Designing Tunable GelMA Hydrogels by Integrating Mammalian and Non-Mammalian Gelatins

1. Introduction

Gelatin is a versatile hydrocolloid derived from the partial hydrolysis and thermal denaturation of collagen [1], obtainable from bovine, porcine, fish, and poultry sources, each yielding materials with distinct physicochemical properties depending on the animal source and extraction conditions [2]. Two principal types are recognized: type A, produced by acid hydrolysis (isoelectric point pH 6–9, predominantly from porcine skin [3]), and type B, produced by alkaline hydrolysis (isoelectric point pH 5-6 [4,5]). At the molecular scale, gelatin comprises α-, β-, and γ-chains [4,6], where α-chains adopt a polyproline II conformation [7,8], sustained by the repetitive Gly-X-Y tripeptide sequence, with proline (Pro) and hydroxyproline (Hyp) most frequently occupying the X and Y positions [9]. This sequence governs the thermoreversible formation of collagen-like triple-helix structures upon cooling, whose extent and thermal stability are directly proportional to total Pro + Hyp content [10,11]. Cold-water fish gelatins, including those derived from salmon skin, contain markedly lower Pro + Hyp levels than mammalian gelatins, resulting in lower Bloom values, reduced gel strength, and melting points that fall well below physiological temperature [11,12,13]. These properties, however, confer distinct processing advantages: salmon skin gelatin remains in the sol state at 5 °C [14,15,16], making it particularly attractive for biofabrication approaches that require low-viscosity inks at near-ambient temperatures. Moreover, its use offers eco-friendly valorization of a major aquaculture by-product and avoids the religious and ethical restrictions associated with mammalian-derived gelatins[10,17,18,19].

Interest in gelatin for biomedical applications has grown considerably, driven by its excellent biocompatibility, biodegradability, and intrinsic cell-adhesion motifs that recapitulate key features of the native extracellular matrix (ECM) [20,21,22]. Nevertheless, its low melting temperature and susceptibility to enzymatic degradation under physiological conditions limit direct use in tissue engineering scaffolds [23,24]. Chemical modification to gelatin methacryloyl (GelMA) overcomes these limitations by introducing photocrosslinkable methacrylate groups that enable the formation of covalently stabilized hydrogel networks with independently tunable stiffness, porosity, and degradation rate [25,26,27]. The resulting hydrogels retain the ECM-mimetic properties of native gelatin, supporting cell adhesion, proliferation, and spreading, and are compatible with a wide range of fabrication strategies including bioprinting and microfabrication [28]. Mechanical and structural properties can be further modulated by adjusting the degree of substitution (DS), polymer concentration, and photocrosslinking conditions [29,30,31], though high GelMA concentrations increase network stiffness at the expense of porosity and cell infiltration [32]. Strategies to decouple these trade-offs include the incorporation of nanomaterials [25,32], optimization of the functionalization protocol [33], and the induction of physical crosslinks through temperature-driven triple-helix formation prior to photopolymerization [34,35]. This last strategy is particularly relevant when working with gelatin mixtures, as the source-dependent differences in Pro + Hyp content create compositionally programmable physical crosslink densities. Geonzon et al. [36] demonstrated that blending porcine skin gelatin with cold-water fish gelatin reinforces gelation temperature and complex modulus even at low porcine fractions, with the extent of reinforcement and the resulting network architecture, ranging from intertwined hybrid triple-helices to isolated, separate polymer networks, governed by the cooling rate applied during gelation [36]. This thermally programmable interspecies co-aggregation mechanism offers a compelling strategy for modulating the physical crosslink network in GelMA hydrogels without altering total polymer concentration or introducing exogenous materials.

Despite the promise of GelMA, its production still lacks standardization: reported synthesis protocols vary widely in reaction pH, temperature, and duration [37,38,39,40,41,42], and these variables have a significant impact on the degree of substitution, photopolymerization efficiency, and the resulting mechanical performance [43]. Thorough characterization of GelMA after synthesis is therefore essential before any inference about hydrogel behavior can be drawn.

