Direct 3D cell-printing of human skin with functional transwell system

Polycaprolactone (PCL) (FDA-approved, molecular weight = 70 000–90 000 g mol⁻¹) was supplied by Polysciences (Warrington, PA, USA). Gelatin powder from porcine skin (Sigma-Aldrich, St. Louis, MO, USA) was dissolved in distilled water (DW) at 60 °C to make solutions with various concentrations of gelatin hydrogel. For the extracellular matrix (ECM)-based dermal compartment, various concentrations (w/v %) of collagen type 1 hydrogels were prepared by dissolving lyophilized collagen sponge derived from porcine skin (Dalim Tissen, Seoul, Korea) in 0.02 M of acetic acid solution and then neutralizing with cold 0.5 M NaOH solution, while avoiding gelation and bubble formation.

We developed the ICBS as a combinative platform with positional accuracy and repeatability of ±2.5 μm and ±1 μm in the x and y motion directions and ±10 μm in the z motion direction. This platform accommodated two dispensing modules: an extrusion-based module and an inkjet-based module. The system provides six individually controllable heads through which nine different biomaterials can be used. Collectively, the ICBS can use five biomaterials with the extrusion-based dispensing module and four biomaterials with the inkjet-based dispensing module. With the extrusion module, biocompatible synthetic polymers could be melted with a heating system (TCD-200EX, Iwashita Engineering, Tokyo, Japan) and extruded. Pneumatic pressure was adjusted with a controller having a range of 30–500 kPa (SuperΣCMII-V5, Musashi Engineering, Tokyo, Japan). Hydrogels were also extruded, using another type of pneumatic controller system with a pressure range of 5–200 kPa (SuperΣCMII-V2, Musashi Engineering, Tokyo, Japan) that enables us to print the hydrogel at a higher resolution than with the pneumatic controller with the range of 30–500 kPa. The thermo-sensitive hydrogels could also be regulated, using a home-made cooling system. Meanwhile, four drop-on-demand PH-46 inkjet heads (MicroFab, Texas, USA) and fluid reservoirs were combined into our cell-printing system, together with fittings and tubing. The four-channel pressure controller was then connected to the components to maintain a slight adverse pressure for optical jetting. Unlike the pneumatic controllers triggered by static pulses, the JetDrive III controller (MicroFab, Texas, USA) is operated under dynamic pulses capable of producing droplets continuously. Therefore, to trigger the JetDrive III controller in the ICBS, a pulse generator was added to it, to convert static pulses into dynamic pulses.

Human primary dermal fibroblasts (HDFs) and epidermal keratinocytes (HEKs) were purchased from ATCC (Manassas, VA, USA). The HDF samples were cultured in Dulbecco's Modified Eagle Medium (DMEM) with 10% fetal bovine serum (FBS) and 1% penicillin-streptomycin (P/S) solutions. The fibroblast growth medium was changed every other day. The HEK samples were cultured in an M154CF (Gibco, UK) medium containing 1% human keratinocyte growth supplement (HKGS) (Gibco, UK), 0.07 mM CaCl₂, and 1% P/S. The keratinocyte growth medium was also changed every other day. The cells were split before they reached 80% confluency. In this experiment, 2.5 × 10⁵ cells ml⁻¹ of HDF at passage 8–10 and 1 × 10⁶ cells ml⁻¹ of HEK at passage 3–4 were prepared for cell-printing. The HDFs were suspended in 1/10 volume of 10× reconstitution buffer with the total amount of pregel and resuspended in collagen hydrogel. Distilled water was then added to the prepared collagen bioink, so that the final collagen concentration reached 2% (w/v). The HEK samples were suspended in the keratinocyte growth medium for printing due to their preference for exposure to air over submersion in an ECM [15]. All components—including syringes, tubing, nozzles, and reservoirs—were sterilized with a 70% alcohol and distilled water mixture and dried under the culture hood with a UV lamp. The collagen bioinks were loaded in a chilled plastic syringe, avoiding any bubbles. The HEK-populated medium was loaded into a 2 ml plastic reservoir for printing.

