Hierarchical biomaterials via photopatterning-enhanced direct ink writing

GelMA was synthesized following established protocols [ 38 ]. Type A gelatin (20 g) was dissolved in dH₂O (150 ml) in a round bottom flask and stirred at 50 °C. Methacrylic anhydride (MA, 12 g) was added dropwise to the gelatin solution and the reaction proceeded for 1.5 h at 50 °C. The solution was transferred to 50 ml conical tubes and centrifuged at 3500 rcf for 5 min. The supernatant containing GelMA was decanted into a glass beaker (500 ml) and diluted 1:2 by volume with preheated (40 °C) dH₂O. The solution was transferred to dialysis tubing (SnakeSkin TM dialysis tubing, 3.5 kDa MWCO; Ref 88 244, Thermo Scientific) and dialyzed against dH₂O at 40 °C for 7 d. The dialysis water was changed twice daily. Later, the solution was diluted 1:10 by volume with preheated (40 °C) dH₂O, sterilized over a 0.2 µm filter, frozen (−80 °C) overnight, and lyophilized for 5 d. ¹H NMR (D₂O) was used to quantify the degree of functionalization (∼70%) by calculating the ratio of the lysine methylene signals (δ = 2.8–3.0 ppm) of GelMA to the phenylalanine signal (δ = 7.1–7.4 ppm) of unmodified gelatin (figure S1 available online at stacks.iop.org/BF/13/044105/mmedia).

PEG-b-PLA was prepared via ring-opening polymerization with tin (II) 2-ethylhexanoate [Sn(Oct)2] as a catalyst using an adapted procedure from literature [39]. The polymerization reaction was performed in an inert and anhydrous environment. Poly(ethylene glycol) methyl ether (5 kDa; 3.5 g, 0.7 mmol) and 3,6-dimethyl-1,4-dioxane-2,5-dione (LA; 11.180 g, 77 mmol) were dissolved in degassed toluene (30 ml) in a three-neck round bottom flask at 100 °C. Upon addition of Sn(Oct)2 (0.32 ml, 0.9 mmol) to the PEG/LA/toluene mixture, the temperature was increased to 140 °C and the reaction proceeded for 6.5 h. Upon completion, the reaction was cooled to room temperature, and the solution was recovered by precipitation from cold diethyl ether, recollected by filtration, and dried under vacuum for 2 d. PEG-b-PLA (7 mg) dissolved in CDCL3 (700 µl) was used for ¹H NMR analysis. The molecular weight of the PLA chain was determined by comparing integration values of the CH and CH3 peaks of the PLA repeat units (1.5 ppm and 5.2 ppm, respectively) to the terminal CH3 peak of the PEG chain (3.58 ppm) (figure S2).

PEG-b-PLA nanoparticles (NPs) were prepared via nanoprecipitation as described [40]. In brief, 140 mg of PEG-b-PLA were dissolved in 2.5 ml of acetone and precipitated dropwise in a stirring solution of milliQ H₂O (10 ml). The acetone was removed overnight by evaporation, and the NPs were recovered via filtration (Amicon filter, 30 kDa MWCO) at 4500 rcf for 70 min. The filtered NPs (figure S3) were collected and resuspended in PBS at a final concentration of 20 wt% and stored at 4 °C until further use.

Lithium phenyl-2,4,6-trimethylbenzoylphosphinate (LAP) was synthesized as described [41, 42]. In brief, 2,4,6-trimethylbenzoyl chloride (3.2 g, 0.018 mol) was added slowly to dimethyl phenylphosphonite (3 g, 0.018 mol) under argon and stirring at room temperature. The reaction proceeded for 18 h. A solution of lithium bromide (6.1 g, 0.072 mol) in 2-butanone (100 ml) was added to the reaction mixture. To induce product precipitation, the reaction was heated to 50 °C for 10 min and then cooled to room temperature. The solution was filtered to recover the precipitate, washed 3× with 2-butanone (100 ml), and dried under vacuum. Product quality was confirmed via ¹H NMR (400 MHz, D₂O): δ 7.75 (m, 2 H), 7.59 (m, 1 H), 7.49 (m, 2 H), 6.91 (s, 2 H), 2.26 (s, 3 H), 2.05 (s, 6 H) (figure S4).

