Introduction

Porous gelatin-based scaffolds have been frequently reported on during the last decade with the main aim to function as scaffolds for tissue engineering purposes [1–6]. To date, a variety of processing techniques have already been applied to introduce porosity into gelatin matrices. Van Vlierberghe et al. reported on the application of a cryogenic treatment and subsequent lyophilization of methacrylamide-modified gelatin (gel-MOD) which was photo-polymerized prior to the cryogenic treatment [6–8]. More recently, rapid prototyping techniques including the Bioplotter technology and two-photon polymerization (2PP) have also emerged. Indeed, Billiet et al. 3D printed gel-MOD in combination with hepatocytes, followed by fixation of the created structure through UV-induced polymerization in the presence of a photoinitiator [9]. Ovsianikov et al. have applied 2PP to structure gel-MOD according to a CAD–CAM approach in the absence and in the presence of cells [10–12]. In addition, combinatory approaches have recently been realized in an attempt to serve a plethora of tissue engineering needs. For example, Van Rie et al. have recently reported on the development and the evaluation of gelatin cryogel-poly-ε-caprolactone combinatory scaffolds for bone tissue engineering as pure gelatin scaffolds elicit insufficient mechanical properties to serve load-bearing applications [13]. Despite recent progress in the field of rapid manufacturing, cost-effective and more conventional processing techniques including cryogelation clearly remain a research area sufficiently versatile to invest into. The present work explores the potential of a series of crosslinking strategies for porous gelatin scaffolds obtained through the application of a cryogenic treatment. In the literature, three mechanisms for cryogel production have already been distinguished including ‘freezing prior to gelation’, ‘freezing after gelation’, and ‘simultaneous freezing and gelation’ [14]. In the present work, the hydrogel gelation can be subdivided into two categories. During hydrogel crosslinking, both physical entanglements inherent to gelatin as well as covalent bonds are introduced.

Since the water-gelatin system is rather fast-gelling, especially at polymer concentrations above 5 w/v%, it is a complicated task to perform a real cryotropic gel formation (i.e. the gel formation within the so-called unfrozen liquid microphase) [15]. For such fast-gelling mixtures, one deals with the kinetic competition between the freezing and the ‘self-gelling’ processes during cooling [16]. Since the physical gelation of a 10 w/v% gelatin solution cannot be avoided when applying a cryogenic treatment, performing ‘freezing before physical gelation’ is impossible. Indeed, the present work will deal with the chemical crosslinking of gelatin after freezing and/or lyophilization. Previous work on gelatin-based cryogels included the chemical crosslinking step (i.e. UV irradiation) to be performed prior to the cryogenic treatment [6–8, 17]. As a consequence, freezing occurred after chemical crosslinking. Plieva et al. demonstrated that polyacrylamide gels prepared by freezing after gelation possessed different mechanical properties compared to cryogels developed by freezing before gelation [14]. The present work aims to study whether ‘freezing before chemical crosslinking’ would result into gelatin scaffolds possessing different pore sizes and architectures compared to the gels produced by ‘freezing after crosslinking’ [6–8, 17]. Indeed, the present work will report on the effect of e-beam, the application of redox initiators and conventional carbodiimide crosslinking using 1-ethyl-3-(3-dimethylaminopropyl) carbodiimide hydrochloride (EDC) on the crosslink efficiency of porous 3D gelatin scaffolds.

Experimental section

Materials

Gelatin (type B), isolated from bovine skin by an alkaline process, was kindly supplied by Rousselot, Ghent, Belgium. Gelatin samples with an approximate iso-electric point of 5 and the Bloom strength of 257 were used. Methacrylic anhydride (MAA), iron (II) d-gluconate dehydrate (98 %) and N-(3-dimethylaminopropyl)-N′-ethyl-carbodiimide hydrochloride were purchased from Aldrich (Bornem, Belgium) and were used as received. Dialysis membranes Spectra/Por® 4 (MWCO 12,000–14,000 Da) were obtained from Polylab (Antwerp, Belgium). Ammonium cerium (IV) nitrate (99 %) was purchased from Avocado Research Chemicals Ltd. (Karlsruhe, Germany). Vitamin C (i.e. ascorbic acid) and N-hydroxysuccinimide (98 %) were, respectively, obtained from Fluka BioChemica (St. Gallen, Switzerland) and Acros (Geel, Belgium).

