Tuning Rheology and Printability of Gelatin Methacryloyl/Sodium Alginate Bioinks: The Role of Sodium Alginate Molar Mass and Sterilization

Abstract

Gelatin methacryloyl (GelMA)/sodium alginate (SA) bioinks are widely explored for various biomedical applications due to their favorable biocompatibility and tunable properties. However, the molar mass (MM) of SA, a critical factor influencing bioink properties, has not yet been systematically investigated in these systems. In this work, we highlight the importance of SA MM as a key parameter that governs the rheological behavior of GelMA/SA bioinks, directly impacting their printability and the structural and mechanical characteristics of the resulting hydrogels. Three types of SA used are characterized by very low (SA-VL), low (SA-L), and medium (SA-M) weight-average MM ranging between 38 and 260 kg·mol^–1. In addition, the variability in the MM of SA is included as a result of SA sterilization by autoclaving. MM exponentially decreases with sterilization time for all SA samples showing a 37% drop for SA-VL and about a 65% drop for SA-M and SA-L after 20 min of sterilization time. MM of SA significantly affects the viscoelasticity of GelMA/SA bioinks, which enables us to recognize the conditions suitable for bioprinting tested by employing various printing conditions. SA-VL did not provide the required quality of printed strands due to high elasticity. The compression analysis of hydrogels suggests that formation of the GelMA covalent network is independent of the MM of SA, as shown from stress relaxation measurements, and the differences in hydrogel mechanical properties can solely be attributed to the SA ionic network. Interestingly, the higher degree of ionic cross-linking is observed in the case of lower MM of SA; however, these differences gradually decrease as the stress relaxation equilibrium is reached. The permeability of hydrogels in saline shows no significant variation across different MMs of SA. This study thus recommends considering the MM of SA as the principal factor allowing for optimization of GelMA/SA bioink printability without compromising the key properties of the resulting hydrogel.

1. Introduction

Polymers serve as fundamental building blocks for biomaterials used for tissue engineering, regenerative medicine, and the encapsulation of active compounds. The polymer characteristics therefore significantly affect the properties of the biomaterials, and, together with a suitable fabrication technique, they determine the biomaterial usability for selected application. Beyond their scientific importance, polymeric biomaterials represent a rapidly expanding sector of the global economy, with the market valued at approximately USD 59.7–70.2 billion in 2024–2025 and projected to reach USD 132.6–182.9 billion by 2029–2030, corresponding to a compound annual growth rate of 14–17%. This remarkable growth reflects an increasing demand for advanced solutions in biomedical applications, highlighting both the technological and economic significance of continued research in polymeric biomaterials.

The 3D extrusion bioprinting is one of the fabrication techniques frequently employed in numerous biomedical engineering fields such as an organ-on-chip preparation or the tissue engineering targeting various tissues, e.g., bone, cartilage, skin, cardiac, or liver. , This technique is based on direct deposition of bioinks in the form of a continuous filament. , The bioink needs to fulfill a number of stringent criteria. The biological criteria include material cytocompatibility, biocompatibility with target tissue after implantation, and in vivo biodegradability. During bioprinting, the bioink is extruded through the printing nozzle onto the substrate. Hence, the rheological properties of a bioink should allow for its ability to deform and flow. This is closely related with the quality of bioprinting − as well as the viability of the bioink-loaded cells due to the stress applied in the printing nozzle. − Together with the stress-induced cell damage, extrusion bioprinting still faces various other challenges, such as lower printing resolution, nozzle clogging, or limited material availability. The development of a bioink thus represents a crucial step in determining the bioprinting process and successful in vivo performance of a bioprinted material.

Nowadays, a number of different polymers of natural and synthetic origin are being used as components of bioinks toward biomaterial formulations. The combination of gelatin methacryloyl (GelMA) and sodium alginate (SA) is viewed as a well-established bioink composition − that has demonstrated its suitability in various applications such as cartilage, − neural tissue engineering, skeletal muscle tissue engineering, and islet encapsulation. GelMA, as a derivative of gelatin, complies with the material requirements with respect to cytocompatibility, biocompatibility, and biodegradability and also provides the sites for the cell attachment. Additionally, thanks to the incorporation of methacryloyl moieties, GelMA is capable of covalent cross-linking controlling the stability of hydrogels at physiological conditions. However, the use of GelMA bioinks alone is limited due to the insufficient resolution in printability aimed at stable and uniform structures. − On the other hand, SA-based bioinks exhibit favorable rheological properties for 3D bioprinting; however, SA by itself is not cell-adhesive. Consequently, the presence of both these components in mixed GelMA/SA bioinks provides the sites for the cell attachment, the capability of both UV and ionic cross-linking, and the rheological properties enabling to achieve high resolution during printing. ,,

Multiple studies previously highlighted the correlation between polymer properties and the bioink rheological behavior, printability, hydrogel properties, and viability of loaded cells. ,,− However, the majority of the studies on GelMA/SA bioinks were primarily focused on performance of a selected system in target application, rather than on the characterization of a bioink by itself., As a result, the standardization of the processing steps across laboratories is limited. The effect of polymer concentrations and of the GelMA:SA ratio on the bioink rheological properties was discussed in several studies. Aldana et al. characterized samples with different GelMA/SA ratios. From the selected set of samples, composition containing 8 wt % GelMA and 5 wt % SA exhibited the highest Young’s modulus of a resulting hydrogel. Wu et al. focused on the printability of different GelMA/SA samples with conclusion that 5 wt % SA concentration provided suitable printability of samples with varying GelMA concentration up to 10 wt %. Similarly, Zhang et al. focused on samples with different SA concentrations ranging between 1 and 4 wt %, where the increase in SA concentration resulted in improved printability. These findings correspond with the work of Costantini et al. who achieved a high printing resolution of GelMA-based bioinks consisting of 4 wt % SA. Other studies reported the use of higher (8%) and lower (2%) SA concentrations for preparation of printable GelMA/SA bioinks. In this work, the bioink composition of GelMA/SA systems with 8 wt % GelMA and 5 wt % SA was selected to correlate our results with previously published data. ,,,

While these studies demonstrate the usability of GelMA/SA bioinks for bioprinting, we identified critical gaps in the current understanding of this widely used bioink system. Surprisingly, several important parameters remain overlooked despite their potential impact on printability and the functional performance of the resulting hydrogels. First, the effect of SA molar mass (MM), which is a fundamental polymer property governing viscosity, gelation, and printability, has not been systematically evaluated in relation to GelMA/SA bioinks. Second, the cross-linking conditions for both components, UV-induced covalent cross-linking of GelMA and Ca²⁺-mediated ionic cross-linking of SA, vary between different laboratories, often without sufficient justification. Third, although sterilization is an essential step for any material intended for in vivo use, protocols for sterilizing GelMA/SA bioinks remain largely missing, and the possible impact of sterilization on bioink properties is unexplored. Only a few publications on GelMA/SA bioinks briefly mention the sterilization process; , however, these studies do not deal with a potential effect of sterilization on the bioink characteristics.

In this work, we systematically address these overlooked parameters. We focus on three critical aspects: (1) the effect of the MM of SA on bioink properties and the resulting hydrogel characteristics, (2) the impact of sterilization on the MM of SA and the resulting hydrogel characteristics, and (3) the optimization of cross-linking conditions. To evaluate the influence of the MM of SA, three different initial MMs of SA (very low, low, and medium) are used, and their degradation under autoclaving conditions is monitored. For the first time, we demonstrate that sterilization by autoclaving leads to a significant decrease in the MM of SA, with the extent of degradation depending on the initial MM. These variations are directly correlated with changes in the rheological properties and printability of the bioinks. Furthermore, we studied the impact of various cross-linking conditions on the GelMA/SA hydrogel characteristics. Hydrogels prepared under selected conditions were characterized with respect to their mechanical properties and permeability. Human adipose tissue-derived stem cells immobilized in selected hydrogels confirm that the cell viability with respect to selected preparation conditions is preserved. This work thus contributes to fundamental information on GelMA/SA bioinks, suggesting that reasonable balancing of the MM of SA, considering both its initial value and changes during sterilization, may be applied to tune the bioink printability without affecting the properties of resulting hydrogels.