UNI–GelMA-5 (HPMC 1 wt%, PEG-b-PLA NPs 15 wt%, GelMA 5 wt%, and LAP 0.1 wt%) and UNI–GelMA-10 (HPMC 1 wt%, PEG-b-PLA NPs 15 wt%, GelMA 10 wt%, and LAP 0.1 wt%) hydrogels were prepared by first dissolving HPMC in PBS at the desired concentration and equilibrating in a syringe overnight (25 °C). GelMA, LAP, and NP solution were loaded at the desired concentrations into a second syringe and kept at 37 °C until further use. Before mixing, the two syringes were centrifuged to remove residual bubbles entrapped in the solution. The syringes were connected by a female-to-female luer lock adapter (Cellink, Sweden) and gently mixed together for 5 min and kept at 37 °C until further use.

Rheological measurements were performed using a strain-controlled shear rheometer (MCR 502; Anton-Paar, Zofingen, Switzerland) equipped with a Peltier stage. A 25 mm cone–plate geometry with a 2° truncation angle was used for all experiments, unless otherwise noted. The experiments were performed at 37 °C and silicone oil was used to isolate the sample during the test to prevent drying.

To investigate the viscoelastic properties of the prepared materials, dynamic oscillatory strain amplitude sweep measurements were performed at a constant angular frequency, ω = 10 rad s⁻¹, and the strain amplitude, γ, was increased from 0.1% to 1000%. Dynamic oscillatory frequency sweep measurements were performed at a γ = 0.3% and the angular frequency, ω, was decreased from 100 to 0.01 rad s⁻¹ (figure S5). Rotational shear rate measurements were performed with initial shear rate dγ/dt = 0.1 s⁻¹ and final shear rate dγ/dt = 100 s⁻¹. The measured viscosity, η, was fit in the linear range (dγ/dt from 1 to 100 s⁻¹) to the power-law model: η = K_PL (dγ/dt)^{n −1} to investigate the shear-thinning index, n, and consistency index, K_PL, at different shear rates (table S1). To measure the self-healing properties of the hydrogel inks, dynamic oscillatory time sweep tests were designed with seven intervals of alternating low strain (γ = 0.3%, ω = 10 rad s⁻¹) and high strain (γ = 1000% or 100% for UNI formulation or UNI–GelMA formulations respectively, ω = 10 rad s⁻¹).

An 8 mm plate–plate geometry with 0.5 mm gap size was used for dynamic oscillatory time sweep measurements (γ = 0.3%, ω = 10 rad s⁻¹) to characterize the cross-linking kinetics of UNI–GelMA. The samples were irradiated (λ = 405 nm, I = 10 mW cm⁻²) after 60 s oscillation to induce photopolymerization of the GelMA (figure S6).

All rheological measurements were performed at least in triplicate (n ⩾ 3).

A pneumatic-driven 3D printer (BioX, Cellink) was used for the 3D printing (DIW) experiments. For all tests, 3 ml cartridges mounted with straight nozzles (22 G, Ø = 410 µm) were used with a heated printhead (37 °C). The printing parameters were set for homogeneous flow for both UNI–GelMA-5 (30–35 kPa, 4 mm s⁻¹) and UNI–GelMA-10 (55–65 kPa, 4 mm s⁻¹). The G-code used to print the 3D rectangular structures for the cell encapsulation experiments was written manually. The 3D models of the ETH logo and the meniscus were designed using NX Siemens CAD software and the G-code was generated with BioX internal software.

