langmuir 26, 6449–6456

39 Koepsel, J.T. and Murphy, W.L. (2009) Patterning discrete stem cellculture environments via localized self-assembled monolayerreplacement. Langmuir 25, 12825–12834

40 Kilian, K.A. and Mrksich, M. (2012) Directing stem cell fate bycontrolling the affinity and density of ligand–receptor interactions atthe biomaterials interface. Angew. Chem. Int. Ed Engl. 51, 4891–4895

41 Benzeev, A. and Raz, A. (1981) Multinucleation and inhibition ofcytokinesis in suspended cells - reversal upon reattachment to asubstrate. Cell 26, 107–115

42 Shin, J-W. et al. (2011) Myosin-II inhibition and soft 2D matrixmaximize multinucleation and cellular projections typical ofplatelet-producing megakaryocytes. Proc. Natl. Acad. Sci. U.S.A.108, 11458–11463

43 Junt, T. et al. (2007) Dynamic visualization of thrombopoiesis withinbone marrow. Science 317, 1767–1770

44 Spinler, K.R. et al. (2015) Myosin-II repression favors pre/proplatelets but shear activation generates platelets and fails inmacrothrombocytopenia. Blood 125, 525–533

45 Holst, J. et al. (2010) Substrate elasticity provides mechanical signalsfor the expansion of hemopoietic stem and progenitor cells. Nat.Biotechnol. 28, 1123–1128

46 Raab, M. et al. (2012) Crawling from soft to stiff matrix polarizes thecytoskeleton and phosphoregulates myosin-II heavy chain. J. Cell Biol.199, 669–683

47 Tse, J.R. and Engler, A.J. (2011) Stiffness gradients mimicking in vivotissue variation regulate mesenchymal stem cell fate. PLoS ONE 6,e15978

48 Isenberg, B.C. et al. (2009) Vascular smooth muscle cell durotaxisdepends on substrate stiffness gradient strength. Biophys. J. 97,1313–1322

49 Wang, N. et al. (2009) Mechanotransduction at a distance:mechanically coupling the extracellular matrix with the nucleus.Nat. Rev. Mol. Cell Biol. 10, 75–82

50 Iyer, K.V. et al. (2012) Mechanical activation of cells induceschromatin remodeling preceding MKL nuclear transport. Biophys.J. 103, 1416–1428

51 Lammerding, J. et al. (2006) Lamins A and C but not lamin B1 regulatenuclear mechanics. J. Biol. Chem. 281, 25768–25780

52 Swift, J. et al. (2013) Nuclear lamin-A scales with tissue stiffness andenhances matrix-directed differentiation. Science 341, 1240104

53 Shin, J-W. et al. (2013) Lamins regulate cell trafficking and lineagematuration of adult human hematopoietic cells. Proc. Natl. Acad. Sci.U.S.A. 110, 18892–18897

54 Harada, T. et al. (2014) Nuclear lamin stiffness is a barrier to 3Dmigration, but softness can limit survival. J. Cell Biol. 204, 669–682

55 Crisp, M. et al. (2006) Coupling of the nucleus and cytoplasm: role of theLINC complex. J. Cell Biol. 172, 41–53

56 Lombardi, M.L. et al. (2011) The interaction between nesprins and sunproteins at the nuclear envelope is critical for force transmissionbetween the nucleus and cytoskeleton. J. Biol. Chem. 286, 26743–26753

57 Buxboim, A. et al. (2014) Matrix elasticity regulates lamin-A,Cphosphorylation and turnover with feedback to actomyosin. Curr.Biol. 24, 1909–1917

58 Akter, R. et al. (2009) Effect of lamin A/C knockdown on osteoblastdifferentiation and function. J. Bone Miner. Res. 24, 283–293

59 Scaffidi, P. and Misteli, T. (2006) Lamin A-dependent nuclear defects inhuman aging. Science 312, 1059–1063

60 Wei, L. et al. (1998) RhoA signaling via serum response factor playsan obligatory role in myogenic differentiation. J. Biol. Chem. 273,30287–30294

