3.2. Morphological Characterization
Pore morphology and distribution play a crucial role since they affect: (i) the cellular adhesion, proliferation and growth; (ii) the permeation of nutrients and oxygen for the cells from the surface towards the core of the scaffold and the elimination of CO2
and other metabolites from the core towards the surface; and iii) the mechanical behavior of the whole scaffold. ESEM images (Figure 2
E–J of the prepared blends show similar pore sizes and pore shape: the similarity in the pore size is due to the fact that porosity is steered by the freeze-drying process which in turn is controlled by the size and geometry of the ice crystals formed during the freezing process [32
]. On the other hand, the pore size of the blends is larger than the ones of the pure polymers (Gel, CNFs) (Figure 2
A–D). The different aspect ratio of pores between the blends and the pure polymers can be assigned to their different viscosities. The viscosities of pure polymers and blends are reported in Figure 3
]. The low viscosity of Gel is responsible for the mono-directional growth of ice crystals, inducing the formation of a channel-like porous structure (Figure 2
B), with a channel diameter of about 50–100 µm. In the case of CNFs, its high viscosity promotes the formation of small randomly oriented ice crystals (Figure 2
D). When CNFs are added to Gel, the ice crystal formation during the freeze drying is retarded and hampered by the increase in the blend viscosity, thus the direction of ice crystals growth becomes more isotropic, and the derived porosity is randomly oriented and characterized by bigger pores (100–250 µm) (Figure 2
However, despite modifying the Gel:CNF ratio, all the biocomposites showed rough surfaces and a porous structure with interconnected pores homogeneously distributed [10
] with sizes ranging from less than 100 up to 250 µm, suitable for cell colonization and proliferation.
Porosity was also evaluated by density measurements and water-squeezing methods, as reported in Table 1
. Porosity tests, according to the SEM images, show high porosity over 90% for all the samples due to the high-water content before lyophilization. The lower value of porosity registered for CNFs was due to the inter-fibril linkages, generating a tight structure. Different polymer ratios did not significantly affect this parameter, but all blends generally showed slightly higher porosity compared to the pure polymers (Gel or CNFs). This behavior is affected by the interactions between the polymers, creating a well dispersed blend able to self-assemble entrapping water molecules [24
Using the water-squeezing method, the porosity of the pure polymers was lower due to the high number of smaller pores trapping the inner water that could not be released by simply squeezing (Table 1
). The difference between the Gel and CNF macro porosity is reflected in the different density observed by the squeezing method and confirms the denser structure of CNFs. CNFs have small rounded-shaped pores whereas Gel has elongated ellipse-like pores with sagittal axis bigger than the transversal axis. The different shapes of the pores and the different pore volumes resulted in lower macro porosity for the CNFs compared to Gel.
3.3. Chemical-Physical Characterization
The TNBS assay, spectrophotometrically determining the amount of unreacted free amines in the analyzed samples (Table 2
), was used to measure the Gel–CNF interaction degree (Int%) and the efficiency of cross-linking degree (CD%) [35
By comparing the blends and pure Gel, it is possible to determine the degree of interaction between the Gel and CNFs during the blend formation. On the other hand, by comparing the cross-linked blends with non-cross-linked ones, it is possible to determine the cross-linking degree due to DHT treatment.
The results (Table 2
) demonstrate that when adding CNFs to Gel the Int% increases, meaning that the number of free amines is reduced as expected. When DHT cross-linking is applied, the efficiency of the procedure is influenced by the interference with CNFs.
Polymer ratio affects the covalent linkage formation, but not in a linear way. In fact, in GCN (Gel:CNFs)-50:50 the CD% is lower than for the GCN-75:25 and the cross-linking seems hampered by the presence of CNFs (Scheme 1
]. However, when the CNF content increases from 50% to 75%, the cross-linking degree increases. This is not easy to understand but the explanation probably lies in the relative amounts of functional groups and the proximity between them. Xuefei et al. [37
] demonstrated that DHT cross-linking of collagen resulted in formation of both covalent and hydrogen bonding. The bonding between the peptide chains seems to be prevalent, as it is proven by the reduction in the number of free acidic and basic residues on collagen molecules after the DHT treatment. They also proved that after DHT cross-linking, both non-covalent and covalent chemical bonds co-exist, and improve the properties compared to the non-cross-linked material. According to Yannas et al. [38
] these chemical bonds can be formed by condensation, and also by esterification and amidation between carboxylic and amine groups. Thanks to the chemical structure of Gel and CNFs, presenting this kind of functional groups, all the aforementioned reactions can also occur within our GCN biocomposite.