Considering the inherent microenvironment of native skin cells [15, 17], we applied the extrusion-based dispensing module for the dermis and the inkjet-based dispensing module for the epidermis. In this way, the benefits of each module in the engineering of the dermis and epidermis were verified. Granular PCL was loaded in a stainless steel syringe to fabricate a supportive mesh. The engineered dermis was fabricated as follows. First, the loaded PCL was heated to 80 °C and extruded with compressed air; the PCL mesh of 22 mm diameter, 400 μm height, and 100 μm pore size was fabricated; and then a 2% solution of collagen bioink was sequentially dispensed onto the mesh. Finally, the entirety of the construct was achieved, of at least 22 mm diameter and 2.5 mm height, including the height of the PCL mesh. We additionally fabricated two dermal constructs as groups for comparison; one was made of PCL anchor and fibroblast-populated collagen, and the other of fibroblast-populated collagen only. The experimental groups are described later in more detail, in figure 3(A). The contraction was next observed for 7 days because it is known to occur within a week [18]. For qualitative analysis, the constructs were imaged inside their respective wells in 12 well plates at 1, 3, 5, and 7 days of culture. For quantitative analysis, the top surface areas of imaged samples were determined through the border using ImageJ software [19].

To verify the superiority of the inkjet-based dispensing module in engineering epidermis layers, two approaches were tested (printed versus manual). A quantity of 1 × 10⁶ cells ml⁻¹ of HEK was loaded in the cell culture medium and uniformly onto the premade collagen matrices. For manual seeding, 10 μl of 1 × 10⁶ cell suspension was added onto the collagen matrices as well [7, 18]; that is, the same number of cells that was used and located onto the collagen matrices, in order to compare our method with the existing manual method. Bright field images were captured every 12 h with a microscope (CX41, Olympus, Japan) until 100% confluency was reached. In addition, the cell proliferation rate was evaluated for 60 h using Cell Counting Kit-8 (CCK-8, Dojindo, Kumamoto, Japan). The CCK-8 stock solution was diluted (1:10) in cell culture medium to constitute a working solution, in which the samples were submerged for 150 min. Then, 100 μl of the submerged solution was extracted from the cell culture plates and moved into a plate of 96 wells. The extracted solution was analyzed through measurement of the absorbance at 450 nm, using a microplate reader (Asys UVM 340, Biochrom, Cambridge, UK). The samples were washed in phosphate-buffered saline (PBS) solution until the working solution was completely removed from the samples, followed by incubation with fresh medium.

We also examined the morphology of the keratinocytes tested in the two approaches (printed versus manual) using immunofluorescence staining. It was assumed that printed keratinocytes could express a tight junction (E-cadherin) with a more uniform distribution than the manually-seeded keratinocytes because of the precise nature of the inkjet module. After fixation, permeabilization, and blocking, the samples were stained with rabbit antihuman E-Cadherin antibody (Abcam, Cambridge, MA, USA) diluted 1:100 in a 1× PBS solution containing 1% bovine serum albumin (BSA) for 1 h at room temperature. A secondary antibody of Alexa Fluor 488 goat antirabbit antibody (1:100; Invitrogen, CA, USA) was subsequently submerged in the samples for 1 h at room temperature and then counterstained with DAPI solution diluted 1:1000 for 5 min. Finally, the stained samples were imaged using confocal laser scanning microscopy (Olympus FV 1000, NY, USA).

A live/dead cell assay kit (Invitrogen, CA, USA) was used to assess cell viability and morphology. The samples were submerged in dye solution composed of ethidium homodimer (EthD-4 μM) and calcein AM (2 μM) in a 1× PBS solution, followed by incubation at 37 °C for 30 min. A fluorescent microscope (Axiovert 200, Zeiss, Jena, Germany) was utilized to capture the stained samples. The live cells appeared in green and the dead cells in red.