For photopatterning of the constructs, we used a 0.47' digital micromirror device DLP^® E4750LC (EKB Technologies Ltd, Israel) with a 405 nm LED light source and 96 mm × 54 mm image size at 118 mm working distance. Based on the display settings (1920 × 1080 micromirrors) and working distance, the pixel size in the printed construct was 50 µm × 50 µm. The grayscale images of Albert Einstein and Marie Skłodowska Curie were downloaded from the internet (https://de.cleanpng.com/png-9069gn/ and https://de.cleanpng.com/png-w8s164/, respectively). The photopatterning geometries were designed in Affinity Publisher and exported as PNG. The DLP device was connected to a portable computer and controlled through a Windows^®-based GUI tool (DLP^® Pico Display and Light Control EVM, TX Instruments) to display the pattern on the printing plate. The light intensity at the surface of the construct was controlled to be 10 mW cm⁻².

To test the potential of tailoring the mechanical properties using photopatterning, uniaxial tensile tests were performed on constructs fabricated with UNI–GelMA-10. To eliminate any effect of the extrusion process on the final mechanical properties, the material was molded into rectangular shaped test-pieces (width × length × thickness: 5 mm × 30 mm × 1 mm). Distinct cross-linking times were combined with different pattern geometries to fabricate composite materials. Uniaxial tests were performed as previously described [43]. Briefly, test-pieces were clamped to two axes of a custom-built tensile testing setup (MTS Systems, USA) and elongated at a nominal displacement rate of 0.02 mm s⁻¹. All tensile tests were performed at room temperature with samples immersed in a 0.15 M NaCl bath. From top-view images of the deforming samples, the local in-plane principal stretches λ₁ and λ₂ were extracted using a custom-written optical flow tracking algorithm [44]. The nominal stress was calculated as P = F/(WH), where F is the measured force, W is the reference width measured from the top-view images during testing, and H is the gel thickness obtained from images of the sample cross-section. The elastic modulus, E, and the Poisson's ratio, v, in the range of small strains were calculated by linear regression of the P − λ₁ and λ₂ − λ₁ curves, respectively, i.e. P = E(λ₁ − 1) and (λ₂ − 1) = −v(λ₁ − 1), using data up to 2% linear strain. Statistical analysis was performed between the matrix condition and the reinforced structures using two-sample t-tests (p < 0.05).

Human bone marrow‐derived stromal cells (hMSCs) were isolated from bone marrow aspirates of healthy donors obtained during orthopaedic procedures with informed consent and in accordance with the local ethical committee (University Hospital Basel; Prof Kummer; approval date 26 March 2007, Ref Number 78/07). Cells were cultured at 37 °C in a 5% CO₂ incubator in minimal essential medium with alpha modification and nucleosides (MEMα) supplemented with FBS (10%) and PS (100 U ml⁻¹). Cells were passaged before reaching 90% confluency and the medium was changed every 2 and 3 d. hMSCs (P7) were added to either UNI–GelMA-5 or UNI–GelMA-10 at a 1:9 ratio to reach a final concentration of 2 × 10⁶ cells ml⁻¹ and gently mixed using a spatula. The mixture was loaded into a preheated (37 °C) 3 ml cartridge mounted with a conical nozzle (Ø = 410 µm).

Bilayer (layer height, 250 µm) rectangular samples (15 mm × 7 mm) were printed with UNI–GelMA-5 using 19–22 kPa pressure at 4 mm s⁻¹ using a pneumatic-driven 3D printer (BioX, Cellink) on a pre-warmed Sigmacote‐coated glass slide. The DLP setup (λ = 405 nm light, I = 10 mW cm⁻²) was used to photopolymerize the scaffolds. Three conditions were tested for samples prepared with photopatterning-enhanced DIW: (a) uniform partial cross-linking (homogeneous exposure for 10 s); (b) uniform complete cross-linking (homogeneous exposure for 30 s); and (c) stepwise composite cross-linking (homogenous exposure for 10 s followed by patterned exposure for 20 s). Two control conditions were used for nonprinted scaffolds, in which samples were casted between two Sigmacote‐coated glass slides (sample thickness = 0.5 mm). These samples were gelled using (a) uniform partial cross-linking (homogeneous exposure for 10 s) and (b) uniform complete cross-linking (homogeneous exposure for 30 s). All samples were placed in cell culture medium (supplemented with 10% FBS, 1% PS, and 5 ng ml⁻¹ FGF-2) and incubated (37 °C, 5% CO₂) until further analysis. The medium was refreshed 15 min after bioprinting or encapsulation, and then once every two days for the duration of culture.

