Sheet-based extrusion bioprinting: a new multi-material paradigm providing mid-extrusion micropatterning control for microvascular applications

The pump is turned on to pass the hydrogels through both syringes at a 1 ml min⁻¹ flow rate, with 0.5 bar supplied to each syringe, to first fully saturate the apparatus. Each KSM is checked visually for air bubbles, which are removed by tilting the seven-input static valve printhead upward while hydrogel is flowing through. Once the whole apparatus is saturated with hydrogel, the printhead is fit into the holder piece on a robotically-controlled 3D printing apparatus developed in previous work in our lab [38] and controlled by LabVIEW from National Instruments (Austin, TX, USA) (figure 4).

Zoom In Zoom Out Reset image size

The pump is then started again while the basement is simultaneously moving at a constant 3.3 mm s⁻¹, a speed calculated to match the 1 ml min⁻¹ flow rate and thereby produce uniformly wide sheets with consistent channels. For handheld printing, the basement was kept in place while the printhead was guided by hand. During manual mixer selection, the stopcocks on the manual valves were manually switched closed on each outlet at uniform intervals in order to direct the flow of hydrogel from one kenics mixer to another midflow. For automated mixer selection, two rotary valves can be programmed via LabVIEW (or other software) to switch after a designated period of time between corresponding tubes leading to each kenics mixer.

Immediately after extrusion, the channeled hydrogel sheet can be cured/solidified. With UV-sensitive hydrogels, such as gelatin methacryloyl (GelMA), the sheets are exposed to 365 nm UV light from an Omnicure S2000 (Laude, GR, NL) for 30 s. While UV exposure can negatively affect cell viability, as mentioned in section 1.2, previous studies on chaotic printing have shown the chosen wavelength and duration to be sufficient for solidifying GelMA throughout hydrogel constructs without a detectable effect on cell viability [34, 36]. For hydrogels that physically crosslink in the presence of calcium chloride (CaCl₂), such as sodium alginate (SA), the sheets are sprayed with a fine mist of 4% (wt./vol.) CaCl₂ using either a handheld sprayer bottle or an ultrasonic atomizer. Applying CaCl₂ in the form of gradually increased misting intensity allows the gel to be crosslinked with minimal impact on physical gel structure, as previously demonstrated [39, 40]. After waiting approximately 30 s, CaCl₂ droplets are then added to the sheet with a Pasteur pipet to completely soak the sheet. At least 5 additional minutes are then allowed to ensure complete crosslinking before handling the sheets. If the hydrogel contains both GelMA and SA, the UV and CaCl₂ crosslinking steps are completed in succession.

The hydrogel sheet constructs produced with this device can be combined with woven thermoplastic polymer fiber scaffolds produced by melt electrowriting (MEW) to combine properties of both materials. This is accomplished simply by extruding directly over a MEW scaffold before conducting the post-print curing steps described in section 2.3.2. This allows hydrogel to crosslink around the MEW scaffold, effectively fusing both together. The 500 µm polycaprolactone (PCL) scaffolds were produced with a MEW device using previously defined biotextile fabrication methods [38].

SA was added to Dulbecco's phosphate-buffered saline (DBPS) from Gibco (Billings, MT, USA) at 4% (wt./vol.) and mixed at 70 °C until fully dissolved. Three to five drops of FluoSpheres polystyrene microspheres from Invitrogen (Waltham, MA, USA) were added to half of the prepared 4% SA, while the other half was kept clear. The final solutions were maintained at 37 °C until use. To create a 'fugitive ink' (a.k.a. 'sacrificial ink'), as developed in previous chaotic printing work [35, 36], a 0.8% (wt./vol.) solution of hydroxyethyl cellulose (HEC) was prepared. HEC was mixed in DI water at 80 °C until fully dissolved, then maintained at 37 °C until use.

