Biophysical regulation of stem cell behavior within the niche

Stem cells reside within most tissues throughout the lifetimes of mammalian organisms. To maintain their capacities for division and differentiation and thereby build, maintain, and regenerate organ structure and function, these cells require extensive and precise regulation, and a critical facet of this control is the local environment or niche surrounding the cell. It is well known that soluble biochemical signals play important roles within such niches, and a number of biophysical aspects of the microenvironment, including mechanical cues and spatiotemporally varying biochemical signals, have also been increasingly recognized to contribute to the repertoire of stimuli that regulate various stem cells in various tissues of both vertebrates and invertebrates. For example, biochemical factors immobilized to the extracellular matrix or the surface of neighboring cells can be spatially organized in their placement. Furthermore, the extracellular matrix provides mechanical support and regulatory information, such as its elastic modulus and interfacial topography, which modulate key aspects of stem cell behavior. Numerous examples of each of these modes of regulation indicate that biophysical aspects of the niche must be appreciated and studied in conjunction with its biochemical properties.


Introduction
Th e concept that the behavior of a stem cell can be modulated by factors in its immediate vicinity arose several decades ago in studies of spleen colony-forming cells, which were later appreciated to be hematopoietic stem and progenitor cells (HSPCs) [1]. It was hypo thesized that these HSPCs and their progeny were distinct cell populations that possessed an 'age structure' , such that once the progeny left their stem cell niche during developmental 'aging' , their stem-like qualities were lost, and entry into a new niche promoted diff erentiation into a more mature, lineage-committed cell type. Subsequent work with Drosophila germ stem cells [2] and other systems demonstrated that the niche is a region that regulates stem cell fate decisions by presenting that cell with specifi c repertoires of soluble and immobilized extracellular factors. It is increasingly appreciated that many of these signals are biophysical in nature, particularly biochemical factors that are spatiotemporally modulated, mechanical cues, and electrostatic cues. Over the past several years, numerous examples in which the fi rst two of these properties in particular have been shown to play key regulatory roles have emerged.

Spatial organization of cues in the niche
Many factors that are often thought of as soluble are known to harbor matrix-binding domains that immo bilize them to the solid phase of tissue. For example, fi broblast growth factors, platelet-derived growth factors (PDGFs), transforming growth factors (TGFs), vascular endothelial growth factors (VEGFs), Hedgehogs, and numerous cytokines contain heparin-binding domains [3][4][5][6]. Immobilization of such factors to the extracellular matrix (ECM) often modulates their activity by promoting sustained signaling via inhibiting receptor-mediated endocytosis [7], increasing their local concentration and establishing concentration gradients emanating from the source [8], and otherwise modulating the spatial organization of factors in a manner that aff ects signaling. As an example, compared with soluble VEGF, VEGF bound to collagen preferentially activates VEGFR2, asso ciates with β1 integrins, and promotes the association of all of these molecules into focal adhesions [9]. Th ere are also strong examples of synthetic systems that harness these phenomena, the fi rst of which involved tethering epidermal growth factor to immobilized poly(ethylene oxide) (PEO) to prolong growth factor signaling in rat hepatocyte cultures [10]. A subsequent study showed that immobiliza tion of Sonic hedgehog (Shh) onto inter penetrating polymer network surfaces, along with the integrinengaging peptide arginine-glycine-asparagine (RGD),

Abstract
Stem cells reside within most tissues throughout the lifetimes of mammalian organisms. To maintain their capacities for division and diff erentiation and thereby build, maintain, and regenerate organ structure and function, these cells require extensive and precise regulation, and a critical facet of this control is the local environment or niche surrounding the cell. It is well known that soluble biochemical signals play important roles within such niches, and a number of biophysical aspects of the microenvironment, including mechanical cues and spatiotemporally varying biochemical signals, have also been increasingly recognized to contribute to the repertoire of stimuli that regulate various stem cells in various tissues of both vertebrates and invertebrates. For example, biochemical factors immobilized to the extracellular matrix or the surface of neighboring cells can be spatially organized in their placement. Furthermore, the extracellular matrix provides mechanical support and regulatory information, such as its elastic modulus and interfacial topography, which modulate key aspects of stem cell behavior. Numerous examples of each of these modes of regulation indicate that biophysical aspects of the niche must be appreciated and studied in conjunction with its biochemical properties.