As a representative biological structure, we 3D printed a meniscus using our biofabrication approach. The meniscus construct was printed with UNI–GelMA-10 using 30–35 kPa at 4 mm s⁻¹ and photopatterned (λ = 405 nm light, I = 10 mW cm⁻²). Three conditions were tested for samples prepared with photopatterning-enhanced DIW: (a) uniform partial cross-linking (homogeneous exposure for 10 s); (b) uniform complete cross-linking (homogeneous exposure for 60 s); and (c) stepwise composite cross-linking (homogeneous exposure for 10 s followed by patterned exposure for 50 s). Two control conditions were used for nonprinted scaffolds, in which samples were casted between two Sigmacote‐coated glass slides (sample thickness = 0.5 mm). These samples were gelled using (a) uniform partial cross-linking (homogeneous exposure for 10 s) and (b) uniform complete cross-linking (homogeneous exposure for 60 s).

Cell viability was analyzed via membrane integrity at days 0, 1, 2 and 5 after encapsulation. A LIVE/DEAD Viability/Cytotoxicity kit (ThermoFisher) was used according to the manufacturer's instruction. In brief, the samples were washed 2× in PBS for 2 min, followed by incubation (37 °C, 5% CO₂) in live/dead solution for 1 h. Before imaging (EVOS M5000, ThermoFisher), the samples were washed 3× with PBS for 2 min. Statistical comparison between the samples and at different time points was performed using one-way analysis of variance with Bonferroni comparison of means (p = 0.05). The Shapiro–Wilk test failed to reject the null hypothesis that the sample population was normally distributed (p > 0.05).

Cell morphology was analyzed at days 0 and 5. AlexaFluor 488 Phalloidin (ThermoFisher) and DAPI (Sigma–Aldrich) were used according to the manufacturers' protocols. In brief, the samples were washed 2× in PBS for 2 min followed by overnight fixation in 4% paraformaldehyde solution (4 °C). After fixation, the samples were washed 3× with PBS for 2 min, incubated in the staining solution for 1 h, and again washed 3× with PBS for 2 min before imaging with confocal microscopy (LSM 780, Zeiss).

In this work, we combined DIW and DLP to fabricate scaffolds with hierarchical control over biomaterial composition and mechanical properties (figure 1). This approach required dual bioinks that allowed for both DIW and DLP in a single material. For DIW, an ink with a high viscosity at low shear rate as well as shear-thinning and self-healing properties was necessary to prevent cell sedimentation, to enable flow, and to provide sufficient shape retention for 3D printing [14]. In addition, shear-thinning inks reduce the shear-stress experienced by cells near the nozzle wall, which can improve cell viability [45–47]. In the case of the sequential process used with our multitechnology printing approach, efficient self-healing was needed to preserve the printed shape until further post-curing. For photopatterning via DLP, the ink should enable light-based changes in mechanical properties through the inclusion of reactive groups, e.g. (meth)acrylates, and a photoinitiator [20, 24]. For patterning in the presence of cells, the polymerization conditions should be cytocompatible. Finally, to provide a conducive environment for cell growth, the ink should include cell-binding motifs, such as the RGD peptide sequence, to facilitate cell–material interactions [38].

