*2.4. Fourier-Transformed Infrared Spectroscopy (FTIR)*

Collagens extracted from both the Ec and Ch were characterized in their IR absorption spectra. Overall, the obtained spectra (Figure 6) are in accordance with the results described in literature [38,42].

**Figure 6.** FTIR spectra of Sigma bovine type I collagen (black); Ec (dark grey); Ch (light grey); Ch supernatant (red). Broken line rectangle indicate Amide A region. Absorbance values have been shifted to facilitate the reading.

In more detail, the analysis of the obtained FTIR peaks indicated the presence of significant differences in the amide A region for both Ec and Ch. The asymmetric NH2 stretching peak in the Ec samples was difficult to identify due to a nearby strong peak associated with the presence of hydrogen bonds. This result is again in accordance to the FTIR collagen spectra describe by Garrone and coworkers [42]. All other major peaks, listed in Table 2, correspond quite well type I collagen (Sigma Aldrich, Merck KGaA, Darmstadt, Germany), used as reference material.

The comparison of the FTIR spectra with Type I collagen (Sigma Aldrich, Merck KGaA, Darmstadt, Germany) revealed a substantial similitude in the peaks; finally the ratio of absorbance measured at 1235 cm−<sup>1</sup> and 1450 cm<sup>−</sup>1, which were all >0.9, suggests the presence of native collagen triple helix in all the samples [49].


**Table 2.** List of representative FTIR peaks and the values obtained from FTIR analyses of Ch and Ec extracts. *S* = stretch; *B* = bend; *s* = symmetric; *as* = asymmetric; *HB* = hydrogen bonds.

#### *2.5. Rheology*

Considering the unusual properties of the obtained collagenous materials and to better understand their mechanical performance the rheological properties of the collagenous materials isolated from both Ec and Ch were characterized using a rheometer. Both materials display a clear shear thinning behavior (Figure 7a,b respectively).

**Figure 7.** Flow curves of ectosome (Ec) and choanosome (Ch) subjected to increasing shear rates. (**a**) = Ec 14.4 mg/mL; (**b**) = Ch 9.6 mg/mL; (**c**) = Ch 8.2 mg/mL (Ch 9.6 mg/mL after centrifugation and resuspension). Square = viscosity; diamonds = shear stress. Values are expressed as the average of three repeated experiments; bars = standard deviation.

In all the samples the viscosity, which at low shear rates tends initially to increase, decreased almost constantly at higher shear rates. The sample extracted from the Ec was much more concentrated than the one obtained from the choanosome, which partially justifies the differences observed in absolutes values when viscosity is considered. Additionally, we cannot exclude that differences in the composition may also be responsible for the observed disparities in the viscosity values.

Figure 7b,c presents the results of the same Ch sample before and after a centrifugation/ resuspension step. In this step, the sample was diluted in dH2O 2:1, mixed thoroughly, centrifuged (6600 g for 1 min) to concentrate the collagenous part and then brought to the initial volume using dH2O. Upon this procedure it is worth remarking that a significant amount of soluble material was extracted. Indeed, the dry weight decreased from 9.6 mg/mL to 8.2 mg/mL. This treatment revealed that the soluble molecules present in the supernatant play an important role in the rheological properties as their absence brings a significant change in the response of the material (Figure 7). In particular, the treatment decreased the viscosity at low shear strain and the overall shear thinning behavior.

The gel nature of the collagenous materials was confirmed by analyzing the storage (G ) and the loss modulus (G) at frequency between 0.1 and 100 Hz. The oscillatory measures provide relevant information on the molecular arrangement of the network of the gel and allowed the determination of the solid-like (elastic) and liquid-like (viscous) regions of the material. In the frequency range studied, the gel formed by the choanosome-derived material is in the plateau zone of the viscoelastic region, i.e., for frequencies lower than 1 Hz both G and G are nearly constant (Figure 8).

**Figure 8.** Storage and loss modulus measurements of choanosome (Ch) and ectosome (Ec) materials. Changes in the storage (G , closed symbols), and loss (G, open symbols) moduli at increasing frequency (**a**,**c**) and loss tangent (**b**,**d**) of Ec 14.4 mg/mL (**a**,**b**) and Ch 9.6 mg/mL (**c**,**d**). Values are expressed as the average of three repeated experiments; bars = standard deviation.

Furthermore, in this region the storage modulus is higher than the loss modulus, which is a characteristic behavior of a viscoelastic fluid. The G/G ratio, called the loss tangent (*Tan ∂*), gives a clear indication of the solid/liquid responses. Values lower than 1 indicate a more solid behavior while at values >1 a liquid-like behavior is prevalent. The curve shown in Figure 8 revealed that at 1.5 Hz the material starts to have a more liquid response. At higher values we observe clear crossing points at 20 and 80 Hz. This phenomenon is related with the entanglements of collagen chains [50]. The entanglements limit the mobility of the chains contributing to a stiffer structure, and hence a higher loss tangent. From our observations the *Tan ∂* values suggest that the material

retained a more solid response also at high frequencies. The storage modulus of Ec was also higher than the loss modulus, at low frequencies (Figure 8).

