Nano- and Microscale Chemical and Topographical Patterning of Synthetic Cell Scaffolds: from Hard to Soft Materials

Abstract

Over the past century, a growing body of work has demonstrated that cellular behavior is impacted by contact with the materials in the surrounding environment, at length scales from centimeters down to nanometers. Soft matter (such as native extracellular matrices) has historically been challenging to pattern with great precision, so early efforts to understand structured cell–material interactions in the 1990s took advantage of hard interfaces, leveraging fabrication methods developed for the electronics industry throughout the 60s and 70s. Ultimately, as it became clear that cells respond to not only topography and chemistry of their environment, but also mechanical properties, patterning methods have been extended to soft materials, although often with lower structural resolution. Here, we provide a historical overview of the development of structured cell scaffold interfaces, highlighting the potential for additional advances in material patterning translated from hard to soft matter.

Introduction

Significant aspects of cell behavior are determined by interactions between the cell and its environment. The earliest examples of the phenomenon known as ‘contact guidance’ date back to the 1910s, when Harrison showed that if spider webs were suspended on cell culture solution, linear arrays of cells grew along strands of the web (Figure a, left). Footnotes in the manuscript describe the necessity of using web from a specific spiderthe stump-dwelling Tiginariaand bemoan the difficulties of finding adequate numbers of spiders early in the season. These early experimental difficulties foreshadowed a central dilemma in modern soft materials: biologically derived materials provide the most realistic scaffolds for cell growth, but are susceptible to batch-to-batch variability and limitations in supply. Ultimately, this has led to the development of key synthetic material analogs discussed here.

1.

(a) Cell–material interactions depend on topography, chemistry, and mechanical microenvironment. (b) Patterning surface topography (left) and chemistry (right). (c) Illustration of differences in structural regularity of hard and soft materials across length scales. (d) Many patterning methods (e.g., lithography) can be applied to either hard or soft materials. (e) Differences in hard and soft material structure and heterogeneity are important drivers of differences in patterning outcomes.

Throughout the 1900s, it became increasingly clear that more precise regulation of cell environments would be required to understand relationships to function, requiring levels of control that did not generally exist for biomaterials, and in many cases remain challenging. Due to these difficulties, many of the approaches used to pattern interfaces for biology have their roots in micro and nanoscale patterning methods developed for hard materials, which rapidly evolved throughout the 60s and 70s to allow micro- and nanometer-scale control in certain materials (Figure b). −

Although clear evidence emerged that topographical and chemical patterns in the cellular environment impact behavior, there was also a growing realization that mechanical environment is an important contributor to cell fate and behavior (Figure a, right). − For instance, a landmark study by Engler and Discher in 2006 showed that mesenchymal stem cells (MSCs) cultured on hydrogels with stiffnesses of ∼1 kPa, ∼10 kPa, and ∼100 kPa committed to different phenotypes (neurogenic, myogenic, and osteogenic, respectively). For comparison, materials such as tissue culture plastic (TCP), silicon, and gold, have stiffnesses on the order of GPa, well above the range typical for most cell types in vivo. Thus, cell responses to chemistry and topography on hard materials may not mirror responses to softer structures in the extracellular matrix (ECM), elevating the importance of achieving greater control over soft material structure. However, the atomic-scale structures of hard and soft materials are typically quite different (Figure c).

As soft material patterning capabilities were developed, it emerged that in many cases, similar patterning strategies (e.g., mask-based photolithography, laser or electron beam scanning lithography, microcontact printing) can be used to control structure and functionality in both hard and soft materials (Figure d). However, spatial resolution, displayed ligand density, and other parameters are often quite different when the same technique is used to pattern hard vs soft materials. For instance, microcontact printing of chemical patterns (Figure e, left) on hard surfaces has typically relied on the formation of alkanethiol monolayers on gold. On such a surface, the lattice structure of the gold atoms provides a framework for orienting and ordering the thioalkanes, and thus also any displayed terminal functional groups, with typical nearest-neighbor distances of ∼0.5 nm. In contrast, a polydimethylsiloxane elastomer with tissue-like stiffness (<40 kPa) typically exhibits structural heterogeneity at scales of >10 nm. Microcontact printing, when applied to such a surface, does not produce a uniform, ordered molecular layer, due to the lack of regularity in the substrate.

Likewise, lithographic and other patterning strategies aimed at controlling topography (Figure e, right) may produce edges with nm-scale roughness on a hard material, but roughness at orders of magnitude greater scales for a soft material. Further complications arise due to differences in mesh structure for different classes of soft materials, which alter ligand display density and other properties. Due to the increasing technological importance of controlling chemical environments on and in soft materials, thoughtful modifications of existing techniques have enabled increasing flexibility in the design of soft materials, discussed in more detail in sections below.

Expansion into high-resolution patterning of soft materials also brings new opportunities to diversify chemical patterning strategies. − For instance, generating initial chemical patterns on a hard crystalline interface that enables high-resolution patterning, then transferring the pattern to a soft material surface − represents one route for combining desirable aspects of hard and soft materials in designing scaffolds for cell growth.

Here, we provide a perspective on the development of cell scaffold interfaces to date, focused primarily on synthetic scaffold materials. We begin with early work focused on patterning of hard material interfaces. To provide context for the additional degrees of complexity introduced in soft material interfaces, we then outline major chemical and physical features of biological extracellular matrices, including relevant length scales and mechanical properties. These then serve as a basis for discussion of soft interface patterning for cell scaffolds. Finally, we examine outstanding challenges for the field. While we focus here on connecting hard and soft material interfaces as they relate to cell scaffold design, interested readers are also encouraged to explore the range of excellent review articles focused on synthetic hydrogels (often including discussion of patterning), − and natural decellularized ECM materials, , as well as other reviews focused on specific processes (including bioprinting), − or properties (such as viscoelasticity). ,

Topographical Patterning of Hard Materials to Control Cell Adhesion, Spreading, and Differentiation

In the 1990s, the routine availability of micro and nanofabrication strategies developed for electronics led to a significant body of work illustrating that small-scale surface topographical patterning (e.g., grooves, grids, and pillars) impacted cell adhesion, spreading and migration. These studies were typically carried out with materials already known to be compatible with lithographic processing (e.g., Si/SiO₂, polyurethane, polycarbonate, quartz) which are much stiffer than biological soft tissue, but offer precise control over fabricated geometries and have low surface roughness. Surface feature geometries tested in such scenarios are often commensurate with cellular component dimensions, with groove depths 100 nm–1 μm common, and some studies using depths in the range 1–10 μm or larger. Groove widths are usually greater than or equal to groove depth.

Broadly, most studies agree that cell growth is modulated by topographyfor instance, most cells elongate and align to some extent with parallel surface grooved structures (Figure a). However, the extent of alignment is impacted by groove height, width, and periodicity, with different dimensions leading to different degrees of alignment based on cell type (e.g., corneal epithelial, myoblasts, fibroblasts, neurons), suggesting that there are advantages to topographical dimensions matching those of cell-specific subcellular structures. Additionally, some cell types are more impacted by topography than others.

