Silk Fibroin Bioink for 3D Printing in Tissue Regeneration: Controlled Release of MSC extracellular Vesicles

Sodium alginate (SA)-based hydrogels are often employed as bioink for three-dimensional (3D) scaffold bioprinting. They offer a suitable environment for cell proliferation and differentiation during tissue regeneration and also control the release of growth factors and mesenchymal stem cell secretome, which is useful for scaffold biointegration. However, such hydrogels show poor mechanical properties, fast-release kinetics, and low biological performance, hampering their successful clinical application. In this work, silk fibroin (SF), a protein with excellent biomechanical properties frequently used for controlled drug release, was blended with SA to obtain improved bioink and scaffold properties. Firstly, we produced a printable SA solution containing SF capable of the conformational change from Silk I (random coil) to Silk II (β-sheet): this transition is a fundamental condition to improve the scaffold’s mechanical properties. Then, the SA-SF blends’ printability and shape fidelity were demonstrated, and mechanical characterization of the printed hydrogels was performed: SF significantly increased compressive elastic modulus, while no influence on tensile response was detected. Finally, the release profile of Lyosecretome—a freeze-dried formulation of MSC-secretome containing extracellular vesicles (EV)—from scaffolds was determined: SF not only dramatically slowed the EV release rate, but also modified the kinetics and mechanism release with respect to the baseline of SA hydrogel. Overall, these results lay the foundation for the development of SA-SF bioinks with modulable mechanical and EV-release properties, and their application in 3D scaffold printing.


Introduction
Three-dimensional (3D) scaffolds for application in tissue engineering (TE) are generally employed as support to improve the proliferation and differentiation of cells seeded on the scaffold or for in vivo colonization when implanted. Among the different techniques employed for scaffold preparation, 3D printing is promising as it allows production of complex scaffolds with intricately sophisticated biomimetic 3D structures capable of promoting functional tissue regeneration [1][2][3]. Specifically, during 3D printing, thin layers of a polymer, natural or synthetic in origin, are deposited in succession to form a 3D structure [4][5][6][7][8] with a fully interconnected porous network that allows cell-to-cell interactions and efficient transport and exchange of oxygen, nutrients, and metabolites [9][10][11]. Preferably, the employed materials should be cytocompatible, mechanically compatible with the target tissue, and biodegradable with a rate that matches neotissue formation so that the scaffold

Silk Fibroin Preparation and Characterization 2.2.1. Silk Fibroin Extraction and Solubilization
To extract SF, the Bombyx mori cocoons were cut into pieces of 1 × 1 cm and boiled in 0.02 M sodium carbonate (Na 2 CO 3 ) for 30 min [46,47], 1 h, 2 h, or 4 h ( Table 1). After being washed in distilled water (37 • C), the degummed fibers were dried at room temperature and then treated with 9.3-M LiBr aqueous solution at 60 • C for 4 h to dissolve SF. LiBr was then eliminated by dialysis against distilled water using dialysis cellulose tubes with a molecular weight cut-off (MWCO) of 3-5 kDa (Spectrum Laboratories, Milan, Italy). The dialysis was conducted at room temperature for 72 h, and the SF final concentration was determined by freeze-drying known SF volumes (Modulyo ® Edwards Freeze dryer, Kingston, NY, USA) at −50 • C, 8 × 10 −1 mbar for 72 h. For the further test, SF was diluted in deionized water, reaching a final concentration of 5% w/v.

Size Exclusion Chromatography (SEC)-UV Analysis
The molecular weights of SF degummed for different times were estimated by SEC-UV analysis. Chromatographic separations were performed on an Agilent HPLC series 1200 system (Santa Clara, CA, USA), equipped with mobile phase online degasser, quaternary pump, autosampler, thermostated column compartment, and diode array detector. For data acquisition, the ChemStation software version Rev. B.04.01 (Agilent Technologies, Santa Clara, CA, USA) was used. The analytical method, previously reported [48], entails the use of a TSKgel SuperSW3000 (4.6 × 300 mm; Tosoh Bioscience, Tokyo, Japan) and a mobile phase composed of 0.1 M Na 2 SO 4 and 0.05% (w/v) NaN 3 in 0.1 M phosphate buffer, pH 6.7. Flow rate, column temperature, and injection volume were set at 350 µL/min, 25 • C, and 5 µL, respectively. UV absorption was monitored at 280 nm. For MW estimation, a calibration curve was constructed using protein standards (ribonuclease A, 14 kDa; β-lactoglobulin (dimeric form), 37 kDa; bovine serum albumin, 67 kDa; rituximab, 145 kDa; thyroglobulin, 660 kDa; y = −0.9949 x + 4.9497; R 2 = 0.9934). SF samples were diluted with water to a final concentration of 0.33% w/v before injection, and analyzed in triplicate.

