Preparation of the bioink

Here, we used our previously developed (HA)-based hydrazone crosslinked bioink [14] with modifications optimal for the bioprinting of cornea endothelium. The synthesis and characterization of crosslinking components, that is carbohydrazide (CDH) conjugated dopamine modified HA (HA-DA-CDH) and aldehyde conjugated HA (HA-ALD), has been previously described [26, 27]. In this study, the crosslinking components were dissolved into 1X Dulbecco’s Phosphate Buffered Saline (DPBS) into concentration of 6 mgml−1. Unmodified sodium hyaluronate (Novamatrix, Norway) with a concentration of 1% (w/v) was used as a primary rheological modifier. Human collagen IV (Col IV) from human placenta (Sigma–Aldrich St. Louis, MO, USA) and human recombinant laminin 521 (LN521™, Biolamina, Sweden) were used as additional rheological modifiers in the bioink and to improve the cytocompatibility and functionality for hPSC-CEnCs. Col IV was dissolved into dilute acetic acid (Merck, Germany) to obtain a solution of 2.5 mgml−1. Just before mixing the bioink, the Col IV was neutralized into pH 7 with 1 M NaOH and 10 × DPBS (Carl Roth, Karlsruhe, Germany). LN521™ was used in a concentration of 0.1 mgml−1. The mixing of the bioink components has been previously described [14]. The bioink was crosslinked at room temperature for 90 min before 3D bioprinting.

Shear thinning

Shear-thinning properties of the bioinks with and without cells were analyzed with viscosity measurements to determine if cells affect the viscosity. Viscosity was measured with HR-2 Discovery hybrid rheometer (TA Instruments, DE, USA) under continuous flow using 20 mm parallel plate geometry, 1 mm gap and shear rate ranging from 0.01 to 10 1 s−1. Bioinks were prepared as previously described [14] and pre-crosslinked in the syringes for 100 min. The cell density of the cell-laden bioink was 6 million cellsml−1. After pre-crosslinking, 400 µl bioink samples were injected onto the rheometer bottom plate for measuring. The measurements were carried out within 80 min after starting the first measurement, and five replicates per bioink were measured (n = 5).

3D bioprinting setup

Extrusion-based 3D bioprinter 3D-Bioplotter® Manufacturer Series by EnvisionTEC (Gladbeck, Germany) was used. Before printing, the bioink was placed in a 30 cc Nordson EFD syringe barrel (OH, USA). Small 32G blunt needles with 100 µm inner diameter were chosen for printing (0.50 inches, CellInk, Sweden, Gothenburg). All the 3D bioprinting experiments were carried out at room temperature. 3D models in stl format were prepared with Perfactory RP Software (EnvisionTEC, Gladbeck, Germany) and printing patterns and parameters were adjusted in Visual Machine (EnvisionTEC, Gladbeck, Germany). Slice interval of 70% of the nozzle inner diameter was used for all printed structures, and the distance of the first layer was set at 0.07 mm.

Two patterns were used for printing hPSC-CEnCs. First, line patterns were used to study the patternability of hPSC-CEnCs. Here, two layers of aligned lines with 1.3 mm distance between lines were printed using printing parameters of 0.4 bar and 13 mms-1 speed. To further study the 3D bioprinting of hPSC-CEnCs, two layers of hPSC-CEnC containing bioink were printed onto fibrin membrane in a form of a low cylinder with a diameter of 10 mm. Line distance of 0.30 mm was used, and the adjacent layers were printed at 90° angle to the previous layer. Printing parameters of 0.6 bar and speed 10 mms−1 were chosen. For these studies, hPSC-CEnCs density in the bioink was 10 million cells ml-1. The fibrin membrane used as printing substrate was prepared from a commercially available two-component surgical fibrin sealant (Tisseel, Baxter, IL, USA). Before use, the fibrinogen component was diluted 1:1 with 2.9% NaCl (VWR), 2.6 mM CaCl2 (Sigma-Aldrich) and the thrombin component 1:166 with 1.1% NaCl; 1 mM CaCl2. For fibrin formation, the fibrinogen and thrombin components were combined, pressed onto a bioassay dish with the help of two sterilized parafilm strips to form a membrane, and left to jellify for 30 minutes RT.