Here we noticed an apparent decrease in both G and G, with the latter slightly more pronounced than the former as demonstrated by the *Tan ∂* values between 0.01 and 0.15 Hz. Ch samples (similarly to the Ec ones) at higher frequencies tend to remain at *Tan ∂* values lower than 1 although rapid changes in the *Tan ∂* values have been observed.

Following these observations, which suggest that the materials have a thixotropic behavior, specific rheological tests were further performed to confirm it. The thixotropic behavior of the Ch sample, less complex in terms of composition with respect to the Ec, was thus evaluated by the hysteresis loop obtained after a two-step experiment in which ramp-up and ramp-down experiments were performed. A large hysteresis area, as it can be observed in Figure 9, indicates the thixotropic properties of the gel, namely the ability of the gel to recover the initial structure after a certain stress is applied [51].

**Figure 9.** Ramp-up and ramp-down experiment showing the thixotropic property of the collagenous materials. Shear stress/rate (**a**) and viscosity/shear rate (**b**) curves of coanosome (Ch) 9.6 mg/mL in a ramp up/down experiment. Open grey symbols = ramp up; closed black symbols = ramp down. Lines connecting the points have been added, the points represent the average values of three different experiments. Standard deviations bars have been omitted to facilitate the reading of the graphs.

The thixotropy of a material is also characterized by the ability to restore its original viscosity after a stress has been applied. Figure 10 presents the thixotropic behavior of the Ch sample in terms of viscosity in a three-step experiment.

**Figure 10.** Viscosity values (grey dots) of coanosome (Ch) 9.6 mg/mL in a three-step experiment. Black dots = shear rate. Values are expressed as the average of three repeated experiments; bars = standard deviation.

We observed that during the ramp down phase, after a first rapid increase of viscosity, the values start to decrease and at very low shear rates are eventually much higher than the ones observed for higher shear rates, in which the viscosity should be around 10 Pa·s.

On the other hand, we found that after a steady increase of shear rate and a following decrease the material was not able to fully recovery its initial viscosity (Figure 10). However the evident initial loss of viscosity observed in the first phase of the experiments suggests that: small but prolonged shear stresses are able to modify the inner structure of the material suggesting that a much longer period to reach a viscosity equilibrium. In this view the viscosity values obtained after the stronger shear stress was removed are likely the real viscosity value of the material at the specific shear stress of 0.1 s<sup>−</sup>1.

#### **3. Discussion**

Collagen, which constitutes the bulk of the extracellular matrix of most of the animals, provides, among others features, the mechanical support of tissues and serves as well as anchoring points for cells and other molecules [52]. In this view, it is not surprising that it has long been proposed as a natural material for tissue engineering and regenerative medicine [53,54]. Among lower invertebrates, and sponges in particular, *C. reniformis* has been considered an interesting source of alternative and biocompatible collagen given its high collagen content [42] and for its interactions with silicic acids (Heinemann et al., 2007b). Moreover, among Porifera, *C. reniformis* is the reference model of dynamic collagenous tissues, which are characterized by the presence of labile and variable collagen crosslinks, and has been extensively investigated [35–37,55].

Standard acidic treatments, able to solubilize most of animal fibrillar collagens, have proven to be inefficient in the extraction of *C. reniformis* collagen and other dedicated protocols have been developed [11,37,41,42,56].

Our method is a substantial modification of the procedure described in Fassini et al. [34] and directly produced a collagenous hydrogel. In terms of FTIR spectra the collagenous hydrogels obtained were similar to data present in the literature. In particular, Ec collagen was very similar to the spectra provided by Garrone and coworkers [42], while Ch profiles were similar to what reported by Heinemann and coworkers [38]. Considering the fact that Garrone investigated the cortex (ectosome) while Heinemann investigated the whole sponge (where choanosome mass is predominant) our results are in line with both the previous publications. Our results provide evidence that some slow chemical modifications, able to increase the dissociation of collagen fibres, occur during the incubation in DS. Furthermore, we observed that the acidic pre-treatment with EDTA, which caused a significant bleaching and shrinkage of the mesohyl pieces, is necessary to obtain the sticky gel materials. Indeed the longest incubation period itself was not sufficient and the final product was a liquid collagen suspension identical to what described by Fassini et al. [37]. Given that acidic treatments do not influence the properties of *C. reniformis* collagen [38], the chelation activity on divalent ions of EDTA is most likely involved in the process. A solution of 4% EDTA is proven to influence the integrity of collagen bundles and to promote the formation of fibrils suspension [42], possibly by competing with the formation of the labile and calcium-dependent bonds that stabilize sponge collagen [36,37]. On the other side, our samples were exposed to significantly lower concentrations of EDTA, which apparently was not enough to dissociate the tissue. The significant increase in the solubilisation of the collagenous materials that was observed after two months is not clear. In the DS solution, EDTA and 2-mercaptoethanol are the active molecules that are mostly involved in the chelation of collagen stabilizing ions and in the reduction of disulphide bonds [43]. The fact that the solubilisation of the materials in reducing conditions occurred only after at least one month in DS, together with the relatively short half-life of 2-mercaptoethanol, signals that some other slow reaction occurred in the mixture during the incubation period. The nature of this possible reaction(s) is also still unclear.