Physico-Chemical Characterization of SF
Infrared spectra were recorded using an Alpha II FT-IR spectrometer equipped with a platinum attenuated total internal reflectance (ATR) module (Bruker, Rosenheim, Germany); data were elaborated using Opus 7.8 software. Collection was performed on SF fibres (after the different degumming times) and the SF solution after exposure to KCl 20% w/v (to induce Silk I → SilK II conformation change) and being left dry.

Hydrogels and Crosslinking Solution Preparation
Four hydrogel formulations were prepared, as reported in Table 2, and analyzed. Three were composed of 10% w/v SA and 5% w/v SF, differing in SF degumming time. The preparation protocol involved adding and dissolving SA powder into an SF solution. The fourth composition (control) was formed of only 10% w/v SA, powdered and dissolved in distilled water. All the formulations were mixed and homogenized manually and using a rotational shaker for 10 min. Finally, considering the importance of sterile scaffolds, all bioink formulations were pasteurized at 72 • C for 1 h according to the procedure previously reported [16]. The crosslinking solution was prepared by dissolving calcium chloride (CaCl 2 ) 2% w/v, protamine 5% w/v, and potassium chloride (KCl) 20% w/v powders in distilled water. The solution was stirred at 40 • C using a magnetic plate. In particular, CaCl 2 and protamine were employed to crosslink SA, and KCl to induce the conformational change of SF from Silk I to Silk II.

3D Bioprinter and Printing Process
Printability and shape fidelity of the SA-SF hydrogel were assessed using CELLINK INKREDIBLE+, an extrusion-based 3D bioprinter. The process involved a 3D virtual geometry translated into machine instructions using a slicing software, which generates the coordinates of the printing head in each layer along with appropriate instructions (i.e., G-code). G-code also defines the velocity of the printhead movement. The positioning system for the printing head has a 10 microns resolution in the three axes. Before starting printing, the printer needs to be homed and calibrated following the process described in our previous work [49,50]. Briefly, the process involves three steps: (a) XYZ homing axes to position the printhead in the middle of the print bed; (b) Z-axis calibration to tune the distance between the nozzle and the printing bed; (c) pressure calibration (manually set through a lateral knob located on the bioprinter), to find the optimal pressure value that enables a proper flow of material.

Definition of Printing Parameters and Shape Fidelity Assessment
Shape fidelity was assessed following the protocols suggested by Paxton and Schwab et al. [51,52] that we already implemented in our previous study [49]. Two tests were carried out based on two different geometries (i.e., serpentine and grid structures) to assess quantitatively the shape fidelity of SA-SF hydrogels. All printings were performed at room temperature (25 • C), and two conical nozzles of 0.41 and 0.25 mm (inner diameter) were used. Moreover, because of the time-variable behavior of the SF, each test was performed at two timepoints, i.e., at 7 and 14 days from SF solubilization. The materials were prepared the day before each timepoint.
A Petri dish was used as printing support. Three samples were printed for each structure (serpentine-like or grid) and formulation. The images were acquired with an iPhone 12 camera and processed using ImageJ software (National Institutes of Health, Bethesda, Maryland, USA) to extract the geometrical dimensions required by the protocol.