Shape fidelity

The uniformity and shape fidelity of deposited HA-based bioinks was investigated to evaluate the printability. To explore the shape fidelity of the HA-based bioink, the filament thickness and pore factor (Pr) of printed grids were determined as a function of time. HA-based bioink was prepared as described above and allowed to pre-crosslink for 90 min. Grids of two printed layers with dimensions of 20 mm × 20 mm were 3D bioprinted onto 35 mm dishes. Distance between lines of 2.50 mm were used and filaments in alternative layers were printed perpendicularly. Printing pressure of 0.4 bar and speed of 17.0 mms−1 was used. The samples were imaged with a high-definition CCD-camera attached to the dispenser head mount immediately after printing and after 7 days submerged in PBS at + 37 °C. The thickness of the printed hydrogel filaments and pore geometry was quantified with Image J image processing and analysis software. For each printed sample (n = 6) and time point, the filament thickness of 2 adjacent layers was evaluated from 9 different filaments. Six randomly selected pores in each printed sample (n = 3) were included in image analysis. Pr was counted according to the following equation, where L means perimeter of the pore and A the pore area.

The statistical data analysis of shape fidelity measurements was determined with non-parametric Mann–Whitney U test (IBM SPSS Statistics software). P-values ⩽ 0.05 were considered statistically significant.

Swelling behavior and degradation of the bioink

To study the degradation and swelling characteristics of the bioink in in vitro setting, three parallel bioink samples (n = 3) were subjected to neutral pH conditions in PBS. 200 µl samples of bioink was injected onto 35 mm diameter dishes. The samples were allowed to stabilize for 90 min and subsequently submerged into PBS. After 30 min, the PBS was removed, and the initial weight of the samples was recorded. Thereafter, the bioink samples were incubated in PBS at + 37 °C. 1, 3, 7 and 10 days of post-printing, the PBS was removed thoroughly from the samples, and weight of the remaining samples was recorded. The swelling behavior of the HA-based bioink was analyzed by calculating the remaining weight percentages of the samples at each timepoint according to the following formula:

$$\mathrm=\frac}}\times 100$$

Differentiation and cryopreservation of hPSCs-CEnC-like cells

The previously established and characterized hPSC lines were used in this study including, human embryonic stem cell (hESC) line Regea08/017 [28] and human induced pluripotent stem cell (hiPSC) line WT001.TAU.bB2 [16]. The hPSC-CEnCs used for the experiments were produced mainly with the hESC line unless indicated differently. CEnC-like cells were differentiated from hPSCs based on our previously established protocol [16] with small retinoic acid (RA) concentration modifications (Fig. 1B). Briefly, hPSC differentiation was carried out on laminin-521 coated 6 or 12 well-plates (Corning) at 20,000–40,000 cells cm−2. The hPSCs were cultured in E8 medium for 24 h. On day 0 of differentiation, the medium was switched to serum free basal medium consisting of KnockOut Dulbecco’s Modified Eagle Medium (KO-DMEM), 15% Knock-out serum replacement (KO-SR), 2 mM GlutaMax-I, 0.1 mM 2-mercaptoethanol, 50 Uml−1 Penicillin/Streptomycin (all from Thermo Fisher), 1% Non-essential Amino Acids (Thermo Fisher, Sigma-Aldrich) supplemented with 10 µM TGF-β inhibitor SB431542 (SB; Stemcell), 4 µM GSK3 inhibitor/WNT pathway activator CHIR99021 (CHIR; Stemcell) and 10 µM RA (Sigma-Aldrich). RA concentration was lowered to 5 µM during the days 3–6 and then lowered further to 1 µM. During the days 7–9, CEnC-like cells had formed, and they were cryopreserved. For cryopreservation, cells were harvested using TrypLE Select (Thermo Fisher) dissociation enzyme and incubated for 3–6 min at + 37 °C. Subsequently, the CEnC-like cells were detached gently by trituration, filtered through 40 µm Cell Strainer (Thermo Fisher) and centrifuged for 5 min at 300 g. The cells were resuspended in cryomedium consisting of basal medium with 40% KO-SR and 10% dimethyl sulfoxide (DMSO, Sigma) in 2 ml cryogenic storage vial (Sarstedt).