2.

(a) MDA-MB-231 cancer cells grown on topographical ridged patterns exhibit greater actin filament and microtubule formation in the channels between ridges. (b,c) Neurons can exhibit either perpendicular or parallel contact guidance when grown on grooved surfaces, with Xenopus spinal neurites exhibiting predominantly parallel guidance (white bars) and rat hippocampal neurites exhibiting predominantly perpendicular guidance (black bars) for narrow grooves, with a transition to parallel guidance for wider (4 μm) and deeper grooves (closer to 1 μm depth). Adapted with permission. , Ref Copyright 1997, The Company of Biologists. Ref Copyright 2018, Cell Press.

Although cells typically align parallel to groove structures, in some cases they can also align perpendicular to grooves (referred to as perpendicular contact guidance, Figure b, center). For neurons, this behavior is more commonly associated with shallow, narrow grooves (e.g. 1 μm wide) while deeper, wider grooves (e.g., 4 μm wide) produce predominantly parallel alignment. This behavior can again depend on cell type: for instance, rat hippocampal neurons exhibit perpendicular guidance, while Xenopus neurons do not (Figure c).

Cell migration (which requires both adhesion and mobility) can also be impacted by topography, with cells that undergo mesenchymal migration (elongated cell morphologies during migration, typically associated with stronger cell adhesion to the surface, e.g. 3T3 and other fibroblasts) exhibiting stronger impacts than cells undergoing amoeboid migration (rounded cell morphologies during migration, typically associated with weaker cell–surface adhesion, e.g. fibrocarcinoma cells). Surface topography has also been shown to shape formation of focal adhesions, nanoscale clusters of proteins that provide mechanical contact between the extracellular environment and the cytoskeleton. This occurs by orienting actin stress fibers, which originate from focal adhesions, and by extension of acto-myosin networks inside the cell (as well as potentially microtubule networks).

Chemical Patterning of Hard Materials to Control Interactions with Cells

Many studies also illustrate that geometric chemical patterns on flat surfaces impact cell growth. − For instance, in 1988, silicon and quartz surfaces lithographically patterned and functionalized with silanes were used to pattern outgrowth of dissociated neurons. Later work illustrated that lithographically defined patterns of increasingly diverse functionalities (e.g., oligoethylene glycol, fibronectin), could be used to control cell adhesion to both silicon substrates (via silane chemistry, Figure a) and metallic substrates (via alkanethiol chemistry, Figure b). − ,− For instance, a broadly used approach on metallic substrates is to pattern areas with a hydrophobic (e.g., methyl-terminated) alkanethiol, then backfill the unpatterned areas with a hydrophilic (e.g., oligoethylene glycol-terminated) alkanethiol. Subsequent exposure of the surface to fibronectin (FN, functionality is discussed more extensively in the section on structural elements of the ECM below) in solution results in physisorption of the fibronectin in the patterned hydrophobic areas, which has been shown to promote cell adhesion. However, the conformation of the fibronectin adsorbed to the surface is not controlled, since it may partially unfold during adsorption.

3.

(a) Murine neuroblastoma cells (top) and osteoblast-like cells (middle and bottom) cultured on linear patterns of amine and alkyl-terminated silanes printed on Si/SiO₂, showing preferential adhesion to amine-functionalized surfaces. (b) Human mesenchymal stem cells display different adipogenesis and osteogenesis profiles based on surface chemistry pattern. (c) Cells cultured on arrays of nanoscale dots and lines of hydrophobic alkanethiol on gold with adsorbed FN, illustrating increased alignment of actin filaments in patterns with nanoscale lines near edges/corners. Adapted with permission. − , Ref adapted under the terms of the CC-BY license, Copyright 1994, Wiley. Ref Copyright 2019, American Chemical Society. Ref Copyright 2010, National Academy of Sciences.

Such functional patterns were used in 1997 to control cell adhesion and spreading to desired shapes and sizes (e.g., squares vs circles, with controlled linear dimensions), in some cases correlated with differences in cell survival. The smallest square patterns (5- and 10 μm edge lengths) were associated with a lack of cell adhesion and/or apoptosis, while larger patterns produced greater cell spreading areas and increased DNA synthesis and nuclear spreading. Simpler functional groups can also be effective in controlling cell adhesion, as shown in Figure a, where stripes of amine-terminated silanes (N-(2-aminoethyl)-3-aminopropyl-trimethoxysilane, EDS) induce preferential adhesion of both neuroblastoma and osteoblast-like cells.

Chemical pattern shape can also be an important factor in influencing cell fate. In Figure b, mesenchymal stem cells cultured on FN patterns displayed preferential osteogenesis when exposed to pointed patterns promoting increased myosin contractility (Figure b, right). In contrast, cells exposed to patterns with rounded edges that promote low contractility (Figure b, left) were more likely to commit to adipocytic lineages.

As nanoscale patterning methods progressed in materials for electronics, parallel advancements led to higher-resolution patterning of hard surfaces for biology. For instance, scanning probe lithographic methods, such as dip-pen nanolithography, , enable molecular inks to be delivered with much smaller lateral feature sizes (typically tens of nanometers). Combinations of nanoscale and microscale functional patterns (e.g., hydrophobic patterning of FN, as described above), have shown that nanoscopic structure, particularly at the edges and corners of microscale adhesive structures (Figure c), are also correlated with the alignment of stress fibers in the cytoskeletons of adherent cells, amplifying the effect associated with microscale pointed geometries.

Structural and Functional Elements of the ECM: Targets for Synthetic Materials

Based on initial work in simplified model systems, it was clear that many cell types exhibit strong responses to their structural and chemical environments. However, such relationships are often complex. This led to a growing interest in designing materials that recreated key elements of the cellular niche more closely, including soft polymeric mesh structures, and more complex ligand display.

To understand the core requirements in chemical and physical patterning of cell scaffolding materials, it is useful to understand structural elements and functions of the extracellular matrix that such materials aim to reproduce. Here, we provide a brief overview of key elements of ECM structure, highlighting nanoto-microscale structural elements potentially important to replicate in synthetic materials, as well as tissue-to-tissue variability that increases challenges associated with material design.

In multicellular organisms, cells exist within a well-defined extracellular matrix (ECM, Figure a) that provides cues for development (e.g., differentiation from stem cell to a tissue-specific lineage such as osteoblast (bone) or myoblast (muscle)). The ECM also provides scaffolding for the interaction of cells to form higher-level structures such as tissues and organs. ,−

4.