For visualization purposes, 4% SA with fluorescent microspheres and 4% SA without microspheres were used as the two hydrogel inputs in the device to produce hydrogel sheet constructs with alternating clear and fluorescent channels. A 365 nm UV light was shined on the resulting hydrogel sheets to illuminate the fluorescent layers while pictures were taken with a Plugable USB Digital Microscope (Redmond, WA, USA). ImageJ software (NIH) was used to measure the width and thickness of the hydrogel sheet constructs, as well as channel widths, with six replicates for each respective measurement.

To create hydrogel sheet constructs with vacant inner channels, 4% SA with fluorescent microspheres and 0.8% HEC were used as the two hydrogel inputs in the device. During the CaCl₂ crosslinking process described in section 2.3.2, SA solidified while HEC remained fluid as a fugitive ink. Resulting constructs were soaked in DI water overnight to allow the HEC to diffuse out, leaving behind vacant internal channels. A 1 ml insulin syringe was then used to inject vacant channels with orange food-colored DI water to demonstrate channel perfusion.

A 3% (wt./vol.) GelMA, 2% SA, and 0.1% lithium phenyl-2,4,6-trimethylbenzoylphosphinate (LAP) solution was prepared in DPBS to serve as a base for the cell-laden bioink. GelMA was chosen for its cell adhesion-promoting properties, while SA serves as structural support for the resulting solidified gel, with minimal added cell interactions [34]. LAP powder was first constituted in DPBS and passed through a 0.22 µm filter via syringe for sterilization. Lyophilized GelMA and SA powder were exposed to UV-C light for 15 min before being added to the LAP solution in a sterile biosafety cabinet. The solution was mixed in sterile conditions at 70 °C until fully dissolved, then maintained at 37 °C until use. To produce a sterile 0.4% HEC solution, HEC powder was similarly exposed to UV-C light for 15 min before being added to sterile DI water. The HEC solution was mixed 80 °C until fully dissolved, then maintained at 37 °C until use.

EndoGRO human umbilical vein endothelial cells (HUVECs) and human brain vascular pericytes (hPCs) were obtained from Millipore Sigma (Burlington, MA, USA) and iXCells Biotechnologies (San Diego, CA, USA) at passage 1 and 2, respectively. Each cell type was thawed, seeded in two T175 flasks at 500 000 cells per flask, and incubated at 37 °C and 5% CO₂. HUVECs were supplied with EBM Endothelial Cell Growth Basal Medium and SingleQuots Supplements and Growth Factors from Lonza (Walkersville, MD, USA), while hPCS received Human Pericyte Growth Medium 5 from iXCells Biotechnologies. Media changes occurred every 72 h. Once the flasks reached approximately 80%–90% confluence, the cells were washed three times with phosphate-buffered saline (PBS) (Gibco) and detached with TrypLE (Gibco). The resulting cell suspensions were pelleted down by centrifuging at 1200 rpm for 10 min and then aspirating the remaining media. Cells were then reconstituted in fresh media. The 10 µl of the cell suspension was combined with 80 µl of PBS and 10 µl of TrypanBlue (Gibco) before being counted with a hemocytometer.

Based on the measured cell densities, the suspensions were transferred to new centrifuge tubes so that there were 20 million cells in each tube. Media was aspirated and 10 ml of 3% GelMA 2% SA 0.1% LAP hydrogel was added to the hPC pellet, while 0.4% HEC was added to the HUVEC pellet. These solutions were aspirated up and down to homogenize the cells within the gels, each at a density of 2 million cells ml⁻¹. The pipet tips used at this step were trimmed approximately 1/3rd the distance from the tip to facilitate transfer of the relatively viscous hydrogel and minimize shear stress on the cells.