Serpentine-Like Structure
This test consisted of printing a single-layer and continuous strand of material following the path represented in Figure 1. Firstly, this test was used to select the optimal printing parameters (i.e., printing velocity and pressure). Then, having defined these values, we quantified the degree of printing accuracy (PA (%)) both for the strand thickness and for the minimum achievable distance between filaments, because adjacent filaments printed in the same layer can fuse. As a result, the degree of printing accuracy for (a) the filament width and (b) the minimal inter-filament distance was calculated as follows: where x r represents (a) the real width of the filament and (b) the real inter-filament distance measured with ImageJ at multiple locations of each printed serpentine structure, then averaged; x i represents (a) the ideal strand thickness, which corresponds to the relevant nozzle diameter (D nozzle ), and (b) the ideal inter-filament distance, which is equal to (see Figure 1):

Grid Structure
Monolayered grid structures were printed with three different infills (10%, 15%, and 20%). Printability factor (Pr) was used as a parameter to describe the precision in printing porous structures, calculated as a function of the pore perimeter (P) and pore area (A) using the following Equation: Pr = P 2 16A Pr was calculated for each pore within the same grid structure and then averaged. High geometric accuracy results in a printability index of Pr = 1 (square transversal pore geometry), while Pr < 1 and Pr > 1 correspond to a rounder or irregularly shaped transversal geometry, respectively.  Table 2 were mechanically characterized. The materials were prepared the day before each timepoint.
Specific molds were designed and manufactured with a commercial 3D FDM printer (Creality 3D technology, China) to prepare samples for compressive and tensile tests. For the compressive test, cylindrical geometry with a diameter of 8 mm and height of 4 mm was used, as reported in [53]; for the tensile test, according to ASTM standard F2900-11 [54], dogbone geometry was used. Next, hydrogels were poured inside the molds and crosslinked for 1 h to ensure proper crosslinking in the whole sample. After that, samples were kept hydrated inside deionized water before testing. Finally, samples were measured (i.e., diameter/platen separation and cross-section/platen separation for compression and tensile tests, respectively) using digital calipers before being mechanically tested.
Different samples were prepared for each condition depending on the material's availability, as described in Table 3. SF-SF hydrogels were evaluated 7 and 14 days after SF solubilization to investigate the impact of time on their behavior, and they were all crosslinked with 2% w/v CaCl 2 + 20% w/v KCl + 5% w/v protamine solution. SA hydrogel (control) is not time-dependent. For the tensile test, two crosslinking solutions (2% w/v CaCl 2 + 20% w/v KCl + 5% w/v protamine and 2% w/v CaCl 2 ) were compared.

Mechanical Tests
Tests were carried out at room temperature (25 • C) using an MTS Insight 30 electromechanical testing system (MTS System Corporation), equipped with a 250 N load cell, following the protocols reported below.

Compression
Mechanical properties of SA-SF hydrogel in different formulations were investigated by performing uniaxial compression tests [55]. A displacement-driven unconfined compression test was performed with preload and test speed set at 0.03 N and 0.1 mm/min [56], respectively. Compression tests were performed at a maximum compression of 2.5 mm, corresponding to 50% compression.

Tension
For tensile test experiments, the gripping pressure was set at 10 bar, and the sample surfaces were covered with cardboard to prevent sample sliding. All tensile tests were performed with a preload and test speed set at 0.01 N and 1 mm/min, respectively [54]. Tensile tests were performed at a maximum tension of 15 mm, corresponding to 50% tension, or up to sample break.

Data Elaboration
Applied force and corresponding displacement were simultaneously recorded. Mechanical properties were computed as follows. Force-displacement data ranging from preload threshold to failure point were considered for analysis. In particular, the nominal stress (σ) was calculated as the ratio between the applied force and the unit area of the sample (σ = F/A), while the strain (ε) was calculated as the ratio between the dimensional variation and the initial height of the sample (ε = L − L 0 /L 0 ). Finally, each stress-strain curve was fitted with a spline adopting a least-squares approach. Finally, the elastic modulus (E) was computed as the average of the spline derivative in the strain range between 0 and 0.1.

Scaffold Fabrication
Porous cylindrical hybrid scaffolds (d: 10.52 mm, h: 9.45 mm) of polycaprolactone (PCL) and lyosecretome-laden SA-SF hydrogel were designed and 3D co-printed according to our previous study [27]. Briefly, CELLINK INKREDIBLE+ was utilized to print in combination PCL pellets heated at 90 • C and lyosecretome-laden SA-SF hydrogel prepared by dissolving 20 mg of lyosecretome for each ml of the respective SA-SF hydrogels ( Table 2). The following printing parameters were used: extrusion pressure 350 kPa; conical nozzle diameter 0.5 mm; printing speed 45 mm/min; and printing temperature 90 • C. The lyosecretome-laden SA-SF hydrogel was printed into one internal well of 6.52 mm diameter and 5.4 mm height using a conical nozzle with diameter of 0.41 mm. Printing pressure was set according to values selected during assessment of hydrogel printability and shape fidelity (Section 2.4.2). After printing, SA was crosslinked with 2% w/v CaCl 2 and 5% w/v protamine, and 20% w/v KCl was applied to induce the conformational change of SF.