(a) Nano and microscale structural components of the extracellular matrix (ECM), including (b) glycosaminoglycans such as hyaluronic acid (HA) and heparin, and (c–e) structural proteins such as (c) collagen (PDB: 1K6F), (d) fibronectin (PDB: 1FNF), and (e) laminin (PDB: 4AQS). Panel (a) adapted with permission. Ref Copyright 2021, American Chemical Society. Electron micrographs in (c–e) adapted with permission. , Ref Copyright 1981, Elsevier. Ref Copyright 2018, Elsevier.

The bulk of the ECM structure is defined by large polysaccharides (i.e., hyaluronic acid (HA), , heparin, and cellulose, Figure b), and fibrillar and other functional proteins (i.e., collagen, elastin, fibronectin, and keratin, Figure c–e), which together are responsible for stiffness, elasticity, adhesion, and other meso-to-microscale structural properties including porosity. Supporting its multiple complex functions in scaffolding cells, the ECM also includes many soluble growth factors (e.g., fibroblast growth factors (FGFs), vascular endothelial growth factors (VEGFs), hepatocyte growth factors (HGFs)). The cellular environment is defined by thousands of tissue-specific ECM components (often collectively referred to as the matrisome); together these enable hierarchical mechanical and chemical control over the cellular placement and function.

In the context of synthetic cell scaffold development, it is typically not feasible to include all elements of biological ECM, so it is important to evaluate material complexity (e.g., number and expense of chemical components as well as difficulty of processing) vs the level of desired function provided.

Polysaccharides

The polysaccharide components of the ECM are often referred to as glycosaminoglycans (GAGs); each GAG is typically comprised of one or more disaccharide repeat units, one of which bears an amine functional group. The most abundant GAG in most ECM is hyaluronic acid (HA, also referred to as hyaluronan). HA is an extremely large (up to 25,000 repeat units), unbranched GAG. Each repeat unit bears one carboxylic acid and an acetyl group on the amine (Figure b, left), meaning there is one negative charge per two sugars along the polymer backbone. Due to the high number of negative charges distributed along the polymer, it has a relatively large persistence length (estimated at ∼4 nm in 200 mM NaCl based on small-angle X-ray scattering), and does not form a globular structure. Individual molecules may occupy a cubic volume with up to 300 nm in edge length. The highly charged structure also draws in large amounts of counterions (e.g., Na⁺) and water, and thus can withstand substantial compressive forces, which is important in many connective tissues such as cartilage. Unlike other GAGs, HA is synthesized and secreted directly from a protein complex in cell surfaces.

Other GAGs differ from HA in multiple aspects that are important from the perspective of chemical and mechanical function (Figure b, right). First, they are comprised of one or more glycans that are appended to a protein core, referred to as proteoglycans. This connectivity generates a branched structure, which may in some cases (e.g., agrin) include 100 or more polysaccharide chains, with chain lengths often ca. 100 repeat units. GAGs other than HA are comprised of more variable 2-sugar unit repeats; however, the unit typically includes one sugar with a carboxylic acid (glucuronic or iduronic acid), and an amino sugar, which typically bears one or more sulfates (e.g., heparin sulfate). The even more highly charged branched structures, and presence of sulfate groups, can be important in binding of growth factors and other functions in the ECM.

The growing understanding of the biological importance of HA and other GAGs has in recent years led to development of a wide range of synthetic glycopolymers with varying architectures, − as well as the direct use of polysaccharides including HA and alginate in scaffolds. , Although many synthetic glycopolymer materials remain relatively expensive to produce, due to the complex stereochemical demands of polysaccharide synthesis, they will increasingly be incorporated into scaffold materials in coming years.

Fibrillar Proteins

Of the fibrillar proteins in the ECM, collagen (Figure c) is the most abundant, but appears in several different forms with variable function. , The core building block of collagen is a Gly-X-Y (where X and Y are often proline and hydroxyproline) repeating sequence that assembles to form a triple helix ∼300 nm long and 1.4 nm in diameter. Collagen molecules are synthesized and assembled inside the cell, with protective peptide groups on both ends that prevent further assembly. Shortly after secretion from the cell surface, the protective peptides are cleaved, and the collagen molecules can undergo further assembly to form fibrils (10–30 nm diameter, lengths up to hundreds of μm). These can then be further assembled into larger fibers, with covalent cross-links forming throughout the network at sites where lysine residues are enzymatically deaminated to form cross-linkable aldehydes. These large fiber networks enable the ECM to resist tensile forces, critical in the function of skin, and in connective tissues like tendon. Estimates of the mechanical properties of collagen vary substantially depending on experimental technique, with persistence lengths of the individual collagen molecule ranging from 14 to 180 nm. , Interestingly, measurements place the Young’s modulus of individual collagen molecules near 5 GPa, while wet collagen microfibrils produce substantially lower values ∼300 MPa, thought to be due to the staggered conformation of molecules in the fibril, which allows for bending.

Although the most abundant variant of collagen is collagen I (the dominant form found in skin), there are many other variants with internal or terminal deviations from the Gly-X-Y motif, producing distinct localization or mechanical properties in hierarchical assembly to form the ECM. For instance, collagen IX, XII, and XIV, referred to fibril-associated collagens with interrupted triple helices (FACITs), contain multiple triple-helical domains separated by nonhelical segments. These collagens do not assemble into fibrils, but instead decorate the surfaces of fibrils and mediate interfibrillar connectivity as well as contacts with other elements of the ECM. Collagen IV, , common in basal lamina (vide infra), also has multiple nonhelical segments embedded in the main collagen molecule, resulting in additional flexibility and smaller mesh structures. Collagen VII forms dimers, with exposed peptide sequences that help anchor basement membranes to surrounding connective tissue. Thus, the basic collagen helical building block can generate a wide range of bulk mechanical properties, porous and sheet structures, and chemically defined adhesion points for cells, which contribute to tissue-specific ECM properties that can be important to replicate in synthetic materials.

In some ECM (e.g., artery epithelium), long-range alignment of collagen fibrils is critical for mechanical function, and in polarization of adherent cell layers. In some cases, collagen fibrils can be aligned by secretion and assembly in membrane folds, which can appear between intercellular actin filaments, meaning that the orientation of the cell cytoskeleton can be used to direct appropriate orientation of the surrounding ECM. ,, In other cases, fibroblasts are known to exert pulling forces on collagen fibrils in the environment, generating longer-range alignment. Thus, there is a bidirectional relationship between cell polarization and ECM alignment that adds to complexity in designing artificial ECM.

Several other structural proteins are also key components of the ECM, providing an additional toolkit for generating mechanical properties, mesh structures, and connectivity with cells. For instance, elastin is a key protein that operates in conjunction with collagen networks to provide elasticity to skin and other ECM environments. ,, Elastin has a large (∼750 amino acid) hydrophobic sequence interspersed with hydrophilic segments containing lysine residues that undergo covalent cross-linking, similar to collagen. The hydrophobic segment is thought to be partially disordered at rest, and can be extended to enable the matrix to stretch. Elastin is present with high abundance (up to 30–60% dry weight) in tissues that undergo regular stretching, such as the aorta and elastic ligaments. Unlike many elements of the ECM, elastin does not turn over regularly in the matrix, and is thought to have a half-life of ∼70 years. Thus, it can be important to generate appropriate elastic properties in synthetic ECM, since these may be more difficult to induce through ECM remodeling by resident cells.