2. Experimental Section

2.1. Materials

The pharma-grade sodium alginate (SA) Protanal LFR 5/60 very low viscosity (VL) with a guluronic acid (G) content of 65–75 mol % and weight-average molar mass, $M_{\mathrm{w}}\circledR$ , of 38 ± 2 kg·mol^–1 (Đ = 1.3 ± 0.2), Manucol LKX low viscosity (L) with a G content of 30–40 mol % and $M_{\mathrm{w}}\circledR$ of 168 ± 10 kg·mol^–1 (Đ = 2.4 ± 0.5), and Protanal CR 8223 medium viscosity (M) with a G content of 30–40 mol % and $M_{\mathrm{w}}\circledR$ of 260 ± 7 kg·mol^–1 (Đ = 2.0 ± 0.1) were obtained from DuPont. The reported G content was provided by the manufacturer. $M_{\mathrm{w}}\circledR$ and Đ values were determined by size-exclusion chromatography (SEC) using the multiangle laser-light scattering and refractive index detectors, with the molar mass distributions (MMDs) shown in Figure S1 of the Supporting Information. Gelatin type A with a gel strength of 300, methacrylic anhydride (≥94%), sodium carbonate (≥99.5%), sodium bicarbonate (≥99.7%), lithium phenyl-2,4,6-trimethylbenzoylphosphinate (LAP, ≥ 95%), acetone, deuterium oxide (D₂O, 99 atom % D), phosphate buffered saline (PBS) powder, 4-(2-hydroxyethyl)­piperazine-1-ethanesulfonic acid (HEPES) powder, low-glucose Dulbecco′s modified Eagle′s medium (DMEM, with 1000 mg·L^–1 glucose, l-glutamine and sodium bicarbonate), fetal bovine serum, fluorescein diacetate, and propidium iodide (≥94.0%) were obtained from Sigma-Aldrich. Sodium hydroxide (Lach:ner), calcium chloride anhydrous (BioReagent, ≥ 96%), sodium chloride (Lachema), and sodium azide (Penta) were used for preparation of solutions. Ethylene glycol was purchased from AFT Bratislava s.r.o. The narrow-distributed pullulan standards with MM ranging from 180 to 805,000 g·mol^–1 were purchased from Polymer Standards Service, dextran standards with MM ranging from 3,500 to 70,000 g·mol^–1 were purchased from Biotika a.s., and the dextran standard of MM equal to 3 800,000 g·mol^–1 was purchased from American Polymer Standards Corporation. Potassium bromide (99+ %, IR grade) was obtained from Acros Organics. Fluid thioglycolate medium (FTM) and tryptic soy broth (TSB), both sterile and ready-to-use media, were obtained from Millipore. Penicillin-streptomycin (PS, 10,000 U/mL) was obtained from Thermo Fisher Scientific. The trypsine-EDTA solution (10×, sterile-filtered, suitable for cell culture, containing 5.0 g of porcine trypsin and 2 g of EDTA, 4Na per liter of 0.9% NaCl) was purchased from BioReagent. Collagenase type I was purchased from Gibco, USA.

2.2. Synthesis and Characterization of GelMA

The gelatin was methacrylated using a previously reported protocol with minor modifications. The schematic representation of GelMA synthesis is shown in Figure S2. Five grams of gelatin powder were dissolved and constantly stirred in 50 mL of carbonate-bicarbonate buffer (0.1 mol·L^–1) at 50 °C and pH around 7. Then, 0.5 mL of methacrylic anhydride was added to gelatin solution in six equal volumes in 30 min intervals while maintaining the pH 7 using 4 mol·L^–1 NaOH. The reaction mixture was additionally stirred at 50 °C for 30 min. The resulting solution was dialyzed against deionized water using a dialysis tubing cellulose membrane with a molar mass cutoff (MMCO) of 14 kDa (Sigma-Aldrich) for 3 days, with frequent water changes to remove unreacted methacrylic anhydride. GelMA was isolated from solution after dialysis by lyophilization as a white powder at a yield of 90%.

The proton nuclear magnetic resonance (¹H NMR) spectroscopy was used to determine the degree of methacrylation (DM) of synthesized GelMA, related to the content of modified lysine moieties of gelatin. The NMR spectra were recorded on a Bruker Avance III HD-400 spectrometer at 50 °C. The spectra were normalized to the phenylalanine signal (7.37–7.44 ppm), which corresponds to the concentration of gelatin. DM values were calculated from eq :

$$\mathrm{DM}(\%)=\left(1-\frac{\mathrm{the}\mathrm{area}\mathrm{of}\mathrm{lysine}\mathrm{methylene}\mathrm{of}\mathrm{GelMA}}{\mathrm{the}\mathrm{area}\mathrm{of}\mathrm{lysine}\mathrm{methylene}\mathrm{of}\mathrm{gelatin}}\right)\times 100\%$$ 1

The resulting DM was ∼80% of the lysine moieties. ¹H NMR spectra of GelMA samples (Figure S3) from three different batches revealed no batch-to-batch variation in the spectra and, hence, excellent repeatability in obtained DM values (SD equal to ± 5).

2.3. Sterilization of Sodium Alginate

SA powders were sterilized using an Elara 11-D autoclave (Tuttnauer). Different SA amounts (0.2, 0.5, and 1.0 g) placed in different containers (5 mL glass vial, 20 × 40 mm, 70271, Macherey-Nagel GmbH & Co. KG, and 25 mL glass beaker, 34 × 48 mm, 08-229.202.03, Chemland) were subjected to three different autoclaving protocols with the same sterilization temperature of 121 °C and different sterilization times of 5, 10, and 20 min. SA samples used for preparation of sterile bioinks were sterilized in an amount of 0.2 g in a glass vial. Figure S4 depicts conditions during autoclaving with respect to the temperature, pressure, and time.

2.4. Characterization of Sodium Alginate

The ¹H NMR spectra were recorded on a Bruker Avance III HD-400 spectrometer in D₂O at 65 °C. The Fourier-transform infrared spectroscopy (FTIR) was performed using KBr discs containing 1 wt % of analyzed sample by an FTS 3000 MX Excalibur Series infrared spectrophotometer (Digilab). Transmission and absorption spectra were collected in the range of wavenumbers from 600 to 4000 cm^–1, using 50 scans for each sample.

The CHNS elemental analysis (EA) was performed to determine the content of carbon, nitrogen, sulfur, and hydrogen elements using a Flash 2000 CHN Elemental Analyzer (Thermo Fisher Scientific).

The thermogravimetric analysis (TGA) was conducted to determine the residual water content using a TGA/DSC 1 instrument (Mettler Toledo). TGA analysis was carried out in a temperature range from 25 to 250 °C with the heating rate of 10 °C·min^–1 in an inert nitrogen atmosphere at a gas flow rate of 50 mL·min^–1. The water content in SA samples was determined as the weight loss corresponding to 105 °C.

The size-exclusion chromatography (SEC) was used to characterize MM of SA samples as previously reported by an aqueous-phase SEC using 0.1 mol·L^–1 Na₂HPO₄ and 200 ppm of NaN₃ as an eluent and a set of Suprema columns (Polymer Standards Service). Ethylene glycol was used as an internal flow-rate marker. The combination of multiangle laser-light scattering (MALS) detector PSS SLD 7000 (Polymer Standard Service) and refractive index detector Waters 2414 DRI (Waters Corporation) was employed to obtain the absolute MM values. The refractive index increment, dn/dc, values were used for data analysis as follows: dn/dc = 0.136 mL·g^–1 for pullulan of a MM of 110,000 g·mol^–1 used for calibration of detectors and dn/dc = 0.145 mL·g^–1 for SA. The pullulan calibration was only used to estimate the dispersity values, Đ, for the SA-VL sample (nonautoclaved and after autoclaving) in order to obtain more reliable Đ values than from MALS detector, which tends to underestimate Đ values due to overestimated number-average molar mass in the case of the low MM samples. WinGPC Unichrome software (Polymer Standards Service) was used for data acquisition and evaluation.

The protocol recommended by European Pharmacopoeia was employed for direct sterility testing of SA samples. Around 100 mg of SA powder was added under sterile conditions into the cultivation flask followed by an addition of 10 mL of sterile cultivation media. Flasks were closed using the vent cups with a 0.22 μm membrane and placed in the thermostat, where they were incubated at 37 °C for 14 days. The positive control (nonautoclaved SA) and negative control (blank sterile medium) were prepared and incubated in parallel to autoclaved SA samples. Each sample was cultivated in two types of media: (1) FTM was used primarily for cultivating of anaerobic bacteria and (2) TSB was used primarily for cultivating of aerobic bacteria and fungi.

2.5. Preparation of Nonsterile Bioink

The bioink without cells, i.e., mixed GelMA/SA solutions, was prepared in three consecutive steps. In the first step, a LAP photoinitiator, in a range between 0.05 to 0.5 wt %, was dissolved while stirring for 10 min (37 °C, 500 rpm) in different media, saline (0.9 wt % NaCl), PBS, HEPES, and DMEM, used as solvents for preparation of bioinks. In the second step, GelMA was added to LAP solution and stirred for 2 h (37 °C, 200 rpm). In the third step, an SA powder was added to the GelMA solution and gently stirred with a magnetic stirrer overnight (37 °C). The GelMA/SA composition containing 8 wt % GelMA and 5 wt % SA was selected as a preferred composition following the literature data. ,,, All of the SA types were used in order to demonstrate the influence of the MM of SA on bioink characteristics and performance. We introduce a sample abbreviation as G8A5x, where x stands for MM of SA (VL, L, and M).

2.6. Preparation of Sterile Bioink

For the preparation of sterile bioink, the same three consecutive steps used for preparing nonsterile bioink were followed. GelMA solution containing LAP was prepared in PBS and sterilized by being filtered through 0.2 μm sterile filters. Subsequently, an autoclaved SA powder was dissolved in a sterile GelMA solution.