In the ICBS, the extrusion-based dispensing module was applied to the fabrication of the transwell system and engineered dermis. The inkjet-based dispensing module was used to uniformly place keratinocytes onto the engineered dermis. The transwell system indicates a 3D PCL construct that enables us to maturate our skin model without additional commercial products. Our strategy was performed as follows. Molten PCL in a 10 ml steel syringe was dispensed through a steel nozzle of diameter 250 μm through pneumatic pressure of 700 kPa; then, 25% gelatin hydrogel loaded in a 10 ml plastic syringe was filled within the internal pores of the fabricated PCL framework, using an extrusion module. This process was repeated until it reached the predesigned dimensions; in sequence, PCL mesh was fabricated onto the hybrid 3D PCL construct. Lastly, HDF-populated 2% collagen bioink was extruded onto the mesh while fabricating a PCL wall that could block medium penetration during air-liquid interface conditions. This process was repeated until the wall height reached at least 3.5 mm. The optimum concentration of gelatin hydrogel was determined from the gel with its intended functional performance being that it could support the pregel collagen for at least 30 min, which is the minimum time needed for crosslinking collagen at 37 °C (figure S1 is available at stacks.iop.org/BF/9/025034/mmedia for supplementary information). While incubating the collagen, the HEK-containing medium was suspended in a plastic reservoir and uniformly distributed onto the cross-linked dermis using the inkjet-based dispensing module. The final construct was submerged in the coculture medium composed of DMEM/F12 (3:1) supplemented with 0.4 μg ml⁻¹ hydrocortisone, 10 ng ml⁻¹ cholera toxin, 180 μm adenine, 5 μg ml⁻¹ human recombinant insulin, 5 μg ml⁻¹ human apo-transferrin, and 5 μg ml⁻¹ triiodothyronine, followed by incubation in a humidified atmosphere of 5% CO₂ at 37 °C. During this incubation period, the sacrificial hydrogel was dissolved and completely removed by washing with PBS solution. The cell culture media was changed every other day for 3–5 days under submerged conditions. The construct was next grown at the air-liquid interface by adjusting medium levels. The dermis and epidermis compartments were also bioprinted without any supportive PCL mesh or transwell system. The compartments were then cultured in commercial transwell inserts (poly (ethylene terephthalate) membranes, 3 μm pore size, Corning, USA) inserted in a 6-deep well plate (Corning). This is a process that may have common ground with existing methods for cell-printing in producing human skin models [13, 20]. We therefore defined this as 'existing fabrication'.

For histological analysis, the matured skin tissue models were fixed in 4% paraformaldehyde solution for 1 h and were then embedded in a paraffin block after dehydration and washing. The paraffin blocks were sectioned (10 μm) and used for an H&E staining process. The staining was performed based on a standard protocol [20]. Specifically, the 10 μm sections were stained with hematoxylin solution at room temperature for 8 min and washed several times with tap water. The samples were sequentially counterstained with eosin Y solution for 5 min. Dehydration was next performed through immersion in 95–100% ethanol. Slides were then mounted with a xylene-based mounting medium. The stained samples were finally examined using an optical microscope (CX41, Olympus, Japan).

The schematic diagrams in figure 1 show the entire procedure for fabricating 3D-printed human skin models with a functional transwell system in a single-step process. First, we developed a 3D cell-printing system that we named the Integrated Composite tissue/organ Building System (ICBS). It is a combinatory fabrication platform that can simultaneously run extrusion and inkjet-based dispensing modules. The 3D CAD images were drawn using a 3D modeling software package (Solidworks, Concord, MA, USA) (figure 2(A)). The hardware of ICBS was initially designed according to the working principle of an extrusion-based dispensing module. A home-made pulse generator was connected to the system to manipulate an inkjet-based dispensing module at the same time. The detailed descriptions regarding the system are shown in figure 2(B). Fundamental experiments on each dispensing module were carried out under various fabrication conditions for leveraging our cell-printing system with optimal printing conditions (supplementary information figures S2 and S3).