The 3% (wt./vol.) GelMA 2% SA 0.1% LAP with hPCs and 0.4% HEC with HUVECs were used as the two hydrogel inputs in the device for the experimental group, while 3% GelMA 2% SA 0.1% LAP without cells and 0.4% HEC with HUVECs were used for the control group. The HUVEC-laden HEC ink was utilized to attempt a new 'endothelial seeding' method, whereas HUVECs could attach around the edges of each vacant channel as the fugitive ink vacates the construct. Sterile bioprinting was conducted by manually dragging the printhead across a well plate lid within a biosafety cabinet. Printhead components were sterilized with autoclaving and ethanol soaks, where appropriate. The kenics mixer with three mixing elements was chosen to create cell-laden hydrogel sheet constructs with approximately 8 alternating channels (i.e. 4 channels of hPC-laden gel alternating with 4 vacant channels lined with HUVECs). The constructs were cured within the biosafety cabinet using the UV and CaCl₂ methods described in section 2.3.2. To prevent CaCl₂ mist from entering the biosafety cabinet air flow, it was sprayed over the constructs within a folded enclosure of sterile aluminum foil. Printed constructs were cut into 1 cm segments with a razor blade before being placed in 24-well plates containing a 50:50 combination of the respective endothelial and PC cell growth media formulations, respectively. The well plates were incubated at 37 °C and 5% CO₂ for 1 week, with media changes every 72 h.

A Live/Dead fluorescent viability assay from Invitrogen (Waltham, MA, USA) was used to determine cell viability percentage in each sample across the experimental and control groups. On days 1 and 7, triplicates from each group were washed three times with DPBS before being submerged in 2 µM and 4 µM of calcein AM and 4 µM ethidium homodimer-1, respectively, in DPBS for 45 min at room temperature in the dark. To degrade the hydrogel and re-suspend the cells, samples were then soaked in TrypLE for 10 min and gently vortexed. The resulting solutions were aliquoted to glass slides, covered with a glass cover slip, and imaged using a Cytation 5 Multi-Mode Reader from BioTek (Winooski, VT, USA) with excitation/emission values of 494/517 nm (green channel) and 528/617 nm (red channel). Three regions from each slide were captured and percent cell viability was calculated as [# of live cells (green)/# of dead cells (red) X 100%]. Select stained samples were also imaged without construct degradation for a quick verification of channeled geometry fabrication.

For a quantitative representation of cell number and proliferation, a PrestoBlue metabolic assay (Invitrogen) was conducted on days 1 and 7 with six replicates from each group. Each sample was incubated in a 1:10 PrestoBlue-to-media solution at 37 °C for 45 min. Four 200 µl aliquots were taken from each sample and pipetted into a 96-well plate. Bottom-read fluorescence intensity was measured from the 96-well plate at 560 nm excitation, 590 nm emission, and 10 nm bandwidth using a Cytation 5 Multi-Mode Reader (BioTek).

Monoclonal cluster of differentiation 31 (CD31) antibodies conjugated with SuperBright 436 from eBioscience (San Diego, CA, USA) and platelet-derived growth factor receptor beta (PDGFRB) antibodies conjugated with PE from Life Technologies (Carlsbad, CA, USA) were chosen to verify CD31 and PDGFRB protein expression, commonly used EC and PC markers, respectively. On day 7, six replicates from each group were washed three times with PBS and fixed with 4% (wt./vol.) paraformaldehyde for 10 min in an incubator at 37 °C and 5% CO₂. The 4% paraformaldehyde was apirated from each sample and they were again washed three times with PBS. Each sample was then incubated with 0.1% (wt./vol.) Triton X-100 from MP Biomedicals (Santa Ana, California, USA) for 15 min to permeabilize the cell membranes. Samples were washed three more times with PBS before soaking in 2% (wt./vol.) bovine serum albumin (BSA) from Fisher Scientific (Waltham, MA, USA) at room temperature overnight to block excess protein binding sites. The 2% BSA was removed before adding the anti-CD31 and PDGFRB antibody solutions diluted in 500 µl of 0.1% BSA to each well in the dark, leaving the plate in a 4 °C refrigerator covered in foil overnight. The next day, the combined antibody solution was removed and samples were washed three more times with PBS. Samples were finally stored in PBS at 4 °C covered in foil until confocal microscopy.