2.7. Rheological Characterization

Rheological properties were measured using a HAAKE MARS iQ Rotational Rheometer (Thermo Fisher Scientific). A cone and plate geometry (diameter 20 mm, angle 2°, titanium) was employed for all the samples except for the 5 wt % SA-VL sample that exhibited a very low viscosity and thus required the use of a double gap concentric cylinder geometry.

The sample was conditioned for 2 min after being loaded into the rheometer. All measurements were performed at a constant temperature of 25 °C, except for the temperature sweep experiments. A solvent trap was used to prevent sample evaporation. All measurements were repeated in triplicate, and the data are presented as an average ± standard deviation. Data analysis was performed with RheoWin Data Manager software.

The following conditions were used for different modes of analysis: (i) for the flow curves, the viscosity was measured as a function of shear rate from 0.1 to 1000 s^–1. Obtained curves were fitted by two rheological models, namely, power law and Cross models, to describe the shear-thinning behavior. The power law model, expressed by eq , is generally used for description of shear-thinning behavior:

$$\eta =k·{(\mathrm{\gamma ̇})}^{n-1}$$ 2

where η is the viscosity, γ̇ is the shear rate, k is the flow consistency index, and n is the flow behavior index.

The Cross model (eq ) is reported as more appropriate for SA solutions with concentrations over 1 wt %:

$$\eta ={\eta }_{\infty }+\frac{{\eta }_0-{\eta }_{\infty }}{1+{(C\mathrm{\gamma ̇})}^m}$$ 3

where η₀ and η_∞ are viscosities extrapolated to zero and infinite shear values, C is the cross time constant, and m is the cross rate constant. (ii) For the amplitude sweep, the storage modulus (G′), loss modulus (G″), and loss angle (δ) were monitored as a function of shear strain from 0.1 to 100% at a constant oscillatory frequency of 10 rad·s^–1. The linear viscoelastic (LVE) range of the samples was characterized by the limit values of strain (in %) and yield stress (in Pa) as a point where G′ deviates from the plateau values by 5%. Subsequently, the strain value of 1% was chosen from LVE to be used as a constant for the frequency sweep measurements; (iii) for the frequency sweep, G′ and G″ were monitored as a function of oscillatory frequency from 0.1 to 100 rad·s^–1 at a constant strain of 1%; (iv) for the temperature sweep, G′, G″, and complex viscosity η* were monitored as a function of temperature from 20 to 40 °C at a heating rate of 2 °C·min^–1 and at a constant oscillatory frequency of 10 rad·s^–1 and a constant strain of 1%; and (v) for the oscillatory recovery, samples were first subjected to a low strain of 0.2% for 120 s, then to a high strain of 500% for 60 s, and then the strain was again decreased to 0.2% for 120 s using a constant oscillatory frequency of 10 rad·s^–1 during the measurement.

2.8. Printability Testing

The bioink printability was tested using an Inkredible Bioprinter (Cellink) at ambient temperature. Three different types of sterile high-precision conical bioprinting nozzles (Cellink) with varying nozzle diameter (D _N) were used as follows: D _N = 580 μm (20 G), D _N = 410 μm (22 G), and D _N = 200 μm (27 G). The UV shielding cartridges were used to prevent the bioink from cross-linking before the printing. The bioink was irradiated by UV at 365 nm by using the built-in bioprinter UV source. Samples were irradiated immediately after printing the full layer in a distance of approximately 3 cm corresponding to a source intensity of about 40 mW·cm^–2. The printed objects were designed in an XYZ Maker software and imported to a Repetier Host software, in which a printing toolpath was analyzed using a Slic3r (G-Code generator). In this step, the printing parameters such as layer height, nozzle diameter, and printing rate were preset to accurately generate the printing toolpath. First, the single layer squares with a side length of 2 cm were printed using different nozzles, printing pressures, and printing rates. The strand thickness (D) was analyzed using a Leica DM 2700 M optical microscope (Leica Microsystems GmbH), and resulting values are presented as an average of four measurements ± standard deviation. Subsequently, the single layer square structures with increasing infill density (ID, %) were printed. These models were prepared in two sets: (i) D _N was input into the slicing software as the D _N of the nozzle used for printing and (ii) D _N was corrected in the slicing software according to the real strand thickness at selected printing conditions.

2.9. Preparation of Hydrogels

Hydrogels of different diameter and thickness of 1 mm were prepared in tailor-made PLA molds (printed in advance using Prusa i3MK3S+ 3D printer) at different cross-linking conditions: (i) LAP concentration ranging from 0.05 to 0.5 wt %, (ii) UV irradiation time between 0 and 10 min, and (iii) ionic cross-linking time between 0 and 20 min using 2 wt % CaCl₂ dissolved in 0.9 wt % NaCl solution. Bioink samples were irradiated by a built-in UV 365 nm LED source of Inkredible Bioprinter in a distance of 3 cm corresponding to conditions during printing. Subsequently, the UV-cross-linked hydrogels were covered by an excess of the CaCl₂ solution for a defined time.

2.10. Mechanical Analysis of Hydrogels

Mechanical analysis was carried out by means of oscillatory rheometry and unconfined compression. Planar hydrogels, prepared in the form of a cylinder with a height of 1 mm and a diameter of 20 mm, were characterized by frequency sweep measurements. Representative values of G′ and G″ at 10 rad·s^–1 were used for comparison of hydrogels and for selection of suitable cross-linking conditions. The compression tests of hydrogel samples were performed in an unconfined compression mode using a TA.XTplus C texture analyzer (Stable Micro Systems Ltd.) equipped with a software Exponent Connect on hydrogel discs of a height of 1 mm and diameter of 4 mm placed in a Petri dish. A small amount of 0.9% NaCl solution, containing 2 mmol·L^–1 CaCl₂, was added to the edges of hydrogel samples to prevent drying during the analysis. The initial height H ₀ of the sample was automatically evaluated as the point at which the force values reached 1 g. Three different types of measurements were performed as previously reported: − (i) ramp loading, in which hydrogels are deformed up to 98% of their H ₀ using a mobile probe moving vertically either at 0.5 or 0.1 mm·s^–1 and which leads to a value of a dynamic Young’s modulus (E _dyn) calculated by the Exponent Connect software from the range of small deformations (<5%), in which it can be defined as a slope of the linear fit according to an eq :

$$E=\frac{\sigma }{\epsilon }$$ 4

where σ represents the stress and ε represents the strain; (ii) relaxation loading, in which hydrogels are deformed up to 5, 10, and 20% with a constant speed of 0.5 mm·s^–1; after reaching a given deformation, this deformation was held constant and the stress relaxation was monitored for 600 s followed by plotting the equilibrium stress against equilibrium strain to calculate the equilibrium Young’s modulus (E _eq) as a slope of the resulting stress–strain curve; and (iii) dynamic loading, in which hydrogels are cyclically deformed up to 20% of their H ₀ with a constant speed of 0.1 mm·s^–1 using 100 consecutive cycles for each sample.

2.11. Inverse Size-Exclusion Chromatography

The molar mass cutoff (MMCO) and pore-size distribution of hydrogels were determined by an inverse size-exclusion chromatography (iSEC). The working protocol followed the principles developed previously for the characterization of hydrogel microcapsules were adapted for characterization of hydrogel samples. Hydrogels were first pressed through the stainless-steel sieve with a mesh size of about 0.5 mm in order to obtain small pieces that were placed into the glass column 10 × 250 mm (Omnifit) fitted with an adjustable plunger on each side. The packed volume of hydrogel pieces was approximately 5.5 mL. Narrow-distributed pullulan and dextran standards were injected on the column at a concentration of 3 mg·mL^–1 in the eluent containing 0.9 wt % NaCl, 2 mmol·L^–1 CaCl₂, and 200 ppm of NaN₃ via the 100 μL loop. The iSEC analysis was carried out using the Waters SEC setup consisting of a Waters 515 pump, a Rheodyne Injector 7725i, and a Waters 2410 DRI detector. The eluent flow rate was 0.2 mL·min^–1. The WinGPC Unichrome software (Polymer Standards Service) was used for data acquisition and evaluation. Elution volume for each standard was determined as the volume at which 50% of the standard is eluted. Obtained values were subsequently used to determine the chromatographic partition coefficient K _SEC for each standard providing the MMCO values as the intercept of tangents corresponding to the exclusion and separation parts of the calibration curve created as a Boltzmann fit of (1 – K _SEC) vs MM. The pore size distribution was determined from the first derivative of the Boltzmann fit of the calibration curve. Two iSEC analyses were performed on newly prepared hydrogels in order to check for repeatability, involving hydrogel preparation and iSEC characterization.