From a sourcing perspective, salmon skin represents an abundant and underutilized co-product of the aquaculture industry, offering collagen contents among the highest reported for fish-derived by-products [44]. The highly controlled feeding and husbandry protocols of intensive salmon farming further reduce the compositional variability that typically complicates batch-to-batch reproducibility in gelatin extraction, making salmon an attractive and sustainable raw material for biomedical-grade GelMA.

Building on this context, the present work comprehensively investigates the gelation behavior and physicochemical properties of hydrogels derived from mixtures of well-characterized porcine and salmon gelatins across a range of compositional ratios, and their corresponding GelMA derivatives. We demonstrate how the PG-SG ratio can be exploited as a tunable structural parameter, independently of total polymer concentration or chemical crosslink density, to produce GelMA hydrogels with systematically varied triple-helix content, mechanical stiffness, swelling behavior, and controlled-release kinetics. The approach offers a simple, cost-effective route to structurally modulated hydrogels based on naturally derived materials.

2. Materials and Methods

2.1. Materials

Atlantic salmon (Salmo salar) skins were provided by AquaChile, Cardonal Plant, Puerto Montt, Chile. Sodium hydroxide (≥99.0%), glacial acetic acid, phosphate-buffered saline (PBS) 10X, chlorohydric acid (HCl), Sodium Dodecyl Sulfate (SDS), and deuterium oxide (≥99.9% deuteration) were purchased from Merck (Germany). Gelatin from porcine skin (gel strength ~300 g Bloom, Type A), methacrylic anhydride (≥94%), 2-Hydroxy-4′-(2-hydroxyethoxy)-2-methylpropiophenone (IRGACURE 2959), glycine, serine, 2,4,6-Trinitrobenzenesulphonic acid (TNBS), sodium bicarbonate, and rhodamine 6G were purchased from Sigma Aldrich (USA).

2.2. Salmon Gelatin Extraction

Salmon gelatin was extracted following Zhou & Regenstein [45] with modifications as described in Padilla et al. [46]. Briefly, salmon skins were cleaned and cut into ~3 cm² pieces, subjected to two pretreatments in 0.1M NaOH at 10 °C for 1h each, then 0.05M acetic acid under similar conditions, with tap water washes between steps. Gelatin was extracted at pH 4.0 (acetic acid) at 60 °C for 4h, vacuum filtered, and dried at 60 °C for 72h. The resulting gelatin was ground and stored until use.

2.3. Gelatin Composition

Composition was determined by proximate analysis and amino acid profiling according to AOAC methods [47]. Moisture content was measured gravimetrically after drying the samples at 105 °C for 24h. Crude fat content was determined using Soxhlet extraction, and total protein content was evaluated via the Kjeldahl method, utilizing a nitrogen-to-protein conversion factor of %N × 5.55. Ash content was measured by calcination in a muffle furnace at 550 °C, while the non-nitrogenous fraction was estimated by difference [17]. For amino acid profiling, key residues (Gly, Pro, and Hyp) were quantified using high-performance liquid chromatography (HPLC)[48]. Samples were subjected to acid hydrolysis using 6N HCl at 110 °C for 24h, followed by pre-column derivatization with phenylisothiocyanate. Separation was performed on a Luna RP18 column (Phenomenex, USA) utilizing an HPLC system equipped with a Waters 600 controller and a Waters 996 photodiode array detector (DAD) set at 254 nm. All proximate and amino acid values were expressed as g/100 g of sample.

2.4. Production of Gelatin Mixtures

Dry salmon (SG) and porcine (PG) gelatins were mixed in different weight proportions. PG weight fractions: 0, 0.1, 0.2, 0.4, 0.6, 0.8, and 1.0, yielding seven samples.

2.5. Gelatin Suspensions Functionalization

SG-PG mixtures were suspended at 10% w/v in PBS 1X (pH 7.4) and functionalized with methacrylic anhydride (MA) at 8% v/v, pH ~4, 60 °C for 3 h. The reaction was stopped by dilution, with three times the initial PBS 1X volume. Suspensions were vacuum filtered (22 µm, Whatman) and diafiltered in a TFF system SartoFlow^® Smart (Sartorius, Germany) with a 10 kDa cutoff Sartocon Hydrosart^® cassette at 20% crossflow rate and 1.5 bar, using 60 °C distilled water, until conductivity stabilized at ~100 µS. GelMA suspensions were freeze-dried and stored at 4 °C.