In detail, the CNFs used in the study have an aldehyde content of 211 ± 60 µmol/g and carboxylic acids of 764 ± 60 µmol/g, while the gelatin used was Type A gelatin, reported in literature to have 800 µmol/g carboxylic acid groups and a primary amine content of 286 µmol/g [39
]. Primary amines and aldehyde groups react easily with each other in a click reaction if they are close enough, but as described, after the DHT all functional groups are involved in cross-linking reactions, so the key point is the distance between the polymer chains/fibrils. From the results given in Table 2
, it is evident that there is no significant difference between the GCN-75:25 and GCN-25:75, while a significant difference is noted for sample GCN-50:50. If we consider the possible distance between the polymers/fibrils we can probably assume that the distance between the Gel chains and nanofibrils are larger in the GCN-50:50 sample than between the Gel chains in the GCN-75:25 sample and the nanofibrils in the GCN-25:75 sample. Thus, when the Gel:CNF ratio is unbalanced, more functional groups are settled in a suitable distance to cross-link. Cellulose nanofibrils have a large tendency to self-assemble and form strong bonds with neighboring fibrils. If the water is involved, this phenomenon is called hornification, and the maximum effect is with a water content of approximately 0.25 g water/g cellulose [40
]. The self-assembling effect is reported to occur even if large amounts of another component are added, e.g., up to 60% maltodextrin [41
]. It is likely that the self-assembling of CNFs, strengthened by DHT, is the main mechanism in the GCN-25:75 sample.
illustrates the possible links between the two polymers as well as the tentative linkages due to DHT cross-linking and is used as a simple way to visualize and describe the nonlinear variation of CD% in the blends.
The evaluation of the cross-linking extent demonstrated that, by blending and cross-linking different polymers, it was possible to engineer new biocomposite blends, capable of different chemical interactions and characterized by tunable physical-chemical properties.
Blend hydrophilicity was measured by static water contact angle measurement on films made from blends and the results are reported in Table 3
. Some significant changes in the contact angles were recorded among the samples, and all of them showed contact angles higher than 70°, (higher with respect to the data found in the literature) indicating a less hydrophilic behavior. As previously reported, this is due to the cross-linking treatment, which reduces the available hydroxyl and amino groups of the polymers (Gel and CNFs), causing the formation of covalent linkages between them [21
The swelling test was carried out to evaluate the capacity of the biocomposites to absorb water and consequently cell medium, which is essential for cell colonization. Each sample showed a distinctive water uptake, influenced by the compositional and morphological differences as illustrated in Figure 4
A. When observing the swelling behavior, a rapid water uptake during the first hour of experiment was clear for the all samples maintaining the same saturation level up to 48 h.
Comparing the different samples, the swelling of all blends was higher than for pure polymers (Gel and CNFs); i.e., the higher values of hydrophilicity and porosity cause a larger degree of swelling of the samples. These data are in contrast with the measured values of contact angles where the porosity was not considered, because the tests were carried out on dried and non-porous films. However, swelling is mainly affected by the porosity of the structure; therefore, the blends showed higher swelling due to a higher porosity than the pure polymers, as already evaluated through the water squeezing method and ESEM microscopy [43
Among blends, higher amounts of CNFs caused a larger swelling, as is in agreement with ESEM microscopy, where a slightly higher pore growth was observed in the GCN-25:75 sample compared to GCN-75:25. Finally, the swelling was fast and did not change during incubation time, meaning that no initial degradation and no loss of structure occurred.