2.12. Isolation and Cultivation of Adipose Tissue-Derived Stem Cells

Human adipose tissue-derived stem cells (hATSCs) were isolated from lipoaspirates by a standard enzymatic protocol. Briefly, lipoaspirates were washed twice with sterile PBS to remove residual erythrocytes. The adipose tissue fragments were digested with 0.1% collagenase type I at 37 °C for 45 min under constant rotation, followed by centrifugation at a relative centrifugal force of 300 × g for 10 min. The cell pellet was resuspended in a complete growth medium consisting of DMEM low glucose, 10% fetal bovine serum, 100 U·mL^–1 penicillin, and 100 μg·mL^–1 streptomycin to neutralize the collagenase and centrifuged at 300 × g for 10 min. The final pellet was resuspended in a complete culture medium and filtered through a 40 μm mesh (Becton Dickinson, USA). hATSCs were plated at a density of 1 × 10⁶ cells·mL^–1 into 75 cm² culture flasks and cultured at 37 °C in a humidified atmosphere containing 5% CO₂. After 24 h, nonadherent cells were removed by aspiration medium. During subsequent cultivation, the medium was changed every 3 days. After 7–10 days, cells were detached by 0.05% trypsin and subcultured up to the third passage.

2.13. Immobilization and Viability of Adipose Tissue-Derived Stem Cells

The suspension of hATSCs in PBS was prepared at a concentration of 10⁷ cells·mL^–1 and immediately used for preparation of cell-loaded bioinks as described below. The sterile bioink were tempered at 37 °C and mixed with the cell suspension in the volume ratio bioink:cells = 9:1, typically using 2.7 mL of bioink and 0.3 mL of cell suspension according to the Cellink mixing protocol. Bioink precursor and cell suspension were loaded into separate sterile syringes connected by using a female double luer-lock connector. Subsequently, solutions were mixed by a cyclic transferring of the solutions from one syringe to another one (typically 10 cycles) resulting in the cell-loaded bioink with a cell concentration of 1·10⁶ cells·mL^–1.

The cell-loaded hydrogels were prepared as follows. First, the 3D printed PLA molds were sterilized by immersion into ethanol (96%) for 4 h, multiple times washing with a sterile ultrapure water, and drying at ambient temperature in a laminar hood. The bioinks were injected from the syringe into the PLA molds, UV cross-linked for 2 min, and, subsequently, ionically cross-linked with sterile-filtered saline solution containing 2 wt % CaCl₂ for 20 min. As a result, cell-loaded cylindrical hydrogels with a diameter of 4 mm and height of 1 mm were prepared.

Immediately after cross-linking, the hydrogel samples were transferred into the sterile well plates with a full growth media and placed in the incubator (Memmert) at 37 °C in a humidified atmosphere containing 5% CO₂. During the cultivation, media was changed every 2 or 3 days.

In the selected time intervals (days 0, 1, 3, 6, 10, 14, 21, and 28), one hydrogel from each set of tested samples was randomly selected and cell viability was visualized using FDA/PI live/dead staining assay. Samples were visualized using a Zeiss LSM 900 (Zeiss) equipped with A-Plan 10x/0.25 objective and excitation/emission settings 493/513 and 305/617 nm for FDA (green fluorescent signal) and PI (red fluorescent signal), respectively. Cell viability was calculated as a ratio between viable cells and the total number of cells:

$$\mathrm{cell}\mathrm{viability}(\%)=\frac{\mathrm{green}\mathrm{cells}}{\mathrm{green}\mathrm{cells}+\mathrm{red}\mathrm{cells}}\times 100\%$$ 5

For quantification, four randomly selected visual fields were manually analyzed with Zeiss Zen 3.7 software (Zeiss).

2.14. Statistical Analysis

Statistical analysis was performed on raw data using one-way ANOVA and Tukey tests at the significance level p < 0.05 within the OriginPro software 2023b.

3. Results and Discussion

3.1. Characterization of Bioinks

3.1.1. Rheological Properties

Figure presents the rheological characteristics of bioinks prepared of SA of different MM. Both amplitude (Figure a) and frequency sweeps (Figure b) reveal the gel-like behavior of all samples, since G′ > G″. The G′ and G″ values strongly depend on the MM of SA. Sample G8A5VL reached significantly lower G′ and G″ values, compared to G8A5L and G8A5M. The strain and stress values limiting the LVE region are estimated as 13 ± 1% and 69 ± 16 Pa for G8A5VL, 21 ± 5% and 327 ± 27 Pa for G8A5L, and 21 ± 5% and 538 ± 119 Pa for G8A5M. The δ values, characterizing the viscoelasticity of the G8A5 samples within the LVE, are determined as 29.1 ± 0.1° for G8A5M, 27.6 ± 1.9° for G8A5L, and 6.8 ± 0.1° for G8A5VL. These values reflect that decreasing the MM of SA results in a significantly more elastic behavior of G8A5VL compared to G8A5L and G8A5M.

1.

Rheological characterization of G8A5 bioink prepared in 0.9 wt % NaCl using different MM of SA (VL, L, and M): (a) amplitude sweep at 25 °C and oscillatory frequency of 10 rad·s^–1, (b) frequency sweep at 25 °C and strain of 1%, (c) temperature sweep at oscillatory frequency of 10 rad·s^–1 and strain of 1%, (d) flow curves at 25 °C, and (e) oscillatory recovery measurements at 25 °C and oscillatory frequency of 10 rad·s^–1.

Different behavior of G8A5VL was observed also during the temperature sweeps between 20 and 40 °C, shown in Figure c. All of the samples show a decrease in G′ and G″ with increasing temperature. As shown in Figure S5a, G′ values decrease by 32, 37, and 52% for G8A5M, G8A5L, and G8A5VL samples, respectively, in a range from 25 to 35 °C. In going further to 40 °C, G′ decreases for all the samples, when G8A5VL approaches the liquid-like character. Overall, the G′ and G″ values from the oscillatory measurements (Figure a–c) reveal that the behavior specifically of G8A5VL bioink is strongly influenced by behavior of GelMA, capable of cross-linking and creating the gel structures with temperature-induced sol–gel transition. Figure S5b shows that at 37 °C, which is the temperature of sample preparation, the complex viscosity of G8A5L and G8A5M samples is 4–8 times higher than that for G8A5VL. This indicates that increasing the MM of SA significantly increases the viscosity of the bioink precursor during preparation and limits the mobility of GelMA chains. This results in decreased efficiency of thermal GelMA cross-linking and, thus, in prevalent SA-like behavior. The rheological behavior under conditions corresponding to Figure for pure GelMA and SA solutions is depicted in Figure S6 in order to understand the contribution of individual bioink components to overall bioink behavior. Aldana et al. reported similar differences for GelMA/SA samples with varying polymer ratios, where sample G6A5 showed more significant decrease of viscosity with increasing temperature, in a range between 20 and 40 °C, compared to more concentrated samples with respect to either alginate (G6A7) or both alginate and GelMA (G8A7) contents.

Also, Wu et al. reported a decrease in G′ and G″ values with increased temperature, in a range between 10 and 37 °C, for GelMA/SA samples containing 5 wt % of SA and either 5 or 10 wt % of GelMA. Although a direct comparison between our and published data cannot be made due to an unknown MM of SA used in works, , the trends observed in our work are similar to those reported previously. In addition to the previously published information on rheological properties of GelMA/SA bioink composition, our results demonstrate how these properties can be further altered by adjusting the MM of SA.

The flow curves shown in Figure d reveal the shear-thinning behavior of all of the samples, which is one of the essential preconditions for successful extrusion through the printing nozzle. Individual curves were fitted with the power law model (eq ) and Cross model (eq ), with characteristic constants summarized in Table S1. Data presented in Figure d and Table S1 show that samples G8A5M and G8A5L exhibit comparable shear-thinning behavior, with differences in k and η₀ attributed to the differences in MM values of SA. The G8A5VL bioink containing SA of very low MM reached a similar η₀ value as G8A5M. However, the shear-thinning behavior was significantly more pronounced, as the shear-thinning of this sample started at a significantly lower shear rate estimated from 1/C values in Table S1. This behavior can also be attributed to pronounced GelMA-like behavior for this system.

Figure e depicts the oscillatory recovery measurements for G8A5 bioinks containing SA of different MM. These data reveal that all the bioinks are capable of fast recovery at a low strain of 0.2% after application of a high strain of 500%. This behavior correlates with the capability of a bioink for recovery after extrusion. G8A5VL was able to recover completely to its original values as a result of its elastic character. As the elasticity of these bioinks decreases with higher MM of SA, the application of a high strain induces an incomplete recovery of G′ and G″ values. G8A5L recovered to around 80% for G′ and G″, and G8A5M recovered to around 60% for G′ and G″. Similar behavior was also reported by Wu et al. for G5A5 and G10A5 samples.

