**3.2. Three-dimensional models mimicking ECM composition**

molecules like proteins. To overcome these issues, porous materials with interconnected pore networks and surfaces or fibers to sustain cell adhesion as well as cell phenotype have been designed. These materials are discriminated in microporous scaffolds where embedded cells will mainly attach on the surface of pores of >100μm diameter and will more represent curved 2D surfaces, and in nanoporous scaffolds where pore structures are in the range of the cell diameter (roughly 10 μm) and represent more the native 3D environment of ECMs [78]. Nevertheless, microporous structures allow a more effective cell penetration and migration into the material while smaller-sized pores can represent a barrier for cell colonization of the material, and limit the cell interactions at the edges of the material [79]. This particular point of cell colonization represents one of the challenges with nanoporous materials, and definitely a limitation when directly compared with both the microporous and the hydrogel materials. Similar to hydrogels, porous materials can be prepared with both natural and synthetic polymers with a large panel of techniques including, but not restricted to electrospinning, phase separation, templating, and vapor-phase polymerization [79]. Electrospinning forms fibrillar materials with control on the fibers diameters, fibers alignment, and fibers interdis‐ tances that dictate the overall porosity of the material. A polymeric solution is injected through a capillary tube into a high electric field that forms a Taylor cone with application of the electrostatic forces. On the opposite side of the capillary, the collector of the electrospun fibers is charged or grounded. The electric field in-between capillary and collector accelerates the flow of the polymer solution that evaporates the solvent and generate coalescence of the polymers that finally form solid fibers arrays. This technique is mainly used with synthetic polymers (poly lactic/glycolic acid and poly caprolactone) but is also suitable for natural polymers such as collagen and silk fibroin [80]. Electrospinning produces thin materials not exceeding millimeters and with a high-packing density of fibers that can impair a deep colonization by cells, but intensive researches are presently conducted to overcome these issues [81]. Collagen porous materials can be formed by thermally induced phase separation leading to the production of the so-called "collagen sponges" due to their high porous structures. Acidic collagen solutions are frozen which induces a phase separation of collagen molecules from the water-based solvent that is then eliminated by lyophilization. Materials are then made of dense collagen walls with unstriated microfibrils [57] or with native cross-striated fibrils [82]. Size and interconnectivity of the pores can be modified by modulation of the length and temperature of the phase separation, by modulation of the collagen concentration or by mixing the collagen solutions with other natural polymers such as GAGs [82] or with synthetic polymers such as poly lactic acids. Freeze-dried collagen materials have weak mechanical properties and are often strengthened by the addition of GAGs, or cross-linked by chemical

310 Composition and Function of the Extracellular Matrix in the Human Body

species (aldehydes) or dehydrothermal processes [83].

The various polymers and processes presented above enable the engineering of different types of 3D materials designed to study *ex vivo* a large variety of cellular phenomenon such as cell differentiation, drug response, tumor formation, cell migration, cell morphology and cytos‐ keleton organization, cell death and proliferation, tissue architecture, and coculture behavior of cells [84]. For that purpose, 3D materials are built either to mimic ECM composition, ECM organization, or ECM mechanical properties. We present here examples of such engineered Recreating *ex vivo* the ECM composition in 3D materials offers the possibility to address in a relative more biologically relevant environment the specific function of one or several of its components on several cell mechanisms. It can be used to recreate cell adhesion and cell migration or to offer adhesive cues that are specially organized.

A minimal system has been designed with a gelatin hydrogel mixed with PEG and function‐ alized with RGD and PHSRN adhesion peptides to evaluate the inflammatory response of adhering immune cells [85]. Monocytes are circulating immune cells which are among the first cells to react at the site of an injury. Once they have migrated and attached to the site of aggression, they start to secrete cytokines and ECM remodeling enzymes. The hydrogel was built with PEG-diacrylate (PEG-da) photopolymerizable polymers mixed with gelatin monomers covalently grafted with PEG-RGD and PEG-PHSRN, two amino acid sequences found in fibronectin and recognized by several types of integrins. The hydrogel mixture was made with different ratios of Gelatin-PEG-peptides compared to the PEG-da polymers for the formation of hydrogels with different adhesive peptide densities (from 30 to 50% of gelatinmodified polymers). The study showed that monocyte adhesion and cytokine secretion reached a plateau when the gelatin-modified polymers concentration was above 40%, indi‐ cating that rising adhesive-peptides densities did not improve their accessibility for mono‐ cytes. The nature of the ligands has also an influence on monocytes behavior, with a higher level of MMP-9 expression when hydrogels are incorporating PHSRN peptides, while more interlekin-1β was secreted in RGD hydrogels. These data suggest a specific monocyte inflam‐ matory response for each of these adhesive cues. This model can be developed and used, for instance, to study the different parameters that initiates the inflammatory response at the site of injury.