induced potent osteoblastic diff erentiation of bone marrow-derived mesenchymal stem cells (MSCs), whereas soluble Shh enhanced proliferation [11]. As another example, crosslinking heparin-binding peptides to fi brin gels along with neurotrophic factor 3 (NT-3) and PDGF resulted in neuronal and oligodendrocytic diff erentiation of mouse neural stem cells (NSCs) with inhibition of astrocytic diff erentiation [12]. Finally, immobilization of leukemia inhibitory factor (LIF) to a synthetic polymer surface supported mouse embryonic stem cell (mESC) pluripotency for up to two weeks in the absence of soluble LIF, indicating the advantage of substrate functionalization in lowering cell culture reagent costs and facilitating future multifactorial cell fate screening experiments [13].
Immobilization of cues to the solid phase -that is, the ECM or the surface of adjacent cells or both -also off ers the opportunity to modulate the nanoscale organization in which these factors are presented ( Figure 1). Growing evidence has indicated that ligand multivalency, or the number of ligands organized into a nanoscale cluster, can exert potent eff ects on cell behavior [14][15][16][17]. For example, seminal work using a synthetic system to present clusters of ECM-derived adhesion ligands showed that the spatial organization of ECM cues can also impact cell responses. Specifi cally, on surfaces functionalized with the integrin adhesion ligand YGRGD in various states of valency, fi broblast attachment did not vary as a function of ligand valency, yet substrates bearing highly clustered or multivalent peptides required signifi cantly lower ligand densities to induce cell spreading and migration [18]. In recent work that explored the behavior of MSCs in a three-dimensional (3D) hydrogel functionalized with RGD peptides, investigators who used a fl uorescence resonance energy transfer technique found that the cells apparently reorganized the peptides into clusters upon integrin binding [19].
Th e role of ligand clustering also extends to growth factors and morphogens. Th e morphogen Hedgehog and its family member Shh, best known for their role in tissue patterning during development, have been shown to require nanoscale clustering to achieve long-range paracrine signaling [20]. Additionally, transforming growth factor-beta (TGF-β) is able to induce distinct diff erential signaling by activating either a homomeric or a heteromeric form of its receptor, which needs to be dimerized or tetramerized before signaling can occur [21]. Furthermore, cell membrane-bound ligands (for example, Delta/ Jagged that activate the Notch receptor and ephrins that activate corresponding Eph receptors) often require oligomeri zation to transduce biochemical signaling cascades [22,23]. Th e creation of synthetically clustered, or multivalent, ligands off ers a useful tool to study basic biological aspects of receptor clustering as well as a reagent to better control stem cell self-renewal or diff erentiation. For example, Shh has been chemically conjugated to the long polymer chain hyaluronic acid at varying stochiometric ratios to produce a range of multivalent forms of Shh, and higher-valency Shh bioconjugates exerted progressively higher potencies in inducing the osteogenic diff erentiation of a primary fi broblast line with MSC characteristics [24]. Th is concept was recently extended to create highly active and multivalent versions of ligands that are naturally integral membrane proteins (A Conway, T Vazin, N Rode, KE Healy, RS Kane, DV Schaff er, unpublished data).
In addition to spatial regulation of cues at the nanoscale, microscale features in the niche can play key roles. Fibrous ECM proteins such as collagen and fi bronectin are present throughout the NSC niche, raising the hypothesis that cells may respond to ECM surface topography. One interesting demonstration of this idea showed that rat NSCs cultured on laminin-coated synthetic polyethersulfone fi bers of 280 or 1,500 nm in diameter preferentially diff erentiated into oligodendrocytes or neurons, respectively. It has also been shown that culturing MSCs atop vertically oriented nanotubes of 70 to 100 nm in diameter (but not less than 30 nm) is suffi cient to induce their diff erentiation into osteoblasts [25]. In an analogous study, culturing MSCs on nanopits of 100 nm also induces osteogenesis but only if the pits are anisotropic, or disordered [26]. Recently, the cytoskeletal scaff olding protein zyxin was shown to play an important role in the response of human MSCs to surface nanotopography [27]. Specifi cally, MSCs expressed zyxin at lower levels when plated on a polydimethylsiloxane (PDMS) surface patterned with a 350-nm grating, which resulted in smaller and more dynamic focal adhesions and increased directional migration of the cells along the gratings.