The correlation between rheological parameters, as the intrinsic property of a material, and printability results were previously reported. − Viscoelastic bioinks exhibiting dominant either viscous or elastic behavior are not suitable for bioprinting applications, as achieving a suitable bioink flow may be extremely challenging. Gao et al. studied different mixtures of SA and unmodified gelatin and correlated their rheological behavior, specifically δ, with printability. The optimum results with respect to printing resolution and structural integrity were reported for samples with tan δ values between 0.25 and 0.45, corresponding to δ interval from 14 to 24°. Then, based on these recommendations, we can assume that sample G8A5VL with δ = 6.8 ± 0.1° is too elastic to provide satisfactory printing results. Values of δ characterizing samples G8A5L and G8A5M, 27.6 ± 1.9 and 29.1 ± 0.1°, respectively, are slightly higher than the recommended values. However, compared to an unmodified gelatin, the viscous behavior of GelMA-based systems can be compensated by fast cross-linking during the bioink deposition. The validity of these recommendations is further directly tested in printability tests.

The results on rheology presented in Figure are obtained in 0.9 wt % NaCl solutions. Ultimately, these bioinks are intended to be tested in vitro in the presence of cells and in vivo in contact with bodily fluids and tissue. Therefore, the sensitivity of a rheological behavior of these GelMA/SA bioinks to various solvents, up to culture media, should be recognized as a part of steps toward sterile systems. Since both GelMA and SA contain charged moieties capable of electrostatic interactions, we assumed that using solvents containing different salts and/or other molecules may affect the bioink properties. Various research groups used different buffers for preparation of GelMA/SA bioinks, such as PBS, HEPES, and culture media, such as DMEM. However, a study comparing the properties of GelMA/SA in different solvents is missing. Bociaga et al. addressed this issue focusing on SA/gelatin bioinks prepared in deionized water and DMEM. The bioink prepared in DMEM exhibited lower viscosity compared with that prepared at the same polymer concentrations in deionized water, which was interpreted as a result of suppressed interaction in DMEM compared to the salt-free aqueous environment.

In this study, G8A5L was chosen as a representative sample to compare the rheological properties of GelMA/SA system prepared in 0.9 wt % NaCl, PBS, HEPES, and DMEM. Data in Figure S7 show similar behavior for G8A5L samples prepared in 0.9 wt % NaCl, PBS, and HEPES. On the other hand, sample prepared in DMEM exhibited lower viscosity at a shear rate of 0.1–10 s^–1, while at higher shear rates, viscosity is comparable to situation in NaCl, PBS, and HEPES (Figure S7a). Also, G′ and G″ values are clearly lower in DMEM than in saline and buffer solutions and, due to G″ > G′, viscous rather than elastic behavior is observed in DMEM compared to solutions prepared in NaCl, PBS, and HEPES (Figure S7b). We assume that salts, such as CaCl₂, or amino acids present in DMEM hinder the interactions between GelMA and SA polymer chains to a significantly higher extent than in the case of NaCl, PBS, and HEPES. In saline solution and buffers, GelMA and SA electrostatic interactions result in higher elasticity and higher viscosity at lower shear rates, while at higher shear rates, these interactions are disrupted and viscosity of G8A5L is similar in all solvents. In this work, PBS was used as a solvent for preparation of sterile bioinks.

3.1.2. Printability

During the initial screening of bioink printability, G8A5M, G8A5L, and G8A5VL bioinks were successfully extruded through three selected conical nozzles with D _N = 580 μm (20 G), D _N = 410 μm (22 G), and D _N = 200 μm (22 G). The individual strands were visualized by using an optical microscope. The uniformity factor, U, was determined as proposed by Soltan et al. The U value provides an option to characterize the printability and quality of the printing results. It is defined as a ratio of the length of a printed strand to the length of a theoretical straight strand. Then, U = 1 represents the uniform strand, while U > 1 represents the nonuniform strand. Figure S8 depicts the shapes of printed strands for all three bioinks by using a 22 G nozzle. G8A5M and G8A5L bioinks provide smooth and uniformly printed strands that are characterized by U ∼ 1. On the other hand, strand thickness of G8A5VL is significantly higher than for G8A5M and G8A5L bioinks, and the uniformity is characterized by U ∼ 1.1. Macroscopically, the strands made of G8A5VL bioink were not straight, and they rather resemble a pearl-necklace-like structure. This behavior originates from high elasticity of the G8A5VL bioink, shown in Figure , resulting in fast elastic reformation due to immediate removal of applied shear stress after bioink exits the printing nozzle. , A lower quality in printability due to increased bioink elasticity has previously been reported for bioinks based on the unmodified gelatin combined with either SA , or oxidized SA. Data in Figure S8 made us to exclude the G8A5VL bioink from the follow-up studies shown further below. It is important to point out that the results on printability are always connected to the bioink composition as well as the printing conditions. A successful use of SA with low MM, similar to SA-VL used in this study, was previously reported by Costantini et al. for coaxial printing of GelMA/SA-based bioinks with CaCl₂.

G8A5M and G8A5L bioinks were printed using all the nozzles in combination with different printing pressures and printing rates in order to determine the effect of printing conditions on strand thickness, D. As shown in Figure , the increased printing pressure at constant printing rate results in an increased D due to the higher flow rate of a bioink precursor. This behavior is more pronounced at lower printing rates as the slow movement of the nozzle causes accumulation of the bioink behind the printhead. On the other hand, increasing the printing rate at constant printing pressure results in a lower D. This is valid for all of the bioink precursors. In case the printing rate exceeds the bioink flow rate, D values decrease below D _N (Figure a). Due to the viscoelastic character of bioinks, D is mostly higher than D _N, as a result of elastic reformation. This effect is pronounced for G8A5L due to higher elasticity revealed by rheological measurements and upon lowering the nozzle diameter. The spreading ratios, D/D _N, based on results in Figure , are contained in Tables S2 and S3 . Figure further shows that it is possible to obtain similar D at the same printing rate for both G8A5L and G8A5M, however, at significantly lower printing pressures for G8A5L. This behavior correlates with yield stress values for these bioink precursors (see discussion to Figure ) and reflects that a higher pressure is needed to induce the flow of the G8A5M bioink precursor.

2.

Strand thickness (D) of (a, c, e) G8A5L and (b, d, f) G8A5M bioink precursors as a function of printing pressure and printing rate, using the conical nozzles with different D _N = 580 μm (20 G), D _N = 410 μm (22 G), and D _N = 200 μm (27 G), represented by the reference line in each plot.

A practical consequence of the results shown in Figure is setting the conditions for printing the square structures with increased infill density either without or with the D _N correction using a G8A5L bioink precursor. Figure shows that employment of the D _N correction clearly affords better printing fidelity, close to the theoretical model, and an infill density of 90%. In the absence of D _N correction, the structure collapses already at the infill density of 60%. These findings correspond to the work of Aldana et al.

3.

Square structures (side length 2 cm) with different infill density (%); structures were printed using G8A5L bioink precursor, 27G nozzle, printing speed of 7 mm·s^–1, and printing pressure of 60 kPa (a) without correction and (b) with correction of the D _N value used for toolpath planning according to real D measured at selected conditions.

Additionally, the printing time of structures with different infill densities varies from a few seconds for infill density of 10% to tens of seconds up to minutes for high infill densities, which also determines the time between the sample deposition and cross-linking. In these time ranges, we did not observe any significant spreading of the bioink on the substrate that would affect the quality of the bioprinted structures.

Here, a comment should be added that printing conditions, such as printing pressure, printing rate, nozzle diameter, and printing time, affect the viability of loaded cells as demonstrated, for example, by Ning et al. showing the effect of printing pressure and by Zhao et al. testing the effect of the printing time. Hence, specific to our study, the variation in MM of SA is considered as a viable approach to lower the stress in the printing nozzle to reduce the risk of cell damage without the need to change the composition of GelMA/SA bioink, the printing resolution characterized by D _N, and the printing time determined by the printing rate.

3.1.3. Cross-Linking

Figure a,b depicts the resulting G′ of UV (UV 365 nm LED source of Inkredible Bioprinter) cross-linked G8A5L and G8A5M hydrogels of 1 mm thickness and 20 mm diameter containing different concentrations of the LAP photoinitiator. No relevant increase in G′ is observed after 2 min of UV irradiation. These data correlate with findings of Nguyen et al., who reported the efficient cross-linking of 10% GelMA solution using 2 min of UV irradiation for similar UV irradiation conditions and LAP concentrations. Additionally, these data show that LAP concentration of 0.1 wt % is sufficient for GelMA cross-linking in this GelMA concentration range, since further increase in LAP concentration does not increase the G′ of hydrogels. It is noteworthy that G′ for both hydrogels is by about an order of magnitude higher than that for respective GelMA/SA solutions. The similar behavior is seen for G″ values in Figure c,d, also significantly increasing after the UV exposure for 2 min due to formation of covalent bonds between the GelMA chains. Based on these results, a LAP concentration of 0.1 wt % and the 2 min irradiation time were selected for preparation of UV cross-linked hydrogels.

4.

