Substrates with Changing Properties for Extracellular Matrix Mimicry

Cell-ECM interactions Fundamental to the success of using biomaterials in medical and health care applications, is the understanding of their interactions with biological tissues and systems. First step towards this end is the elucidation of cell-ECM interactions, which has attracted considerable interest in recent decade. Cellular decision-making process is driven by the internal genetic program and external factors comprising primarily other cells and extracellular matrix (ECM) via soluble factors and direct physical connections such as focal adhesion [1, 2]. Three key features of ECM have been identified of great significance in affecting cells, namely, chemical and biological composition, dimensionality (twovs. threedimensional), and physical properties [3-6]. These features can be sensed by cells via cellECM linkages, and the resulting signals subsequently follow intracellular pathways and trigger a cascade of events leading to alterations in gene expression and manifestation in phenotype. In contrast to the long recognized chemical composition and adhesive character‐ istics of the ECM, physical cues including topography, pore size, geometric patterns, and mechanical stiffness and their significance has just started being appreciated [7-10]. Whilst characteristics of ECM have profound effect on cells, cells may also actively exert impact on ECM by secretion of soluble factors or modify properties of ECM, or contribute to maintain‐ ing integrity or properties of ECM. At a larger scale, biological systems may actively interact with biomaterials to maintain or re-establish homeostasis.

especially for load-bearing and mechanically active tissues (e.g., heart, cartilage, lung) [15]. Not only do these changes naturally occur, but there are also benefits associated with them from a tissue engineering viewpoint, as highlighted in the series of discussions in the March 2005 issue of MRS (Materials Research Society) Bulletin [16,17] and later studies. Whitesides [18] and Mrksich [19] and their coworkers among a number of investigators pioneered the work on engineering cell growth by using dynamic substrates. Their work and later reports on differential cell responses to materials with different properties suggest that it is beneficial for biomaterials to have controlled changing properties [20]. These facts make it very desirable for the bio-mimetic materials to have the capability of undergoing controlled remodeling with respect to time. They also raised caution in interpretation of the observations made from the majority of the biological studies, where properties of the substrates (e.g., culture flask, Petri dish, and hydrogels) remain unchanged throughout the process.
The scope of this work A significant number of reviews are available on the changes in soluble factors of ECM that may affect cellular behavior (e.g., [21]) and particularly on the changing environment in bioreactors [22] (e.g., nutrients concentration, oxygen level, temperature). Thus, they are not covered in this review. Moreover, flow conditions and the resulting traction forces, and their effect on certain cell types including blood cells (i.e., endothelial cells and red blood cells) have also been intensively reviewed and hence are not discussed here. This is also the case for mechanical forces, strain and stress applied directly to the cells (e.g., [23,24]) in load-bearing tissues such as bone, cartilage and lung (for reference, see, e.g., [25][26][27] ). Therefore, this review is focused on the latest studies and current knowledge of two-or three-dimensional substrates with changing or dynamic mechanical and adhesive properties, design and conditions to trigger and achieve designed dynamics, and the impact of insitu change of these properties on cell behavior, which provides guideline for design of biomaterials for their applications in medical and healthcare applications. Note that mechanical stiffness and elastic modulus were used interchangeably in this work.
Pathological tissues Diseased tissue may possess properties such as mechanical stiffness different from those of the normal tissue [40,41]. As a typical example, it has been found that tumor cells display enhanced movement towards ECM with lower mechanical rigidity, which is interesting considering the general stiffening phenomenon of tumor tissues [42], and biomechanical characteristics of tissues play a crucial role in tumor development [41]. It has also been shown that during the surgical procedures such as radio-frequency (RF) ablation, tissue properties can be modified [43]. Moreover, changes in ECM composition and relative quantity of ECM molecules can be correlated to pathology. For instance, ECM composition change that occurs during sub-epithelial tissue remodeling proved associated with asthma [33]. ECM remodeling in diseased heart valves is correlated to myofibroblast contractility [44], and certain cell types such as valvular interstitial cells can be activated and contribute to further tissue remodeling [45]. Additionally, ECM remodeling affects tissue mechanical properties in addition to inflammatory responses [46]. Moreover, mechanical forces, as experienced in traumatic brain injury or even under normal conditions, could potentially cause protein aggregation, giving rise to various diseases including neurodegenerative diseases [47]. Furthermore, early investigation of properties of central nervous (CNS) tissue under impact yielded modulus values with considerable variation. As an example, Fallenstein and coworker reported storage modulus of human brain tissue of 0.6 ~ 1.1 kPa under sinusoidal shear stress input mimicking head impact [48].
Development and aging During development, synthesis and degradation of ECM is a controlled process (e.g., [8,49]), and mis-regulation contributes to many forms of diseases [30] Particularly, the microenvironment for embryonic and adult stem cells is regulated both temporally and spatially [2,34], and is involved in various developmental processes including responses to soluble factors, cell differentiation, and morphogenesis [12]. ECM in musculoskeletal and other tissues adapts to increasing mechanical requirements by altering the size of tissue components [50] during development. Structural dynamics of ECM components such as collagen, laminin, and fibronectin coincides with estrous cycle and developmental progression [51]. Besides development, aging is also accompanied by changing ECM composition and structures. For example, in connective tissues, aging has been reported to be associated with increase in type I collagen content and decrease in both type III collagen and proteoglycans content, and with collagen fiber disruption and unraveling [50].