The weight loss test was carried out to evaluate the efficacy of DHT that is essential for maintaining the shape and slowing down the degradation rate of the scaffold when in vivo [9
]. As reported in Figure 4
B, after 28 days at 37 °C in PBS, all samples lost 10–20% of their overall weight, macroscopically preserving their shape and structure. The degradation rate is a crucial point in the choice of the scaffolding strategies, because a too fast degradation would not support the tissue remodeling process [12
]. A compromise between the tissue growth and the biocomposite degradation rate is essential to guarantee a sustainable scaffold throughout the cellular regenerative process [44
3.4. Mechanical Characterization
The mechanical behavior of the blends was evaluated in simulated body fluids since the biocomposites were expected to be used in a hydrated environment. Figure 5
shows the mechanical properties of cross-linked biocomposites in comparison to the pure polymers, evaluated under compression mode. Pure polymers showed a very different Young’s modulus, higher for CNFs compared to Gel (0.0774 MPa versus 0.0025 MPa) confirming the use of CNFs for nano-reinforcement. The blends show similar mechanical properties; GCN-25:75 shows slightly higher Young’s modulus (0.0065 MPa), but far from the one of pure CNFs. When the CNF amount in the blend is reduced, Young’s modulus decreases in a linear way passing from 0.0036 MPa to 0.0029 MPa for GCN-50:50 and GCN-75:25 (Figure 5
A shows the viscoelastic properties of the biocomposites when they were subjected to different frequencies in the range 0.1–10 Hz. Specific frequencies were chosen to simulate in vivo stress conditions, evaluating the biocomposites’ behavior in term of storage (E’) and loss (E’’) modulus [48
]. E’’ are not reported because the values observed were very low in comparison to E’, indicating a predominantly elastic behavior [48
]. For all the biocomposites, the storage modulus (E’) did not show significant differences by varying the frequency, except for the CNFs where E’ increased from 0.0359 MPa to 0.0786 MPa. Although the storage modulus (E’) does not significantly change along with frequency, it changes depending on the Gel:CNF ratio. In agreement with the stress–strain data, the CNF sample displayed the highest storage modulus E’, and accordingly, the lowest values were obtained in the blends with a lower CNF amount. In detail, GCN-75:25 and GCN-50:50 revealed a storage modulus E’ of 0.005 MPa and 0.009 MPa, respectively. Only GCN-25:75, due to its higher amount of CNFs, revealed a storage modulus (0.0221 MPa) between the one of pure CNFs and Gel (0.0132 MPa).
Finally, a creep test was performed to evaluate the materials’ behavior when subjected to high stress for a certain time [12
]. The preliminary linearity study (data not showed) allowed to identify the suitable stress value to apply on all samples. Creep tests (Figure 6
B) showed that all the biocomposites, after being subjected to a static stress of 0.01 MPa for 15 min, can recover almost all their shape, showing an elastic behavior. Compared to the pure polymers, which displayed a strain recovery of 71% (Gel) and 59% (CNFs), only GCN-50:50 revealed a lower strain recovery of about 55%. Contrariwise, GCN-25:75 and GCN-75:25 highlighted a good strain recovery, higher than the pure materials (85% and 84%, respectively).
3.5. Evaluation of Biological Performance of Biocomposites
In order to evaluate the biomaterials’ cytotoxicity, both quantitative and qualitative cell viability tests were performed. The quantitative viability test (MTT assay) showed for all the biomaterials an overall increment of absorbance over time, directly related to the quantity of metabolically active cells (Figure 7
). GCN-50:50 and GCN-25:75 achieved the highest cell viability values at day 7, showing a significant statistical difference compared with all the other biocomposites (** p
≤ 0.01). The biocomposites had in general a higher cytocompatibility compared to the pure materials (Gel, CNFs), with a significant difference of p
≤ 0.0001; ** p
≤ 0.01 at day 1 and 3, respectively.
The qualitative viability test was performed using Live/Dead kit one day after cell seeding (Figure 8
). Here, the living cells are labelled in green and the necrotic calls are labelled in red.
For all biocomposites, the green-labeled cells were qualitatively higher in ratio compared to the red-labeled cells, which were almost not detected. This indicates a lack of cytotoxicity. The morphology of the cells cultured on the most promising biocomposites was analyzed through SEM observations. The analysis conducted at day 3 after cell seeding, showed anchored and healthy cells growing on the GCN-25:75, as their stretched cytoskeletons testify (Figure 9
In conclusion, this work well defines the key role of blends of gelatin and CNFs. All the discussed characterizations demonstrate that properties of the Gel–CNF blends are not the result of a physical mixing characterized by intermediate features only. In fact, amino functional groups of Gel react with aldehydic or carboxylic functional groups on CNFs through covalent linkages capable of generating interesting blends with new and different features [50
]. Especially, by changing the Gel:CNF ratio, it is possible to modulate the morphological, chemical and mechanical features of the biocomposites. This is essential for the creation of different biomaterials with tunable properties, and capable to guide and favor cells processing to regenerate or form new tissues [27
]. Although in vitro tests are only at a preliminary stage and are not able to evaluate a specific differentiation depending on Gel:CNF ratio, they revealed that each blend creates a different microenvironment in which the cells can grow and proliferate [53