Methods

Crosslinking using E-Beam

Gelatin was chemically modified with methacrylamide side groups, as described in a previous paper [18]. A gel-MOD derivative with a degree of substitution of 60 % was selected for this work. The degree of substitution is defined as the percentage of the free amino groups being modified. 1 g gel-MOD was dissolved in 10 ml distilled water at 40 °C. The solution was then injected into the mould of the cryo-unit, after which the solution was allowed to gel for 1 h at room temperature. Finally, the samples were treated cryogenically (cooling range on top side: 21 °C until −30 °C, cooling rate: 0.15 °C/min). In addition, a temperature gradient (T top–T bottom = 30 °C) between top and bottom of the scaffold was applied during the freezing step. After incubating the samples for 1 h at the final freezing temperature, part of the frozen hydrogels were transferred to a freeze-dryer (Christ freeze-dryer alpha I-5) to remove the ice crystals resulting in a porous scaffold.

Both frozen as well as freeze-dried samples were irradiated in the absence of oxygen. The samples were irradiated with electrons from a 15 MeV linear electron accelerator. This accelerator delivered electron beams with a well-defined energy in the range of 3–15 MeV and with an average power up to 5 kW. For electron irradiation, an 80 µA 10 MeV beam traversed a water-cooled vacuum window and 80 cm of air, so that the lateral dose distribution was flattened by scattering. The electrons that interacted with the samples had an energy of 8 MeV (that is, 10 MeV minus the energy-loss in the vacuum window). The average duration for a 25 kGy irradiation was 400 s. Before every irradiation, a dosimetric calibration was performed.

Crosslinking using redox initiators

First, the potential of cerium ammonium nitrate (CAN) was evaluated to initiate the crosslinking of 10 w/v% gel-MOD hydrogels prepared as described higher. Freeze-dried scaffolds (1 g) were immersed in 20 ml double-distilled water in the presence of 10 mol% (13 mg) CAN for varying incubation times (1–24 h). Alternatively, 10 mol% CAN (13 mg) was added to 10 ml double-distilled water, containing 1 g gel-MOD at 40 °C. Finally, a cryogenic treatment was applied, followed by lyophilization and 1H-NMR spectroscopy analysis. 1H-NMR-spectra of the gelatin derivatives were recorded at 50 °C in deuterated water with a Bruker WH 500 MHz instrument. The chemical shift was expressed in ppm as a function of tetramethylsilane as internal standard.

As a second redox-initiating system, a combination of ammonium persulphate (APS) and tetramethylethylenediamine (TEMED) was evaluated. More specifically, 1.4 g gel-MOD was dissolved in 14 ml double-distilled water at 40 °C in the presence of 10 mol% APS (8 mg) and 26 µl TEMED. Next, a series of samples was treated cryogenically as described higher, followed by overnight incubation at −30 °C and lyophilization. A second series of samples was incubated overnight at 5 °C without applying a previous cryogenic treatment. Finally, the gel fractions were determined after incubation of the freeze-dried samples in double-distilled water at 37 °C.

A final redox system which was assessed was Fenton’s reagent. Cryo-treated and freeze-dried gel-MOD scaffolds (1 g) were incubated for 1–24 h in 20 ml double-distilled water in the presence of 10 mol% iron (II) d-gluconate dihydrate (12 mg) and 7.2 µl H2O2. Additionally, 44 mg ascorbic acid was added in a second series of experiments. The gel fractions were determined after incubation of the freeze-dried samples in double-distilled water at 37 °C. Alternatively, 10 mol% iron (II) d-gluconate dihydrate (12 mg) and 7.2 µl H2O2 (with and without 44 mg ascorbic acid) were added to 10 ml double-distilled water, containing 1 g gel-MOD at 40 °C. Finally, a cryogenic treatment was applied, followed by lyophilization and 1H-NMR spectroscopy analysis.

Crosslinking using carbodiimide

Cryogenically treated and freeze-dried gelatin scaffolds were incubated overnight in 9:1 acetone/H2O mixtures (25 ml) in the presence of 45 mg EDC with or without 5.4 mg NHS. Next, the scaffolds were washed two times in double-distilled water, followed by freeze-drying and characterization using scanning electron microscopy (SEM) and micro-computed tomography (µ-CT).