Metastatic cancerous cells escape the original tumor site to attach and invade a distant organ. The switch of the cellular microenvironment from primary tumor site to the metastatic organ can affect the metastatic cells response to therapeutics which was originally designed to act on the original cancerous cells in the specific microenvironment of the primary tumor site. Blehm and coworkers have engineered a 3D model of the ECM of the perivascular niche of the brain, a metastatic target of cutaneous melanomas, to address the effect of anti-MAPK therapeutics (ERK inhibitor and Darafemib) on metastatic cutaneous melanomas cells lines [86]. Cells were seeded in a huyaluronic acid-PEG-diacrylate hydrogel supplemented with purified full-length proteins, laminin-111 and fibronectin, or with RGD-cyclic peptides. Diminution of Darafemib drug efficacy is reported for one melanoma cell line in hydrogels supplemented with fibro‐ nectin, whereas laminin-supplemented hydrogels protected against ERK inhibition. With another cell line, drug efficacy was weakened for both therapeutics only on laminin-supple‐ mented hydrogels, reinforcing the importance of the cellular microenvironment for drug testing.

Natural polymers, including ECM proteins, GAGs, and growth factors, with a high degree of purity were successively assembled to form molecularly defined materials evaluated by subcutaneous implantation as well as in a cartilage reconstruction model [87]. Collagen type I solution purified from bovine Achilles tendon was mixed with insoluble elastin, purified from equine ligamentum nuchae and homogenized in a grinder before molding and freezedrying of the solution. This process formed a porous scaffold (collagen sponge-based material) with both collagen and elastin fibrils. Porosity of the material was controlled by the freezing temperatures, as higher rates of freezing provided materials with smaller pore diameter. Complexity and ECM mimicry of the material is extended by incorporation of chondroitin sulfate or heparan sulfate. Collagen-elastin material is soaked with either chondroitin or heparan sulfate solution, and GAGs are covalently bounded to the collagen-elastin fibers by carbodiimide cross-linking. The amount of GAGs effectively grafted to the materials is a function of GAGs structures and their respective reactiveness to the cross-linker, with a final grafting of 10% of the original heparan sulfate solution and 6% of the chondroitin sulfate solution. Finally, growth factors such as bFGF and VEGF were loaded into the Collagen-Elastin-GAG material to form a valuable model of the different classes of natural polymers found in ECMs. Incubation of growth factor solution with the material only allowed a partial adsorption as growth factors are rapidly lost after a few hours of cultures. However, heparan sulfate bears native biding sites for bFGF, enabling a more efficient loading of bFGF into the material and a progressive and extended release during cell culture. Two-weeks after subcu‐ taneous implantation in rat, collagen-heparan sulfate materials showed a higher vasculariza‐ tion than collagen alone. After 10 weeks, collagen materials loaded with bFGF exhibited a lower cell invasion and vascularization than the collagen-heparan sulfate-bFGF material. Even evaluated *in vivo*, where cells are in a more positive environment than cultured *in vitro*, this study showed the possibility to engineer a complex material reproducing the different class of natural polymers found in ECMs. A critical parameter for the use of such porous material as an *ex vivo* ECM model is the possibility to seed evenly the cells within the material as compared to hydrogel-embedding. However, this can be overcome by creation and control of an open and interconnected porosity within the material.

#### **3.3. Mimicking ECM organization and mechanics**

Two-dimensional cell cultures on TCPS have been coated with ECM molecules to give more biologically relevant ligands for cell adhesion, spreading, migration, and expression of phenotypes. This has partly raised the idea that instead of coating, ECM-derived adhesive cues should be displayed in 3D to be more close to the cell environment, and we have presented above 3D materials that recreate ECM composition. However, this approach does not generally consider the reconstruction of the spatial organization of ECM that is of equal importance as the ECM composition [88].

