Human fibrosarcoma cells (HT-1080; ATCC; Manassas, VA) were cultured in the appropriate growth media recommended by ATCC. All the cells were passaged according to standard practices, in which they showed >80% confluency for each passage and were seeded at desired concentrations for our experiments. All chemicals, media, and substrates were mainly purchased from VWR (Radnor, PA), with some exceptions mentioned in the text. HT-1080 cells were cultured at 37°C with 5% CO2. Dulbecco’s Modified Eagle Medium (DMEM, VWR, Radnor, PA) was mixed with 10% v/v FBS and 1% v/v Pen/Strep to feed cells every 2 days. Required cell density for experiments was collected after trypsinization using 1 mL of trypsin-EDTA solution (0.25% Corning, Manassas, VA) centrifuged (Beckman Coulter, Avanti J-15R) for 3 min at 900 rpm and 4°C for the experimental setup.

In addition to HT-1080, epithelial, human breast cancer cells (MDA-MB-231) and human mammary epithelial cells (HMEC; ATCC; Manassas, VA) were also cultured in appropriate growth media recommended by ATCC. After trypsinizing, cells were suspended in DPBS with varying cell densities (0.1–10 × 10⁶ cells/ml) and homogeneously pipetted. The samples were loaded in a 96-well plate (Storage Plate 96-Well Flat Bottom Ltd., New York, NY) for light absorbance analysis to observe the effect of cell type on the cross-linking. The data were recorded at a multi-well scanning spectrophotometer using a wavelength of 490 nm.

GelMA was synthesized according to an established protocol (Van Den Bulcke et al., 2000). In a 100 mL glass flask of Dulbecco’s phosphate-buffered saline (DPBS; Sigma-Aldrich, St. Louis, MO) and 10% w/v Porcine Skin Gelatin (CAS Number 9000-70-8; Sigma-Aldrich, St. Louis, MO) were mixed. The flask was covered to prevent evaporation and was stirred using a magnetic stir bar on a hot plate at 60°C until fully dissolved (approximately an hour). Following gelatin dissolution in the DPBS, methacrylic anhydride at 3 ml and 8 ml were slowly pipetted into the solution to obtain “low MA” and “high MA,” respectively. The temperature was reduced to 50°C, and the solution was stirred and allowed to react for an hour. Pre-warmed DPBS was 5x the volume of the initial solution and was added to the solution after an hour to stop the reaction. Dialysis tubing (12–14 kDa cut-off molecular weight) was used to seal the solution. The dialysis tubes were then submerged in DI water for a week at 40°C. After a week, the solution from the dialysis tube was collected into a glass flask and freeze-dried until the synthesized GelMA demonstrated a porous foam structure. GelMA solution (5, 7, and 10% w/v) was prepared with DPBS and pre-warmed at 40°C with constant stirring. Once fully dissolved and homogenized, a final concentration of 0.07% w/v lithium phenyl-2,4,6-trimethyl-benzoyl-phosphinate (LAP; Sigma Aldrich) was used as the photo-initiator. After sterilizing using 0.2 µm filters, GelMA solutions were prepared at the desired concentration and agitated to allow cells to disperse through the GelMA volume. The orange color in bioprinted samples (Figure 1A) was used for better visualization here and to demonstrate our simple geometry. Photoabsorber was excluded from this study to minimize any potential counter-effects.

The degree of methacryloyl functionalization was quantified by using ¹H NMR according to the previously described method (Hoch et al., 2013; Li et al., 2016). ¹H NMR spectra were collected using an NMR spectrometer, Bruker AVIII-500-MHz equipped with RT TXI probe (H1, C13, N15), B008T (Bruker, Billerica, MA) at a frequency of 500 MHz. Before the measurement, 20 mg of GelMA macromers was completely dissolved in 1 mL deuterium oxide containing 0.05% w/v 3-(trimethylsilyl)-propionic-2,2,3,3-d4-acid sodium salt for calibration (Sigma-Aldrich, St. Louis, MO, United States). Data was processed using Bruker TopSpin 3.6.5 software (Bruker, Billerica, MA). The methacryloyl substitution was quantified using the following equation (see Supplementary Table S1; Supplementary Figure S1): D e g r e e   o f   f u n c t i o n a l i z a t i o n   D o F = 1 − l y s i n e   m e t h y l e n e   p r o t o n   o f   G e l M A , δ = 2.9   p p m / l y s i n e   m e t h y l e n e   p r o t o n   o f   g e l a t i n , δ = 2.9   p p m × 100 % . (1)