Scaffold physico-chemical characterization

Freeze-dried gelatin samples (1 × 1 × 0.5 cm3) were weighed and then immersed in 80 ml double-distilled water at 37 °C, in the presence of NaN3 to prevent bacterial growth. At regular time points, the hydrogels were removed from the solution, dipped gently with paper to remove excess water and weighed again. The hydration properties, expressed as the  % water uptake, were calculated using the following equation:

$$ \% {\text{ Water uptake }} = \, \left[ {\left( {W_{\text{t}} {-} \, W_{0} } \right)/W_{0} } \right] \, \times { 1}00\;\%, $$
(1)

where W 0 is the initial weight of the dry scaffold and W t is the weight of the hydrated/swollen hydrogel at time point t.

All data points are the mean ± standard deviation of three separate measurements.

After water uptake analysis, the hydrogels were freeze-dried for the determination of the gel fraction (i.e. the polymer fraction that remains after incubation), using the following equation:

$$ {\text{Gel fraction }}\left( \% \right) \, = \, \left( {W_{{{\text{d}}t}} / \, W_{0} } \right) \, \times { 1}00\;\%, $$
(2)

where W dt is the dry mass of the scaffold after degradation at time point t and W 0 is the initial weight of the dry scaffold.

Scaffold visualization

Scaffold visualization occurred by examining the morphology of gold-sputtered samples using scanning electron micrographs obtained on a Fei Quanta 200F (field emission gun) SEM. For µ-CT, a “Skyscan 1072” X-ray micro-tomograph was used. This compact desktop system, consisting of an X-ray shadow microscopic system and a computer with tomographic reconstruction software, generated high-resolution images for small samples (7-mm diameter). During the measurement, both the X-ray source and the detector were fixed, while the sample rotates around a stable vertical axis. Samples were scanned at a voltage of 130 kV and a current of 76 µA. Random movement and multiple-frame averaging were used to minimize the Poisson noise in the images. The spot size of the Hamamatsu micro-focus tube limited the spatial resolution of the reconstructed slices to 10 µm in the X, Y and Z directions. During acquisition, X-ray radiographs were recorded at different angles during step-wise rotation between 0° and 180° around the vertical axis.

After reconstruction of the 2D cross-sections, 3D software µCTanalySIS was used in order to segment and analyse the images. Octopus, a server/client tomography reconstruction package for parallel and cone beam geometry, was also used for analysis.

Statistical analysis

All results are expressed as a mean value ± the standard deviation for each group of samples.

Possible significant differences among groups were studied with the Student’s t test by a two-population comparison. Statistical significance was considered at a probability of P < 0.05.

Results and discussion

Crosslinking using E-Beam

Hydrogels produced by high-energy electron irradiation possess various advantages [19]. For example, sterilization and crosslinking occur in one step. In addition, physical properties can be easily controlled by varying the applied dose, while the electron penetration profile can be fine-tuned (i.e. the crosslink density depends on the electron penetration depth).

E-Beam after lyophilization

First, the influence of the applied irradiation dose on the mechanical properties of freeze-dried, uncrosslinked gel-MOD hydrogels was evaluated by means of compression tests. The compression moduli were obtained from the slope of the initial, linear part of the stress–strain plots. In Fig. 1, the compression moduli are plotted as a function of the irradiation dose. The results showed that the compression moduli increased linearly with increasing irradiation dose applied. The e-beam treatments were performed in the absence of oxygen, since the latter functions as a crosslinking inhibitor.

Fig. 1
figure 1

Compression moduli of gelatin scaffolds as a function of the irradiation dose applied

Full size image

After e-beam irradiation, the hydrogels were incubated at 37 °C in double-distilled water. Both the equilibrium swelling degree and the gel fraction of the incubated gelatin samples were determined and plotted as a function of the applied dose (Fig. 2).

Fig. 2
figure 2

Equilibrium swelling degree and gel fraction of incubated hydrogels (37 °C) as a function of the applied dose

Full size image

Figure 2 shows that applying an irradiation dose of 25 kGy to the freeze-dried gelatin samples resulted in scaffolds possessing a gel fraction of around 50 %. The latter implies that 50 % of the polymer chains dissolved after incubation in double-distilled water at 37 °C. The applied dose (i.e. 25 kGy) was not increased further, based on the current international standards, EN 552 and ISO 11137, recommending an irradiation dose of 25 kGy as a reference dose for terminal sterilization [20, 21].