61 Miralles, F. et al. (2003) Actin dynamics control SRF activity byregulation of its coactivator MAL. Cell 113, 329–342

531

Review Trends in Cell Biology September 2015, Vol. 25, No. 9

62 Sotiropoulos, A. et al. (1999) Signal-regulated activation of serumresponse factor is mediated by changes in actin dynamics. Cell 98,159–169

63 Connelly, J.T. et al. (2010) Actin and serum response factor transducephysical cues from the microenvironment to regulate epidermal stemcell fate decisions. Nat. Cell Biol. 12, 711–718

64 Ho, C.Y. et al. (2013) Lamin A/C and emerin regulate MKL1-SRFactivity by modulating actin dynamics. Nature 497, 507–511

65 Duester, G. (2008) Retinoic acid synthesis and signaling during earlyorganogenesis. Cell 134, 921–931

66 Gudas, L.J. and Wagner, J.A. (2011) Retinoids regulate stem celldifferentiation. J. Cell. Physiol. 226, 322–330

67 Piccolo, S. et al. (2014) The biology of YAP/TAZ: hippo signaling andbeyond. Physiol. Rev. 94, 1287–1312

68 Low, B.C. et al. (2014) YAP/TAZ as mechanosensors andmechanotransducers in regulating organ size and tumor growth.FEBS Lett. 588, 2663–2670

69 Dong, J.X. et al. (2007) Elucidation of a universal size-controlmechanism in Drosophila and mammals. Cell 130, 1120–1133

70 Vassilev, A. et al. (2001) TEAD/TEF transcription factors utilize theactivation domain of YAP65, a Src/Yes-associated protein localized inthe cytoplasm. Genes Dev. 15, 1229–1241

71 Zhao, B. et al. (2008) TEAD mediates YAP-dependent gene inductionand growth control. Genes Dev. 22, 1962–1971

72 Aragona, M. et al. (2013) A mechanical checkpoint controls multicellulargrowth through YAP/TAZ regulation by actin-processing factors.Cell 154, 1047–1059

73 Dupont, S. et al. (2011) Role of YAP/TAZ in mechanotransduction.Nature 474, 179–183

532

74 Yang, C. et al. (2014) Mechanical memory and dosing influence stemcell fate. Nat. Mater. 13, 645–652

75 Musah, S. et al. (2014) Substratum-induced differentiation of humanpluripotent stem cells reveals the coactivator YAP is a potent regulator ofneuronal specification. Proc. Natl. Acad. Sci. U.S.A. 111, 13805–13810

76 Kloxin, A.M. et al. (2009) Photodegradable hydrogels for dynamictuning of physical and chemical properties. Science 324, 59–63

77 Huebsch, N. et al. (2010) Harnessing traction-mediated manipulationof the cell/matrix interface to control stem-cell fate. Nat. Mater. 9,518–526

78 Khetan, S. et al. (2013) Degradation-mediated cellular traction directsstem cell fate in covalently crosslinked three-dimensional hydrogels.Nat. Mater. 12, 458–465

79 Dembo, M. and Wang, Y.L. (1999) Stresses at the cell-to-substrateinterface during locomotion of fibroblasts. Biophys. J. 76, 2307–2316

80 Rehfeldt, F. et al. (2012) Hyaluronic acid matrices show matrixstiffness in 2D and 3D dictates cytoskeletal order and myosin-IIphosphorylation within stem cells. Integr. Biol. 4, 422–430

81 Klingberg, F. et al. (2014) Prestress in the extracellular matrixsensitizes latent TGF-beta 1 for activation. J. Cell Biol. 207, 283–297

82 McBeath, R. et al. (2004) Cell shape, cytoskeletal tension, and RhoAregulate stem cell lineage commitment. Dev. Cell 6, 483–495

83 Pajerowski, J.D. et al. (2007) Physical plasticity of the nucleus in stemcell differentiation. Proc. Natl. Acad. Sci. U.S.A. 104, 15619–15624

84 Majkut, S. et al. (2013) Heart-specific stiffening in early embryosparallels matrix and myosin expression to optimize beating. Curr.Biol. 23, 2434–2439

85 Binnig, G. et al. (1986) Atomic force microscope. Phys. Rev. Lett. 56,930–933