Other proteins including fibronectin (Figure d) are responsible for generating connections between structural proteins and polysaccharide elements of the ECM, as well as connecting both elements to cells. Here, we will focus largely on fibronectin, which is commonly used in synthetic materials to control cell adhesion, either as the full protein, or often simplified to one of the short peptide sequences (RGD) that is central to adhesion. − Fibronectin is a very large protein with a complex structure that enables it to form modular connections between many chemically different elements of the ECM environment, including cells. , Two large subunits are joined by a disulfide bridge; each subunit has several domains with distinct functions (binding to collagen, heparin, cell surface receptors, etc.). Each domain, in turn, consists of smaller modules. The most common of these is the type III fibronectin repeat, which is ∼99 AA in length, appears at least 15× per subunit, and in some cases includes an amino acid trimer “RGD” that is a key binding motif for cell surface integrin proteins, and is presented as a flexible loop extending from the face of the type III repeat. In some cases, synthetic materials append a cyclic form of the RGD peptide (cRGD) to more closely mimic the native presentation. Fibronectin proteins mostly appear in the ECM as fibrils, which are assembled as fibronectin is secreted from fibroblast cells, and are typically aligned with intracellular actin networks. In basement membranes, laminin proteins (Figure e) are also very common interconnection elements , for this cross-shaped heterotrimeric protein, the three short arms often connect to other laminin molecules, helping to form the sheet-like lamina structure (vide infra), while the longer fourth arm binds to cell surfaces. Motifs from the cell-binding arm of laminin (e.g., IKVAV) are also fairly common in material design.

Overall, the programmable design of chemically appropriate synthetic materials with the degree of tunable mesh structure, porosity, and local member stiffness, combined with complex peptide ligand display and growth factor storage, remains challenging to fully replicate, particularly with appropriate tissue-specific variations.

Specific Structural Elements of ECM Including Basement Membranes

Certain structural elements of ECM have nanoscale features that are challenging to recreate using existing synthetic materials. Basement membranes are thin sheet-like structures found in many tissues, ,− with important examples including skin, skeletal muscle, vasculature, lung, kidney, adipose tissue, and peripheral nerve axons. From the perspective of materials design, basement membranes are especially interesting, since they highlight the importance of precise structural and chemical control across length scales.

Basement membrane (BM) is primarily composed of filamentous networks of collagen IV (for mechanical stability), laminin (providing connectivity to adjacent cells, often endothelial or epithelial cell layers) and glycans such as hyaluronic acid and heparan sulfate (providing functions including filtration) ,, As described earlier, collagen IV has >20 nonhelical regions along the molecular length, increasing flexibility, and retains solubilizing peptide end groups after secretion, meaning that it does not form the thicker fibrils observed in most connective tissue. However, it does form dimers through headgroup interactions, and subsequent lateral interactions to generate the nanoscopic porous mesh of the BM. Although collagen IV is abundant in mature BM, in the early stages of development, laminin is the predominant protein structural constituent, forming the initial sheet architecture.

The BM typically resides at a boundary between two different cellular environments in a tissue, so perhaps unsurprisingly, the chemical structure of the BM varies throughout its thickness, even though the entire BM may be only tens of nanometers thick. Early observations of the BM via electron microscopy identified a layer adjacent to cells that exhibited limited electron scattering (and was thus referred to as the lamina lucida) and a layer further from the cell that exhibited stronger electron scattering (lamina densa). , Thickness of the lamina densa is typically 10–100 nm, , but can be as great as 5–10 μm (e.g., lens capsule), with pore structures ranging from 10–100 nm in different types of BM (e.g., ∼10 nm for kidney glomerular (endothelial) and tubular (epithelial) BM, and ∼100 nm for human corneal epithelial BM). Young’s modulus is difficult to measure directly for many BMs, since they are very thin and difficult to separate cleanly from surrounding cells. When such measurements are possible, values are often in the range of 1–5 MPa, thought values can be as low as 0.5–5 kPa. In some cases (e.g., skin epithelium), modulus can also be different on opposite sides (epithelial and stromal) of the membrane.

The BM performs diverse functions that differ based on the type of tissue, and include providing a framework for tissue growth and regeneration, separation of epithelial and endothelial cell layers from surrounding connective tissue, and filtration (in the kidney and lung), as well as providing a reservoir for soluble growth factors, and potentially also oriented multivalent presentation of growth factors. Together, these represent significant, but potentially tractable challenges for synthetic materials hierarchical chemical and mechanical patterning.

For instance, muscle has a multinucleate fibrous structure, with a BM surrounding individual myofibers (Figure a, right). The BM provides a structured niche environment for satellite (stem) cells around the fiber periphery. When fibers undergo damage due to exercise, the BM provides a framework within which the existing fiber structure is degraded, and satellite cells migrate into the junction to repair the fiber. The BM also contains differentially structured regions enriched in the matrix protein agrin, which coordinate positions of neuromuscular junctions across the lamina, so that neurons can induce myofiber contraction. If the BM is breached at large scales (for instance, in volumetric muscle loss injuries, where a section of tissue is removed), fibroblasts migrate into the opening and deposit collagen, limiting the capacity of the myofiber to regenerate. Thus, the capability to generate ultrathin constructs with controlled 3D geometries that support myofiber alignment would be of significant value in regeneration following volumetric muscle loss injuries.

5.

(a) Schematics of basement membrane structures in muscle fibers and in the cornea. (b) Electron micrograph of microscale structure in the kidney glomerulus endothelium. (c–e) Electron micrographs of (c) corneal epithelial basement membrane microstructure, (d) corneal basement membrane nanostructure, and (e) Matrigel, a commercial basement membrane product extracted from murine epithelial sarcomas. Adapted with permission. − Ref Copyright 2013, Academic Press. Ref Copyright 1999, Cell Press. Ref Copyright 2000, Lippincott Williams & Wilkins.

In the cornea, two spatially distinct basement membranes form frameworks for epithelial and endothelial layer structure (Figure a­(left),d), − separating them from the dense, oriented layers of collagen that comprise the stroma layer. The thin structures remain transparent to light, and control diffusion of nutrients to endothelial and epithelial layers (of particular importance since the stroma lacks vasculature), while preventing overhydration and swelling of the stroma, allowing for correct intraocular fluid pressure. The endothelial basement membrane (often referred to as Descemet’s membrane) also prevents migration of growth factors into the stroma, where they can cause undesired activation of fibroblasts, producing scarring. Here, simultaneous control over optical properties, porosity, and mechanical robustness in an ultrathin chemically patterned layer are an important material challenge.