The bioprinting applies a digital micromirror device (Texas Instrument, Dallas, TX) with an ultraviolet (UV) light system (Visitech; Wetzlar, Germany) of wavelength ∼380 nm and light intensity of ∼0.7 W/cm² (at the focal plane) to print hydrogels (Bhusal et al., 2021). The platform was equipped with a tri-axial stage to move the printing platform and a rotational movement to rotate the Petri dish platform controlled by a custom-written program. The fabrication process starts with layer-by-layer slicing. Then, the bioink is kept in the UV-grade Petri dish, and the printing process is initiated. The layer thickness and exposure time were selected as 50 μm and 2 s for each layer (Bhusal et al., 2021). The printing platform then moves to the printing position, and a linear z-stage lowers the platform. The program turns on the UV light, and the first layer is cured. The platform then moves to the second layer position, and the bioink flows inside between the first layer and the base of the reservoir to print the second layer. The process was used to ensure the uniformity of the samples.

Standard compression testing was performed to determine the material’s stiffness under physical load. We followed an established protocol (Schuurman et al., 2013) and made disk-shape samples of 10 mm in diameter and 5 mm in height. We analyzed the elastic moduli for different hydrogel sets (Table 1, n = 4) to evaluate the roles of mass concentration (5, 7, and 10% w/v), MA degree (low and high), and cell density (0–10 × 10⁶ cells/ml) in gel stiffness. The samples were placed between the flat metal plates of the mechanical tester (Model 5848; Instron Inc., Norwood, MA) using a 10 N load cell, and the tests were performed using the displacement rate of 1 mm/min (or strain rate of 20% per min). The rate was selected high enough to minimize the effects of hydrogel relaxation. The elastic modulus was calculated as the slope of the stress-strain curve from up to 8%–10% strain (i.e., almost linear range).

Diffusion experiments were conducted to assess the diffusion coefficient of RhD-B in GelMA. We evaluated the apparent diffusion coefficients of the RhD-B (Sigma-Aldrich) agent in crosslinked bioinks (n ≥ 4) with varying cell density (0–10 × 10⁶ cells/ml, see the example in Figures 1A–v) through the relative intensity method using the Nikon Eclipse Ti2 microscope and analyzed according to Fick’s second law (Shkilnyy et al., 2012). A 1.2 × 10⁻⁵ M RhD-B solution was prepared after several trials with different RhD-B concentrations to obtain a smooth intensity profile in GelMA samples. 3D bioprinted samples (Bhusal et al., 2021) for diffusion experiment consisted of a polyethylene glycol (PEGDA) circular frame in 8 mm inner and 10 mm outer diameter and a half-cylindrical GelMA construct 8 mm in diameter and 2 mm thick, as shown in Figure 1B. After, we printed a half inner GelMA core, placed it under a Nikon fluorescent microscope, and filled the half gap with RhD-B solution. The intensity profile was recorded for 30, 60, 90, and 120 min during the experiment with an exposure time of 30 ms. To take an intensity reading, a straight line of 4 mm was drawn in the Nikon software from the edge of the GelMA, which meets the RhD-B towards the center of the GelMA sample. Experiments were performed in triplicate (n = 3). The intensity profiles along the line were exported, and these steps were repeated for all the samples to gather initial and interval readings. The raw data were imported into Excel, turned into calibration curves, and processed to evaluate the results (Supplementary Figures S2 and S3). Diffusion coefficients (µm²/s) (Caló and Khutoryanskiy, 2015) for RhD-B were obtained by fitting concentration gradient data using Fick’s second law Eq. 7.