Tissue-implant interfaces
With the growing interest in developing biomimetic materials for tissue engineering applications, tissue-implant interfaces have been subject to considerable research effort. Previous reports showed that cells can actively modify ECM at the interfaces, and cause drastic changes in tissue or construct mechanics using fibroblast-populated construct and other biomaterials [52,53]. The study by Lee and co-workers suggested that dynamic moduli of an alginate material may be due to the bioactivities of the chondrocytes encapsulated in the scaffold [54]. In a similar study, different substrate composition and architecture gave rise to distinct levels of modulus increase owing to chondrocytes responses [55]. To take another example, smooth muscle cells (SMCs) in contact with engineered arterial construct displayed distinctive responses in protein synthesis and consequently the mechanical properties of ECM were significantly different [56]. Additionally, biodegradable materials used in various tissue engineering applications possess changing properties associated with specific degradation profiles.
Engineering advantages It has been suggested that temporal control over substrate or scaffold properties may entail great benefits in engineering cell growth. Among the notable examples is the stem cell differentiation and proliferation. A recent work showed enhanced hepatic functions from differentiated stem cells on softer substrates and improved expansion of undifferentiated cells on stiffer ones [57]. Therefore, it is promising to use stiffer substrates for optimal proliferations and subsequently soften them to gain better hepatic functions once differentiation completes. Langrana group also found that different neurite properties (e.g., axonal length and primary dendrite number) show differential preference towards substrate stiffness [58], suggesting the strategy of promoting nerve regeneration with scaffold of varying properties. Similar approach can be adopted to take advantage of differential cell responses (e.g., migration and functions) to adhesive properties.
The recurring indication from the above discussions is that in vivo ECM interacts with cells in many ways, and that the alteration in ECM composition or structures leads to changes in adhesive properties (hence cell adhesion) and/or mechanical properties. This potentially affects a variety of cell types and their properties and functioning, at different developmental stages, under normal or pathological conditions, or upon impact or injury. It also holds promises in offering novel approaches to tissue engineering applications. As a result, it is imperative to understand cellular responses to changing substrate properties for basic biology and biomimetic material (including biodegradable materials) design.

Types of dynamic substrates and stimuli
Here we consider two major classes of dynamic substrates that are based on self-assembled monolayer, or SAM, and hydrogels, as well as other types of substrates with surface or structural modifications. Since the focus of this work is on mimicking dynamic nature of ECM to examine cellular responses, those dynamic materials that are developed for other specific applications such as drug delivery [59] and do not involve changes that mimic dynamic ECM are beyond the scope of the review.

