**3. Discussion**

The ability of the micro-environment and the adhesion substrate to affect various biological processes, i.e., differentiation and proliferation, has been shown in the evolution since the low metazoans [58]. The extracellular matrix (ECM) of multicellular animals is a complex, dynamic system. It is mainly formed by collagen fibrils structured in a highly organized, three-dimensional scaffold supporting cell adhesion and playing a crucial role in differentiation and tissue remodelling [59,60]. Indeed, collagen is considered to be extremely attractive to the manufacturing of biomaterials for TERM applications. The production of collagen-based biomaterials has intensively grown over the

past decades. Besides, always new techniques are being developed to improve their performance in terms of mechanical properties and the resistance to enzymatic degradation. Collagenous biomaterials are used as composite materials [61], or as de-cellularized derivatives, obtaining scaffolds that closely mimic the ECM structure and its properties [62]. Marine collagen is considered as one of the most promising sources for these purposes. In particular, the collagen that was derived from the marine sponge *C. reniformis*, due to its dynamic plasticity [43], is particularly interesting for biotechnological applications. In fact, it is considered as a marine biopolymer that is suitable for the production of dynamic biomaterials thanks to its ability to rapidly change stiffness and viscosity similarly to what was observed in the mutable collagenous tissues of echinoderms [63]. Collagenous membranes that are made from ECM extracts, containing native collagen fibrils of echinoderms with good mechanical/biological properties, have been already produced [55]. Conversely, although *C. reniformis* collagen has been studied for a long time [44–46], collagenous membranes that are derived from this species have not yet been described.

The aim of the present work was to obtain marine collagen membranes using intact collagen fibres of this animal. In the production of de-cellularized tissue scaffolds, the purification procedures have a key role, since the removal or maintenance of certain residues may strongly affect their biocompatibility or mechanical properties [64]. In this study, intact sponge collagen fibres were isolated using four different extraction methods (see Scheme 1), and then SCMs were produced from each of them (Figure 4). Finally, the chemical and physical properties of the four extracts were related to the mechanical and biological performances of the respective membranes. These evaluations allowed for selecting the optimal procedures and obtaining the best compromise among yield, 3-D fibre integrity, resistance, and biocompatibility of said membranes.

When comparing the yield values that were obtained in the different FSs (Table 1), the highest total yield and the highest collagen/GAGs recovery was obtained in the F3 suspension. The strongly reducing conditions, due to the presence of 2-mercaptoethanol in the F3 and F4 method, likely contributed to the higher sponge tissue disaggregation and FS yields of these two extraction methods. These results could be explained by *C. reniformis* presence of peculiar short-chain collagen types, which were related to the mammalian type IV, which seem to be involved in the tissue stiffness. In particular, the five conserved cysteines at the C-terminal of these proteins [46], suggest the presence of a disulphide-bridge crosslinking system for a supramolecular arrangement, as already reported in type IV collagen organization in mammalian basal membranes [65]. Thus, the addition of the reducing agent is likely to act at this disulphide-bond level, promoting the complete sponge tissue disruption in the F3 and F4 extraction procedures.

Conversely, for what concerns collagen extraction total yield (Table 1), although the F3 and F4 methods were very similar, the total ionic strength of the extraction buffer seemed to be fundamental to enhancing collagen fibril release. Indeed, the presence of a double NaCl concentration in F3 determined a proportional double release of collagen fibrils from the sponge tissue with respect to the F4 extract.