In the kidney, a somewhat thicker basement membrane (∼200 nm, Figure b) in the glomeruli lies between endothelial and epithelial cell layers, and is responsible for allowing free passage of water and small-molecule solutes, while restricting passage of macromolecules such as proteins, as the kidney carries out its normal filtration functions. Highly charged heparan sulfate glycans support filtration, and with super-resolution optical microscopy techniques, it has been shown that ECM proteins including agrin are localized to specific subsections of the membrane, suggesting they are active in filtration as well. From a materials perspective, understanding of kidney glomerular membrane structure is of potential importance in diseases such as diabetes, in which thickening of the BM due to increased collagen IV synthesis impairs waste filtration.

Challenges in Patterning Soft Materials

Much of the work in microscale and nanoscale patterning of synthetic soft materials has been aimed at the design of artificial ECM with nanostructural elements, which is an important challenge in modern materials chemistry, both for fundamental studies of cell culture, and for in vivo applications including regenerative medicine. ,,, Application-driven studies target implantable materials suitable for tissue repair (e.g., muscle, cornea, , bone, and cardiac tissue , ). In parallel, there is also significant focus on materials that enable in vitro cell culture under conditions that deconvolve impacts of competing chemical and mechanical factors, providing a fundamental understanding of cell biology that informs processes occurring in more complex in vivo environments.

Trade-offs in Material Selection for In Vivo and In Vitro Applications

For cell culture in vitro, materials such as collagen hydrogels can be used to recapitulate basic aspects of the ECM. Since details of collagen structure vary widely from tissue to tissue as described above (both blend of collagen sequences present, as well as their assembled structures), this represents an important consideration in selecting a collagen source; batch-to-batch variations in composition may also impact gel structure. Products such as Matrigel (Figure e), a homogenized soluble basement membrane extract of Engelbreth–Holm–Swarm murine epithelial tumor tissue, , are often used in cases for which a more realistic ECM composition is required. A convenient aspect of Matrigel and similar products is that they contain major protein and polysaccharide components of basement membrane (e.g., laminin-111, collagen IV, entactin, and heparan sulfate proteoglycan), as well as complex mixtures of growth factors (e.g., FGF, EGF, TGF beta, IGF, and PDGF) and proteases; together these components often support cell growth more successfully than simpler matrices. However, compositional and structural variability (e.g., variable abundance of growth factors) can become even more significant for these complex mixtures of natural products.

Challenges including batch-to-batch variations in natural products and the need to limit immunogenic responses for implantable materials have meant that fully synthetic materials suitable for regenerative medicine and tissue engineering are of growing importance, with synthetic hydrogels among the most commonly used materials. , Synthetic hydrogels (e.g., acrylamide, polyethylene glycol diacrylate, etc.) (Figure a) are typically polymerized through radical addition (vs condensation to form biopolymers), yielding polymers with different backbone structures, cross-linking architectures, degradation chemistries, and other properties at the molecular and mesoscale in comparison with biological ECM. For instance, most synthetic polymers with saturated sp³-hybridized carbon backbones have persistence lengths on the order of 1 nm, smaller than values for GAGs such as hyaluronic acid (persistence length ∼4 nm) and much smaller than structural polymers such as collagen (16–180 nm). At the same time, the diversity of synthetic monomers available means that parameters including hydrogel wettability, bulk stiffness, dimensionality, porosity, and degree of cross-linking can be chosen to match desired bulk ECM scaffold properties. ,,,− For instance, in Figure b, average pore sizes from >1 μm to 100 nm are created in PAAm hydrogels by altering the percentage of bis­(acrylamide) (from 0.03% to 0.3%). However, the mesh structure is noticeably different than the more fibrillar native ECM (e.g., Figures and ), in part due to the lower persistence length of the synthetic polymer. For instance, in Figure c, cross-linked clusters with diameters ∼4 nm form higher-order aggregates detectable in both small-angle X-ray scattering and dynamic light scattering, but are not visible in cryoEM.

6.

Overview of some synthetic hydrogels commonly used as cell scaffolds, including (a) synthetic hydrogel monomers, and (b) cryoEM of polyacrylamide with pore size varied based on cross-linker percentage ((left) 0.03% bis­(acrylamide); (right) 0.3% bis­(acrylamide)), (c) hierarchical PAAm gel structure including nanoscale architectural elements not visible after cryopreparation. (d) Schematic of hydrogel including sites cleavable by matrix metalloproteinases. (e) Fibroblast radial invasion of MMP-cleavable hydrogel over 7 day period. Adapted with permission. , Ref adapted under the terms of the CC-BY-NC license, Copyright 2020, Royal Society of Chemistry. Ref Copyright 2017, American Chemical Society.

The Need for Chemical Functionalization

Because synthetic scaffolds materials often bear very little chemical similarity to native ECM, the bulk material is frequently functionalized with biomolecules (e.g., polysaccharides, peptides) to mimic specific chemistries of the ECM, including cell–ECM contact sites. , For instance, in tissues, ECM structural proteins often undergo targeted degradation by matrix metalloproteinases (MMPs), to facilitate cell migration and tissue growth or regeneration. Saturated carbon backbones present in many synthetic hydrogels are not amenable to MMP degradation, therefore some gel materials designed for implantation incorporate peptide linkers with sites for enzymatic degradation (Figure d), which can improve cell growth into the material. Figure e illustrates the radial growth of fibroblasts into an MMP-cleavable PEG hydrogel (also functionalized with RGD adhesion sites), over the course of 7 days.

However, structural differences between synthetic hydrogels and native ECM also creates challenges in achieving adequate control over chemical functionalization. For instance, for many synthetic hydrogels, one straightforward method of modification is to use unreacted acrylate groups on the hydrogel mesh as functional handles for addition of new chemistries, producing a stochastic distribution of the functionality throughout the postfunctionalized volume of the gel mesh. For soft materials containing appropriate functional groups, (1-ethyl-3-((3-(dimethylamino)­propyl)­carbodiimide/N-hydroxysuccinimide (EDC/NHS)) , coupling and other standard coupling chemistries can also be used to immobilize biomolecules including RGD HA, laminin, heparin, and collagen. Stochastic functionalization methods, while chemically straightforward and broadly useful, typically produce ligand densities that vary with gel mesh size (Figure a), which can make it difficult to deconvolve chemical and mechanical contributions to cell behavior.

7.

Challenges in surface patterning of soft materials, including (a) structure-dependent ligand density achieved through stochastic ligand addition, (b) deformation of soft or sticky materials during microcontact patterning creates stiffness-dependent patterning defects. Panel (b) adapted with permission. Ref Copyright 2012, Elsevier.