Self-assembled monolayer (SAM)
SAMs are formed by adsorption of molecules in solution or gas phase onto substrates in a spontaneous and organized fashion [60], and have emerged as an important candidate of materials in studying cellular responses to dynamic substrates [60,61] where modifications could be made in situ. One of the major research focuses in this direction is to examine the effect of dynamic adhesive property of the substrate on cells, particularly by leveraging the ability to selectively capture or release cells upon application of a variety of stimuli (Table 1). These stimuli, applied to initiate substrate dynamic, include light [62,63], electricity [16,17], pH, temperature, and others [16,64] (Fig. 1). These approaches generally involve photochemical or electrochemical conversion, redox reactions, or stimulated configuration change of surface proteins, which leads to the attachment, detachment, shielding, or exposing of cell adhesion molecules, among which a popular choice is RGD peptide.
Mrksich group has been actively engaged in the development of SAM-based dynamic substrates by integrating surface chemistry, micro-patterning, and cell microenvironment engineering [17,19,61,65,66]. Based on an elegant design of SAM with electrochemically responsive group on a micro-patterned substrate, they first applied electrical stimulation to release 3T3 fibroblasts from designed areas on the substrate, and subsequently encouraged migration of neighboring cells to those areas with newly added adhesion molecules [65]. Refining this design by adding responsiveness to both negative and positive electric potentials, they demonstrated selective control over cell release [67] (Fig. 1C). Other groups have also engaged in the effort along this direction. By employing a hydroquinone terminated SAM based on re-dox reactions, Chan and colleagues proposed a SAM on gold surface that enables attachment and release of cell adhesion molecules such as those with RGD motif [68], and selectively released adherent 3T3 fibroblasts bound through RGD motifs but not those adherent based on hydro-phobic interactions (Fig. 1A). Reversibility of cell adhesion is attractive in studying cellular responses and cell-ECM interactions [60]. As an example, a surface chemistry involving azobenzene capable of switching between two configurations was utilized to expose or hide adhesion sites (e.g., RGD motif) upon photochemical stimulation [69] (Fig. 1B). While the finding is interesting, the long exposure of cells to UV may be problematic despite the reported negligible impact of light with wavelength over 320 nm on cells [63].  [69] and application to cell culture (B2) where NIH 3T3 fibroblasts initially adhere to adhesive surface (a) which was inhibited upon surface modification (b) followed by recovery of adhesion due to azobenzene configuration change (c). Extracted from [69]. (C) Illustration of a SAM that allows different modifications with positive and negative electric potential (C1) and its application in selective release of Swiss 3T3 cells (C2). Extracted from [67]. All with permission from publishers.
The above surveys part of the key advancement using SAM in modifying adhesion properties of the substrates mimicking those of natural cellular microenvironment. For a complete analysis of SAMs and their various applications, readers are referred to other reviews (e.g., [60,70]). It suffices to point out that SAMs possess advantages in the precision (down to molecular level) of the control that can be applied in mechanistic studies [60,66] of cell-ECM interactions, and are potentially useful for cell-based diagnostics among many applications. However, this approach has certain limitations. First, it mostly relies on coupling between electrical, chemical (including pH), mechanical, thermal, optical and biochemical (e.g., protein conformation) cues whose applicability under in vivo conditions is problematic. Next, the resulting changes in these studies are mostly of surface biochemical properties or of the presentation and biological activities of the surface ligands. Nevertheless, SAMs have greatly facilitated the probe and understanding of cell-ECM interactions and particular interplay between cells and ECM with dynamics in adhesive properties.