In the F2 extraction method, where the chaotropic conditions were used in addition to the reducing agent, the total FS yield and low collagen content were significantly different than F3 and F4, which is probably due to the shorter time of incubation in this extraction buffer (i.e., 1 day vs. 4 day, respectively). Conversely, the F1 extraction method provided the highest collagen percentage with respect to the total FS content, indicating trypsin digestion as the best strategy to obtain highly collagen-enriched sponge FSs (Table 2). Furthermore, when comparing the total protein and the collagen percentages in the four FSs (Table 2), we observed that the four methods provided very different values, with F1 suspension showing the highest content of soluble proteins. Conversely, all of the other FSs showed lower and variable percentages of soluble proteins with respect to the total FS. This indicates that these methods (especially F2 and F4) also provided significant amounts of uncharacterized, microscopic materials that are retained in the final FS, despite centrifugation. Notwithstanding the variability of the soluble protein percentage, the collagen/protein ratio seems more homogeneous in the four FSs. Indeed, all of the extraction preparations showed a collagen/protein ratio in a range that was close to 1:2, except for

F2, in which the proportion was ≈1:6, indicating that the F2 extraction method retrieves the lowest amount of collagen fibrils from the sponge tissue. Overall, the four methods allowed for obtaining collagen fibrils in good state of integrity, as showed by the ultrastructural qualitative analysis by TEM (Figure 2). Although, in F2 suspension, the presence of many curled fibrils (not shown) likely indicate a slight damage of the extracted collagen. These observations strengthen the idea that the F2 extraction method is less efficient than the others.

The qualitative evaluation of the four FSs by histological staining indicated that the collagen component in F2 preparation had thicker collagen fibres than the other three extracts (Figure 3, PicroS series: panel J vs I-K-L, respectively). Since fibres of such dimensions are supposed to be removed in the centrifugation steps during the extraction, it is probable that the strong PicroS F2 coloration is due to the artificial clotting of the collagen fibres during the acetic acid precipitation step. This again suggests that F2 may not be optimal for subsequent scaffolding procedures. Conversely, a low-diameter fibrillar organization was observable in F1, F3, and F4 PicroS staining, with the bigger collagen bundles that are typically present in *C. reniformis* native tissues [66] being completely removed during the first centrifugation steps.

In *C. reniformis*, tissue collagen fibres are closely associated to complex carbohydrates [44]. Besides, it is known that this component improves the mechanical and the biocompatibility properties of the collagen scaffolds [67]. Thus, the GAG content was also evaluated quantitatively and qualitatively in the four FSs that were obtained from *C. reniformis*. Our data indicate that, even if GAGs are present in low percentages in the four FSs (Table 2), from a qualitative evaluation they seem associated to the collagen fibres extracted with the various methods (Figure3 Alcian and PAS panels). Although, the F2 suspension differs from the others, showing, respectively, the highest percentage of GAG and the lowest percentage of collagen extracted. In this case, the Alcian histological staining could highlight a significant concentration of GAGs (panel B) probably associated to non-fibrillar proteins. Interesting differences were also observed in the RC/GAG, with F1 and F3 showing significantly higher ratios than the other two (Table 2). Considering that these two FSs were the only ones able to generate suitable SCMs and nc-SCMs, we could infer that the ratio between collagen and GAG content in the FS may be determinant for a successful reticulation during the membrane formation. Indeed, our results suggest that a RC/GAG <10 is inappropriate for SCM production.

The viscosity values that were measured in the four extracts (Figure 1B), do not seem to reflect the different collagen or GAG percentages, likely indicating that other factors (i.e., insoluble protein content, electrostatic interactions) could come into play, thus influencing this parameter. These results are in conflict with that observed visually in F1 and F3 suspensions, which appear as dark coloured viscous hydrogels respect to the other FSs. These features probably indicate a higher thixotropic behaviour and the presence of aggregates/micro-gels in solution. Conversely, the hydrogel consistency may be related to the collagen/FS percentage, which is the highest in these two samples, and where the fibrillar network could give a hydrogel aspect to the suspension [68].

SCMs were produced by EDC/NHS crosslinking of the four FSs; however, only F1 and F3 extracts could generate manageable membranes (Figure 4, SCM-F1 and SCM-F3, respectively). Conversely, F2 and F4 tentative membranes lacked structure and texture, were extremely fragile, and were difficult to recover from the silicon mould, also increasing the FS concentration (data not shown). These data indicate that *C. reniformis*-derived FSs with a collagen/FS percentage that is lower than 35% (Table 2) are unsuitable for membrane production.