2.6. Degree of Substitution by TNBS

The TNBS assay was performed as described by Claaßen et al. [49], with minor modifications. Samples were dissolved at a concentration of 20mg/mL in a sodium bicarbonate solution (4% w/v, pH 6.0). Aliquots of 25µL were subsequently mixed with 25 µL of the sodium bicarbonate buffer and 25 µL of a 0.1% (v/v) TNBS solution. The reaction mixture was incubated at 37°C for 2.5h protected from the light. After incubation, the reaction was stopped by adding 25µL of a 10% (w/v) SDS solution and 25µL of 1M HCl. The absorbance of the resulting solution was measured at 335 nm using a microplate reader (Spark, Tecan Männedorf, Switzerland). A glycine standard curve (12.5 - 100 µg/mL) was used. Measurements were performed in triplicate. The DS of amine groups in GelMA samples was determined in respect to an average value of amine groups of the corresponding unmodified gelatin as follows:

$$DS\%=\left[1-\left({\displaystyle \frac{GelF}{Gel}}\right)\right]x100$$

where GelF= functionalized gelatin, Gel = unmodified gelatin.

2.7. Degree of Substitution by ¹H-NMR

Samples at 6mg/mL in D₂O were analyzed on a Bruker Avance II+ 500 MHz spectrometer (B500, Bruker, Karlsruhe, Germany). All measurements were performed at 25 °C and 16 scans were acquired. Data was processed with MestReNova14 (Mestrelab, Spain). Spectra was normalized to the phenylalanine signal (6.9–7.5 ppm). The lysine methylene peak (2.95-3.05 ppm) was integrated, using manual integration of peaks of interest with automatic lineal correction. Measurements were performed one time, to confirm TNBS results. The DS was calculated as described by Hoch et al. [50]:

$$DS\left(\%\right)=\left(1-{\displaystyle \frac{lysinemethylenepeakareaofGelMA}{lysinemethylenepeakareaofgelatin}}\right)·100$$

2.8. Rheological Characterization

Gelatin mixtures and GelMA samples prepared at 7% w/v in PBS 1X pH 7.4 were characterized using a Discovery HR-2 rheometer (TA Instruments, USA). After the determination of the linear viscoelastic region (LVR), oscillatory measurements during cooling (40 to -5 °C, 3 °C/min) at 1% deformation and 1Hz werw performed. A 50 mm parallel plate geometry with a 300µm gap were used, as well as and a solvent trap to prevent water evaporation during the analysis. Gelation temperature (Tgel) determined as the elastic modulus (G’) and loss modulus (G’’) crossover. Viscosity was measured at 800 s⁻¹ using a 40 mm, 0.5° cone geometry (50 µm gap). Measurements were performed at least in triplicate. The viscosity of the suspensions was determined using the same temperature range and rate as oscillatory measurements, with a shear rate of 800 s⁻¹. A 40 mm diameter and 0.5º angle cone geometry with a 50µm gap were used.

2.9. Mechanical Characterization of Gelatin Mixtures

Gelatin mixtures at 7% w/v PBS 1X pH 7.4 were loaded on a 40 mm parallel plate (2 mm gap) preheated to 37 °C using a Discovery HR-30 rheometer (TA Instruments, USA), conditioned at 4 °C for 480 s, then subjected to unconfined linear compression at 10 µm/s at 4 °C. Compression modulus was calculated from the initial linear region up to 4% strain. Measurements were performed at least in triplicate.

2.10. Thermal Properties by DSC

Thermal transitions were determined with a DSC-1 (Mettler Toledo) calibrated with indium (T_m = 156.6 °C, ΔH = 28.4 J/g). Approximately 90µL of a 10% w/v pH 7.4 suspension was loaded into 100µL aluminum pans. The thermal program was as follows: heating to 40 °C at 10 °C/min (hold 3 min); cooling to -15 °C at 3 °C/min (hold 3 min); heating to 40 °C at 3 °C/min. ΔHm was estimated from the main endotherm. Measurements were performed at least in triplicate.