The structural complexity of biopolymers also enables precise clustering and placement of chemistries in the matrix for cell adhesion and other functions, which can be challenging to replicate using synthetic hydrogels. For instance, ligand accessibility to adherent cells can also vary with pore size (Figure a, bottom). At the same time, patterning methods such as contact lithography and photolithography, which have been broadly successful in chemical and topographical patterning of hard materials must often be modified significantly for soft materials, and/or produce much larger minimum feature sizes. For instance, microcontact printing of soft materials can be challenging, because if the PDMS stamp (often Sylgard 184, Youngs modulus ∼2 MPa) is stiffer than the receiving substrate (Figure b shows examples 2–300 kPa), the receiving substrate can deform during printing, creating pattern defects. In response, methods including force-induced micropatterning and patterning using surface standing acoustic waves (SSAW) on soft substrates have emerged as alternative strategies to achieve micron-scale resolution topographical control for cell alignment on surfaces. ,

Methods for Chemical and Topographical Patterning of Soft Materials

Despite the significant challenges in controlling and characterizing structure in soft materials, as described above, the great value in doing so has driven rapid progress across a variety of patterning techniques, often requiring ingenious modifications to protocols originally developed for hard materials Here, we provide an overview of photopatterning and printing methods applied to control local structure of soft materials.

Photolithography and Photopatterning

Photopatterning, either using a mask (Figure a), or using one or more scanned laser beams in a confocal microscope (Figure b), has increasingly been used as a means of patterning soft materials, ,− due to the degrees of freedom in patterning geometry, and the capability of patterning both inside and on the material. Such approaches can be used to add new chemical functionality and/or induce local polymerization or polymer network degradation.

8.

(a,b) Schematics of (a) photomasking and (b) laser scanning lithography for reactive patterning of hydrogels. (c) Human dermal fibroblasts cultured on PEGDA hydrogels photopatterned with microscale linear patterns of RGD peptides. (d) 3D pattern of two fluorophore-conjugated peptides photoconjugated to the hydrogel using thiol–ene click chemistry. (e) Example photopatterning chemistries utilized in reactive patterning of hydrogels. Panels (c–e) adapted with permission. , Ref Copyright 2006, Elsevier. Ref Copyright 2011, Nature Portfolio.

While photolithography routinely creates feature sizes <10 nm in semiconductor chip fabrication, chemical patterning of soft materials raises new challenges. For instance, in lithographic processes on hard materials, photoexposure is often carried out on very thin (<100 nm), densely cross-linked polymer films, with exposed areas subsequently dissolved or resisting solvent exposure, rather than carrying out advanced chemical functions. This distinction is important in regards to the use of short-wavelength UV/EUV radiation, which is routinely used to achieve small feature sizes in hard materials. Soft materials undergoing chemical functionalization through photolithographic processes are typically meant to host cells or carry out other chemical functions following exposure, potentially limiting the use of high-energy photons and high photon fluxes, which have the potential to create unwanted side reactions that alter material function.

A common example of chemical function added through photopatterning is a reaction to append RGD adhesion peptides based on the integrin binding sequence of fibronectin. One of the earliest examples of mask-based photolithography was carried out on PEGDA hydrogels, generating 40 μm-wide linear patterns of acrylate-conjugated RGD peptides and 2,2-dimethoxy-2-phenylacetophenone as a photoinitiator. Peptide conjugation densities of 0.02–0.15 mg/cm² were reported by cleaving peptides from the gel and quantifying using a ninhydrin assay. Patterns were subsequently used to control human dermal fibroblast adhesion (Figure c).

Later approaches applied new classes of reaction chemistries, including high-efficiency click reactions aimed at maximizing functionalization yield with expensive peptide and biomolecular conjugation reagents. An early notable example used locally photogenerated Cu­(I) to increase reaction rates for an azide–alkyne cyclization, generating 35 μm-wide and 4.2- μm deep features based on 30 μm-wide mask features, with the diffusivity of Cu­(I) contributing to pattern broadening. Building blocks such as dibenzocyclooctyne (DBCO)-conjugated biomolecules also have emerged as alternatives for reacting with azide-containing counterparts via strain-promoted, metal-free azide–alkyne cycloaddition under physiological temperature at pH 8; photoclick cyclization reactions have also been reported for hydrogel functionalization.

Two-photon lithographic processes , have been increasingly broadly used since 2008, when it was demonstrated that the process could be used to pattern RGD peptides in microscale focal volumes in 3D. This approach can also be used with a wide variety of chemistries; for instance, thiol–ene click reactions (Figure d,e) using an eosin Y photoinitiator and multiphoton excitation (860 nm) with a scanning laser in a confocal microscope were subsequently shown to generate patterns with ∼1 μm x–y resolution and 3–5 μm z resolution in a 300 × 400 × 400 μm 3D structure, with peptide patterning concentrations 0–1 mM.

While synthetic hydrogel chemistries can be chosen to include reaction sites for subsequent patterning, most biopolymers (e.g., collagen and gelatin) used in hydrogels for cell culture are not intrinsically photoreactive. However, straightforward reaction chemistry introduced by Van Den Bulcke in 2000, using methacrylic anhydrides, enables installation of reactive methacrylate moieties on amino side chains, generating biopolymers that can be self-cross-linked and undergo photopatterning reactions as described above. It is important to note that the density of reactive sites produced on native collagens is highly tissue-specific, again underlining the importance of controlling collagen source in designing cross-linked collagen gels.

Inkjet, Droplet-Based, and Extrusion Printing

Inkjet, droplet-based and extrusion printing methods are inexpensive, rapid, and scalable, and can be used to deliver a wide range of material inks. ,− In general, the relatively large feature sizes and fluid dynamic constraints (e.g., upper and lower limits on ink viscosity) on such printing processes have meant that, in the context of cell scaffolds, these methods are most successful in generating relatively large 3D soft matter constructs with features at multicellular scales. ,−

In standard inkjet printing applications (e.g., printing documents on paper), droplets are typically ∼20 μm in diameter, and are jetted from the cartridge either by local heating, creating a bubble that forces the droplet out of the cartridge, or using a voltage pulse to a piezoelectric actuator, which deforms, creating pressure on the ink in the cartridge and forcing a volume of it out of the nozzle. The jetting process places certain constraints on the ink material (e.g., viscosity and surface tension). Additionally, the droplets undergo significant fluid dynamics both during jetting and during impact on the substrate. Inks outside a specific range of viscosity and surface tension fail to form discrete droplets with controlled volumes when exiting the print head, or splash, creating satellite droplets when impacting the substrate. For inks commonly used in nonbiological applications, droplets spread on impact to form a rounded feature ∼50 μm in diameter, meaning that the best realistic printing resolutions are on the scale of one or more cell diameters. Printing of multiple adjacent droplets can result in fusion to form linear patterns. However, precise edge structures on the cellular or subcellular scale are difficult to achieve, which can be problematic given the sensitivity of many cells to the geometry of chemical and physical edges.