ESEM analysis indicated that SCM-F1 and SCM-F3 (as well as also the fragmented SCM-F4) had a fibrillar organization. This further confirms that F1 and F3 extraction procedures were able to maintain an intact fibrillar structure with a significant difference in random pore dimensions between the two (Figure 9 A,C,D). Indeed, the higher mechanical resistance that was showed by SCM-F3 (Figure 5) could be explained by the presence of a smaller mesh of the fibrillar interlace with respect to SCM-F1.

Using the same FS concentration, F1 and F3 were also casted in silicon moulds without crosslinking. Surprisingly, both F1 and F3 formed collagenous membranes, named nc-SCM-F1 and nc-SCM-F3, respectively. These membranes showed macroscopic (Figure 4B) and ultrastructural (data not shown) features that were similar to their crosslinked counterparts. These data suggest that the collagen extracted from *C. reniformis* is naturally provided with a complex system of preformed fibrillar crosslinks, which may justify either their strong mechanical properties [44] or their insolubility in acidic solutions [49,50].

Another important consideration arising from the results of the membranes mechanical tests is about collagen/total protein content in F1 and F3 suspensions. Due to the higher resistance of SCM-F3 (Figure 5), we can also infer that the higher percentage of non-collagenous proteins present in F3 respect to F1 may have positive impact on the mechanical properties of the sponge FS-derived membranes. Although their exact nature is still unknown, it has been reported that these proteins constitute an amorphous inter-fibrillar matrix in the sponge ECM and are actively involved in the formation of the sponge fibre network [53]. These proteins are susceptible to trypsin treatment, being consequently digested in the F1 extraction procedure, and their absence could help to explain the significant reduction of the mechanical properties that were observed in SCM-F1. Moreover, while no differences were observed for SCM-F1 with respect to nc-SCM-F1 in terms of resistance improvement, a better performance was instead observed in SCM-F3 when compared to nc-SCM-F3. These data indicate that the higher non-collagenous protein content in F3 may participate to the formation of the crosslinks in the membrane, further strengthening this collagenous film.

Biomaterials originating from collagen still have some limitations to their use in human tissues due to inflammation arising from their biodegradation and relatively short durability [69]. It is known that, both living sponge tissues as well as isolated collagen fibres that are derived from *C. reniformis* are particularly resistant to bacterial collagenases [44]. In our strong experimental conditions (five times higher collagenase concentration than Garrone et al. [44], and 6 day incubation instead of 2 day), all of the membranes, crosslinked or not, maintained their intact structure. This evidence, in addition to confirming previous data on *C. reniformis* collagen, tell us that thin collagen membranes that are derived from this species possess a high stability in strong enzymatic degradation conditions. Conversely, in the same conditions, a commercial not crosslinked collagenous membrane, Bio-Gide®, was completely dissolved (not shown). Since this commercial product, which is mainly used in dental surgery, shows an in vivo resorption rate of 2–4 weeks [70], it is reasonable to believe that SMCs may show significantly longer resorption rates. If this was the case, another important requisite for implantable tissue engineering scaffolds would be met by these membranes. Also, extensive membrane treatment (15 day) with native FBS did not show any sign of disaggregation or enzymatic digestion. However, a higher opaqueness and thickness was observed with respect to control membranes that were incubated in PBS (Figure 6A,B). This was likely due to serum protein adsorption on the SCM surface. Indeed, also this feature is considered to be important in the long-term performance of implants [71]. In fact, once the proteins are adsorbed onto the material surface, cell adhesion and growth are facilitated. This adsorbed protein layer can also mediate the type of cells that adhere to the surface, which ultimately can determine the type of tissue that develops. The data collected in the present work indicate that F1 and F3 SCMs, either crosslinked or not, show strong in vitro enzymatic degradation resistance and are able to interact with serum proteins. Thus, they are suitable for the production of biomaterials needing long-term stability in guided tissue and bone regeneration applications.