The study of collagen-rich connective tissues such as skin, tendons, and bones by polarizedlight microscopy and TEM revealed that type I collagen is highly structured, with a spatial geometry specific for each tissue. This spatial organization of collagen I in tissues can be reproduced *in vitro*, using the autoassembling properties of collagen molecules in acidic solutions which will self-organize in different liquid-crystal organizations dependent on the collagen concentration [89]. Regulated evaporation of solvent raise progressively the collagen concentration and modify collagen molecules organization. These organized molecular textures are stabilized by neutralization of the pH of the solution and further triggering the assembly of soluble collagen molecules into insoluble cross-striated native collagen fibrils [54,90]. The supramolecular assembly of collagen molecules into collagen fibrils also induce the transition from the original collagen solution to a solid hydrogel structure (i.e., so-called "sol/gel transition") that retains the molecular tissue-like organization of collagen molecules. This leads to materials with collagen fibrils architecture which may mimic those observed in tendon, skin, and bones [21,91] but that also reproduce the mechanical features of collagenrich tissues from dermis to bone, depending on the collagen concentration of the material [75]. Ordered collagen hydrogels have been used to study myofibroblasts migration as a model of wound-healing [92], behavior of human osteoblasts in long-term *in vitro* cultures [57,93] and also as a model to study the kinetics of noncollagenous ECM proteins secretion in relation to the biomineralization processes [94]. However, tissue-like organization of collagen fibrils resulted in mild (20–40 mg/mL) or high (>80 mg/mL) collagen concentrations, which mainly restricted cell seeding to the surface of the materials.

Natural polymers, including ECM proteins, GAGs, and growth factors, with a high degree of purity were successively assembled to form molecularly defined materials evaluated by subcutaneous implantation as well as in a cartilage reconstruction model [87]. Collagen type I solution purified from bovine Achilles tendon was mixed with insoluble elastin, purified from equine ligamentum nuchae and homogenized in a grinder before molding and freezedrying of the solution. This process formed a porous scaffold (collagen sponge-based material) with both collagen and elastin fibrils. Porosity of the material was controlled by the freezing temperatures, as higher rates of freezing provided materials with smaller pore diameter. Complexity and ECM mimicry of the material is extended by incorporation of chondroitin sulfate or heparan sulfate. Collagen-elastin material is soaked with either chondroitin or heparan sulfate solution, and GAGs are covalently bounded to the collagen-elastin fibers by carbodiimide cross-linking. The amount of GAGs effectively grafted to the materials is a function of GAGs structures and their respective reactiveness to the cross-linker, with a final grafting of 10% of the original heparan sulfate solution and 6% of the chondroitin sulfate solution. Finally, growth factors such as bFGF and VEGF were loaded into the Collagen-Elastin-GAG material to form a valuable model of the different classes of natural polymers found in ECMs. Incubation of growth factor solution with the material only allowed a partial adsorption as growth factors are rapidly lost after a few hours of cultures. However, heparan sulfate bears native biding sites for bFGF, enabling a more efficient loading of bFGF into the material and a progressive and extended release during cell culture. Two-weeks after subcu‐ taneous implantation in rat, collagen-heparan sulfate materials showed a higher vasculariza‐ tion than collagen alone. After 10 weeks, collagen materials loaded with bFGF exhibited a lower cell invasion and vascularization than the collagen-heparan sulfate-bFGF material. Even evaluated *in vivo*, where cells are in a more positive environment than cultured *in vitro*, this study showed the possibility to engineer a complex material reproducing the different class of natural polymers found in ECMs. A critical parameter for the use of such porous material as an *ex vivo* ECM model is the possibility to seed evenly the cells within the material as compared to hydrogel-embedding. However, this can be overcome by creation and control of

Two-dimensional cell cultures on TCPS have been coated with ECM molecules to give more biologically relevant ligands for cell adhesion, spreading, migration, and expression of phenotypes. This has partly raised the idea that instead of coating, ECM-derived adhesive cues should be displayed in 3D to be more close to the cell environment, and we have presented above 3D materials that recreate ECM composition. However, this approach does not generally consider the reconstruction of the spatial organization of ECM that is of equal importance as

The study of collagen-rich connective tissues such as skin, tendons, and bones by polarizedlight microscopy and TEM revealed that type I collagen is highly structured, with a spatial geometry specific for each tissue. This spatial organization of collagen I in tissues can be reproduced *in vitro*, using the autoassembling properties of collagen molecules in acidic

an open and interconnected porosity within the material.

**3.3. Mimicking ECM organization and mechanics**

312 Composition and Function of the Extracellular Matrix in the Human Body

the ECM composition [88].