The materials obtained from the two different regions of the sponge have quite similar compositions although material obtained from Ch region has a higher collagen/non-collagen ratio than the one obtained from the Ec. The SDS-PAGE profile of digested samples revealed the presence of hydrolysis products, one single putative band for collagen at around 110 kDa and another three main bands of other unidentified proteins that appears to be stably associated with collagen. Indeed, proteins such as pepsin and peptides originating from the enzymatic digestion were more abundant in the supernatant obtained after the centrifugation than in the pellets while, on another hand, no proteins were found in the supernatant of untreated samples. Collagen extracted and solubilized from sponge was composed of a homotrimeric chain of around 105/115 kDa, the small molecular weight differences found between digested and undigested collagens suggest that the telopeptides are maintained during the extraction while differences between Ec and Ch might result from a different glycosylation level, different transcriptions, or by the presence of two different collagen types. Indeed, while the presence of type I collagen has been demonstrated in the sponge *Ircina fusca* [57], recent transcriptomic data from several sponges revealed the presence of the fibrillar type XI collagen [58].

The gel structure of both Ch and Ec was also transferred to the surrounding media and can be disrupted by the enzymatic digestion suggesting that some pepsin-sensitive protein is involved in the gel structure formation. However our efforts to investigate the nature of the molecules present in the supernatant were not conclusive.

The rheological properties of the materials obtained are difficult to compare with other described collagenous gels due to the complex nature of the extracts obtained. Moreover, *C. reniformis* collagen fibrils are rather different from type I collagen having an aspect ratio of 1:5000 [35], thus significantly inhibiting a proper comparison with the shorter and bigger mammal type I collagens.

Our data demonstrates the shear thinning and pseudoplastic properties of the materials, which exhibited gel like behavior. Moreover, considering the thixotropic behavior and the dramatic increase of its viscosity once the shear rate is decreased (i.e., recovering the viscosity properties when the external stimulus—stress—is removed), the possibility of modulating the mechanical properties while keeping the collagenous matrix unaltered suggests that the materials might be further investigated in order to design an injectable collagenous hydrogel.

In the recent years much effort has been focussed on collagenous hydrogels and on the strategies to control their biochemical and physical properties [9,59,60]. Hydrogels are particularly interesting for several tissue-engineering applications given their capability to provide a suitable substrate for cell growth and encapsulation and their high water content [61,62]. Other interesting and regarded properties of new designed hydrogels concerns their injectability and the capability to strongly adhere to wet surfaces [60], as well as the possibility to control their stiffness with simple changes in the pH, temperature or ionic strength, remaining inside the physiological boundaries of human tissues. In this sense, the extracts obtained with the methodology proposed in this work may see exciting developments in the future, namely in the context of advanced tissue regeneration therapies.

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

#### *4.1. Sponge Sampling*

*C. reniformis* specimens were collected at Paraggi (Eastern Ligurian Sea, 44◦18 40 N, 9◦12 46 E), transferred in a thermic bag to a tank filled with artificial sea water (Instant Ocean®, Blacksburg, VA, USA) and left rest for at least one night. Specimens were then frozen at −20 ◦C until further processing.

#### *4.2. Collagen Extraction*

Sponges were thawed and cut in small (1 × 1 × 1 mm) cubes with a scalpel; the outer brownish layer—the ectosome (Ec)—was separated from the inner yellowish core—the choanosome (Ch). Briefly 5 g of dry material (Ec or Ch) were transferred into two 50 mL tubes and in each tube 40 mL of 5× phosphate buffer saline (PBS, Sigma Aldrich, Merck KGaA, Darmstadt, Germany) were added before putting the samples on a vertical rotator at room temperature (RT). After 24 h, 10 mL of supernatant were removed and 100 mg ethylenediaminetetraacetic acid (EDTA, Sigma Aldrich) per

gram of sponge material were added in the each tube. After three days on the vertical rotator, the pieces were transferred into a new tube containing the disaggregating solution (DS) composed of 0.1 M Tris, 0.5 M NaCl, 0.05 M EDTA, and 0.2 M 2-mercaptoethanol. Ten milliliters of DS/g of fresh sponge material were used, and samples were left in DS for two months at RT. The resulting mixtures were then centrifuged (15 min at 500 g) to remove undissolved parts and the extracts obtained were then extensively dialyzed against dH2O (ratio about 1:1000, two changes per day for 5 days) using a 14 kDa molecular weight cut off (MWCO) membrane tubing (Sigma Aldrich, Darmstadt, Germany) in order to remove all the 2-mercaptoethanol.