In addition to nanoscale features, cell-cell interactions at the microscale aff ect behavior. Specifi cally, the assembly of stem cells themselves into multicellular aggregates exerts strong infl uences on cell self-renewal or diff erentiation, as the cells actively secrete factors and modulate local biological transport properties in ways that impact their neighbors. For example, several groups have created controlled 3D culture systems to generate human embryonic stem cell (hESC) embryoid bodies (EBs) -or cell clusters -of defi ned sizes. Th ese involved centrifugalforced aggregation [28] as well as microfabricated PDMS wells surrounded with functionalized protein-resistant self-assembled monolayers [29]. Th ese methods produced more consistent sizes than EB suspensions, and in the latter example a tighter distribution of EB volume was accompanied by a higher level of expression of the pluripotency marker Oct-4. In another key study, hESC culture inside microfabricated poly(ethylene glycol) (PEG) wells yielded EBs from 40 to 450 μm in diameter [30,31]. Greater endothelial cell diff erentiation was observed in smaller EBs (150 μm), which was shown to be due to higher Wnt5a expression, whereas larger EBs (450 μm) enhanced cardiogenesis as a result of higher Wnt11 expression. Interestingly, another group used microcontact printing of adhesive islands on twodimensional substrates to control hESC colony size and showed that smaller hESC colonies became more endoderm-biased, whereas larger colonies exhibited greater diff erentiation into neural lineages [32]. Within the endoderm-biased colonies, cardiogenesis was found to be more pronounced in larger EBs as opposed to the neural-biased colonies, which had higher levels of cardiogenesis in smaller EBs. Collectively, these results demonstrate that spatial organization of molecules and cells can play critical roles in modulating stem cell fate and can therefore serve as important tools to exert exogenous control over these processes.

Mechanoregulation in the niche
Th e mechanical properties of tissues have been studied for a number of decades. In the 1950s, it was observed that cells of the mesenchyme grow preferentially toward regions that are under higher mechanical stress, indicating a fundamental contribution of mechanical properties to biological function [33,34]. Aberrant tissueelastic mechanical properties have also been shown to play a pathological role in certain cases, such as causing increased contractility of arterial resistance vessels within hypertensive rats, leading to elevated blood pressure and eventual heart failure [35]. Th ere is a strong rationale for why mechanical properties may also modulate stem cell behavior. Tissues in the body range over several orders of magnitude in stiff ness, from the softness of adipose to the toughness of bone, hinting at the possible importance of mechanics in maintaining diff erent adult organs. In addition, there is local heterogeneity within individual tissues, as it has been shown, for example, that the hippocampus -a brain region that harbors adult NSCsspatially varies in stiff ness, as assessed by atomic force microscopy [36]. Th ese various diff erences are not captured in the hard tissue culture surfaces typically used for in vitro study.
Engler and colleagues [37], in pioneering work, demonstrated that substrate elastic modulus aff ects stem cell lineage commitment, in which MSCs cul tured on polyacrylamide substrates of varying elastic moduli diff erentiated into cell types characteristic of tissues with the corresponding stiff ness: neurons, myo blasts, and osteoblasts. A later study extended this concept to another stem cell type by showing that NSCs cultured on variable modulus substrates diff erentiate preferentially into neurons on softer substrates and astrocytes on harder materials [38]. Recently, it was shown that soft substrates enhance the ability of human embryonic and human-induced pluripotent stem cells to diff erentiate into neural lineages [39].
Th e fi nding that increased matrix rigidity can modulate cell diff erentiation has also been extended to analysis of the epithelial-mesenchymal transition (EMT) of both murine mammary gland cells and canine kidney epithelial cells, where more rigid substrates promoted EMT via upregulating the Akt signaling pathway [40]. In addition to diff erentiation on a single stiff ness, durotaxis -the ability of cells to migrate in response to a stiff ness gradient -and mechanosensitive diff erentiation can be integrated. For example, upon seeding of MSCs on a surface with a gradient in stiff ness, cells migrated preferentially toward the stiff er region of the gel and then diff erentiated according to the local stiff ness [41]. Finally, stem cells can, in turn, strongly infl uence their mechanical environment. MSCs cultured on non-linear strainstiff en ing fi brin gels have been shown, upon application of local strain via cytoskeletal rearrangement and cell spreading, to globally stiff en the gel [42]. Th is eff ect led to long-distance cell-cell communication and alignment, thus indicating that cells can be acutely responsive to the non-linear elasticity of their substrates and can mani pulate this rheological property to induce patterning.