The ability to bind water is another fundamental aspect of biomaterials. The WBC of SCMs and nc-SCMs indicated a slightly higher WBC in the F3-derived membranes than F1. Moreover, SCMs showed lower WBC than nc-SCMs, likely because most of the functional groups in the membranes are crosslinked and are less available for water interaction. Furthermore, the WBC values of the two nc-SCMs seem quite remarkable if compared to the value of the mammalian collagen not-crosslinked membrane Bio-Gide®. The latter, in fact, was 33% and 37% less hydrated than nc-SCM-F1 and nc-SCM-F3, respectively. The increased WBC in the nc-SCMs as compared to the commercial membrane could be explained by the presence, solely in the sponge membranes, of the ECM-derived GAGs. Indeed, their addition has been reported to proportionally improve the membrane's ability to bind

water [72]. However, although nc-SCMs were much more hydrophilic than SCMs and commercial membranes, they appeared to be less easy to manipulate when hydrated (Figure 6C), which is a feature that has to be accurately considered when designing new biomaterials.

Preliminary biocompatibility assays showed good and encouraging results for a further employment of these biomaterials for medical purposes. Both fibroblasts and keratinocytes were able to adhere and grow on all coated plates when compared to control samples (Figures 7 and 8). In particular, for fibroblasts, the slight cell decrease and the rounded shape onto F4 coating at 16 h indicated a lower short-term biocompatibility with respect to the other FSs and to the controls that were grown onto standard rat tail collagen-coated plates. Conversely, for keratinocytes, a cell decrease was observed only at the 6 day mid-term for the F3 and F4 coatings, maybe indicating the necessity of a period of adaptation onto these two matrices with respect to the other FSs and to the controls. Anyway, this apparent difficulty of adhesion of fibroblasts at 16 h and the growth of keratinocytes at 6 day was only temporary. In fact, long-term analyses at 15 day showed similar growth rates for the two cell lines onto all of the FSs and all closely comparable to the controls (Figure 8). In addition to this, a good cell-SCM surface interaction was observed by ESEM analysis for both cell types on SCM-F1 and SCM-F3, as evidenced by the presence of a tight adherence and a multitude of cell processes interacting with the collagenous membranes (Figure 9E–H).

Marine sponges are also important sources of bioactive metabolites [36], including compounds with antioxidant properties [73,74], and it is also known that marine collagen-derived peptides possess radical scavenging activity. All of the four FSs that were extracted from *C. reniformis* showed antioxidant properties (Figure 10). These properties, however, did not reflect their collagen percentage; hence, the antioxidant activity may be determined by other molecular types, which were probably co-extracted during collagen fibril isolation. The significantly higher scavenging value of F1 suspension could be due to the absence of reducing conditions in this extraction method, as compared to the other three, which could have partially inactivated the antioxidant component. Surprisingly, the antioxidant properties were also retained in SCM-F1 and SCM-F3, even if reduced with respect to the radical scavenging values of the respective FS. This suggests that these marine membranes are suitable for wound healing applications, for skin repair after superficial cancer treatments, or for the prevention of skin photo-damage and photo-ageing.

In conclusion, as previously reported [20,21,75,76], collagen derived from marine sponges is an extremely performant biopolymer that is suitable for biomedical applications. Here, for the first time, a thorough analysis and chemical characterization of four different sponge collagenous extracts allowed for generating crosslinked thin collagenous membranes from *C. reniformis* demosponge that is suitable for TERM purposes. The two types of SCMs that were obtained showed good mechanical properties, enzymatic degradation resistance, water binding capacity, and biocompatibility. In addition to this, our results demonstrate that it is possible to adapt the extraction procedures in order to alternatively improve the mechanical properties or the antioxidant performances of the derived biomaterials thanks to the versatility of *C. reniformis*-derived extracts.