The fluorescent properties of the selected agent allowed for the visualization of mass transfer into the GelMA porous structure (Figure 1C). We assumed mediums to be homogeneous, and the fluorescence signal is linearly related to the concentration of the RhD-B particles. Fick’s second law (1D) is applicable to model the diffusion of the particles: ∂ c ∂ t = D ∂ 2 c ∂ x 2 (2)

In Eq. 2, c is the solute concentration, x is distance, t is time, and D is the diffusion coefficient of the structure. The variable x represents the perpendicular distance measured from the center of our hydrogel sample. In this case, the center corresponds to the center of the circle, where the edge of the hydrogel is adjacent to the RhD-B solution. The distance extends from this edge to infinity, which in our printed sample, measures 4 mm (radius). When x equals zero x = 0 , it indicates the edge of the hydrogel, as shown in Figure 1C. As our experiments start with a tracer concentration of zero within the gel sample, and the tracer is introduced from the source at x = 0 , with a constant concentration of c o , we have applied the following boundary conditions 3): c x = 0 , t = c o     c x , t = 0 = 0     c x = ∞ , t = 0 (3)

To solve Fick’s second law, the error function equation and error function complement equation were used as shown below: erf z = 2 π ∫ z ∞ e − t 2 d t (4) erf ⁡ c z = 1 − erf ⁡ ⁡ z (5) z = x 2 D t (6) C x , t = C o   *   erf ⁡ c z (7) C x , t C o = erf ⁡ c z (8)

We used the above formulation to translate our intensity readings into diffusivity (Supplementary Figures S2 and S3) and calculate the apparent diffusion coefficients of RhD-B. According to Einstein’s equation of Brownian motion (Einstein, 1905) and semi-empirical tortuous flow in porous media equations (Koponen et al., 1996), the effective diffusion coefficient through hydrogels D e f f , can be expressed as a function of a structure factor, p , the porosity, φ , and bulk solution diffusion coefficient, D 0 D e f f = D 0 1 + p 1 − φ (9)

We conducted an established release experiment to validate our approach for the diffusion experiments (Ma et al., 2018). Hydrogel precursor solutions were prepared in DPBS, with a final concentration of 5%–10% w/v for high- and low-MA GelMA, 0.07% w/v for LAP, and a final concentration of 0.1 mM RhD-B. Disk-shape samples of 10 mm in diameter and 5 mm in height were made, and we conducted a couple of trials to optimize RhD-B concentration and ratio of DPBS to sample amount. RhD-B-loaded hydrogels were immersed in 3 mL DPBS by shaking at 200 rpm and 37 °C. At different times, the release medium (1 mL) was removed and replaced with an equal volume of DPBS. RhD-B concentration was quantified by SpectraMax i3 (Molecular Devices, San Jose, CA) at 485/535 nm. Seven readings from each well were collected and averaged. The total amount of RhD-B was calculated by a standard curve of RhD-B in DPBS versus fluorescence intensity (Supplementary Figure S4). The release experiments were performed in triplicate (n = 3).

The viability of encapsulated cells and spheroids in GelMA was assessed using one Live/Dead assay (PromoCell GmbH, Heidelberg, Germany). At endpoints, the samples were rinsed with PBS (Sigma-Aldrich), incubated for 40 min with Calcein-AM (1 mM; live cells in green) and ethidium homodimer 1 (6 mM; dead cells in red), and then rinsed again. Encapsulated cells in GelMA were then imaged with a fluorescence microscope (Nikon, Melville, NY) through FITC and TRITC filters. The combined Live/Dead images were processed in ImageJ software to estimate the percentages of live and dead cells.

Statistical analysis was performed in GraphPad software (GraphPad Software, Inc., San Diego, CA) statistical tool. A one-way analysis of variance (one-way-ANOVA) and two-way analysis of variance (two-way-ANOVA) tests were used for data analysis. A value of p < 0.05 was considered statistically significant. Finally, a linear correlation analysis was performed among the governing parameters using Excel (Microsoft).