Drug Release Studies
The release tests were conducted on scaffolds immersed in pH 7.2 phosphate-buffered saline (PBS, USP). The quantification of proteins was assessed using a BCA protein assay kit (from Thermo Fisher Scientific, Milan, Italy), while lipids were dosed using the Nile red method that was previously validated for this purpose [57]. The protein and lipid concentrations were extrapolated from a concentration vs absorbance plot obtained from standard protein solutions (bovine serum albumin) or standard lipid solutions (phosphatidylcholine) using a third-degree polynomial equation with R 2 = 0.99. Both analyses were performed on aliquots of PBS withdrawn at fixed time intervals. After each removal, an equivalent amount of fresh PBS was added to maintain the sink conditions. The cumulative amount of released proteins or lipids was calculated as a percentage using Equation (1): where C i is the amount of the proteins or lipids released at a definite time interval, and C 0 is the loaded protein or lipid amount. All the experiments were conducted in triplicate.

Drug Release Kinetic Study
As reported below, the in vitro drug release data were interpolated using different kinetic models. Higuchi: where F(t) is the amount of drug dissolved at time t and k is the release constant. Equation (3) is Equation (2.12) from [58]. Peppas-Sahlin: where F(t) is the amount of drug dissolved at time t, k 1 is the diffusion constant, k 2 is the erosion constant, and m is the diffusional exponent, indicative of the drug-release mechanism. Ritger-Peppas: where F(t) is the amount of drug dissolved at time t, k is the release constant, and n is the release exponent, indicative of the drug-release mechanism. Zero-order: where F(t) is the amount of drug released in time t, and k is the release constant. Korsmeyer-Peppas: where F(t) is the amount of drug released at time t; k KP is the release constant; n is the release exponent, indicative of the drug-release mechanism; and Q 0 is the initial amount of drug.

Statistical Analysis
Raw data were processed using Statgraphics XVII (Statpoint Technologies, Inc., Warrenton, VA, USA). A generalized linear analysis of variance (ANOVA) model was used, followed by Fisher's least significant difference (LSD) procedure to estimate the differences between the means. Comparisons among geometrical dimensions of the printability and shape fidelity tests were conducted using a one-way analysis of variance (ANOVA). Drugrelease results were analyzed considering the batch as a fixed factor and the drug loading as the response variable. The statistical significance was set at p < 0.05.