E-Beam prior to lyophilization

Petrov et al. already described the UV irradiation of moderately frozen cellulose derivatives [22]. In that study, the mechanical properties of samples, UV-treated at room temperature or irradiated at −30 °C were compared. The cryogels exhibited a gel fraction of around 50 %, while the higher crosslink efficiency after UV irradiation at −30 °C was ascribed to the polymer accumulation in the non-frozen liquid microphase (NFLMP). Due to the non-frozen solvent, the NFLMP remained liquid and the polymers retained sufficient segmental mobility to enable crosslinking [22].

Since hydrogels possessing a gel fraction of 50 % were obtained after applying e-beam (25 kGy) on freeze-dried samples (see "E-Beam after lyophilization" section), the effect of e-beam on frozen samples was also evaluated at −5 and at −30 °C.

Interestingly, applying high-energy irradiation at −30 °C did not result in a gel fraction increase. Probably, segmental mobility of the pending methacrylamide moieties was too much reduced at −30 °C thereby hindering sufficient crosslinking to take place. As a result, the cryoconcentration effect typically still occurring at −30 °C did not result in an increase of the gel fraction for the cryogel versus its counterpart developed at room temperature. It should be noted that Petrov et al. applied a polymer concentration of 3 wt% in combination with a photoinitiator concentration of 5 wt% [22]. They stated that crosslinking occurred by hydrogen atom abstraction from the polymer chains and subsequent intermolecular recombination of two macroradicals. It can be anticipated that herein, the methacrylamide moieties present in the gelatin side chains were not numerous enough or insufficiently mobile to enable crosslinking in the absence of an initiator, in contrast with the work performed by Petrov et al. [22]. Moreover, in addition to crosslinking, degradation of polymer chains can also occur during e-beam treatment as already described earlier [23, 24], while Petrov et al. applied UV irradiation to create radicals thereby initiating crosslinking [22].

Conversely and surprisingly, when the crosslinking reaction was performed at −5 °C, high gel fractions ranging from 85 to 99 % were accomplished, even in the presence of oxygen. It can be anticipated that the cooling kinetics played a pivotal role in this observation. Indeed, upon cooling at −5 °C, a system can be obtained consisting of polycrystals of the frozen solvent and a non-frozen liquid microphase (NFLMP), while incubation at −30 °C results in faster cooling with a less pronounced phase separation (cfr. cryoconcentration effect) resulting in lower gel fractions or even the lack of cryogel formation depending on the chemical functionalities applied for crosslinking together with their number.

It should be noted that in contrast with the work from Petrov et al. focussing on the use of UV irradiation [22], herein attention is paid to the application of e-beam given the lack of necessity for material transparency and limited sample thickness. To the best of our knowledge, this is the first paper describing the use of irradiation of electrons, generally corresponding with an energy of 5–12 MeV, to develop crosslinked cryogels.

The scaffolds crosslinked at −5 °C were visualized after lyophilization using µ-CT (see Fig. 3).

Fig. 3
figure 3

µ-CT image of an e-beam-treated scaffold at −5 °C after freeze-drying. The scale bar represents 1000 µm

Full size image

Figure 3 shows that the pore size was in the same range compared to scaffolds prepared by ‘freezing after gelation’ (i.e. 330 ± 48 µm), as described earlier by Van Vlierberghe et al. [6].

Crosslinking using redox-initiating system

A very effective method to generate free radicals under mild conditions is by one-electron transfer reactions, such as redox initiation [25]. A large variety of redox systems was already applied before to crosslink polymer systems [26, 27]. In the present work, three redox systems (including cerium ammonium nitrate (CAN), a combination of ammonium persulphate (APS) and tetramethylethylenediamine (TEMED) and Fenton’s reagent) will be applied and compared in order to crosslink gel-MOD scaffolds after implementation of a cryogenic treatment.

Application of cerium ammonium nitrate

Ceric salt was already applied earlier to initiate the polymerization of acrylamide [28]. The mechanism was explained by a radical process based on Ce4+-coordinated acrylamide. As a result, CAN was evaluated as an initiator for the polymerization of gel-MOD. Uncrosslinked gel-MOD hydrogels were treated cryogenically, followed by lyophilization. In a subsequent step, the freeze-dried scaffolds were swollen in an aqueous solution containing various ceric salt concentrations (5–10 mol%). Since the resulting scaffolds dissolved, a 1H-NMR-spectrum of gel-MOD was recorded after addition of 10 mol% CAN to study the effect of CAN on the double bonds present. From the spectrum (data not shown), it could be concluded that only 5 % of the double bonds had reacted. The latter can be explained since native, unfunctionalized gelatin already possesses a large variety of functional groups which are inherently part of the amino acids constituting the gelatin backbone (e.g. hydroxyl moieties) which are sensitive to CAN, thereby hampering/limiting its reactivity towards the methacrylamide moieties introduced into gel-MOD in order to develop chemically crosslinked networks [29].