G′ and G″ values determined for hydrogels prepared at different cross-linking conditions determined at 1% strain and 10 rad·s^–1 frequency by oscillatory rheological measurements: (a) G′ of G8A5L hydrogels as a function of LAP concentration and UV irradiation time, (b) G′ of G8A5M hydrogels as a function of LAP concentration and UV irradiation time, (c) G″ of G8A5L hydrogels as a function of LAP concentration and UV irradiation time, (d) G″of G8A5M hydrogels as a function of LAP concentration and UV irradiation time, (e) G′ and G″ of G8A5M hydrogels prepared by 2 min UV irradiation using 0.1 wt % LAP followed by Ca²⁺ cross-linking of different duration, and (f) G′ and G″ of G8A5M hydrogels as a function of Ca²⁺ cross-linking time (without previous UV cross-linking).

As a second cross-linking step, the UV-cross-linked hydrogels were exposed to 2 wt % CaCl₂ solution (180 mmol·L^–1 CaCl₂). Figure e depicts the evolution of G′ and G″ values as a function of exposure time of hydrogels made of the G8A5M bioink precursor to Ca²⁺ gelling ions (an identical behavior was seen for the G8A5L sample). Both these values sharply increase in the first 5 min of cross-linking followed by a gradual increase up to 20 min. Figure f shows a reference experiment, in which G8A5M hydrogels were stabilized only by Ca²⁺ cross-linking. These hydrogels exhibit the increase, almost by a 1 order of magnitude, only in G′ during the first minute of cross-linking, and no additional increase is observed upon extending the cross-linking time up to 5 min. The G″ values are only slightly increased, by a factor of 1.4, during this time interval, which suggests that the mechanical energy is dissipated due to GelMA chains, not being influenced by gelling Ca²⁺ ions, present in the SA hydrogel. Data in Figure thus suggest that each individual network, ionic as well as covalent, contributes to the increase in the G′ value by about an order of magnitude. Then, a duration of 20 min was selected for ionic cross-linking as a second step for preparation of hydrogels. Other groups previously reported the use of shorter cross-linking times such as 2 min and 10 min using 100 mmol·L^–1 CaCl₂ or 5 min and 15 min using 50 mmol·L^–1 CaCl₂. However, the optimization of the cross-linking conditions was not discussed in these studies.

3.2. Sterile GelMA/SA Bioinks

Sterility is a mandatory requirement for every material that is intended to be in direct contact with cells or living tissues. However, the sterilization of bioinks is not a trivial task due to their high viscosity that disables a commonly applied sterile filtration for sterilization of polymeric solution. Although various strategies are in place for sterilization of hydrogels, we realized that the sterilization strategy and, specifically, its influence on GelMA/SA bioink properties have been insufficiently covered in the literature. A few studies ,, discuss the effect of SA sterilization by autoclaving of SA solutions, generally resulting in SA degradation and decreased viscosity of sterilized solutions. Below, we describe the efficiency of SA sterilization using different durations of sterilization cycles and the impact of sterilization on MM and chemical composition of SA and on properties of resulting bioinks. We opted for sterilization of SA in a powder form because (i) in our approach, GelMA/SA bioinks are prepared by dissolving SA in a sterile GelMA solution and (ii) data in Lorson et al. indicate that sterilization of SA by autoclaving lowers MM less significantly in powder form than in solution.

3.2.1. SA Sterilization by Autoclaving

Figure shows the results on sterilization of all SA samples, including SA-VL despite not considering this SA suitable for bioink formulations, with weight average molar mass, $\overline{M_{\mathrm{w}}}$ , values and dispersity contained in Table . Figure a shows that $\overline{M_{\mathrm{w}}}$ values exponentially decrease as a function of sterilization time for all samples as expressed by eqs –:

$$\mathrm{SA}‐\mathrm{VL}:y=25+13{\mathrm{e}}^{-0.44x}$$ 6

$$\mathrm{SA}‐\mathrm{L}:y=56+112{\mathrm{e}}^{-0.34x}$$ 7

$$\mathrm{SA}‐\mathrm{M}:y=88+172{\mathrm{e}}^{-0.34x}$$ 8

5.

SEC measurements of autoclaved SA: (a) $\overline{M_{\mathrm{w}}}$ as a function of sterilization time for SA of different initial $\overline{M_{\mathrm{w}}}$ , where the exponential fits are represented by eqs –, (b) normalized decrease of $\overline{M_{\mathrm{w}}}$ of SA as a function of sterilization time represented by the exponential fits, and (c) $\overline{M_{\mathrm{w}}}$ as a function of sample volume for samples M10 and L10 sterilized in different amounts and containers and samples M0 and L0 used as a reference. Statistical significance was calculated separately for SA-M (blue) and SA-L (green) sets using one-way ANOVA and Tukey test (n = 4, * p < 0.05).

1. Summary of $\overline{M_{\mathrm{w}}}$ and Đ (average ± SD) of SA after Autoclaving Treatment at Different Sterilization Time .

	SA-M		SA-L		SA-VL
autoclaving cycle	$\overline{M_{\mathrm{w}}}$ (kg·mol–1)	Đ	$\overline{M_{\mathrm{w}}}$ (kg·mol–1)	Đ	$\overline{M_{\mathrm{w}}}$ (kg·mol–1)	Đ
0	260 ± 7	2.0 ± 0.1	168 ± 10	2.4 ± 0.5	38 ± 2	2.2 ± 0.1
5	119 ± 4	1.8 ± 0.1	76 ± 5	2.4 ± 0.4	26 ± 2	2.0 ± 0.1
10	90 ± 16	2.1 ± 0.3	59 ± 4	2.2 ± 0.1	27 ± 3	2.0 ± 0.1
20	88 ± 6	2.0 ± 0.3	60 ± 18	1.9 ± 0.3	24 ± 2	2.0 ± 0.1

^aEvaluated toward pullulan calibration

^bCycle 0 represents the untreated sample as a control.

The extent of $\overline{M_{\mathrm{w}}}$ decrease for individual SA samples is revealed by plotting of $\overline{M_{\mathrm{w},t}}$ determined at a given sterilization time, normalized to the initial $\overline{M_{\mathrm{w},0}}$ for nonautoclaved SA (Figure b). Interestingly, this normalized decrease is identical for SA-M and SA-L samples. $\overline{M_{\mathrm{w}}}$ values for these samples are reduced to about 45% of $\overline{M_{\mathrm{w},0}}$ within the first 5 min. A prolongation of sterilization time to 10 min results in additional decrease to about 35% of $\overline{M_{\mathrm{w},0}}$ , with only minor decrease observed between 10 and 20 min of sterilization, demonstrated by a visible shift in MMDs to lower $\overline{M_{\mathrm{w}}}$ shown in Figure S9. Thus, after 5 and 20 min of sterilization, the $\overline{M_{\mathrm{w}}}$ for SA-L decreases from $\overline{M_{\mathrm{w},0}}$ = 168 to 76 and 60 kg·mol^–1, respectively, and for SA-M from $\overline{M_{\mathrm{w},0}}$ = 260 to 119 and 88 kg·mol^–1, respectively. The dispersity values of around 2 are not affected by sterilization. Also interestingly, a decrease in $\overline{M_{\mathrm{w}}}$ for SA-VL upon autoclaving is significantly lower compared to SA samples of higher $\overline{M_{\mathrm{w},0}}$ (Figure a,b and Figure S9). The initial $\overline{M_{\mathrm{w},0}}$ = 38 kg·mol^–1 reaches a value of 24 kg·mol^–1 after 20 min of sterilization.

To interpret the data presented in Figure , we need to understand the possible mechanisms responsible for the observed decrease in MM for SA samples during the autoclaving process. Autoclaving involves hydrothermal conditions that break down the polymer chains through a reaction with water at elevated temperatures and pressures. Hence, hydrolysis is considered as the principal mechanism responsible for depolymerization of SA. Reaching the limiting MM for all the alginate samples shown in Figure a is an interesting phenomenon that has been described for SA by employing other depolymerization techniques, including thermal, ultrasound, and γ-irradiation. Dodero et al. suggested the pseudo-first and second order models to evaluate the depolymerization of SA by ultrasound, when the first model considers the nonspecific random scission and leads to the formation of polymer chains of high dispersity, while the latter one considers the midchain scission of polymer chains with dispersity similar to the original sample. Since in our case the dispersity of autoclaved samples does not change compared to the starting materials (Table ), and assuming that the ultrasound degradation mechanism can be applied to the hydrothermal degradation, the midchain scission may be proposed to be the principal mechanism responsible for observed decrease in MM of SA in our study. The reason for reaching the limiting MM can be interpreted in view of the applied autoclaving conditions that are not too severe to result in complete SA degradation as reported, for example, by Aida et al. for the hydrothermal degradation of SA.