Polymeric hydrogels
Hydrogel materials are gaining popularity in the development of biomimetic materials, primarily due to the hydrated nature of natural ECM [14,71]. Implantable hydrogel materials are increasingly being used in cardiovascular disease, nerve regeneration, and other conditions [59]. With careful design, hydrogel materials can have tunable materials properties, which have been demonstrated in a myriad of examples (Table 1). For instance, different than SAM-based approach, a polymer with both thermo-and photo-sensitivity was used to reversibly control adhesion of a group of cells [72]. Kim and colleagues took advantages of the thermo-responsive swelling behavior of copolymer between NIPAM and sodium acrylate, and created a hydrogel film that can be used to control cell encapsulation with surface topography [73]. Moreover, biomaterials responsive to the natural stimuli such as those experienced by biodegradable materials were found useful in mimicking biological events under physiological conditions, as illustrated in cell invasion to a MMP-responsive hydrogel scaffold [74]. This finding, among others, exemplifies the strategy of triggering material dynamic from bio-responsiveness to potential site-or disease-specific cues. The information from these studies is instrumental to the design of biodegradable materials in optimizing degradation profile for target cellular responses [75]. Naturally, in order to achieve desired outcome in adopting these strategies, it is important to gain thorough understanding of the natural environment, and minimize risks associate with biodegradable materials such as premature degradation, and potential toxicity of intermediate products from degradation.
Using a popular polyacrylamide hydrogel culture system with modifications that impart it with photo-sensitivity, Wang and colleagues [76] showed that upon UV induced substrates softening, spreading of 3T3 fibroblasts was hindered in contrast to that under static conditions ( Fig. 2A). More interestingly, localized softening at anterior and posterior of cells yielded differential cellular morphology and migration responses [76]. Meanwhile, a PEG based polymer (PEGA) crosslinked by photosensitive crosslinker (PEGdiPDA) has been developed by Kloxin et al. [77], and used to lower gel stiffness upon UV exposure, which resulted in de-activation of myofibroblasts (Fig. 2B). Although UV radiation is preferentially avoided, these methods made possible high precision in applying changes of cellular mechanical microenvironment, and potentially allow creation of dynamic stiffness gradient. Similar observations were made by Burdick group for human mesenchymal stem cells (hMSCs) by using hyaluronic acid (HA) hydrogel degradable from hydrolytic and enzymatic reactions [78] (Fig. 3A). Very recently, a new material platform has been constructed by this group [79] and others [80] where the stiffness of a methacrylated hyaluronic acid hydrogel is increased via addition of photo-initiator and UV light exposure. In response to elevated stiffness, human mesenchymal stem cells (hMSCs) spread more and exert greater traction forces in hours (almost one magnitude of difference), and the rate of stiffness elevation dictates fate of cell differentiation towards adipogenic (slower) or osteogenic (faster) lineage. Their work highlighted that cellular behavior on dynamic gels is not the same as that on static gels with same initial or final properties, underlining the significance of dynamics in gel properties. This has been echoed in the concurrent work [81], where, for instance, the fibroblasts on 100% crosslinked hydrogels demonstrated different morphology from that on 100% crosslinked gels modified from 50% gels. Therefore, it is conceivable that the previous state of the cells and their ECM is also among the determinants of their current state, and that time dimension of ECM is of great importance.
Factors other than environmental conditions (e.g., light, pH, temperature) can also be delivered to stimulate dynamics in substrate properties. Incorporating DNA as crosslinker, Jiang and colleagues have developed a hydrogel system for cell attachment where mechanical properties of the substrates can be altered in situ in a controlled fashion when the cell culture is present [20,81]. These DNA crosslinked hydrogels may also be designed to be potentially responsive to bio-stimuli, such as temperature or enzymes. Two representative cell types were chosen for the study of cell responses to dynamic substrate: fibroblasts whose sensitivity to mechanical cues is well documented, and neurons whose mechanosensing capability has recently just started being appreciated. The reports [20,81] offered evidence that both cell types do respond to dynamic alternations in the mechanical characteristics of ECM, and suggest that the alternations in the mechanical stiffness may be involved in disease progression (Fig. 3B). It has been shown that the stiffness change resulting from pathological processes, may also aid in further progression of diseases [82].
The same material system was employed by Previera and co-workers, and they firstly proved the dual mechanical stimuli, namely strain and stiffness drop, during the dynamic processes, and secondly contrasted cell behavior to stiffness decrease with that for hardening of the substrates [83]. On hardening gels (from 12 kPa to 22 kPa), cells spread more than those on static substrates of higher stiffness (22 kPa), whereas on softening ones, they have greater spreading area than that on either starting or ending stiffnesses. In these studies, cell responses are determined by the range of rigidity change (due to crosslinking density), starting and ending rigidity, and specific cell properties (e.g., projection area vs. aspect ratio and protrusion for fibroblasts). The stress generation may also be involved in affecting cell behavior [83].