2.11. GelMA Hydrogel Fabrication

GelMA suspensions were prepared at 7% w/v in PBS 1X with 0.3% w/v of IRGACURE 2959 and adjusted to pH 7.4. The samples were maintained at 37˚C for at least one hour and immediately poured into PDMS molds (10mm diameter, 2mm height) for crosslinking or poured into PDMS molds, preincubated at 4˚C for 1h (FOC 215E, VELP Scientifica, Usmate, Italy) and then immediately crosslinked. Crosslinking was achieved by exposing samples to UV light (365 nm, 12.0 W/cm², Cool Cure 395, Lesco UV, American Ultraviolet, Paramount, CA, USA) at 4 cm for 2 min.

2.12. Mechanical Characterization of GelMA Hydrogels

Hydrogels were tested using a TA.XT2 Plus texture analyzer (Stable Micro Systems, Surrey, UK) in an unconfined compression setting. Hydrogels were compressed 1mm at a compression rate of 0.2mm/s using a 1cm diameter cylindrical probe at room temperature. The compression modulus was calculated from the initial slope of the stress/strain curve that included data up to 5% strain. Measurements were performed at least in triplicate.

2.13. Swelling

The swelling degree was quantified by first freezing the hydrogels through a stepwise cooling protocol (4 °C for 2h, -20 °C for 24h, and -80 °C for 24h) and subsequent freeze-drying for 24 h. Upon recording the dry weight, the samples were rehydrated in 1X PBS at 37 °C for 24 h. After reaching equilibrium, the hydrogels were removed from the medium, gently blotted to remove excess surface moisture, and re-weighed to obtain the swollen mass. The final swelling ratio was calculated as follows:

$$Swelling\%=\left({\displaystyle \frac{Weq-Wi}{Wi}}\right)x100$$

where, Weq and Wi are the weights of swollen sample after equilibrium and initial dried sample, respectively. Measurements were performed in triplicate.

2.14. Controlled Release Assays

Rhodamine 6G (R6G) was used as a model molecule, as it has been previously employed in controlled release assays from hydrogels [51]. Hydrogels were prepared as in Section 2.11, cooled through a stepwise cooling protocol (4 °C for 2h, -20 °C for 24h, and -80 °C for 24h), freeze-dried, and rehydrated in PBS 1X for 24 h at 37 °C. After rehydration, hydrogels were immersed in R6G loading solution (0.5 mg/mL in PBS 1X,) for 72 h at 37 °C. The release at 37 °C was monitored in 2.5 mL PBS 1X by withdrawing 100 µL aliquots at 10, 20, 30, 50, 110, 170, 230, 350 min, replacing with an equal volume of fresh PBS 1X to maintain sink conditions. R6G was quantified spectrophotometrically at 525 nm (Infinite 200 Pro, Tecan). Cumulative released mass was calculated as:

$$M_{\mathrm{cum}}\left(n\right)=V_t{C}_n+{\displaystyle \sum _{i=1}^{n-1}}{V}_s{C}_i$$

where V_t = 2.5 mL (total volume), V_s = 0.1 mL (aliquot volume). Measurements were performed with a n=6 different hydrogels.

2.15. Weibull Kinetic Modeling

To characterize the release kinetics of R6G from the GelMA hydrogels, the Weibull model [52] was fitted to the cumulative release data normalized to the release value at 350 min, which was assumed to represent the plateau (maximum) release since no significant increase was observed at longer times. The Weibull equation corresponds to:

$$M=M_{\infty }\left[1-e^{-{\left({\displaystyle \frac{t-T}{\alpha }}\right)}^{\beta }}\right]$$

where α is the characteristic release time (time for 63.2% of drug release), β is the shape exponent that governs the release mechanism, and T is the lag time prior to detectable release [53,54], which was fixed at 0 due to the immediate release of rhodamine upon contact with PBS 1X. Model parameters were estimated by nonlinear regression, and the fit accuracy was assessed by the coefficient of determination (R²).

2.16. Statistical Analysis

Data obtained was summarized in tables by calculating the media average and standard deviation. The number of replicates for each experiment is depicted on each figure description. For inferential statistics analysis, specific analysis between two experimental groups was applied using a Mann-Whitney or Unpaired t-tests. All statistical analysis was performed with a confidence level of 95%, using the software Graph Pad Prism V.10.