Zoom In Zoom Out Reset image size

Based on these design constraints, we formulated dual bioinks by combining our previously developed UNI platform with GelMA. The UNI platform is composed of a PNP hydrogel, in which interactions between HPMC and PEG-b-PLA NPs form a transient physical network with suitable rheological properties for DIW [37]. Previously, the UNI platform was combined with different secondary polymers (e.g. collagen, methacrylated hyaluronic acid, PEG diacrylate, and alginate) to formulate functional and post-curable bioinks with shear-thinning and self-healing properties. In this project, we used GelMA as the secondary polymer (UNI–GelMA) as it is conducive to cell growth and facilitates photopolymerization. Gelatin is a denatured form of collagen that contains cell binding domains, including RGD [48]. The functionalization of gelatin with methacryloyl moieties has been exploited broadly to synthesize photo-curable biomaterials for tissue engineering and bioprinting [24, 49, 50]. We included LAP as the photoinitiator as it can be activated upon exposure to visible light (λ = 405 nm) and is commonly used for cell encapsulation [42].

Formulating GelMA with our PNP-based rheological carrier fluid (UNI platform) generated a dual bioink suitable for DIW and DLP within a single material. Alternative approaches can be used to tailor the viscoelastic properties of GelMA beyond our UNI platform, potentially increasing the spectrum of bioinks compatible with this multitechnology printing approach [51–53].

The base formulation of the UNI platform consisted of HPMC (1 wt%) and PEG-b-PLA NPs (15 wt%; figure S3). To generate dual bioinks, GelMA was added to the base UNI formulation at either 5 wt% (UNI–GelMA-5) or 10 wt% (UNI–GelMA-10). All photocurable formulations contained LAP (0.1 wt%) as a photoinitiator. Shear rheometry was used to characterize the viscoelastic properties of the different biomaterial ink formulations. GelMA alone (10 wt%; GelMA-10) exhibited liquid-like behavior (G'' > G') at 37 °C, indicating that without a carrier ink this GelMA formulation was not suitable for DIW. In contrast, the UNI formulation displayed solid-like properties (G' > G'') over the whole frequency range tested (figure S5), indicating the formation of a transient physical network based on the formation of a PNP hydrogel (figure 2(a)) [37, 47, 54, 55]. Combining GelMA (5 wt% and 10 wt%) with the UNI platform generated inks that retained solid-like properties; the storage moduli of the inks increased with GelMA concentration (G'_UNI = 34.4 ± 1.4 Pa, G'_{UNI–GelMA-5} = 265.5 ± 50.4 Pa, and G'_{UNI–GelMA-10} = 939.7 ± 129.9 Pa at ω = 1 rad s⁻¹).

Zoom In Zoom Out Reset image size

In order to use the UNI–GelMA inks for extrusion-based printing, we investigated the shear-thinning and self-healing properties of the dual bioinks. Similar shear-thinning properties were observed for the different formulations (n_UNI = 0.31 ± 0.01, n_{UNI–GelMA-5} = 0.35 ± 0.09, and n_{UNI–GelMA-10} = 0.18 ± 0.07; figure 2(b)). In addition, each of the dual bioinks exhibited rapid self-healing during step strain measurements. Thus, the combination of the UNI platform with GelMA provided dual bioinks with shear-thinning and self-healing properties suitable for DIW, which were not observed for GelMA alone at 37 °C.

In addition to having rheological properties for DIW, we also demonstrated the ability for the dual bioinks to undergo secondary photo-cross-linking to replicate the process of photopatterning with DLP. We investigated the cross-linking kinetics of UNI–GelMA-5 and UNI–GelMA-10 upon exposure to visible light (λ = 405 nm, I = 10 mW cm⁻²) for varying time intervals. Photopolymerization of both dual bioinks induced a rapid increase in mechanical properties, which then slowed with increasing exposure time (figure 2(b); figure S6). Based on the kinetics of the photopolymerization, we defined exposure times for partial cross-linking, t_partial = 10 s, and complete cross-linking, t_complete = 30 s or 60 s for UNI–GelMA-5 and UNI–GelMA-10, respectively. The storage moduli of both formulations increased after partial cross-linking (G'_{UNI-GelMA-5,partial} = 3.2 ± 0.4 kPa and G'_{UNI-GelMA-10,partial} = 15.6 ± 3.6 kPa). The moduli continued increasing with additional light exposure until the time of complete cross-linking, beyond which no relevant change in the modulus was observed. We also applied a stepwise photopolymerization, in which the samples were partially cross-linked (t_partial = 10 s) and after 2 min of equilibration (light turned off), the cross-linking reaction was continued for an additional 20 s or 50 s, in order to reach complete cross-linking of UNI–GelMA-5 (t_stepwise = 10 s + 20 s) and UNI–GelMA-10 (t_stepwise = 10 s + 50 s), respectively. No differences in mechanical properties were observed between stepwise and complete cross-linking (G'_{UNI-GelMA-5,stepwise} = 4.5 ± 0.3 kPa, G'_{UNI-GelMA-5,complete} = 4.4 ± 1.0 kPa, and G'_{UNI-GelMA-10,stepwise} = 29.1 ± 7.3 kPa, G'_{UNI-GelMA-10,complete} = 26.8 ± 3.2 kPa).