#### **4. Materials and Methods**

#### *Chemicals*

All reagents were acquired from SIGMA-ALDRICH (Milan, Italy), unless otherwise stated.

#### *4.1. Sponge Sampling*

Specimens of *C. reniformis* were collected in the area of the Portofino Promontory (Liguria, Italy) at depths of 10–20 m and were transferred in laboratory in a thermic bag. During transport, the temperature was maintained at 14–15 ◦C. A short-term stabulation was performed, as described in [27]. In particular, the sponges were stored at 14 ◦C in 200-L aquaria containing natural sea water that was collected in the same area of the Portofino Promontory with a salinity of 37‰ and was equipped with an aeration system. Finally, the sponge specimens were frozen at −20 ◦C until further processing.

### *4.2. Fibrillar Collagen Suspension Extracts*

*C. reniformis* fibrillar collagen suspensions (FSs) were obtained using four different extraction procedures, obtaining, respectively, F1, F2, F3, and F4 extracts. As indicated in Scheme 1, for each procedure, about 25 g of frozen sponge tissue was thawed, extensively rinsed with cool deionized water, cut in small slices, and then minced in a blender in ice, with five volumes of the respective extraction buffer (step 1).

F1 collagen suspensions were obtained, as described in Gross et al. (1956) [53], with some modifications. Briefly, sponges tissue was minced in step 1 in five volumes of 100 mM ammonium bicarbonate pH 8.5, then 0.1% trypsin was added and the sample incubated overnight at 37 ◦C on a horizontal shaker. Afterwards, the fluid was removed by filtration with a metallic strainer and the solid material was suspended in three volumes of cool deionized water and incubated at 5 ◦C for three days in a rotary disk shaker aliquoted in 50 mL-tubes. The dark and viscous suspension was then filtered with a metallic strainer and the remained solid material was subjected at a second round of 3 day of water extraction. The viscous fluid that was obtained from the two rounds of water extraction was pooled and centrifuged at 1200× *g*, 10 min at 4 ◦C, in order to remove cell debris and sand particles. The supernatant fluid containing the collagen suspension was frozen at −20 ◦C for long-term storage.

F2 collagen suspensions were obtained using the protocol that was described by Diehl-Seifert et al. [52], with some modifications. Here, the sponge specimens were minced in step 1 in five volumes of 100 mM Tris–HCl buffer, pH 9.5, 10 mM EDTA, 8 M urea, and 100 mM 2-mercaptoethanol. The sample was incubated at room temperature (RT) continuously stirring for 24 h. Afterwards, the viscous extract was centrifuged at 5000× *g*, for 5 min at 4 ◦C. The pellet was discarded and the collagen precipitated from the supernatant by adding 1/3 of its volume of glacial acetic acid, and finally centrifuged at 20,000× *g* for 30 min at 4 ◦C. The collagen pellet was washed twice with distilled water until the pH was neutral and was finally suspended in 50 mL of 100 mM Tris–HCl buffer pH 9.0, homogenized for 30 s at 24,000 rpm with a T25 basic ULTRA-TURRAX® (IKA®-WERKE, Verke Staufen, Staufen im Breisgau, Germany) and stirred overnight at 4 ◦C. The collagen suspension that was obtained was frozen at −20 ◦C for long-term storage.

For F3 collagen suspension extraction, as already reported [46], sponge tissue was minced in step 1 in presence of 5 volumes of 50 mM Tris–HCl buffer pH 7.4, 1 M NaCl, 50 mM EDTA and 100 mM 2-mercaptoethanol. The sample was incubated at 17 ◦C for 4 days in a rotatory shaker disk aliquoted in 50 mL-tubes. The viscous extract was then centrifuged at 1200× *g*, for 10 min at 4 ◦C in order to remove cell debris and sand particles and the supernatant was extensively dialyzed against deionized water (ratio about 1:20, two changes per day for 5 days at 4 ◦C) using a 12 kDa molecular weight cutoff membrane tubing, in order to remove excess of 2-mercaptoethanol. The collagen suspension obtained was frozen at −20 ◦C for long-term storage.