In the depth of a tissue, different ECM organization could be found, like in the skin with different ECM compositions, organizations, and properties of epidermis and dermis, or in articular cartilage from the GAGs-rich upper surface to the deeper interface with the under‐ lying bone tissue. A layer-by-layer approach, based on successive freeze-drying of different natural polymer solutions to modulate material organization, was used to recreate the discrepancies in the ECM organization observed in the cartilage tissue [95]. The first layer intends to recreate main features of the deeper osteochondral tissue architecture. Collagen type I in acidic solution is mixed with hydroxyapatite, the mineral phase found in bones, and freezedried at a constant rate to obtain a dense porous structure. The subsequent porous material was cross-linked with carbodiimide to strengthen its structure. The second layer mimics the bone–cartilage interface and is made with an equal amount of mixture of the collagen type I, main collagen of bone tissue, and of the collagen type II that is the main collagen of cartilage tissue. Collagen I and II mixture is supplemented with hydroxyapatite at a final concentration five-times lower than on the first layer. This mixture is poured on the surface of the first layer and freeze-dried to finally form a two-layer material, with two distinct compositions and porous structures. The last layer mimics the articular cartilage tissue. This layer is made from a mixture solution of 25%/75% of collagen type I and II, respectively, and then further supplemented with hyaluronic acid to recreate the GAGs content of the articular cartilage. The last layer is again freeze-dried with time-prolonged freezing and drying steps to ensure the proper porous structure. The overall material is strengthened by a dehydrothermal process that creates amine-based cross-links. This process forms a material with three specific porous structures as well as a molecular composition and with the opened and interconnected porosity that allows an effective cell colonization of each of the three layers.

The specific composition and spatial organization of the ECM dictate its overall mechanical properties, that cells are able to sense through their integrin receptors and their cytoskeletons. On 2D surfaces, it has been shown that cells are responsive to surface rigidity and that it influences the commitment of mesenchymal stem cells toward differentiation in a specific lineage [96]. Again, surface stiffness applies forces which are unevenly distributed on the cells (i.e., only located at the cell-surface interface). Developing a 3D material with tuned and controllable mechanical properties will generate a more biologically relevant environment to evaluate the role of ECM mechanobiology on cells functions and differentiation processes. To study the influence of mechanical stiffness on mesenchymal stem cells differentiation, a series of alginate gels with elastic modulus ranging from 2.5 to 110 kPa has been developed [97]. Mechanical properties of alginate gels are modulated by the percentage of alginate polymers in the final hydrogel. Because alginate is not sensitive to the degradation of hydrolytic enzymes of mammalian cells, the elastic modulus of the mechanical properties of the material are expected to remain constant all along the study (7 days of cell culture). RGD-peptides are covalently grafted to alginate polymers prior to hydrogel formation to give to the cells adhesive cues. The more rigid materials trigger mesenchymal stem cells differentiation toward bone lineage with an expression of the bone-related molecular markers such as alkaline phosphatase and osteopontin after 7 days of culture. On the other hand, alginate gels with the lower elastic modulus (softer material) triggers an accumulation of oil-droplets into stem cells, indicating adipose tissue differentiation. The density of RGD-peptides incorporated into these materials did not modify the cell fate related to the elastic modulus of the material, but induce a higher level of expression of the lineage markers for both bone and adipose-committed cells.

#### **3.4. Conclusion: toward a gold-standard of 3D model of ECMs?**

As mentioned at the beginning of this section, the ECM-derived Matrigel® represents currently the most often used material for 3D experiments in cell biology. Despite that Matrigel®-related drawbacks are of importance, Matrigel®is a widely and available model to investigate many fundamental questions in cell biology, from cell adhesion and tumor formation, to drug testing. We have presented in this chapter a large panel of techniques, methodologies, and engineering processes that allow the exploration of ECM organization and permit to recreate *ex vivo* some of their key features. At the conclusion of this chapter and after the review of several studies investigating various 3D materials, it appears that no material can represent the unique and ideal answer for all cell investigations in 3D [63]. A modular approach should be taken by rationalizing the biological question to be studied and the parameter of ECM intended to be recreated. Nevertheless, more and more complex materials are engineered that will finally be able to mimic simultaneously several key factors of ECM composition, architecture, or mechanical properties, and so enabling investigation of multiple parameters for cell biology experiments. An important drawback with engineered 3D materials is to create a "black-box" where undefined and uncontrollable parameters may influence the cellular outcomes to be investigated. To avoid part of this problem, rigorous attention should be paid on the purity of polymers used to build the material, in particular with biological polymers. The development of DNA-recombinant production of ECM proteins can overcome this problem, even if this will raise ultimately the cost of the final material. The structural characteristics of the final material (porosity, polymers distribution, and fibrils diameter) should be consistently reproducible and addressed. To do so, an "easy," meaning straightforward, process of the material should be sought and preferred rather than a more complex multistep fabrication process. Biomaterial scientists propose continuously new design and approaches to engineer *ex vivo* ECMs. The production of a gold-standard material may become possible for a specific biological question. It might happen with the existence of a deep and intelligible dialog in-between material scientists, whom brings engineering strategies, and the cell biologists, that implement the material design to mimic the biological process that has to be investigated *ex vivo*. This collaboration may result in major advances for science and medicine.