Results and Discussion
The present work aimed to obtain an optimized SA-SF hydrogel for 3D bioprinting applications in the TE field. In terms of its ideal characteristics, this bioink should be easily printable with good mechanical properties. While the presence of fibroin in Silk I conformation is good for printability, because the low abundance of β-sheet network avoids clogging in the needles [44,45], its mechanical properties are not ideal; to this end, a Silk II conformation is indeed preferred to achieve high tensile strength and modulus [29]. Therefore, ideally, SF must (i) remain in Silk I conformation, i.e., in solution, during printing to prevent the nozzle from clogging, and then, after being printed, (ii) the conformational change from Silk I to Silk II must still be possible, to obtain a material with appropriate mechanical properties.
Regarding point (i), the degumming process (i.e., the separation of SF from sericin) was successfully modified. The standard degumming method (boiling the cocoons for 30 min in 0.02 M sodium carbonate) followed by fibroin dissolution in 9.3 M LiBr allowed us to obtain a regenerated SF that, despite being in Silk I conformation, was not printable, as the protein clotted the needle (data not shown). Therefore, the treatment with the alkaline solution was increased from the standard (30 min) to 1, 2, and 4 h. In agreement with previous reports [59], more hydrolysis of fibroin was observed: the SEC-UV analyses showed a decrease in the MWs of SF. Specifically, a broad peak was observed in each of the chromatographic profiles ( Figure S1) of SF that had been degummed for 1, 2, and 4 h. The elution volumes calculated at the peak apex allowed estimation of the average MWs, reported in Table 4. Unlike the other samples, the 30 min degummed sample eluted mainly in the void volume, demonstrated to be outside of the top MW limit of the column (> 500 kDa). Peak area values showed acceptable precision for all the samples (RSD < 10%), even when repeatability diminished with reduced degumming time and thus diminished SF hydrolysis. This evidence may be related to precipitation and/or aggregation phenomena. Following this first step, regarding point (ii), it was assessed whether SF even with a low molecular weight could change its conformation from Silk I (in solution) to Silk II (insoluble). To evaluate this, the formation process of fibroin filaments in the silkworm was considered. Specifically, Chang and colleagues demonstrated that the pH gradient and cations in the anterior silk gland both play important roles in changing fibroin conformation: lower pH and Ca 2+ levels and high K + concentrations promote β-sheet formation [60]. Accordingly, fibroin samples were treated with KCl 20% w/v to induce conformational change from SilkI to Silk II. After being drying, the infrared spectra were recorded and compared with native SF fibers to assess the presence of the β-sheet network (Figure 2). Amide I (1600-1700 cm −1 ) and amide II (1500-1600 cm −1 ) bands were used for the conformational analysis of the secondary structure. The absorption peaks in the amide I at around 1620 cm −1 are assigned to the β-sheet and at 1680 cm −1 to the random coil and/or helical conformation, while in the amide III bands, peaks at 1266 cm −1 and 1242 cm −1 are assigned to the β-sheet and random coil and/or helical conformation, respectively [61,62]. As observed in the spectra of the SF fibers, amide I and amide II bands were detected at about 1620 cm −1 (C = O stretching) and 1510 cm −1 (N-H bending), respectively, while amide III was present at about 1230 cm −1 (C-N and N-H functionalities). The amide I bands were still detectable for the regenerated SF treated with KCl but shifted towards 1640 cm −1 , while the amide II and II bands remained unaltered. The shifting of amide I in the regenerated SF treated with KCl suggests a reduction of β-sheets and an increase in random coil conformations with respect to the native fibre. In this regard, it should be considered that SF chains have intramolecular or intermolecular interactions in which molecules are entangled. Therefore, it is likely that because of the reduced molecular weight following solubilization, the chain-chain interaction in the solution has been hindered, causing an increase of randomness in regenerated SF treated with KCl compared with the native fibers. However, no differences among the different degumming times were revealed, suggesting that all SFs were transited after treatment with KCl.
Then, a printing assessment was performed (Figure 3). To this end, the three different formulations of SA-SF hydrogels were evaluated in terms of shape fidelity at two different timepoints (7 and 14 days after SF production) to assess the time-dependent behavior of SF. Two different nozzle diameters (0.41 mm and 0.25 mm) were used during this characterization. Firstly, various combinations of printing speed and pressure were tested using a trial-and-error approach, as described in [51,52], with the final selection of the best pairs of values reported in Table 5.  It can be observed that 7 days after the preparation of SF, all three formulations were printable. The only difference was in SA-SF-1h, for which a lower pressure and a greater speed were selected; this is a typical feature of hydrogels with lower viscosity. However, 14 days after SF preparation, SA-SF-1h was not printable, probably due to a conformational change of the SF from Silk I to Silk II, leading the material to become stuck in the nozzle, thus making it hard to extrude [44,45]. Conversely, SA-SF-2h and SA-SF-4h were still printable with the same parameters determined at t1.
Then, once the printing parameters were defined, we quantitatively assessed the shape fidelity by measuring filament width, inter-filament distances, and the printability index (Pr) value. The following measures are shown in Figure 4. For inter-filament distance and the Pr, we report only the values related to the minimum resolution that we were able to reach, i.e., 1 mm − D nozzle and 20% infill, respectively. Results show statistically significant differences in filament width, inter-filament distance, and Pr among the three formulations evaluated at the same timepoint with differences within the same formulation printed on different days after SF synthesis. Moreover, as expected, due to the pooling effect typical of viscous material such as hydrogel, it was impossible to satisfy the ideal values of diameter and inter-filament distance. All of the measured Pr values were slightly lower than the ideal value of 1, which describes good shape fidelity in the pores, while highlighting the viscous and potentially collapsible nature of the material.