In addition to diff erentiation, modulus can infl uence stem cell self-renewal. For example, it was shown that substrate stiff ness strongly impacts the ability of muscle stem cells, or satellite cells, to undergo self-renewal in culture. Upon implantation, cells isolated from muscle and grown on soft substrates were able to expand and contribute to muscle to a much greater extent than stem cells cultured on stiff surfaces [43]. Furthermore, mESC self-renewal is promoted on soft substrates, accompanied by downregulation cell-matrix tractions [44].
Mechanobiologists have begun to elucidate mechanisms by which stem cells undergo mechano regulation, building on advances with non-stem cells. Several mechanotransductive proteins involved with producing traction forces via cytoskeletal rearrangements are thought to be implicated in translating mechanical signals into changes in gene expression in stem cells [37,45,46]. For example, it has been shown that inhibition of myosin II diminishes the eff ect of ECM stiff ness on MSC diff erentiation [37]. Furthermore, decreasing ECM stiff ness decreases RhoA activity and subsequent calcium signaling in MSCs [47]. Recent work also indicates that Rho GTPases, specifi cally RhoA and Cdc42, enable NSCs to adjust their own stiff ness as a function of the substrate modulus and thereby regulate the cells' stiff nessdependent diff erentiation into either astrocytes or neurons in vitro and potentially in vivo [46]. Furthermore, an important study demonstrated that the transcriptional coactivator YAP undergoes nuclear localization in MSCs on higher-stiff ness substrates, thereby narrowing the gap in our understanding of how microenvironmental mechanical properties may ultimately modulate gene expression and, as a result, cell diff erentiation [48]. Finally, while mechanosensitive stem cell behavior has been demonstrated on several materials in addition to the original polyacrylamide, recent work broaches another possible mechanism for cell behavior on diff erent stiff nesses. Specifi cally, investigators found that MSCs exhibited diff erent behavior on polyacrylamide but not PDMS gels of variable modulus, and additionally found that the porosity of the polyacrylamide but not the PDMS gels varied with stiff ness. Th is raised the intriguing possibility that diff erences in ECM conjugationspecifi cally the number of anchoring points of collagen to the gel surface -could subsequently aff ect integrin binding and thereby modulate cell responses [49]. Th is possibility should be explored further, potentially in comparison with fi ndings that NSCs and MSCs on polyacrylamide-based materials behave similarly as a function of modulus for materials presenting either ECM proteins [37,46] or simple RGD peptides [19,38].
In addition to the static mechanical properties of cells and surrounding tissue, dynamic biomechanical processes can regulate stem cell function. For instance, stress and strain from local tissue contraction and expan sion, including processes such as contraction of muscle, tendons, and ligaments as well as cyclic deformation of tissue surrounding vasculature and the lungs, are prevalent in vivo. Furthermore, organismal development is a highly dynamic process that exposes cells and structures to mechanical forces. In Drosophila embryos, for example, compression of cells induces expression of Twist, a protein involved with regulating germ layer speci fi cation and patterning [50]. Similarly, in zebrafi sh, tensile strains were shown to regulate gastrulation during early develop ment [51]. Such basic studies extend to mammalian stem cells. For example, cyclic strain of lung embryonic MSCs stimulates expression and nuclear localization of tension-induced/inhibited protein-1 (TIP-1) and inhibits expres sion of TIP-3, thereby promoting myogenesis and inhibiting adipogenesis [52]. Cyclical stretching also inhibits diff erentiation of hESCs through upregulation of Nodal, Activin A, and TGFβ1 [53]. Diff erential eff ects of equiaxial versus uniaxial strain have also been observed, with equiaxial primarily down regulat ing smooth muscle cell promoting factors in MSCs and uniaxial upregulating them [54].