Overall, these results demonstrated that light exposure could be used to tailor the mechanical properties of the dual bioinks and that the sequence of cross-linking did not affect the final modulus of the material. In addition, a tunable t_partial could be used to modulate the local mechanical properties in the material by varying exposure time and a short t_complete was beneficial to reduce the total fabrication time for our multitechnology approach.

In order to fabricate 3D hierarchical biomaterials with defined properties at both micro- and macro- scale, we coupled a pneumatic-driven 3D printer for DIW with a digital micromirror device for DLP (figure S7). Using this approach, we printed different biomaterial inks sequentially to control their spatial organization (figure 3(a)). We then used DLP to control the micro-scale mechanics of the construct by projecting photopatterns onto the printed material. We manufactured a stainless steel plate adapter to ensure alignment and orientation of the printed construct during transfer from the 3D printer to the DLP (figure S8).

Zoom In Zoom Out Reset image size

The integration of DIW and DLP within a single print technology through the design of dual bioinks enabled multiple fabrication opportunities, including grayscale patterning in distinct print layers, differential mechanical reinforcement within distinct regions of a printed construct, as well as multimaterial objects with multilayer mechanical reinforcement. To demonstrate grayscale patterning, we printed a bilayer construct. After printing the first layer, the material was patterned via DLP. An initial uniform exposure (t = 10 s, λ = 405 nm, I = 10 mW cm⁻²) partially solidified the whole layer. Thereafter, a grayscale image of Albert Einstein was projected on the left side of the printed layer (t = 50 s, λ = 405 nm, I = 10 mW cm⁻²; figure 3(b)). Following a similar approach, a second layer was printed and a grayscale image of Marie Skłodowska Curie was patterned on the right side of the construct.

To provide differential mechanical reinforcement, a multimaterial scaffold was printed with two dual bioink formulations (UNI–GelMA-5, and blue; UNI–GelMA-10, clear). Each ink was loaded into a separate heated cartridges (37 °C) and mounted with straight nozzles (22 G; Ø = 410 µm). The ETH Zurich logo was printed in UNI–GelMA-5, and surrounded by printed rectangular frame of UNI–GelMA-10 (figure 3(c)). After DIW, the multimaterial construct was reinforced via DLP. An initial uniform exposure (t = 10 s, λ = 405 nm, I = 10 mW cm⁻²) partially solidified the whole construct. In a 2nd step, distinct patterns (parallel lines for the surrounding frame; z-shaped design for the ETH logo) were simultaneously projected onto the construct (t = 50 s, λ = 405 nm, I = 10 mW cm⁻²). To confirm polymethacrylate cross-linking at the site of photopatterning acryloxyethyl thiocarbamoyl rhodamine B was used for visualization of patterns crosslinked for t_stepwise (figure S10). With this approach, it was possible to control the spatial deposition of different biomaterial inks and to pattern specific features in distinct regions of the final construct.