3. Results and Discussion

3.1. Gelatin Characterization

Both porcine and salmon gelatins exhibited high purity, with protein contents exceeding 90% in both cases (Table S1). Salmon gelatin (SG) showed slightly lower protein and higher ash content compared to commercial porcine gelatin (PG) (Table S1). Consistent with previous reports [46], SG contained significantly lower concentrations of proline (Pro) and hydroxyproline (Hyp) compared to PG (Table S2), reflecting the well-established compositional differences between cold-water fish and mammalian gelatins.

3.2. Gelatin Mixtures Characterization

Gelatin mixture samples were prepared and characterized at 7% w/v, a concentration chosen for its comparability to standard Bloom value determinations [18]. Increasing PG proportions in gelatin mixtures increased viscosity across all evaluated temperature ranges (Figure 1A), accompanied by increases in elastic modulus (G’), compression modulus at 4 °C (Figure 1B, 2A, and S1), and in the gelation temperature (Tgel) (Figure 1C, and S1). This suggests that PG effectively reinforces the gelatin mixture molecular structure even at low contents, as evidenced by G′, Tgel, and compression modulus of the 0.9SG0.1PG sample. Interestingly, this reinforcement is not linear (Figure 1C), suggesting that molecular interactions or entanglements may be occurring.

Thermal properties determined by DSC revealed no unique transition temperature at the tested cooling rate; instead, separate transitions from SG (0 - 15 °C) and PG (20 - 40 °C) were identified (Figure 2B and Table 1), suggesting a low inter-species chain interaction, where it is probable that SG concentrations may be hindering PG triple-helix formation. Since our DSC data show separate endothermic transitions for PG and SG under the tested conditions, we cannot confirm the presence of interspecies chain aggregation. This contrasts with the work of Geonzon et al. [36], who showed a single endothermic transition dependent on the relative fractions of fish scale and porcine gelatins using micro-DSC. However, in that study both gelatins had similar molecular weights, whereas our SG has a lower molecular weight than PG [46], potentially influencing the extent of SG-PG interspecies chain interactions. Overall, under the cooling conditions used (3 °C/min) in this study, it is possible to obtain hydrogels with mixed and separate PG and SG triple-helix structures.

3.3. GelMA Suspensions Characterization

After functionalization, DS was determined by TNBS and ¹H-NMR, with both methods yielding very similar results and near-complete functionalization (95–98%) across all samples (Table 2). Rheological characterization showed similar trends to gelatin mixtures: increasing PG weight fraction increased viscosity, G′ at 4 °C, and Tgel of GelMA samples (Figure 3A-C and Figure S2). However, functionalization decreased overall viscosity, G′ at 4 °C, and Tgel compared to original gelatin mixtures (Figure 1 and Figure 3), attributed to steric impairment due to the presence of methacryloyl groups and decreased molecular weight during the functionalization reaction [38,41,46]. DSC showed similar trends, showing an overall lower triple-helix amounts due to chain destructuring (Table 3 and Figure 4A). For example, in the case of the 0.8SG0.2PG GelMA sample, the transition related to porcine gelatin cannot be detected. Another evidence of this destructuring can be identified in the 0.2SG0.8PG GelMA sample, were no significant differences in rheological behavior or triple helix content were observed compared to porcine GelMA (Table 3 and Figure 4A), suggesting that the maximum triple helix content was reached in the evaluated system.

Pre-incubation at 4 °C for 1 h prior to photocrosslinking produced hydrogels with significantly increased compression modulus with increasing PG content (Figure 4B). Although no differences were found in rheological or thermal characterization between 0.2SG0.8PG GelMA and porcine GelMA, a higher compression modulus was observed in the latter, potentially due to the contribution of salmon GelMA triple helices. The presence of PG in the GelMA hydrogels also decreased swelling degree (Figure 4C), particularly at higher PG contents. Overall, under the cooling conditions used (3 °C/min) in this study, GelMA hydrogels with differential triple-helix content, mechanical properties, and swelling degree can be obtained by modulating the SG-PG ratio: higher PG content yields hydrogels with higher compression modulus and lower swelling, suggesting potential for modulating drug release profiles, as further explored in Section 3.4.