Because inkjet printers are so ubiquitous in the workplace, they were examined as early as the 1980s for synthetic polymer and subsequently bioprinting applications. In 1988, Klebe first demonstrated that standard inkjet printers or graphical plotters could be used to dispense solutions of fibronectin onto plastic substrates (Figure a), which could then control cell adhesion (Figure b). Standard inkjet printers had several limitations (including, as noted by Klebe, the necessity to use thin, flexible substrates that could be rolled through the printer like paper). However, minor adjustments to printer structure, including use of a flat printing surface, ,, have enabled materials printing (e.g., polymer thin films for photovoltaics), , and there have been many examples of surface patterning of biomolecules using inkjet printing. For instance, inkjet printed growth factors (Figure c) on polyacrylamide have been shown to impact the differentiation of neural stem cells, with cells cultured on fibroblast growth factor 2 (FGF2, Figure d) continuing to express nestin, a marker for neural stem/progenitor cells (stained green), and cells cultured on areas of printed ciliary neurotrophic factor (CNTF, Figure e) producing a significant fraction of cells expressing glial fibrillary acidic protein (stained red), a marker for neural differentiation.

9.

(a,b) First report of inkjet printing of fibronectin on thin plastic films, subsequently used to pattern adhesion of SV-T2 cells; (b) higher-magnification view of a patterned area in (a), showing cell adhesion. (c–e) Inkjet printing of growth factors FTF2 and CNTF on polyacrylamide uniformly coated with amines and fibronectin; (c) image showing spatial control over molecular patterning using a fluorophore-modified protein, (d,e) images illustrate neural stem cell differentiation on surfaces modified with (d) CNTF, but not (e) FGF2, as expected. Adapted with permission. , Ref Copyright 2007, Elsevier. Ref Copyright 1998, Elsevier.

Over time, more significant customization of printers for biological applications has enabled 3D printing of materials for cell culture, either with or without suspended cells. ,, Such printing is typically not performed using inkjet and similar printing methods, due to upper limits on solution viscosities that can successfully form jettable droplets. However, extrusion and related printer types have been more successful. For these techniques, volumes for individual printed features are typically even larger (≥100 μm diameter). In contrast with inkjet printing, which requires relatively low viscosity inks, much higher viscosities are typically necessary to generate free-standing 3D structures. High viscosities in turn require high pressures for extrusion, which can negatively impact cell survival. Thus, many recent bioprinting approaches extrude softer material (e.g., granular hydrogel microparticles) into a support bath using lower pressures. − In one recent report, the use of a gelatin support bath with evenly sized 25 μm particles, in conjunction with a rapid pH change during ink extrusion, enabled printing of unmodified collagen (i.e., not synthetically cross-linked) inks in conjunction with cardiomyocytes, to generate replicas of heart components. Modifications to the extrusion process can also benefit scaffold design. For instance, electrowriting applies a strong electric field at the nozzle to produce a Taylor cone, enabling printing of fibers 2–50 μm in diameter, with a porous structure that can be used to support cell culture. −

Microcontact Printing

Soft lithography techniques such as microcontact printing (μCP) ,− have been utilized to pattern soft biomaterials with peptides, proteins, and other biomolecules at a resolution of 750 nm–10 μm, ,,− influencing cell adhesion and migration. − A significant challenge in using μCP to transfer molecular inks to soft materials is the potential deformation of soft (<40 kPa) materials due to the pressure required to transfer molecules from stamp to substrate.

To address this issue, trans-printing methods have been developed, in which biomolecular patterns are first transferred from a topographically patterned PDMS stamp to a hard, flat substrate (e.g., glass coverslip, poly­(vinyl alcohol) (PVA) film), and subsequently transferred to the soft material. This approach avoids placing anisotropic pressure on the soft material substrate, which can lead to deformation and patterning defects. , Figure a shows an example of this approach, with FN first inked on a PDMS stamp and printed onto a glass coverslip. PAAm (here, 5.6 kPa) is then cured in contact with the patterned glass, causing the proteins to be transferred to the PAAm when the glass coverslip is exfoliated from the hydrogel. The right panel of Figure a illustrates that this approach (left column) is more effective in transfer of both large patterns (top) and smaller microscale patterns (bottom) in comparison with direct stamping of FN on a SANPAH-activated PAAm surface. NIH-3T3 fibroblasts cultured on patterns prepared in this way (Figure a, bottom left) exhibited aligned actin stress fibers near the corners of the square patterns, enabling quantitation of traction forces. For very soft gels, exfoliation of a hard template such as a glass slide can become problematic. In these cases, the use of a sacrificial hard trans-printing substrate can be useful. FN and collagen patterns with dimensions of 2–20 μm have been transferred to soft PDMS (2.1 kPa) and PAAm (2.6 kPa) substrates using a sacrificial PVA film as a trans-printing medium that can be dissolved after transfer.

10.

(a) Microcontact printing of FN on glass; subsequent curing of PAAm in contact with glass produces pattern transfer to soft (5.6 kPa) PAAm surface. Images at bottom left illustrate alignment of actin stress fibers in NIH-3T3 fibroblasts cultured on square FN patterns; images in the right two columns illustrate efficacy of transfer patterning in comparison with direct patterning. (b) Subtractive patterning of PDMS with a 1 cm × 1 cm array of linear patterns (top left) with varied alignment, line density, and line width (lower left). NIH-3T3 fibroblasts cultured on nonaligned and aligned patterns (right two columns) exhibit differences in actin stress fiber alignment. Adapted with permission. , Ref Copyright 2018, American Institute of Physics. Ref Copyright 2011, Elsevier.

Subtractive transfer can also be used to achieve reproducible patterns on soft surfaces (Figure b). Although the minimum features sizes achieved through subtractive μCP patterning on soft surfaces is large relative to features on hard surfaces (to date, ∼500 nm on soft materials vs ∼5 nm on hard surfaces), narrow protein patterns on soft surfaces can also direct cell adhesion and formation of subcellular features such as stress fibers. For instance, NIH-3T3 fibroblasts adopt an elongated uniaxial morphology and enhanced migration speed on linear patterns of FN on PDMS with line widths as little as 500 nm, although increased orientation is observed with somewhat wider lines (2 μm).

Molding, in which the soft material intentionally deforms to adopt the surface topography of the stamp, can also be used generate microscale features in surface topography, which can further modulate cell behavior. For instance, hMSCs cultured on PDMS micropost arrays created by molding exhibited preferential differentiation based on post geometry differences that modulated effective stiffness, with taller posts (which deform to a greater extent based on cell traction forces) favoring adipogenic lineage commitment, and shorter posts (which allow for less deformation) favoring osteogenic commitment. This allowed for modulation of substrate mechanical response without changing the molecular network structure of the substrate. More recently, wells (80 μm diameter) molded in soft (2 kPa) and stiffer (35 kPa) hydrogels produced differences in endothelial cell migration, with cells migrating to the edges of the soft wells to form ring structures. Likewise, microfluidic channels in a mold comprised of PDMS or other materials can be used to confine deposition of a new layer of soft matter from the liquid phase. , Thus, it is possible to utilize both topography and chemistry of soft materials to substantially impact cell behavior.