To control structure and composition at both micro- and macro-scale, a multimaterial 3D object (length = 7 mm, width = 7 mm, and ten layers of 0.35 mm thickness) was printed with patterned light-exposure to form a reinforced cuboid within the multimaterial construct. UNI–GelMA-10 (green and clear) was used for both inks. The inks were loaded into separate heated cartridges (37 °C) and mounted with straight nozzles (22 G; Ø = 410 µm). First, the materials were deposited sequentially to form the 1st two layers. The printed layers were then treated with a stepwise photopatterning step; the surface was homogeneously cross-linked (t = 2 s, λ = 405 nm, I = 10 mW cm⁻²) and then an internal square geometry (length = 4 mm; width = 4 mm) was patterned (t = 50 s, λ = 405 nm, I = 10 mW cm⁻²) in the center of the scaffold. This process was repeated to form the 3D construct. Despite the scattering of the thick construct, the internal 3D cuboid pattern was visible at the end of the process (figure 3(d)).

Overall, our printing approach that combined DIW and DLP with dual bioinks provided a simple and user-friendly strategy to fabricate multimaterial scaffolds with controlled properties at both micro- and macro-scale. DIW enabled the controlled deposition of multiple materials with a resolution of ∼410 µm, which was determined by the nozzle diameter. DLP provided tunable control over the mechanical properties and reinforcement of the constructs. The pattern design, grayscale intensity, and time of exposure dictated the local degree of GelMA cross-linking and the spatial resolution (∼100 µm, figure S9). A further advantage of DLP photopatterning is that grayscale images can be exploited to simultaneously generate structures having different cross-linking degrees without the need for stepwise photopolymerization, accelerating fabrication.

Another useful application of photopatterning-enhanced DIW is spatially controlled mechanical reinforcement of 3D constructs. To achieve this, the freedom of design provided by DLP was combined with the phototunable mechanical properties of the dual bioinks to fabricate composite scaffolds with high resolution mechanical reinforcement (figure 4(a)). We exploited the difference in modulus for UNI–GelMA-10 after two distinct intervals of photopolymerization. Partially cross-linked regions of the scaffold were labeled as matrix (t_partial = 10 s) and fully cross-linked regions were labeled as reinforcement (t_complete = 60 s). Initially, bulk materials were cross-linked and the mechanical properties of the constructs were characterized by tensile testing after equilibration. The elastic modulus increased with the time of cross-linking (E_partial = 6.4 ± 2.9 kPa and E_complete = 28.9 ± 4.4 kPa) with similar Poisson's ratios (v_partial = 0.4 ± 0.1 and v_complete = 0.3 ± 0.1). Composite structures were fabricated in a sequential approach—in an initial step the matrix was cross-linked homogeneously (t_partial = 10 s), and in a 2nd step the reinforcement structures were patterned (t = 50 s). We produced constructs with different designs and volume fractions of reinforcement (lines, 16% and 33%, and z-shaped; figure 4(a)).

Zoom In Zoom Out Reset image size

To determine the efficiency of reinforcement, the elastic moduli of the reinforced composites (16% and 33% lines) were compared with standard composite theory [56]. When the reinforcement was photopatterned as lines parallel to the direction of applied stress, the longitudinal strain in the matrix and fiber reinforcement should be equal, whereas the reinforcement structures should experience larger stress than the matrix due to the larger elastic modulus (E_complete > E_partial). In this case, the upper bound on the elastic modulus of the composite can be described by the rule of mixtures (E_// = E_r ϕ + E_m(1 − ϕ)) for a perfectly reinforced composite, where ϕ denotes the volume fraction of reinforcement and the subscripts r and m denote reinforcement and matrix, respectively [56, 57]. The measured elastic moduli of the patterned constructs (E_{16% reinforcement //,Exp} = 9.2 ± 2.8 kPa and E_{33% reinforcement //,Exp} = 12.9 ± 1.2 kPa) were comparable to the upper bound estimated based on the rule of mixtures (E_{16% reinforcement //,Model} = 10.1 kPa and E_{33% reinforcement //,Model} = 13.8 kPa) (figure 4(b)). The scaffold fabricated with 33% line reinforcement tested parallel to the line orientation resulted in a statistically significant increase (p < 0.05) in elastic modulus as compared with the non-patterned matrix. In contrast, minimum reinforcement should occur when the tensile load is applied perpendicular to the orientation of the reinforcement. In this case, the load should be distributed uniformly between matrix and reinforcement structures, and the resulting lower bound on the elastic modulus of the composite can be estimated using the transverse rule of mixtures (E_⊥ = E_r E_m/(E_r(1 − ϕ) + E_m ϕ)). Again, observed elastic moduli followed the rule of mixtures (E_{33% reinforcement ⊥,Exp} = 8.8 ± 2.8 kPa vs E_{33% reinforcement ⊥,Model} = 8.6 kPa; figure 4(b)). In all cases, the Poisson's ratio was not affected by the reinforcement structure used.