3.4. Rhodamine 6G Loading and Release

Rhodamine 6G (R6G) was used as a model molecule to evaluate loading and release capacity of the GelMA hydrogels, as it has been previously employed in analogous systems [51]. Similar cumulative release profiles were obtained for all GelMA formulations (Figure 5). The pre-incubation temperature prior to photocrosslinking markedly modulated R6G delivery: hydrogels pre-incubated at 4 °C exhibited slower cumulative R6G release compared to those pre-incubated at 37 °C (Figure 5).

The Weibull model was selected to describe the release kinetics of the different GelMA formulations, yielding an excellent fit across all datasets (R²$\ge$0.995) (Table 4). The model parameters considered most critical for our samples are α and β. The parameter α represents the time required to reach a cumulative molecule release fraction of 63.2%, a value derived from the mathematical structure of the exponential Weibull distribution. Meanwhile, β defines the physical transport mechanism of the model molecule and dictates the geometric shape of the release curve [52,53]. This model showed an increase in α values when physical crosslinking was promoted, indicating an overall slower release kinetics and greater resistance to the diffusion process, likely arising from physical or chemical barriers such as matrix tortuosity or polymer-molecule interactions [52,53].

The values obtained for β ranged between 0.75<β<1, suggesting that the transport mechanism within these hydrogels is not purely dominated by Fickian diffusion, instead it corresponds to an anomalous transport mechanism that combines Fickian diffusion with structural relaxation or changes in the polymer network [54,55]. This behavior reflects the dual network architecture of GelMA hydrogels, in which covalent crosslinks established during photopolymerization coexist with physically associated triple-helix domains derived from the gelatin chains

Taken together, these results demonstrate that the pre-incubation temperature, and thus the extent of physical crosslinking prior to photopolymerization, is a simple and effective lever for tuning the drug release profile of SG-PG GelMA hydrogels. While the overall differences among different SG-PG proportions remained relatively subtle, the 0.4SG0.6PG formulation exhibited a higher increase in β at 4 °C, with a more pronounced disparity between its 37 °C and 4 °C kinetic profiles. Given that higher SG contents appeared to hinder the assembly of PG triple-helical structures, as corroborated by our DSC and compression tests, the 0.4SG0.6PG formulation stands out as the most promising candidate and an interesting starting point for further optimization.

4. Conclusions

This study demonstrates that blending porcine skin gelatin (PG) and salmon skin gelatin (SG) at controlled compositional ratios constitutes an interesting additive free strategy for engineering GelMA hydrogels with tunable physical crosslinkuing density, mechanical stiffness, swelling behavior, and compound release kinetics, without modifying total polymer concentration or the degree of chemical functionalization. Gelatin source composition determines the physical crosslink landscape. SG, with its lower Pro + Hyp content, remains largely in the sol state at biofabrication relevant temperatures and contributes minimal physical crosslinking, while porcine gelatin forms thermally stable triple helices that reinforce the network in a dose dependent, non-linear fashion. DSC confirmed that under the tested cooling conditions, PG and SG chains form largely independent helical populations, with high SG fractions partially hindering PG triple-helix nucleation, behavior distinct from co-aggregation systems where both gelatins share similar molecular weights. Gelatin mixtures functionalizationn attenuates their ability to form physical crosslinks, reducing viscosity, G′, and Tgel in all samples relative to parent gelatins, consistent with steric impairment of triple-helix formation. However, compositional trends were preserved after functionalization, validating source ratio as a reliable pre-functionalization control parameter. The pre-incubation temperature prior to photocrosslinking acts as a direct lever for modulating molecule delivery in SG-PG GelMA hydrogels. Promoting physical crosslinking via low temperature pre-incubation prolonged the characteristic release time, probably by introducing structural barriers like matrix tortuosity. Modeled release profiles confirmed, an anomalous transport governed by network chain relaxation, which reflects a dual network architecture of the hydrogels, covalent crosslinks with physically associated triple-helix domains. Future work should address cell compatibility across the formulation space, explore slower cooling rates to evaluate the promotion of interspecies triple-helix formation, and validate release behavior with biologically relevant small molecules.