F4 collagen suspensions were obtained as described by Di Benedetto et al. [55] with some modifications. Sponge tissue was minced in step 1 in five volumes of disaggregating solution that was composed of 100 mM Tris–HCl buffer pH 8.0, 0.5 M NaCl, 50 mM M EDTA, and 0.1 M 2-mercaptoethanol. The sample was incubated at 17 ◦C for 4 days in a rotatory shaker disk that was aliquoted in 50 mL-tubes. The viscous extract was then centrifuged at 1200× *g*, for 10 min at 4 ◦C in order to remove cell debris and sand particles, and the supernatant was extensively dialyzed against deionized water, as described for F3 collagen extraction, and at the end, it was frozen at −20 ◦C for long-term storage. Finally, in order to obtain the fibrillar concentration of each FS extract, 1 mL of suspension was lyophilized and weighted. All of the procedures were carried out two times in duplicate.

### *4.3. FS Characterization*

#### 4.3.1. BCA Total Protein Quantification

0.2 mL of each FS sample (F1, F2, F3, and F4) were centrifuged at 18,000× *g*, for 5 min at RT, the supernatant was discarded and the insoluble collagenous pellet was solubilized in 0.2 mL of 8 M urea pre-heated at 50 ◦C. The samples were then centrifuged at 18,000× *g*, for 2 min at RT in order to remove any insoluble residues. Total protein content was assayed in the soluble supernatant with Bicinchoninic Acid Protein Assay kit, following the manufacturer's instructions. Absorbance of each sample was read at 562 nm using a Beckman spectrophotometer (DU 640, Beckman Coulter SpA, Milan, Italy), in comparison to a skin porcine gelatin standard curve. The procedure was carried out in duplicate.

#### 4.3.2. Collagen Quantification

The total collagen present in each FS was determined by the estimation of the hydroxyproline content using a modified method that was based on the Cloramine T reaction [77].

0.2 mL of each FS was hydrolyzed with 2 N NaOH by autoclaving at 120 ◦C for 20 min. Samples were neutralized by adding one volume of 2N HCl and were diluted fourfold in deionized water. The hydroxyproline concentration evaluation was obtained by adding Cloramine T and Ehrlich's reagent, as described previously [77]. Absorbance of each sample was read at 550 nm using a Beckman spectrophotometer (DU 640), in comparison to a cis-4-hydroxy-L-proline standard curve, and, finally, the content of hydroxylated proline residue was used to infer collagen content of each FS using the proportion factor of 1 g of hydroxyproline per 10 g of collagen [44]. The procedure was carried out in duplicate.

#### 4.3.3. Alcian Blue GAG Assay

The glycosaminoglycan (GAGs) content was measured in each FS by the Alcian blue GAG assay, as described by [78]. To 20 μL of each FS extract, 20 μL of 0.027 M H2SO4, 0.375% Triton X-100, and 4 M guanidine-HCl were added, and then GAGs were stained with 0.2 mL of working dye solution containing 0.25% Triton X-100, 0.018M H2SO4 and 0.005% Alcian blue. All samples were incubated 10 min at RT in a horizontal shaker and then centrifuged at 18,000× *g*, for 10 min at 4 ◦C. The stained GAG pellet obtained was solubilized with 0.4 mL of 4 M guanidine-HCl and the absorbance of each sample was read at 620 nm using a Beckman spectrophotometer (DU 640), in comparison to shark cartilage chondroitin sulfate standard curve. The procedure was carried out in duplicate.