Finally, as reported in Table 6, the printing performance of each SA-SF hydrogel was quantified. We can state that SA-SF-2h and SA-SF-4h, printed at the first timepoint with a 0.41 mm nozzle, showed the better results. However, at the second timepoint, there was a reduction in printing performance when the same speed and pressure values were maintained. Mechanical evaluation was performed of SA-SF-based hydrogels characterized by different degumming times, and the results are shown in Figure 5. SA hydrogel was used as a control, with compressive and tensile elastic modulus values consistent with other scientific works [54,63]. Regarding the presence of SF in the hydrogel, ANOVA analysis showed that at t1 the degumming process did not affect the compression response for the three SA-SF formulations (p > 0.05), and their compressive moduli were three times higher (p < 0.01) than that of the hydrogel formed only with alginate, in accordance with previous findings [64]. At t2, SA-SF-2h and SA-SF-4h maintained their initial compressive modulus, while the compressive modulus of SA-SF-1h significantly decreased, reaching the Young's modulus value of SA hydrogel. Conversely to the compression response, the presence of SF in SA hydrogel did not affect the tensile elastic modulus, and this was observed for each type of degummed SF at both timepoints ( Figure 5B). Moreover, we performed a tensile test on SA hydrogels crosslinked with only 2% w/v CaCl 2 and with 2% w/v CaCl 2 + 5% w/v protamine + 20% w/v KCl to assess how the crosslinking method could influence tensile performance. As indicated in Figure 5C, the value significantly changed between the two crosslinking methods. Indeed, the CaCl 2 solution formed a structure with a tensile modulus (~660 KPa) three times higher than the other method (~215 KPa). Therefore, it was hypothesized that during the crosslinking process the K + ions of KCl compete with the Ca 2+ ions in the formation of the alginate hydrogel, making it less compact overall. This observation suggested how to change the crosslinking method: at first with CaCl 2 , then with protamine, and only when completely gelified was the last treatment step performed with KCl. Analyzing the literature, we found that improvements in tensile performance for SF-based materials have mainly been obtained when fibroin is present in the form of fibers [65] or as physically crosslinked hydrogels [66]. However, no tensile mechanical characterization studies have been published for SF/SA blends chemically crosslinked with CaCl 2 , protamine, and KCl and few studies have characterized chemically crosslinked SF-based hydrogels with tensile testing. Indeed, as Yu Zhao et al. reported [67], chemical crosslinking leads to rapid gelification of the material without inducing in the SF network an immediate formation of β-sheet structures, which are responsible for conferring the structure's high mechanical performance. In contrast, a physical crosslinking method stimulates the β-sheet's nucleation, possibly reaching compressive and tensile modulus ranges of MPa.
Finally, we investigated how the presence of SF and its molecular weight modified the release of lyosecretome-a freeze-dried formulation of MSC secretome-from the SA-SF hydrogels. The results are reported in Figure 6 for only the lipidic component of the lyosecretome, i.e., the EVs, as it was impossible to discriminate if the protein released were of lyosecretome or fibroin. The time on the x axis is not in scale, to better visualize the curves and the error bars. The addition of SF significantly modified the release of lipids with respect to the baseline of SA-only hydrogel, which was considered a control, depending on the degumming time. Specifically, the release of EVs was faster than the control when 1 h degummed SF was used, while the use of SF degummed for 2 or 4 h significantly slowed the release of lipids (p < 0.001). In detail, after 2 h, SA-SF-1h hydrogel had a burst release of almost 50% vs 30% for the control and 16% and 11% for SA-SF-2h and SA-SF-4h, respectively.  The release data were further elaborated by the kinetic models employed, including Higuchi, Peppas-Sahlin, Ritger-Peppas, and zero-order, which were applied to investigate the release mechanism in greater depth (Table 7). Table 7. Results of release model fitting. Kinetic elaborations were performed on release data obtained from at least three independent experiments for each batch.~indicates that the analysis performed was "ambiguous"; therefore, the fit does not confirm the values of all the parameters, and 95% confidence bounds cannot be reported. These latter data were not considered in the interpretation of results.   SA-SF hydrogels can control the release of EVs in lyosecretome by a combination of diffusion and erosion. According to this statement, the data fit the Ritger-Peppas and Korsmeyer-Peppas models with an R 2 close to or greater than 0.9. The data also fit the Peppas-Shalin model, which considers a Fickian contribution (first term) and case-II relaxation (second term), i.e., diffusion and erosion, respectively [67,68]. Regarding the diffusion contribution (Ritger-Peppas and Korsmeyer-Peppas models), the diffusion speed decreases as the degumming time of the fibroin increases. For example, the K values were high when using 1 h degummed SF (29.56) and low (4.572, 1.982) when using 2 h and 4 h degummed SF, respectively. Unfortunately, as the results of the analyses were ambiguous, no further information can be drawn regarding the erosion mechanism. Furthermore, no interpretation of the release exponent (n in Equations (5) and (7) or m in Equation (4) is provided, as these exponents are valid only for drug-release systems with defined geometries (planar as thin films, cylindrical or spherical), which does not apply to the geometry of the scaffolds considered in this study. Based on observations of the effects of SF in the release profile, it is likely that its presence modifies the polymeric network and affects the interactions among polymer chains, depending on the molecular weight of SF. When the molecular weight of the protein is high, it offers a higher barrier to forming a compact polymeric network of alginate polymer chains, resulting in faster release. A key role in the diffusion process is played by the polymeric network and the steric hindrance it offers to the diffusion path of nanometric lipid assembly, including EVs. In general, an increase in the compactness of the polymeric network, consequent to a higher concentration of the polymer or higher interaction and crosslinking among polymer chains, slows diffusion. Conversely, when the molecular weight of SF is low, it does not hamper the formation of a solid and compact polymeric network of alginate polymer chains, and indeed the release is slower precisely due to the high steric hindrance the hydrogel offers. The slower release of SA-SF-2h and SA-SF-4h in comparison with the control may be due to the presence of SF in Silk II conformation reducing the erosion of the hydrogel, thus offering for longer a steric hindrance to lipid diffusion.