Application of ammonium persulphate and TEMED

Ammonium persulphate (APS) is typically used for polyacrylamide gel electrophoresis (PAGE). A mixture of acrylamide and bisacrylamide can form a chemically crosslinked polymer network in the presence of APS. N,N,N′,N′-tetramethylenediamine (TEMED) functions as an accelerator for the polymerization reaction.

As already described above, chemical crosslinking was intended to occur after the cryogenic treatment. Consequently, APS and TEMED were added in various concentrations (5–40 mol%) and ratios to two series of samples, of which the first was incubated at 5 °C and the second cryogenically treated, followed by incubation at −30 °C. The concentrations were selected based on the literature data describing the synthesis of polyacrylamide gels at subzero temperatures [14, 30, 31]. After the lyophilization, the samples were incubated in double-distilled water at 37 °C to determine the crosslink efficiency. In Table 1, chemically crosslinked (cryo)gels are indicated by ‘×’, while the uncrosslinked samples are marked by ‘–‘.

Table 1 Formation of chemically crosslinked gelatin (cryo)gels in the presence of APS
Full size table

The results show that 10 mol% initiator is sufficient to enable chemical crosslinking at −30 °C. Moreover, incubating these samples at 5 °C did not result into the formation of chemically crosslinked hydrogels. Incubating the samples at 5 °C did not thus result in the so-called cryoconcentration effect to increase the apparent polymer concentration in the NFLMP thereby triggering crosslinking below the actual critical polymer concentration. The results are therefore in agreement with among other observations described by Petrov et al. [22].

The results also show that the optimal cryogelation conditions can differ depending on several parameters including whether or not an initiator is used, the initiator concentration, etc. Indeed, in the absence of an initiator and upon applying e-beam for crosslinking purposes, an incubation temperature of −30 °C did not result in crosslinking, as already described above.

As a consequence, a concentration of 10 mol% was selected to enable ‘freezing before gelation (i.e. chemical crosslinking)’ and/or ‘simultaneous freezing and gelation’. The obtained gel fraction (i.e. 76 %) was similar to the value obtained before, when applying a cryogenic treatment to chemically crosslinked gel-MOD hydrogels, as reported earlier by Van Vlierberghe et al. [7]. It should be noted that a disadvantage of this approach was that the obtained scaffolds sometimes collapsed. Moreover, the reproducibility was rather low compared to the conventional matrices.

Application of Fenton’s reagent

The metal catalysed Haber–Weiss reaction, also called Fenton reaction, has already been applied earlier for the free radical polymerization of hydrogel systems [32]. This redox system involves the combination of H2O2 and ferrous salt to generate radicals [33]. The mechanism involves a one-electron transfer from the ferrous ion to the peroxide with the dissociation of the oxygen–oxygen bond and the generation of one hydroxyl radical and one hydroxyl ion: [25]

$$ {\text{Fe}}^{{2 + }} + {\text{H}}_{2} {\text{O}}_{2} \to \left[ {{\text{HO}}^{ \cdot } - {\text{OH}}} \right]{\text{Fe}}^{{3 + }} \to {\text{Fe}}^{{3 + }} + {\text{OH}}^{ - } + {\text{OH}}^{ \cdot } . $$

In a subsequent step, the hydroxyl radicals can initiate the polymerization of methacrylamide-substituted gelatin. Freeze-dried gelatin scaffolds were incubated during different time periods (1–24 h) in aqueous solutions containing various amounts of initiator (2–10 mol%) in the absence or in the presence of ascorbic acid functioning as catalyst. Since all hydrogels studied dissolved when incubating them at 37 °C, it could be concluded that the scaffolds were not crosslinked chemically by the applied redox system. In order to investigate the effect of the initiator system on the double bonds incorporated, 1H-NMR-spectra were recorded before and after the addition of Fenton’s reagent to gel-MOD (data not shown). The spectra showed that only 5 % of the double bonds present had reacted after addition of the redox system (data not shown). Probably, the latter can be explained, since gelatin possesses many other functionalities being sensitive for the redox system applied [34].