Samples were imaged using a 10X objective on a Nikon AXR (Minato City, Tokyo, Japan) confocal microscope, with a 405 nm excitation laser and 423–476 nm emission range for CD31, and 488 nm excitation laser and 503–660 nm emission range for PDGFRB.

The CEVIC device was shown to successfully extrude adjacent, alternating channels of two hydrogel inks within sheet geometries. Sheets were fabricated containing 8, 16, 32, 64, 128, 256, and 512 alternating internal channels, with average channel widths ranging from 621.5 ± 42.92% µm to 11.67 ± 14.99% µm, respectively (figures 5(A)–(G)). While the largest and smallest channel widths had the highest and lowest relative standard deviation and percent error, respectively, there was no trend in these values across all channel widths. The average percent error between measured and theoretical channel widths was calculated to be 10.74%. The measured channel widths were relatively close to their 'theoretical' widths (e.g. an 8-channeled 5 mm sheet should average 625 µm-thick channels) (figure 7). The smallest channel width was measured as 10 µm within a 512-layered sheet. It is possible that smaller channel widths could be achieved in future work. These constructs demonstrate the CEVIC device's ability to pattern alternating channels into sheet constructs at resolutions in the range from surgically manipulatable small-diameter arteries and veins (i.e. ∼1–6 mm) to microvasculature (i.e. <100 µm) down to capillaries (i.e. ∼10 µm) [23, 24]. Channel orientation was also switched between vertical and horizontal by rotating the fanning outlet 90 degrees (figure 5(J) and (K)). This capability could allow for multi-cellular, multi-layered sheets of tissue (e.g. skin, organ walls) with a gradient of layer patterning to be produced with a single extrusion.

Zoom In Zoom Out Reset image size

Mid-extrusion material switching from a single printhead was also demonstrated with traditional filament extrusion (figure 5(L)). Additionally, mid-extrusion pattern switching was demonstrated by transitioning from 8, to 16, then 32 channels within a continuous hydrogel sheet (figure 5(M)). This capability can allow for a single, surgically manipulatable hydrogel sheet to contain complex micropatterning reminiscent of natural tissue structures such as the hierarchically branching microvascular tree, a well-documented challenge for bioprinting [41]. The transition lengths between 8, 16, and 32-channel regions achieved with manual valve switching were approximately 1 cm. Complete automation of valve switching, as well as reducing flow rate, should allow for shorter transition lengths to be obtained. This is the subject of an ongoing study.

Sheet dimensions were measured on average to be 4.78 ± 0.07 mm wide by 1.22 ± 0.33 mm thick. Sheets tended to be thickest in the center, bloating beyond the 0.5 mm outlet thickness, while thinning to their edges. However, this variation was likely driven by wetting between the hydrogel and polystyrene well plate lid being used as a printing surface for demonstration purposes. This effect may have contributed to the 8-channel sheet having the highest variation in measured channel widths, as channels toward the middle of the construct bloated to become wider than those toward the construct edge. Sheets with more uniform thickness could be produced with more rapid crosslinking upon deposition (e.g. UV and/or CaCl₂ exposure while printing takes place) or by using a printing surface with higher affinity for the hydrogel. Early tests of extruding sheets directly into bulk CaCl₂ solutions produced more consistent thicknesses (not without expected elastic swelling inherent to extruding hydrogels into crosslinking solutions [42]) but were more prone to disrupted flow. This justifies the method of depositing sheets before subjecting them to curing. Other existing strategies for improving structural stability, such as gelatin embedding [43], could be integrated with the CEVIC device in future studies.

Sheets were also produced with HEC fugitive ink that diffused away to leave vacant channels in between solid hydrogel channels. Orange-dyed water injected into vacant channels was able to exit the opposite end of the sheet, demonstrating their perfusability (figure 6). Not every vacant channel was shown to be perfusable with this method, likely due to channel sinking, inexact needle placement, and variable flow rate. However, even vacant channels that are not fully perfusable immediately upon fabrication could provide a framework for in vitro development of perfusable, endothelialized microvascular structures [44]. Further study is planned to investigate and optimize printed structures for developing prevascularized constructs.