Additionally, we investigated the potential changes in the chemical structure induced by autoclaving by ¹H NMR, FTIR, and elemental analysis. The ¹H NMR spectra of SA samples in Figure a,b show that a proton signal located at 8.8 ppm appears for autoclaved samples and increases with the time of sterilization. Previous studies on the depolymerization of SA under hydrothermal , and thermal treatments suggest that acid-catalyzed hydrolysis and β-elimination are the mechanisms that could also be responsible for the scission of SA chains. These mechanisms are illustrated in Figure c. While in acid-catalyzed hydrolysis, the chemical structure of chain-end functional groups is not changed, and β-elimination results in different chemical structures of chain-end functional groups. This proton signal is proposed to be associated with the hexenuronic acid chain-end unit formed through a β-elimination mechanism, resulting in the formation of a double bond between C-4 and C-5. The content of the functional moiety related to this signal is at the level of 1–2 mol %. The FTIR spectra (Figure S10) demonstrate the typical SA structure with characteristic peaks at 3440, 1610, and 1089 cm^–1 attributed to −OH, −COO^–, and C–O, respectively. No significant chemical changes induced by autoclaving were detected by FTIR, likely due to the similarity of the functional groups present after SA depolymerization and the low content of hexenuronic acid chain-end units formed via the β-elimination mechanism.

6.

(a) Schematic representation of the proposed mechanism and resulting product upon cleavage of glycosidic linkage in SA by acid-catalyzed hydrolysis and β-elimination mechanism. (b) ¹H NMR spectra (in D₂O at 65 °C) of SA with medium MM (SA-M0) and the corresponding SA autoclaved for 5 (SA-M5), 10 (SA-M10), and 20 (SA-M20) min. (c) ¹H NMR spectra (in D₂O at 65 °C) of SA with low MM (SA-L0) and the corresponding sodium alginate autoclaved for 5 (SA-L5), 15 (SA-L15), and 20 (SA-L20) min.

The initial white color of SA powders changed to a brownish color during autoclaving, as shown in Figure S11. Regarding this change of the SA color after autoclaving, it should be noted that autoclaving of ultrapure SA in our lab results in a light-yellowish color. This suggests that impurities (polyphenols, endotoxins, and proteins) in the pharma-grade SA used in this study are responsible for appearance of brownish color after autoclaving. To monitor any alterations in SA chemical composition with respect to elements, the results from CHNS elemental analysis contained in Table S4 did not reveal any detectable changes in SA samples following autoclaving.

We also explored the effect of sample amount and SA layer thickness on degradation of SA during autoclaving in order to identify the possible influence of experimental conditions on degradation of SA. Three different amounts of SA of 0.2, 0.5, and 1.0 g placed to a 5 mL sample vial and 1.0 g placed to a 25 mL beaker (Figure S11) were subjected to autoclaving using the sterilization time of 10 min. The increased SA amount in a vial provides a higher layer thickness, from approximately 1.5 mm at 0.2 g to 3 mm at 0.5 g and 5 mm at 1.0 g, whereas SA placed to a beaker experiences a higher surface area than SA placed to the vial and the thickness ∼2.5 mm similar to that for 0.5 g SA placed to the vial. Quite interestingly, these at glance minor changes in sample amount and selection of a container affect the MM of SA samples after sterilization. Figure c shows that the SA layer thickness, affecting the penetration of water vapor to the sample, is a factor that needs to be considered since the $\overline{M_{\mathrm{w}}}$ for a layer of 5 mm is 1.5-fold higher than that for a layer of 1.5 mm, valid for both SA types. Regardless of the container used, the 3 mm thick SA layer provides similar $\overline{M_{\mathrm{w}}}$ values that are between those obtained for lower and higher SA thickness.

During autoclaving, the samples are placed in the steam environment. Therefore, it is relevant to determine the water content of each sample after sterilization. The TGA measurements and resulting water content are shown in Figure S12 . The water content varies between 5 and 25 wt % and in most cases increases after autoclaving. However, there is neither direct trend correlating the water content with the sterilization time nor repeatable water content when using the same autoclaving protocol, as seen from two repeats, sets A and B, shown in Figure S12c. The repeatability of the water content measurement is confirmed for nonautoclaved samples showing the same water contents in both sets. Then, the determination of the water content for each autoclaved sample is recommended in order to prepare SA solutions of a correct concentration.

Finally, the efficiency of autoclaving was evaluated by testing the sterility according to European Pharmacopoeia. Figure S13 depicts the samples after cultivation in FTM and TSB for 14 days at 37 °C. Tested samples are compared to the positive control (PC), represented by a nonautoclaved SA, and the negative control (NC), represented by a blank sterile media. In the case of PC in both media, the turbidity and foam at the surface are observed as a sign of the microbial growth already after 24 h of incubation and become more evident over time. On the other hand, none of autoclaved SA samples exhibits any indication of the microbial growth after 14 days of incubation in both media. All these samples remain clear and transparent identically to the NC. Hence, already 5 min of sterilization appears sufficient for sterilization of SA in the powder form. Similarly, Lafuente-Merchan et al. reported the efficient sterilization of nanocellulose-SA-hyaluronic acid bioinks using short autoclaving cycles with sterilization time of 3 min and overall duration of autoclaving for about 1 h, including heating and drying.

Considering the importance of high MM of SA from the points of view of bioink printability and the time needed for sterilization, the SA-M samples sterilized for 5 (M5) and 20 (M20) min were selected for preparation and characterization of sterile bioinks, reported further below, testing G8A5M5, G8A5M20, and G8A6M20 compositions. The latter sample containing 6 wt % of M20 is selected in order to examine whether the increase of the content of SA with lower MM can be used to suppress high elasticity of the bioink. This is expected due to significantly lower MM of M20 compared to M0, i.e., the nonautoclaved SA-M sample.

3.2.2. Rheological Properties

Figure S14 shows the amplitude sweeps for sterile bioink, prepared in PBS, resulting in δ values of 14 ± 2° for G8A5M5, 14 ± 2° for G8A5M20, and 56 ± 2° for G8A6M20. The δ value for a nonsterile control G8A5M0 is 27 ± 1°. Obtained δ values therefore indicate that the decrease in MM from 260 ± 7 kDa (M0) to 119 ± 4 kDa (M5) and 88 ± 6 kDa (M20) shifts the viscoelastic properties toward a more elastic behavior for samples with SA concentration of 5 wt %. The δ values for G8A5M5 and G8A5M20 bioink precursors are identical within the LVE and reach the lower limit of the ideal printability region defined by Gao et al. Therefore, satisfactory printability may be expected for these systems. Additionally, Figure S14 shows that the increase in M20 concentration to 6 wt % for sample G8A6M20 leads to a decrease in elasticity as given by a higher δ value. This behavior can be ascribed to the increased amount of SA polymer chains that prevent the thermal cross-linking of GelMA. At this point it should be emphasized that relatively minor increase in SA content from 5 to 6 wt % abruptly changes the bioink precursor rheological behavior. This further demonstrates the need to understand the relationship between the polymer, both MM and content, and the printability. In regards of the G8A6M20 sample, since the upper limit for ideal printability toward viscous behavior is exceeded, the satisfactory printability may be achieved only by UV or ionic cross-linking during deposition.

3.3. Mechanical Analysis of Hydrogels

Hydrogels were made of sterile bioinks G8A5M5, G8A5M20, and G8A6M20, and a nonsterile control G8A5M0. Cylindrical samples with a diameter of 4 mm and thickness of 1 mm were prepared by UV cross-linking for 2 min followed by 20 min cross-linking in CaCl₂ solution. Figure shows dynamic (E _dyn) and equilibrium (E _eq) Young’s modulus values determined from compression measurements.

7.

Young’s modulus (E) of sterile hydrogels G8A5M5, G8A5M20, and G8A6M20 and the nonsterile control sample G8A5M0, all prepared in PBS: E _dyn determined from the ramp loading measurements with two different loading rates of 0.5 and 0.1 mm·s^–1, and E _eq determined from the stress relaxation measurements at strains of 5, 10, and 20%. Statistical significance was calculated using one-way ANOVA and Tukey test (n = 4, * p < 0.05). Group marked as “ns” was not significantly different.

Figure depicts E _dyn and E _eq values for hydrogels depending on the MM of SA and the loading rate. For sample G8A5M0, there is no difference between the values for both E _dyn and E _eq. On the other hand, lowering of MM of SA leads to the increase in E _dyn at a higher loading rate of 0.5 mm·s^–1, whereas E _dyn at a loading rate of 0.1 mm·s^–1 remains similar to E _eq and independent of MM of SA. The similarity between E _dyn values at a lower loading rate of 0.1 mm·s^–1 and E _eq values can be explained by fast stress relaxation of hydrogel networks. As shown in Figure S15, for all of the samples, the stress relaxes after the fast initial decrease (Figure S15a) to a finite level of σ_eq between 2 and 3 kPa (Figure S15b). Stress relaxation was very fast for all samples, with relaxation modulus decreasing to 50% of its original value within the 1–2.4 s (determined for 10% strain). This behavior can be interpreted in view of viscoelastic contributions of individual networks that respond to applied stress differently. The initial decrease can be attributed to rearrangement of the SA ionic network, whereas reaching the σ_eq values corresponds to the covalent GelMA network that is of similar character for all hydrogels. This behavior is in agreement with studies of Zhao et al, who propose that in ionic gels, the stress relaxes mainly through breaking and reforming of the ionic cross-links with tendency to relax to zero stress values, while permanent covalent networks relax to finite equilibrium. Then, the increased E _dyn values obtained at a higher loading rate of 0.5 mm·s^–1 for hydrogels containing SA of lower MM, M20, can be understood as a result of more effective ionic cross-linking compared to SA of higher MM, M5. The higher degree of ionic cross-linking for shorter SA chains can be attributed to enhanced mobility of SA polymer chains during the formation of SA-Ca²⁺ network within the UV-cross-linked GelMA/SA hydrogels. Affected by similar mechanism as stress relaxation, the hydrogel fatigue related to the bond type and network topology supports the previous statement of more efficient cross-linking of SA of lower MM within the GelMA/SA hydrogels. Figure S16 shows a decrease of maximum stress, σ_max, during the cyclic deformation, which is more pronounced (up to 50–60%) for samples containing M20. This likely reflects a higher relative content of ionic bonds in hydrogels containing SA of lower MM. This result appears to be specific to the GelMA/SA systems, since the opposite trend is observed in case of ionically cross-linked SA hydrogels with E _dyn values decreasing upon lowering MM below 100 kg·mol^–1 determined at a low loading rate of 1 mm·min^–1.