The conjugation of methacryloyl to the gelatin molecules was confirmed by the ¹H NMR spectra (see Supplementary Figure S1). The increased signal at δ = 5.4 and 5.7 ppm (i.e., the protons of the methacrylate vinyl group) and decreased signal at δ = 2.9 ppm (i.e., the protons of the methylene of lysine signal) confirmed the rate of modification degree with the MA (Li et al., 2016). The intensity of the proton signal of the aromatic amino acid moieties in gelatin was used to normalize the intensity of other protons in various samples because their signal stayed constant over time. As a result, the DoF was determined by comparing the proton signal of GelMA and unmodified gelatin at δ = 2.9 ppm. By adjusting the feed ratio of MA to gelatin, the DoF of the three types of GelMA macromers was found to be 50.14% ± 2.04% for low-MA, 65.32% ± 2.06% for high-MA (see Supplementary Table S1). Increased MA is associated with the introduction of more crosslinking sites that expand the connectivity within the network. As a result, the MA factor provides control over the hydrogel’s mechanical and biological properties (Zhu et al., 2019). ¹H NMR analysis validation shows the efficiency of our protocol in preparing GelMA samples and a baseline for biotransport analysis.

The unconfined compression testing of GelMA disk samples with different MA degrees and mass concentrations was used to quantify the relationship between the stiffness and cell density. The trend can also be used to understand the transport properties of GelMA. We selected 7% low MA for simplicity and evaluated the elastic moduli of crosslinked samples (n ≥ 4) with a cell density of 0–10 × 10⁶ cells/ml. Figure 2A shows the elastic modulus for GelMA 5%, 7%, and 10% w/v at low and high MA. The elastic modulus was found to be ranging from 1.633 ± 0.522 kPa for 5% w/v low-MA GelMA to 15.454 ± 0.657 kPA for 10% w/v high-MA GelMA. The cell encapsulation led to an increased level of stiffness (although not statistically) and then lower values at high cell densities, such as 10 × 10⁶ cells/ml (p < 0.01), while MA degree and GelMA concentration had a significant impact on stiffness, as shown in Figures 2A, B. This would support the notion that a lower photo-crosslinking density can occur in the presence of large cell volumes. In summary, the cell density insignificantly impacts the hydrogel stiffness after bioprinting.

The unconfined compression testing has been used to measure the modulus of soft samples in the field. The linear modulus may not be sensitive enough to accurately capture the effect of cell density. The cell presence may contribute to the material resistance after the linear deformation (i.e., when the network chains are deformed to their limits). There might be other reasons for the lack of sensitivity in such testing methods. For instance, the mechanism involves load transfer through the interconnections of the fiber cross-sections, providing an additional contribution to the reinforcement process (Castilho et al., 2018).

The fluorescent properties of the selected agent allowed for visualization of mass transfer into the GelMA porous structure. We assumed mediums to be homogeneous, and the fluorescence signal is linearly related to the concentration of the RhD-B particles (validated by our preliminary tests). We used Eqs 5–9 to translate our diffusion experiments into numbers (see the examples in Supplementary Figure S3). The average diffusion coefficient of the agent through GelMA was higher at lower mass concentrations and low MA, as shown in Figures 2C, D; Table 1. The diffusivity of RhD-B in high-MA samples with varying GelMA concentrations demonstrates a low significance level. The diffusivity is different in low-MA samples with varying GelMA concentrations (Figure 2C). Interestingly, these results indicate that the high-MA case dominates over GelMA concentration in terms of forming a dense and less diverse hydrogel network. The higher value means a larger pore size or ease of fluid movement within the microstructure. We also tested 7% GelMA loaded by cells in a 1–10 × 10⁶ cells/ml range. At a cell density of 5 × 10⁶ cells/ml, the diffusion coefficient decreases ∼33% for low MA compared to acellular GelMA. From 5 × 10⁶ to 10 × 10⁶ cells/ml, an increase in the diffusion coefficient up to ∼31% was observed in Figures 2C, D; Table 1. Light reduction can occur due to high cell density, which decreases the degree of GelMA crosslinking. A similar pattern in high-MA GelMA was also observed. The MA factor impacts the degree of photo-crosslinking; thus, high- and low-MA cases were used to test the role of cell densities for the physical properties of GelMA 7% samples. Three different mass concentrations of GelMA with and without the cell presence were used: 5%, 7%, and 10% GelMA. It was observed that the average diffusion coefficient of the agent was higher at lower mass concentrations.