Zoom In Zoom Out Reset image size

Zoom In Zoom Out Reset image size

PCL MEW scaffolds and 16-channelled hydrogel sheets were successfully fused into single constructs (figure 5(H)). While MEW has been combined with extrusion-based and inkjet bioprinting in past work [45], this is the first study combining MEW with chaotic printing, as well as with extrusion-based hydrogel sheet printing. Combining the two technologies allows for the tuned extracellular environment of hydrogels to be mechanically reinforced by the melt electrowritten woven fiber network [46]. Using the CEVIC device to deposit a channeled hydrogel sheet over a prefabricated MEW scaffold allowed the uncured hydrogel to first penetrate the MEW scaffold pores before the curing process effectively bound the two materials together. The uncured hydrogel was drawn into space within the MEW pores via capillary action between the hydrogel and PCL fibers. With larger pores (e.g. 1 mm), this disturbed enough hydrogel volume to disrupt channel structures within the deposited sheet. However, sheets deposited onto MEW scaffolds with 350 µm pores or less maintained their channel patterning until curing. We are working to combine CEVIC and MEW printheads on one robotically controlled 3D printer to allow production of hybrid constructs within a single fabrication period.

Viscosity measurements across varying shear rates for the hydrogel inks utilized in this study are presented in figure 8, providing a useful reference to compare gel viscosities while choosing materials as combined inputs for the CEVIC device. Gels inputted into the device must have relatively similar viscosities to successfully flow adjacently (i.e. not through or past each other) during channel production. Further investigations of varied concentration combinations could quantify a threshold for required viscosity 'percent similarity' between each device input. All inks demonstrated shear-thinning behavior, an important ink characteristic for extrusion-based bioprinting to allow flow through the printhead with minimum shear stress on cells [47]. The 4% (wt./vol.) SA and 0.8% HEC inks used for acellular testing had relatively comparable viscosities across all shear rates, justifying the choice of 0.8% as a concentration for HEC to be extruded adjacent to 4% SA to produce perfusable channels alongside SA channels. Likewise, the 3% GelMA—2% SA ink compared more closely to the 0.4% HEC fugitive ink flowed concurrently in the cell experiment. Interspersing cells into the inks raised viscosity in each tested material, but apparently not enough to disrupt the adjacent flow profiles.

Zoom In Zoom Out Reset image size

Average percent cell viability was measured to be 92.47 ± 0.02% and 66.43 ± 0.12% on day 1 for the experimental and control groups, respectively. On day 7, it was measured to be 94.48 ± 0.01% and 75.48 ± 0.06% for each respective group (figure 9(E)). These results indicate strong viability across the week of culture for the experimental group containing both hPCs and HUVECs, although the viability of each cell type could not be distinguished within the experimental group with the methods used. By contrast, viability was significantly lower in the HUVEC-only control group. This suggests that the endothelial seeding method attempted in this proof-of-concept was not an efficient way to seed HUVECs into the constructs. It is expected that the fugitive ink diffused out of the constructs quickly during exposure to CaCl₂, washing most of the HUVEC population away before attachment could occur. As a result, the HUVEC populations in both groups were much lower than intended, contributing to the observably lower viability in the control group. Future efforts should attempt to seed both hPCs and HUVECs within the solid channels surrounding vacant channels as an alternative to the endothelial seeding method. To potentially improve efficiency of the endothelial seeding method, future CEVIC device studies could experiment with higher-viscosity fugitive inks, temperature-induced fugitive ink dissolution [48], or strategies that do not require CaCl₂ immersion to cure the constructs (e.g. using UV-crosslinkable materials only). Viability would then likely be comparable between hPC and HUVEC populations. Early work seeding HUVECs alone in CEVIC-produced GelMA-SA constructs suggested strong HUVEC viability via fluorescent microscopy of calcein AM staining, although these preliminary cell viability results were not quantified (figure S1).