3.4. Permeability of Hydrogels

The results on mechanical analysis of hydrogels under equilibrium conditions, i.e., E _dyn at low loading rate and E _eq in Figure , suggest that the macroscopic hydrogel characteristics are not influenced by the MM of SA, at least in the studied range. Another possibility to demonstrate this feature is to analyze these GelMA/SA hydrogels from the point of view of their permeability by iSEC. This method provides information on the molar mass cutoff (MMCO) and pore-size distribution , of hydrogels in their native liquid environment (saline, buffer, media) and a dependence of these characteristics on hydrogel preparation conditions. − , The iSEC is based on injecting a series of polysaccharide (pullulans, dextrans) standards on the chromatographic column with a hydrogel forming the column packing. Figure S17 shows elution curves for G8A5M0, G8A5M5, G8A5M20, and G8A6M20 hydrogels, which are further processed to determine the calibration curves, with the exclusion limit providing the MMCO values, and the pore-size distributions, shown in Figure . The data and cumulative distributions in Figure a are overlapped suggesting that the permeation of polysaccharide standards is similar for all these hydrogels, with the mutual MMCO value close to 70 kg·mol^–1. This value corresponds to a viscosity diameter of around 14 nm. The full width at half-maximum values in differential distributions (Figure b) are between around 1 and 100 kg·mol^–1, which provide the information about the pore-size distribution between 2 and 16 nm for these hydrogels. While these quantities can be associated with the diffusion characteristics of hydrogels, the principal message from iSEC is that MM of SA and the concentration of M20 in tested hydrogels does not play a significant role.

8.

iSEC measurements for sterile hydrogels G8A5M5, G8A5M20, and G8A6M20 and the nonsterile control sample G8A5M0, all prepared in PBS: (a) cumulative pore-size distribution and (b) differential pore-size distribution. The eluent contains 0.9 wt % NaCl and 2 mmol·L^–1 CaCl₂.

3.5. hATSC-Loaded Hydrogels

The experiments with hATSC-loaded hydrogels were designed with two goals: (i) to evaluate the effect of cross-linking conditions on cell viability, and (ii) to observe the cell proliferation in selected hydrogels. _N.

Figure depicts the hATSC viability for the first set of hydrogels during the 28 day period of hydrogel cultivation in full culture media with the fluorescence images shown in Figure S18. Data collected at day 0 (2 h after the hydrogel preparation) confirm the cytocompatibility of selected cross-linking conditions, as hATSC viability in hydrogels was evaluated as 80 ± 3% for G8A5M5, 89 ± 3% for G8A5M20, and 91 ± 1% for G8A6M20. Similar cell viability after preparation, between 80 and 90%, was previously reported also for bone marrow mesenchymal stem cells encapsulated in GelMA/SA-based hydrogels. , A lower viability of hATSCs in G8A5M5 may be related to rheological data showing that samples containing SA of higher MM require higher stress (Figure , Figure S12) to induce the material flow, which may negatively affect the cell viability. As can be seen in Figure , hATSC viability initially decreases during the first 24 h, most likely as a result of stress imposed on cells during the sample preparation. The viability starts to increase between 1 and 3 days of cultivation. At day 3, the images shown in Figure S18 reveal the first changes in cells toward the fibroblast-like morphology. The cells elongation suggests that cells start to attach to the GelMA binding sites represented by RGD and MMT motifs. After 1 week, differences between the samples disappear and hATSCs reach viability close to 95% in all three hydrogel types. These results correlate with high cell viability of different cell types at day 7 of cultivation in GelMA/SA hydrogels reported by other studies. , After 10–14 days, the cell clusters are recognized (Figure S18) indicating that the cell proliferation and high cell viability of around 95% are maintained until the end of experiment. Thus, obtained data confirm that selected hydrogels represent a suitable environment supporting the cell proliferation.

9.

hATSC viability in G8A5M5, G8A5M20, and G8A6M20 hydrogels in selected time intervals during 28 days of cultivation. Statistical significance was calculated using one-way ANOVA and Tukey test (n = 4, *p < 0.05).

Conclusions

This study shows that despite gelatin methacryloyl (GelMA)/sodium alginate (SA) bioinks being one of the frequently used bioink systems in the focus of the biomedical applications, there are still open questions and variables that require attention. The molar mass (MM) of SA was found to significantly affect the bioink rheological properties, and thus, it directly impacts the bioink printability. We showed that the printed structures with similar quality can be obtain for bioinks containing SA of lower as well as of higher MM. The lower printing pressures correlating with the lower yield stress values are associated with bioinks containing SA of lower MM. The reduction of MM of SA to very low values of about 38 kg·mol^–1 leads to a further decrease of the yield stress; however, the high elasticity of this bioink results in a compromised printability. Based on these observations, the variation in MM of SA should be considered as one of the strategies in the bioink design to balance the quality of the printing outcome with the printing conditions.

The sterilization of GelMA/SA bioinks, with the specific focus on SA sterilization in the powder form by autoclaving, was addressed as one of the challenges associated with translation of these bioinks toward the in vivo testing. While no significant changes of chemical composition were captured as a result of autoclaving, MM of SA showed an exponential decrease with increasing the sterilization time for all tested samples. Additionally, we showed that while autoclaving of SA at the same conditions provided reproducible results with respect to SA degradation, even a slight variation in the sample amount and layer thickness may affect the resulting MM of SA and, therefore, also the reproducibility of bioink properties.

In addition, the rheological properties of GelMA/SA hydrogels were studied as a function of cross-linking conditions including photoinitiator (LAP) concentration, UV 365 nm irradiation time, and Ca²⁺ cross-linking time. As a result, the hydrogel preparation conditions involving 2 min of UV cross-linking followed by 20 min of Ca²⁺ cross-linking using 0.1 wt % of LAP can be recommended as optimal. The mechanical analysis of hydrogels prepared using SA of varying MM revealed that formation of GelMA covalent network, determining the equilibrium hydrogel properties upon application of stress, is not affected by MM of SA. No significant differences in the hydrogel permeability under physiological conditions between samples prepared using SA of different MM were observed from the iSEC measurements. In the case of SA of lower MM values, the higher degree of ionic cross-linking was suggested based on the immediate response of hydrogel to fast deformation. Finally, the high viability of hATSCs during 28 days of in vitro cultivation was confirmed for all tested hydrogels.

Overall, this study contributes to the topic of GelMA/SA bioinks by describing the relationships between the MM of SA and the principal characteristics of both GelMA/SA bioinks and the resulting hydrogels. Moreover, our findings highlight that these characteristics are also significantly affected by, at a glance, minor factors such as selection of buffer or conditions for sterilization. Altogether, this work emphasizes the importance of a systematic approach in bioink development that is obviously not limited to the GelMA/SA bioink systems and can be extended to other bioink types. Future work should focus on systematically coupling MM-dependent rheological behavior with the studies on cell–material interaction to understand how subtle differences in SA MM influence differentiation, extracellular matrix deposition, or biodegradation, thereby advancing the translational potential of the GelMA/SA bioinks.

All data relevant to the paper are included in the article and its Supporting Information. Further details and information can be provided by corresponding authors upon request.

The Supporting Information is available free of charge at https://pubs.acs.org/doi/10.1021/acsomega.5c05476.

• SA molar mass distributions (Figure S1); GelMA synthesis scheme (Figure S2) and ¹H NMR spectra (Figure S3); autoclaving conditions (Figure S4); rheological measurements (Figures S5–S8 and S14); images of printed strands (Figure S8); autoclaved SA characterizationimages, SEC, FTIR, TGA, sterility (Figures S9–S13); compression measurements (Figure S15 and S16); elution curves iSEC (Figure S17); cell fluorescence imaging (Figure S18); power law and Cross constants (Table S1); spreading ratios (Tables S2 and S3); elemental analysis (Table S4) (PDF)

The authors declare no competing financial interest.