A summary of the diffusion coefficients and elastic moduli is shown in Table 1. The trend of the diffusion coefficient is significant when compared to diverse cell densities indicating a possible interference in the photo-crosslinking through absorption of light energy and possible inhibition by oxygen-based bioproducts around the cells (oxygen is a major inhibitor of our crosslinking). The authors speculate that cell pellets in the hydrogel precursor absorb light through the membrane (which is practically challenging to validate). Cell membranes, mostly of lipids and proteins, have refractive indices in the range of 1.46–1.54 where the refractive index of the suspending medium is assumed to be 1.33 (Meyer, 1979). The higher cell density raises the overall refractive index of the hydrogel. A higher refractive index means a slower traveling speed of light and an increased change in the direction of the light (or less focus of light energy). From this perspective, by an increase in cell density, we may anticipate a consistent reduction of the elastic modulus (as seen in Figure 2B) and an increase in the diffusion coefficient. To assess whether cell-light interaction is specific to the select phenotype, we compared the selected type with 3 cell types. The data in Figure 3A shows a dependency of light absorption capacity on the cellular type, being the highest in HEMC. This would suggest that the HMECs will lead to more diverse data considering their larger size (60–100 μm in diameter) (Romeo et al., 2006) when compared to HT1080 (10–15 μm in diameter) and MDA-MB-231 (8–2 µm in diameter) (TruongVo et al., 2017). Figure 3B demonstrates diffusion coefficients of RhD-B in HT-1080, MDA-MB-231, and HMEC encapsulated in 7% w/v GelMA samples with 10 million cell density. Our results demonstrated a direct correlation between cell size and light absorbance rate. Larger cell size can raise the solute diffusivity. This indicates an impaired crosslinking, which produced light absorption of larger cells and the hindrance for GelMA chain connectivity.

Another speculation about the role of the cell density is that the lower density causes pore-clogging and/or reduces the pore size of hydrogel, lowering the possibility of the diffusion of solute particles. This phenomenon can be explained by Brownian particle motion through porous structure decreases via more obstacles (Alsaid et al., 2021). Therefore, the impact of cell density on diffusivity can be inferred from Eq. 9 for the specified cell densities. On the other hand, we observed that high cell density increased solute diffusivity, which may be due to high cell density causing light scattering during photo-crosslinking that causes a reduction in crosslinking energy to form a covalent bond between methacrylate side chains. This is illustrated in Figure 3C. The lower crosslinking denotes a larger pore size, which acts against the cell-induced frictions or drags onto the fluid flow.

The release behavior represents the mass transport behavior of bioprinted hydrogel and is related to a large group of bioprinted models: drug delivery scaffolds. We used this established approach to validate our methodology. The results in Supplementary Figure S5 demonstrate a similar trend to the diffusivity results, indicating that the high-MA case dominates over GelMA concentration in terms of forming a dense hydrogel network. While low-MA 5% GelMA releases 83% of the loaded RhD-B, there is a decreasing trend with increased GelMA concentration, which shows a higher significance level compared to the high-MA samples with varying GelMA concentrations. The release test sensitivity is questionable compared to the diffusion test since spectrometers can only detect significant differences in a narrow range of concentrations. After reaching saturation levels of higher concentrations, it is difficult to detect a significant difference between higher molar concentrations of RhD-B for lower concentrations. The resolution and sensitivity of microplate readers are inadequate to monitor calcium flux in live cells compared to the fluorescent microscopy method (Meijer et al., 2014).