#### 4.3.4. Transmission Electron Microscopy: Negative Staining

FS samples (F1, F2, F3, and F4) were fixed in 2% PFA in PBS for 20 min at RT and were washed out in PBS. 5 μL drops of fixed FS were placed on formwar-coated grids for 20 min. When the suspension was partially dried, grids were washed by touching them three times to the surface of a drop of distilled water. Grids adsorbed with FS samples were then stained with 2% uranyl acetate in 0.15 M oxalic acid for 5 min and an additional 5 min in a 9:1 mixture of 2% uranyl acetate and metylcellulose 25 ctp. FS samples were imaged with a CM10 Philips transmission electron microscope equipped with Megaview 3 camera and Olympus SIS iTEM software for digital image acquisition. Representative images of the four FS preparations were taken at 92,000× magnification (scale bar: 200 nm).

#### 4.3.5. FS Qualitative Evaluation by Histological Methods

100 μL of each FS were smeared in triplicate on histological slides and were dried for 30 min at 37 ◦C. The slides were stained similarly to standard histological sections with various methods, (all products by Bio-Optica SpA, Milan, Italy): Periodic Acid Schiff (Hotchkiss-Mc Manus) (PAS) that produces a red staining reacting with glycol-containing cellular elements, e.g., glycogen or

neutral mucopolysaccharides; Alcian (pH 2.5), whih stains in blue acidic polysaccharides, such as glycosaminoglycans and some types of mucopolysaccharides, and finally, Picro-Sirius Red (PicroS), which stains specifically collagen fibres: in bright-field microscopy collagen appears red, when examined through crossed polarized light the larger collagen fibres are bright yellow or orange, and the thinner ones, including reticular fibres, are green [79]. The sections were observed through a Leica DMRB light and epifluorescence microscope equipped with DIC (Leica microsystems srl, Milan, Italy). Images were acquired using a Leica CCD camera DFC420C.

#### 4.3.6. Rheological Characterization

The rheological measurements were performed with an Anton Paar Physica MCR 301 Rheometer (Anton Paar, GmbH, Ostfildern, Germany), which was equipped with a 50 mm cone/plate geometry (CP50). The viscosity curves were carried out using a shear rate range between 0.1 and 2500 s<sup>−</sup>1, and each sample was tested twice to check for repeatability. The Rheometer was used with a Peltier heating system for an accurate control of the temperature. All of the measurements were performed at 20 ◦C.

#### *4.4. SCM Production*

For biocompatibility tests, all four FS were used to directly coat 24-well and 96-well plates. 300 μL (for 24-well plates) or 50 μL (for 96-well plates) of 2 mg/mL of each FS and of a standard rat tail collagen in the presence of 0.01% TritonX-100 were placed on the plates and were left to dry at 37 ◦C overnight. The coated plates were then incubated with 300 μL (24-well plates) or 50 μL (96-well plates) of EDC/NHS cross-linker solution: 30 mM (EDC 1-ethyl-3-(3-dimethylaminopropyl) carbodiimide/15 mM NHS N-hydroxysuccinimide in MES (N-morpholinoethanesulfonic acid buffer 100 mM, pH 5.5) for 4 h at RT in the dark. Crosslinked coated wells were then washed twice with 0.1 M Na2HPO4 for 30 min and twice with deionized H2O for 15 min, and then finally dried/sterilized with 70% ethanol solution.

SCMs were produced using silicone molds as rectangular (25 × 28 mm) sheets for in vitro biodegradation, water binding capacity, and ultrastructural analyses that were filled with 3.3 mL of 2 mg/mL of each FS and as rectangular (10 × 45 mm) sheets for mechanical tests, filled with 2.25 mL of 2 mg/mL of each FS. The FSs were left to dry at 37 ◦C, successively incubated, as previously described with EDC/NHS cross-linker solution for 4 h, washed with Na2HPO4, and deionized H2O and finally dried/sterilized with 70% ethanol solution.

Negative controls, lacking the cross-linker step, were prepared as well using 3.3 mL of 2 mg/mL F1 and F3 fibrillar suspensions casted in the same molds without adding the cross-linker solution.