Zoom In Zoom Out Reset image size

Zoom In Zoom Out Reset image size

We fabricated three different collagen-based constructs using the extrusion-based dispensing technique (figure 3(A)). To examine the effects of the supportive PCL mesh on construct shrinkage, the samples were cultured in a fibroblast culture medium for 7 days and imaged at days 1, 3, 5, and 7 (figure 3(B)). A graph was plotted for quantification analysis (figure 3(C)). Collectively, these results revealed only a contraction of 32.1% ± 0.5% for the fibroblast-populated collagen in the initial area after 7 days, whereas the fibroblast-populated collagen with the PCL anchor and mesh showed much smaller contraction values of 77.4% ± 0.3% and 89.4% ± 0.6%, respectively. More importantly, the collagen with PCL mesh displayed not only the minimum contraction rate but also a more stabilized cellular morphology than the other groups (figure 3(D)). These results indicate that the collagen with PCL mesh could stabilize our skin tissue model during tissue maturation, potentially leading to improved morphology and long-term maintenance [21].

Zoom In Zoom Out Reset image size

The inkjet-based dispensing module was used to enhance the epidermis formation process through positioning keratinocytes with high spatial resolution. To verify its potential feasibility, two approaches were examined (printed versus manual). On a microscopic level (figure 4(A)), we observed that the printed keratinocytes provided a further uniformly distributed area after 12 h compared with the manually-seeded keratinocytes arranged in clusters. The printed keratinocytes showed a uniformly distributed proliferation rate throughout the collagen matrices, leading to nearly 100% confluency within 60 h. Our quantitative analysis also revealed that the printed keratinocytes provided a much higher proliferation rate, probably because the uniform distribution could enhance cell-to-cell communication (figure 4(B)). Furthermore, we investigated morphology characters of HEKs before the formation of epidermis layers, which is the presence of a tight junction that plays a central role in the process of epithelial morphogenesis [15]. The samples were stained for the adhesion protein E-cadherin antibody at day 3. We observed that the junction protein was highly visible in both approaches (figure 4(C)). More importantly, the protein of printed cells was evenly expressed throughout the collagen matrices, whereas the empty region was observed in the samples prepared by the manual method. This observation indicates that the inkjet-based dispensing module may be capable of positioning cells with higher spatial resolution, leading to enhancement of keratinocyte differentiation.

Zoom In Zoom Out Reset image size

Zoom In Zoom Out Reset image size

Taking these features of each module together, we established a cell-printing strategy to produce 3D human skin tissue models with a functional transwell system in a single-step process (figure 5). The characterization of our transwell system and commercial transwell insert was first compared in air-liquid interface conditions to verify its feasibility (figure S4 of supplementary information). The observation indicated that our transwell system could also maintain air-liquid interface conditions. In addition, we revealed that human skin models of varying shapes could also be fabricated in accordance with our potential purpose through designing our transwell system (figure S5 of supplementary information). This strategy was also highly cost-effective, with a 98 percent reduction in costs compared with the use of commercial transwell inserts and a 90 per cent reduction in the amount of medium used.

After culturing our tissue model in the air-liquid interface conditions for 14 days, the H&E-stained sections at days 7 and 14 were examined (figure 6(A)). The histological appearance showed that fibroblasts in the matured skin were stretched over day 14, comparable to what is observed in a native dermis [17]. This appearance also revealed that keratinocytes were stratified at day 14. The thickness of stratified epidermis (97 ± 3 μm) also remained within the range of native human epidermis (75–150 μm) [22]. The results indicate that our transwell system could successfully make matured skin tissue models. More importantly, our bioprinted skin model included significantly improved structural features of the dermis and epidermis when compared with the skin model fabricated based on the existing method, although both were comparable in epidermis thickness (figure 6(B)). Moreover, we observed expressions of proteins of 3D cell-printed dermis and epidermis by examining representative respective markers; collagen type I was produced in matured dermis and early/late differentiation markers were expressed in stratified epidermis in our system (figure 6(C)). Collectively, it should be noted that our bioprinted skin model could include structural morphology similar to that of native skin.

Zoom In Zoom Out Reset image size