Other types of materials
The approach of employing polymeric hydrogel to study dynamic changes has certain limitations, one of which is the coupling of mechanical stiffness and forces (e.g., [83]). To address this concern and others, different from the approach by using SAM or polymeric hydrogel, AFM based method put forth by Webster and co-workers [84] probed cellular response to instant step change in stiffness excluding influence from stress or strain in the substrates (Table 1). It has been confirmed that indeed individual 3T3 fibroblasts are able to sense and respond to the stiffness in a scale of seconds as demonstrated in traction rate and contraction velocity [84]. However, this approach is most likely with inherent limitation in mimicking natural cell environment while remains an interesting tool in probing cellular responses to instantaneously change in stiffness. Additionally, this approach is applicable mostly to cells with dynamic morphology.
Common cell culture substrates (e.g., glass coverslip) modified with common photo-cleavable agents (NPE-TCSP) were shown to be useful for controlling cell adhesion selectively and temporally [62]. In this method, target areas were first irradiated to remove BSA known to prevent cell adhesion, and then pro-adhesive molecules (e.g., fibronectin) were added, followed by cell seeding. It is useful to study dynamics in interactions between single cells. Petersen et al. [63] used light to stimulate photosensitive surface modification resulting in uncovering of the RGD motif upon release of a caging group (Fig. 4A). In doing so, adhesion of 3T3 fibroblasts was first inhibited and then encouraged, although this process is not reversible. With a sequential activation of adhesive sites upon application of electric potential, a recent study [85] demonstrated the utility of substrates with ITO (indium tin oxide) microelectrodes modified with poly(ethylene glycol), or PEG, in co-culture of two cell types (hepatic cells and fibroblasts) in a controlled manner (Fig. 4B). . Extracted from [63]. (B) With PEG-modified ITO microelectrodes on glass substrates (B1), co-culture of two different cell types, HepG2 (hepatic) and 3T3 cells, was made possible. Extracted from [85]. Both with permission from publishers.

Promising substrates
By applying an oscillation to a resilin-like polypeptide network crosslinked by THPP (β-[Tris(hydroxymethyl) phosphino] propionic acid (betaine), Li and colleagues were able to observe dynamic mechanical stiffness of the gels varying with regard to oscillatory frequency mimicking the load from human vocal tissues [86] Oscillatory shear induced stiffening and softening of the collagen network might also serve as good substrates for mimicking cellular microenvironment particularly that in mechanically active tissues [87]. Ion concentration may be another stimulus to allow for temporal modification of hydrogel properties as exchange of ions between monovalant and divalant cations [88], and further work is needed to confirm it. Temperature-dependent substrate softening has been demonstrated by Krekhova et al. [89] and 3D complex with temperature-mediated crosslinking hence mechanical properties has been proposed by Stahl et al. [90], while applicability of these systems in mimicking cellular microenvironment remains to be seen. Zustiak et al. [91] reported mechanical stiffness drop, from approximately 1 kPa and at different rate, along with degradation of a poly(ethylene glycol), or PEG, hydrogel which might serve as not only drug delivery vehicle but also biomaterial construct, and they have offered preliminary evidence of good viability of 3T3 balb fibroblasts on the hydrogel substrate. Similarly, rigidity change from ~180 kPa to tens of kPa in 3-week period of degradation from a photo-crosslinked alginate hydrogels based on alginate methacrylation were presented by Jeon et al [92], and cyto-toxicity has been found to be low. In summary, these substrates holds promises as substrates with modifiable properties in situ, and need to be carefully tuned and evaluated for use as substrates with dynamic properties (Table 1). Other materials responsive to various stimuli including pH, temperature, and biochemical factors for a variety of applications, including can be found in the earlier reviews [64,70,93], and thus are not discussed in detail here due to the focus of the current analysis.

Dynamic properties of the substrates
As indicated in the discussions in Background and Motivation, the progression in changes of ECM properties is also critical in addition to changes per se in light of the observations in normal and pathological tissues, development and aging, and potential engineering benefits. Towards this end, rate of change (e.g., gradual vs. abrupt), range of change (e.g., small perturbation vs. drastic modifications), and change profile (e.g., monotonic increase vs. fluctuation) characterizing the nature of changes and their impact on cellular processes are subject to research effort, apparently adding to the complexity of the problem (Table 1). Take biodegradable material (e.g., [94]) as an example. It would be relevant to understand how mechanical and adhesive characteristics evolve with degradation and how the degradation profile affects the changes in the cellular micro-environment. Experimental design along this line may include, for example, different rates of release of RGD motif decreasing adhesiveness while keeping the same range of change (e.g., half of the total RGD presenting sites), or altering the range of change while maintaining the same rate of change. Furthermore, it is not clear at this point whether cellular responses to opposite changes (e.g., increase vs. decrease in adhesiveness or rigidity) of substrate properties are symmetric, thus their behavior to one direction of dynamic alterations may not be a reliable predictor of that to the opposite changes.