Conclusions
We optimized the preparation protocol for obtaining an SF-based solution by investigating three different degumming times to develop a 3D (bio)printable SA-SF-based hydrogel for TE applications. First, the extraction process of fibroin was optimized by increasing the degumming time in Na 2 CO 3 0.02 M (from 30 min to 1, 2, or 4 h) to reduce the fibroin's molecular weight and thus achieve a printable protein solution that remained capable of the conformational change from Silk I (random coil) to Silk II (β-sheet). From a printing point of view, SA-SF hydrogels with a degumming time of 2 h and 4 h resulted in the best performance in term of shape fidelity. Furthermore, adding SF to the alginate hydrogel increased the compressive response, especially when degummed for 2 and 4h (e.g., at 7 days, Young's modulus: 24.7 ± 10.7 KPa, 54.6 ± 17.3 KPa, and 70.9 ± 12.7 KPa for SA, SA-SF-2h, and SA-SF-4h hydrogels, respectively), probably caused by the formation of denser and condensed networks due to increased polymer content. However, it did not influence tensile performance (e.g., at 14 days, Young's modulus: 215.5 ± 44.2 KPa, 198.6 ± 64.2 KPa, 223 ± 38 KPa, and 281.3 ± 103.8 KPa for SA, SA-SF-1h, SA-SF-2h, and SA-SF-4h hydrogels, respectively), for which a physical crosslinking method could be adopted for future development. Indeed, we found that the crosslinking method in the material strongly influenced the tensile response. For example, for SA hydrogel treated with CaCl 2 + KCl + protamine solution and CaCl 2 solution, Young's modulus results were 215.5 ± 44.2 KPa and 659.4 ± 31.5 KPa, respectively. Finally, degumming of SF for 2 and 4 h dramatically slowed the EV release and modified the kinetics and mechanism of release with respect to the SA hydrogel baseline.
Overall, these results lay the foundation for further development of SA-SF bioinks with modulable mechanical and EV-release properties, and their use in scaffold 3D printing.

Data Availability Statement:
The data presented in this study are contained within the article.

Conflicts of Interest:
The funders had no role in the design of the study; in the collection, analyses, or interpretation of data; in the writing of the manuscript; or in the decision to publish the results. S.P. and M.L.T. are co-founders and members of the advisory board of PharmaExceed S.r.l.; F.S. and M.C. are co-founders and members of the advisory board of P4P S.r.l.