Application of carbodiimide crosslinking

Chemical crosslinking agents can be classified into non-zero-length and zero-length crosslinkers [35]. Non-zero length agents introduce poly- and bifunctional crosslinks into the network structure by bridging free amine groups of lysine and hydroxylysine, or free carboxylic acid residues of glutamic and aspartic acid of the protein molecules.

Zero-length crosslinkers introduce crosslinks without incorporation of foreign substances into the polymer, for example, by activation of the carboxylic acid groups with N-(3-dimethylaminopropyl)-N′-ethyl-carbodiimide (EDC) enabling the reaction with free amines [36]. N-Hydroxysuccinimide (NHS) is often added to the reaction mixture to improve the crosslink efficiency by reducing side reactions [37–39]. Crosslinking of gelatin with EDC and NHS takes place according to the mechanism in Fig. 4.

Fig. 4
figure 4

Reaction scheme for the crosslinking reaction of gelatin with EDC and NHS

Full size image

The carbodiimide activates carboxylic acid residues of aspartic and glutamic acid, with the formation of O-acylisourea groups. Next, amine groups of lysine and hydroxylysine residues can nucleophilically attack the activated carboxylic acids, with a urea derivative as a leaving group. In order to diminish possible side reactions, such as hydrolysis and rearrangement to a stable N-acylurea, NHS can be added. All residues are water soluble and can be easily washed out after the crosslinking procedure. It was already described earlier that the incubation of uncrosslinked gelatin-based scaffolds in aqueous solutions often resulted in matrix collapse [40, 41]. As a consequence, a mixture of acetone and water (9:1) was selected as reaction medium. Native, unmodified gelatin scaffolds were incubated overnight in the presence of various EDC/NHS concentrations. Visual observation showed that an EDC concentration of 1.8 mg/ml was minimally required to avoid dissolution of the scaffold in water upon incubation at 40 °C after crosslinking, irrespective of the presence or absence of NHS. After washing in double-distilled water, the crosslinked hydrogels were freeze-dried and characterized using SEM and µ-CT (Fig. 5).

Fig. 5
figure 5

Images of the EDC-crosslinked gelatin scaffold obtained with SEM (a, scale bar 100 µm) and µ-CT (b, c, d, scale bar 1000 µm). b Represents 2D cross-section, while c and d depict 3D reconstructions of the scaffold developed from a different angle

Full size image

The SEM image showed that the pore walls showed higher roughness compared to conventional hydrogels. Small pores seemed to be present on the surface of larger pores (Fig. 5, upper left). An average pore diameter of 123 µm (±10 µm) was calculated, based on 10 pores randomly selected throughout the scaffold.

Unfortunately, µ-CT analysis showed the occurrence of various inhomogeneities within the 3D scaffolds (Fig. 5, bottom). The average pore diameter was 244 µm (±75 µm), based on 10 pores randomly selected throughout the entire scaffold. The pore size, obtained using µ-CT, was higher compared to the pore diameter calculated from SEM images. The latter can be attributed to the presence of large fractures visualized using µ-CT. The relatively high deviation (i.e. 75 µm) of the pore diameter, obtained by µ-CT, supports this explanation.

The gel fraction of the scaffolds developed was 70 %, thus somewhat lower compared to conventional hydrogels.

In order to enable proper comparison between the various crosslinking strategies applied, the obtained results are outlined in Table 2.

Table 2 Overview of the various crosslinking strategies applied and their resulting gel fractions
Full size table

Conclusions

In the present work, alternative approaches have been elaborated to enable chemical crosslinking of the frozen and/or freeze-dried scaffolds developed.

E-beam did not enable chemical crosslinking of lyophilized gel-MOD scaffolds nor of frozen cryogels at −30 °C. The lack of sufficient segmental mobility hindered the reaction of the methacrylamide moieties present on the gelatin backbone. However, when performing the high-energy irradiation at −5 °C, enough segmental mobility within the non-frozen liquid microphase was retained enabling the chemical crosslinking after performing the cryogenic treatment. Alternatively, the potential of various redox-initiating systems and their concomitant crosslinking efficiency was evaluated. Cerium ammonium nitrate and Fenton’s reagent were unable to chemically crosslink gel-MOD, while the combination of ammonium persulphate and TEMED resulted in the formation of scaffolds with a reasonable gel fraction (75 %). However, poor reproducibility and matrix collapse indicated the need for alternative crosslinking procedures. Interestingly, carbodiimide-coupling chemistry was found suitable to efficiently crosslink freeze-dried gelatin matrices.