The Potential for 2D + 3D Soft Material Patterning Approaches

Chemical patterning methods applied to soft materials are more limited in terms of spatial resolution than similar methods applied to hard materials, in part due to the characteristic length scale of heterogeneity in soft materials commonly used as cell scaffolds. However, to achieve microenvironments that more closely mimic the hierarchy of chemical structures found in biological ECM, it would be useful to combine the open meshwork and low modulus common in hydrogels with precision structural elements such as nm-scale surface ligand clustering and ultrathin basement membranes. These and many other structural elements are not native matches for many hydrogel architectures, but more closely resemble the precise structures in layer chemistries typically assembled on hard materials. Thus, there is substantial room for new chemistries that combine strengths of hard and soft material patterning.

Toward this end, one approach involves assembly of molecular layers on hard materials, which can be covalently cross-linked to lock nanometer-scale chemical patterns in place, and then undergo further covalent reactions to transfer them to the surfaces of hydrogels and other soft materials (Figure ). For instance, long-chain alkanes with internal diacetylenes and functional (typically polar) headgroups have been used to generate 2D layers and fibers that expose the polar headgroup chemistry, ,− enabling bioactive coatings, as well as sensors based on the red-blue color transition of the PDA. − However, these same building blocks also undergo efficient molecular assembly on HOPG with the alkyl chains lying flat on the 2D material surface, generating surfaces with 1 nm-wide functional patterns based on the assembled headgroups (Figure a). The assembly also aligns internal diacetylenes for topochemical photopolymerization with UV irradiation, ,,− which functions as a locking mechanism for the functional patterns. With appropriate assembly protocols, , molecular ordering in individual domains can in some cases extend over areas >10,000 μm², similar to the area of multiple cells, and can be combined with processes including μCP to achieve straightforward geometric control over chemical patterning (although with smaller individual ordered molecular areas). Reactivity of the polydiacetylenes (PDAs) also enables them to participate in reaction chemistries (radical polymerization of hydrogels, hydrosilylation reactions in PDMS) that are common in soft material solidification. ,,,, Thus, by cross-linking soft materials such as hydrogels or PDMS in contact with the PDA layer, it is possible to covalently transfer chemical patterns in the PDA layer to the surface of the soft material (Figure b). While the complexity of such layer structures does not yet match that in the ECM, layers of surface-templated PDA glycopolymers have been shown to produce selective multivalent binding to lectin protein binding partners with dissociation constants 10–40 nM (e.g., N-acetyl-glucosamine with wheat germ agglutinin, but not concanavalin A, Figure d). More broadly, such layer structures can also be used to impact interfacial wetting and adsorption of nanoscopic and microscopic structures at the interface. ,,,−

11.

(a) Schematic of chemical pattern assembly on hard surface (HOPG), photopolymerization to covalently lock functional pattern, and covalent transfer to soft material (here, PAAm). (b) SEM image of sPDA surface-functionalized PAAm. (c) Cartoon representation of accessible ligand clustering at surface of soft material through sPDA transfer, in contrast with stochastic functionalization. (d) Application of sPDA functionalization of PAAm to cluster carbohydrates for multivalent binding of lectins (wheat germ agglutinin, WGA). Adapted with permission. , Ref Copyright 2022, American Chemical Society. Ref adapted under the terms of the CC-BY-NC-ND license, Copyright 2023, American Chemical Society.

Such approaches to high-resolution control over interfacial (2D) structure also have the potential to be combined with 3D patterning approaches that replicate other intrinsic morphological properties of natural tissues, such as muscle fibers and blood vessels. For instance, a number of 3D soft materials constructs have been developed by folding 2D materials in response to different triggers, such as surface tension and pH. Cell traction forces can also be used to induce 2D-to-3D transitions, with cell morphological changes induced by modulating the geometry of the 2D microplates. High-resolution interfacial patterning can also be combined with molecular or polymer-assembly approaches, , or with larger-scale 3D patterning strategies such including additive manufacturing , and templating. ,

Summary and Prospects

It has been over 100 years since the first discovery that cell behavior is changed based on interactions with materials in the environment. While early observations relied on naturally occurring patterns of materials, and visual characterization of structure, dramatic advances in both fabrication techniques and instrumental characterization methods have transformed our understanding of how and why such interactions occur. For instance, we now know that cells not only respond to orientations of protein fibers around them by orienting mechanical fiber networks inside the cell, but also that cells are constantly remodeling the matrix, secreting and orienting their own protein and polysaccharide networks. However, much remains to be understood about ECM structure and functionthe large meshworks of macromolecules that provide a stable scaffold for multicellular function are also typically insoluble, and challenging to characterize at the level of detail that has been possible for many other biomolecules (e.g., crystallization for structure determination of many globular proteins).

At the same time, our ability to design precisely controlled topographical and chemical frameworks to test specific relationships between cell and environment has also grown. Regular arrays of geometric features such as ridges and posts can be created with tightly controlled dimensions in hard materials, and this type of work has suggested that congruence between length scales of structures outside and inside the cells is one of the important guiding principles, although the complete biological pathways that lead to such changes in growth patterns are often not entirely clear yet, and the relationships between physical structure and cell response can be complex.

Likewise, the ability to precisely structure synthetic soft materials has improved. Increasingly diverse chemistries can be patterned into scaffold materials, although spatial resolution is not typically equivalent to that achieved in hard materials, and artificial soft scaffolds do not yet fully recreate structure and function of real ECM. While it is known that cell fate and other outcomes during cell culture can differ based on mechanical properties of the matrix, the extent to which the mesh structure itself must match that of the original ECM is less clear. For instance, given the disparate structures of fibrillar meshes common in biological ECM and heterogeneously cross-linked structures typical in synthetic hydrogels, is a bulk metric such as elastic modulus adequate to describe the needed mechanical properties of the matrix? Newer work is illustrating important roles of viscoelasticity; other aspects of the nanoscale network structure will likely also emerge as important. More broadly, significant challenges remain in homogenizing ligand display in scaffold materials with ranges of elastic modulus, due to differences in mesh structure, making it difficult to independently determine the roles of the convolved structural factors.

These challenges are particularly critical in nanostructured elements of the ECM such as basement membranes, which have thicknesses just tens of nanometers, and in this small volume create distinct nanoscale chemical and mechanical structure in comparison with the surrounding matrix. In the coming years, it will likely remain challenging to fully recreate chemical and mechanical patterns present in soft materials in the cell membrane. However, because cells can in many cases remodel their environment to generate new ECM, interesting opportunities emerge to pattern elements of soft materials that induce ECM deposition that recapitulates the function of the native ECM. Overall, we suggest that there are also significant new opportunities to translate high-resolution patterning strategies developed for hard materials by covalently transferring them to soft materials to control cell adhesion and growth.

The authors declare no competing financial interest.

Published as part of ACS Materials Au special issue “Design of Complex Materials”.