DLP enables arbitrary reinforcement patterns beyond simple parallel reinforcement. We also prepared constructs with a z-shaped reinforcement pattern. In contrast to the line reinforcement, composite structures reinforced with z-shaped patterns exhibited similar elastic moduli (E_{z-shape reinforcement //} = 13.5 ± 1.3 kPa and E_{z-shape reinforcement ⊥} = 13.9 ± 3.9 kPa; figure 4(c)) and Poisson's ratios independent of the orientation of the tensile load. Thus, pattern design was used to tailor the extent of anisotropy in the mechanical reinforcement. Overall, we demonstrated that the spatiotemporal control provided by DLP enables facile and tunable reinforcement of 3D constructs. In the future, z-shape reinforcements could be used to fabricate auxetic materials with negative Poisson's ratio; however, a larger difference in elastic moduli between the matrix and the reinforcements would be necessary [58].

Finally, we demonstrated that cells remained viable following photopatterning-enhanced DIW toward applying this multitechnology printing approach for biofabrication. We encapsulated hMSCs in both UNI–GelMA-5 and UNI–GelMA-10 bioinks and produced cell-laden 3D-printed constructs. In case of UNI–GelMA-5, all samples, except the non-printed controls, were printed as a bilayer rectangular shape using DIW. In a 2nd step, the scaffolds were photo-cross-linked via DLP. Different time intervals (t_partial = 10 s, t_complete = 30 s, and t_stepwise = 10 + 20 s) and pattern designs (uniform or 16% fiber reinforcement) were used to photopattern (λ = 405 nm, I = 10 mW cm⁻²) the 3D-printed constructs. In addition, as a representative biological structure, we 3D printed a meniscus using UNI–GelMA-10 (figure 5(a)). The meniscus constructs were printed and photopatterned with mechanical reinforcement inspired by the natural arrangement of the collagen fibrils (circumferential and radial alignment) [59–61]. The extruded filaments were deposited in a circumferential orientation and then followed by photopatterning. Different time intervals (t_partial = 10 s, t_complete = 60 s, and t_stepwise = 10 + 50 s) and pattern designs (uniform or radial mechanical reinforcement lines) were used to photopattern (λ = 405 nm, I = 10 mW cm⁻²) the 3D-printed constructs. All samples, UNI–GelMA-5 and UNI–GelMA-10, showed similar cell viability compared to non-printed control scaffolds at days 0, 1, 2, and 5 (figures S11–S14).

Zoom In Zoom Out Reset image size

We also investigated if local mechanical properties in the different constructs or within the patterned regions had an effect on cell-material interaction. Similar cell spreading was observed in both non-printed control and bioprinted samples after 5 d in culture (figures 5 and S15). Moreover, the spatial control over the cross-linking via photopatterning and the slow but constant release of HPMC from the scaffold (figure S16) did not significantly influence cell morphology. Overall, these results demonstrated the cytocompatibility of our approach to fabricate hierarchical composite biomaterials. In the future, this feature could be useful to better mimic tissue anisotropy and replicate tissue organization in cell-laden constructs at the micro- and macro-scale.