Even temporal variation of the ECM on slower timescales may play a role in regulating stem cell function [55]. For example, matrix metalloproteinases (MMPs), enzymes that remodel the ECM through cleavage of key constituent proteins, can modulate stem cell diff erentiation. Interestingly, it has been shown that, in response to two injury-induced chemokines, SDF-1 and VEGF, NSCs in the subventricular zone of the lateral ventricles in the adult rodent brain diff erentiated into migratory cells that secreted MMPs at elevated levels [56]. Blocking the expression of these proteins inhibited diff erentiation of the NSCs, indicating that the cells require matrix remodeling to proceed with their diff erentiation and subsequent migration into injured areas of the brain. MSCs localized to bone marrow have also been shown to secrete MMPs to facilitate infi ltration of sites of tissue damage, infl ammation, or neoplasia before undergoing diff erentiation [57]. In addition to experiencing a decrease in ECM integrity, cells can experience ECM stiff ening (for example, an approximately 10-fold increase in stiff ness during cardiac maturation). Young and Engler [58] created a hyaluronic acid poly(ethylene glycol) hydrogel that could undergo stiff ening over a two-week period and found that pre-cardiac cells within the gel underwent a signifi cantly higher increase in maturationboth expression of muscle markers and assembly into muscle fi bers -than corresponding cells seeded on static hydrogels. Th e development of hydrogels in which crosslinks are photosensitive has enabled investigators to vary stiff ness in time and space, powerful capabilities that will enable further advances in the fi eld [59,60].
Another form of dynamic stress is shear fl ow, most often associated with the circulatory system. Th e earliest study of shear on stem cell fate determined that fl ow promotes maturation and capillary assembly of endothelial progenitor cells [61]. Subsequent studies showed that shear fl ow can induce diff erentiation of other stem cell types, including endothelial cell specifi cation from murine embryonic MSCs [62] and vascular endothelial cell lineage commitment from ESCs [63,64]. Each of these properties and parameters of the niche (summarized ECM, extracellular matrix; EGF, epidermal growth factor; EMC, embryonic mesenchymal cell; EMT, epithelial-mesenchymal transition; hEPC, human endothelial progenitor cell; hESC, human embryonic stem cell; Hh, hedgehog; hMSC, human mesenchymal stem cell; hPSC, human pluripotent stem cell; LIF, leukemia inhibitory factor; MAPK, mitogen-activated protein kinase; mEMC, mouse embryonic mesenchymal cell; mESC, mouse embryonic stem cell; mMuSC, mouse muscle stem cell; MSC, mesenchymal stem cell; NPC, neural progenitor cell; NT-3, neurotrophic factor 3; PDGF, platelet-derived growth factor; RGD, arginine-glycine-asparagine peptide; SCF, stem cell factor; Shh, sonic hedgehog; SM, smooth muscle; STAT3, signal transducer and activator of transcription 3; TGF-β, transforming growth factor-beta; TIP, tension-induced/inhibited protein; VEGF, vascular endothelial growth factor; VEGFR2, vascular endothelial growth factor receptor 2. in Table 1) off ers oppor tunities to control cell fate for down stream therapeutic application.

Conclusions
Understanding the properties and eff ects of each complex component of a local stem cell microenvironment is an essential step toward understanding the stem cell itself.
In particular, the ability of a stem cell to respond to spatiotemporally varying biochemical cues and distinct mechanical and physical stimuli within its surroundings is being increasingly recognized and will continue to be elucidated in the years to come. Th e eff ect of substrate stiff ness on stem cell fate has been increasingly appreciated in recent years, and other facets of the niche's solid phase -including spatial organization in the presentation of biochemical information, electrostatics [65], and biomolecular transport [66] -will increasingly be investigated. While technological limitations in the ability to control, quantify, and image these properties currently exist, advances in super-resolution microscopy may be combined with stem cell research to enable considerable progress [67]. Furthermore, an appreciation of these interactive processes in natural tissue may greatly aid the development of stem cell therapies to treat numerous human diseases. For example, this basic knowledge may enable therapeutic modulation of endogenous stem cells via alterations in the niche as well as off er opportunities to create more eff ective large-scale culture systems and bioreactors to expand and diff erentiate stem cells. Furthermore, the creation of in vitro cell and tissue equivalents of therapeutically relevant organs, enabled by the technological advances and optimized model culture systems, will enable both basic and therapeutic investi gations of human disease biology. Th erefore, as is evidenced by an increasing number of important studies, a blend of biology, chemistry, physics, and engineering can empower progress in both basic and translational directions.

Competing interests
DVS is an inventor on a patent involving polyvalent ligands as potent activators of cellular signalling, and holds founder's stock in a company that is developing this commercially. AC declares no competing interests.

Authors' contributions
AC wrote the manuscript with helpful feedback and editing by DVS.