After cryopreservation, the cells were thawed and passaged on LN521™ coated 6 or 12 well-plates at 200,000–300,000 cellscm−2 in basal medium supplemented with 10 µM SB, 4 µM CHIR, 1 µM RA and 10 µM ROCK inhibitor Y27632 (ROCKi, Stemcell). After 24 h, ROCKi was removed from the medium. Cells were cultured for 5 days and then harvested for bioprinting. Contrast light microscope Nikon Eclipse TE2000-S with a DS-Fi1 camera (Nikon Corp. Tokyo, Japan) was used to capture images of the cell morphology. The data in this study was produced from 8 individual batches of hESC-CEnCs and 2 individual batches of hiPSC-CEnCs.

Flow cytometry

Flow cytometry (FC) was used to quantify the expression of CD166 10 days from the beginning of the differentiation of the hPSC-CEnCs (WT001.TAU.bB2 hiPSC-line). For FC staining, cells were detached with TrypLE Select and Defined Trypsin Inhibitor (DTI, Thermo Fisher), centrifuged 5 min at 1000 rpm, resuspended in medium and counted. 1 × 105 cells/100 µL per sample were divided to 5 ml sample tubes and the cells were washed twice by centrifuging 2 min at 1900 rpm in FC wash buffer containing 0.5% BSA and 1 mM ethylenediamine-tetraacetic acid (EDTA; Gibco, Thermo Fisher Scientific) in DPBS. Next, the samples were incubated with PE-conjugated mouse anti-human CD166 primary antibody (#559,263, BD Biosciences, NJ, USA) for 20 min in the dark on ice. BD Pharmingen™ PE Mouse IgG1, κ Isotype Control (#555,749, BD Biosciences) was used along with unstained control. Samples were washed twice with the FC washing buffer and immediately analyzed with the CytoFLEX S Flow Cytometer (Beckman Coulter Life Sciences, IN, USA) collecting 10 000 events of the primarily gated population of interest. The collected data were further analyzed with the CytExpert Software. In CytExpert, the negative control was used to gate the population of interest containing the cells. After excluding doublets from the analysis, the negative vs. positive gates were set with histograms using 0.5% marginal. Finally, the established gates were copied to each sample of the experiment.

Rat cornea organ culture for the biocompatibility and integration test of the bioink

Corneal tissues (n = 6) were obtained from the eyeballs extracted from three euthanized Sprague Dawley rats. Briefly, the ocular surface was cleaned using saline solution before excision. For excision, the rat eyes were held on the bulbar conjunctiva using a toothed forceps and gently pulled outward. Thereafter, fornix conjunctiva was first excised all around the eye using sterile curved ophthalmic scissors with serrated tips. Later, intraocular muscles and optic nerve were gently excised to extract the eyeball. During the procedure, care was taken to avoid bleeding to prevent additional corneal contamination. Eyeballs were immediately placed in the basal medium and were transferred to a sterile hood. Further, the eyeballs were placed in the basal medium containing 400 units ml−1 of penicillin–streptomycin and were incubated for 30 min at 37 °C and 5% CO2.