Potential effect on cell-cell interactions
Changes in adhesive or mechanical properties of ECM can stimulate cells, which, in response, secrete soluble factors and ECM molecules, and this further impacts neighboring cell types. Additionally, some cell types such as neurons may use other cells (e.g., astroglia) as substrate [95], and stiffness change of 'underlying' cells per se due to ECM stimulation may give rise to further alternations thanks to cell-cell interactions. For instance, during asthma, ECM stiffening contributes to stiffness increase of airway smooth muscle (ASM) cells, which potentially affects other cell types in the close proximity [33].

Design parameters for biomaterials and outlook
The design parameters of dynamic substrates from current studies are summarized in Table  2, which includes, but are not limited to, material system to consider (e.g., SAM or polymeric hydrogel materials), nature of change (mechanical stiffness or adhesion), rate of change (e.g., transient or gradual change, controllability of the rate of change), range of change (e.g., at different stiffness range) as well as potential issues in further investigations and applications to medical and healthcare applications. If the interest is in understanding the cellular behavior to mechanical stiffness alone, then an AFM based approach might be more attractive [84] as others will involve stress or strain as part of the stiffening or softening process. If precise control over stiffness range is desired, the DNA crosslinked PAM hydrogel system will serve the purpose better [20,58,81]. Polymeric hydrogel materials with controllable degradation profile and hence mechanical stiffness dynamics during degradation (e.g., [88]) will serve the purpose best when biodegradable materials are applied. Some of the material systems do offer unique benefits such as reversible property change or without using environmental factors (by applying oscillation, crosslinker, or ssDNA).
Meanwhile, there are inherent limitations to each of the material system under discussion (Table 2). UV exposure generally causes concern to its impact on cellular activities despite the findings of little impact from a number of studies based on a range of biological assays. Under physiological conditions, application of certain cues (e.g., ssDNA, light, or ion) might be too difficult or it might be greatly limited (e.g., temperature triggered changes). However, it is still possible to find ways to apply these cues with careful design. For instance, ssDNA design based on pre-screening using BLAST search against targeted specie or tissue type may minimize the chance of interfering with normal biological activities. Local heating/cooling may be carefully applied to induce dynamic changes to achieve cellular responses. Three dimensional system may better mimic natural cellular micro-environment than their 2D counter parts. Highly dynamic Strain/stress that is associated with oscillation [86] Note: N/A: not available; This is a partial list of the current work under examination. A few new material system have been identified with the potential as dynamic cell culture platform as well as choice of biomaterials ( Table 1). Many of them have demonstrated good cyto-compatibility, and investigation of impact of in situ changes to cells will be desired.

Concluding remarks
There is an increasing recognition of the discrepancy between static nature of the current cell culture substrates or scaffolds and the dynamics in ECM in natural or diseased tissue, during development and aging, or at tissue-scaffold interfaces. This has motivated the development of materials with controlled changing properties that mimic those of ECM. An array of stimuli, including environmental factors (temperature, pH, light, electrical potential) and non-environmental cues including enzyme and DNA, have been implemented to trigger dynamics in a number of material platform such as SAMs, polymeric hydrogels and other substrates with surface chemistry and modifications.
To date, most of the effort along this line has been devoted to in vitro models, and in vivo studies of the effect of dynamic tissue properties on cellular behavior are still rather limited, which awaits further development in cell biology and proper tools such as imaging techniques [12,14,29].
Understanding the interplay between cells and the extracellular matrix (ECM) including its dynamic aspect is fundamental to biology, development, aging and pathology, and can aid in the design of biomaterials. Ultimately, the system enabling both spatial and temporal control [96] of cells would be most relevant in terms of bio-mimicry and tissue engineering applications. Some of the potential directions include creating dynamic adhesive gradient to guide cell migration or neurite outgrowth at desired time point, constructing scaffolds with suitable mechanical rigidity to inhibit glia cell growth (thus hinder scar formation) while promoting nerve regeneration with compliance gradient, and developing dynamic platform for stem cell harvesting and differentiation for cellbased therapies.