Zoom In Zoom Out Reset image size

This theory on ineffective HUVEC seeding is supported by the significantly lower metabolic activity readings in the HUVEC-only control group on both timepoints, as compared to the hPC and HUVEC combined experimental group (figure 9(F)). While these results suggest significant combined cell populations in experimental group constructs, there was no significant increase in net metabolic activity from day 1–7 in the experimental group. This suggests either low proliferation rate or high turnover rate within the experimental group constructs, despite their relatively high percent viability.

Based on these viability results, shear stresses associated with extrusion through the CEVIC device, as well as the 30 s 365 nm UV exposure, do not appear to influence cell viability to an extent that could be detrimental to the cellular functionality of resulting constructs. However, a more in-depth analysis of the device could be performed to correlate optimal cell viability and proliferation to the maximum shear stresses upon cells (e.g. determined via finite element modeling analysis of device geometries under various hydrogel flow conditions) and UV exposure (i.e. intensity and duration).

Calcein AM staining illuminated living cells under a fluorescent microscope to demonstrate cells positioned in solid hydrogel channels around vacant channels, as intended 1 d after printing (figure 9(A)). CD31 and PDGFRB protein expression was observed from HUVECs and hPCs, respectively, in the constructs at day 7 via confocal microscopy. Both cell types are observed surviving in proximity to each other around vacant channels on day 7, despite a visibly lower number of HUVECs due to the endothelial seeding method employed (figures 9(B)–(D)).

The images shown in figure 9 are generally representative of all the samples observed, although some variations do occur across samples and between each channel. These variations are likely attributed primarily to the timespan where sheets have been printed but are not yet solidified via curing. During this time, small fluctuations can occur in the still-fluid hydrogel and fugitive inks. Additionally, sheets with vacant channels may be prone to channel collapse without sustained flow through the channels during the culture period. Fluctuations appeared to have been more substantial with the lower viscosity inks and vacant channels of the in vitro cell experiment, as compared to the acellular tests and preliminary in vitro tests of seeding HUVECs into solid channels within a higher viscosity GelMA-SA hydrogel (figure S1). These fluctuations could likely be reduced by providing instant UV and CaCl₂ exposure upon deposition in future iterations of the CEVIC device, as well as by quickly providing media flow during in vitro culture.

No significant observations were made regarding changes in cell number or morphology across the 1-week culture period. However, it is likely that a longer culture period (e.g. 28 d) will be necessary to fully investigate cell behavior in CEVIC-produced constructs. Future in vitro experiments should investigate how to best align parameters for inducing the development of vessel-like cell structures in CEVIC-produced constructs, such as growth factor doses/combinations (in nutrient media or within the channels), cell seeding densities, hydrogel concentrations, and channel widths, as well as exploring other relevant biomaterials and cell types (e.g. mesenchymal stem cells, EC and PC sources better specified to tissue application of interest) [16]. It should be noted that promoting some of these parameters may require consideration of the impacts on others (e.g. increasing cell density could increase hydrogel viscosity to the point of producing high shear stresses during printing that decrease cell viability, thus necessitating a decrease in hydrogel concentration). Additionally, it would be relevant to run a small animal model that compares in vivo vascularization results with constructs that are implanted directly after printing vs constructs that are 'pre-cultured' in vitro to form functional microvasculature prior to implantation. A pre-implantation culture period is expected to be beneficial for producing a sheet construct that can form anastomoses with adjacent host vessels as quickly as possible, or even suture directly to host vessels and receive immediate perfusion (i.e. in an ideal outcome for this application). However, culturing time introduces drawbacks such as opportunity for contamination, cell senescence, and treatment delay (i.e. in the event of an eventual CEVIC-based clinical therapy). Such optimization work is well-justified by this proof-of-concept in vitro experiment, which demonstrates that the CEVIC device can produce viable multicellular constructs with microvascular cell types seeded around vacant microchannels.