For corneal extraction, eyeballs were placed in a petri dish containing medium to prevent drying. Using a stereo zoom microscope, a 21-gauge needle was used to pierce the rat retinal area, after which the peripheral cornea was nicked using a scalpel blade no. 15. Later, curved Vannas scissors with sharp tips were used to cut the corneal tissue, which were then transferred to a new petri dish with the medium. Rat corneas were treated with TrypLe Select over night to remove native corneal endothelium. Corneas were moved into 96-well plate wells with cut bottom part of sterile 1.5 ml Eppendorf tube to give the well U-shaped bottom which helps the cornea to stay endothelial side up. Next, hPSC-CEnCs in bioink were injected on three corneas (n = 3) and hPSC-CEnCs without bioink were seeded on the other three corneas (n = 3) at 300 000 cells cm−2 according to 96-well plate surface area. After that, hPSC-CEnCs were cultured 5 days in hPSC-CEnC differentiation medium with 1 µM RA and first 24 h also with 10 µM ROCKi at + 37 °C in 5% CO2. For analyses, two corneas with bioink and cells, and two corneas seeded with cells were fixed for flat mount immunofluorescence staining. In addition, one cornea from each sample type were fixed for cryosections and staining.

Porcine cornea organ culture for biocompatibility and integration test of the bioink

The corneal organ culture using excised porcine corneas was performed as previously described [10, 29] with minor modifications. Briefly, the excess tissue was stripped from the whole fresh porcine eyes after which they were disinfected with 2% povidone iodine (Betadine®, Leiras, Helsinki, Finland). Corneas were then dissected from the eyes in aseptic conditions and further disinfected in 1% povidone iodine. The corneas were put overnight in Advanced DMEM supplemented with 1% GlutaMAX™, 1% Penicillin–Streptomycin and 0.25 μgml−1 amphotericin B (Thermo Fisher Scientific) at + 37 °C in 5% CO2 to prevent bacterial contamination and then overnight in trypsin–EDTA (0.25%) (Thermo Fisher) to dissociate porcine primary CEnCs and corneal epithelial cells (CEpC) from the cornea. Remaining primary CEnCs and CEpCs were removed gently with blunt forceps without detaching the DM. Porcine corneas (n = 2) were put on 12 well plates DM side up and the bioink comprised with hPSC-CEnCs was injected to cover the whole DM. Corneas were cultured in the hPSC-CEnC differentiation medium with 1 µM RA and first 24 h with 10 µM ROCKi at + 37 °C in 5% CO2 for 10 days before freezing for cryosections.

Human DM from donor cornea for biocompatibility and integration test of the bioink

The human donor corneas (Regea tissue bank, Tampere University, Finland) not suitable for clinical use (n = 6) were stored in CorneaMax storage medium (Eurobio Scientific, France). To separate DMs from the corneal stroma, each cornea was handled in the following manner. Storage medium was injected using a 27-gauge needle into the corneal stroma underneath the DM to create a liquid filled bubble. The injection site was enlarged using the same needle to deflate the bubble and to prevent the membrane from bursting during cutting. Then, with an 8 mm biopsy punch, the DM was cut off and transferred to the DPBS solution. Long storage time (minimum of 3 months) and rinsing with DPBS had denuded the DM from primary CEnCs. DMs were moved to 24 well plate and they were pinned on the bottom of the well with CellCrown™ 24NX (Scaffdex, Tampere, Finland) or with small metal ring to prevent floating. The bioink comprised with hPSC-CEnCs was injected on the DM. As a control, hPSC-CEnCs were seeded without bioink on the DM with density of 150 000 cellscm−2. Cells on DM were cultured in the hPSC-CEnC differentiation medium with 1 µM RA and first 24 h with 10 µM ROCKi at + 37 °C in 5% CO2 for 6 days before fixing for immunofluorescence staining.

Bioink cytocompatibility with bioprinted hPSC-CEnCs

The cytocompatibility of the bioink was evaluated using LIVE/DEAD® viability/cytotoxicity kit with bioprinted hPSC-CEnCs on fibrin membrane. Briefly, the cells were washed with DPBS and incubated with Live/Dead staining solution containing 2 µM Calcein AM and 1 µM Ethidium homodimer diluted in DPBS in room temperature for 30 min. The cells were washed and imaged using a fluorescence microscope (Olympus IX51; Olympus, Tokyo, Japan) or confocal microscope (Zeiss LSM 800, Carl Zeiss AG, Germany). The percentage of dead cells was manually calculated from the images.

Immunocytochemistry of the hPSC-CEnCs

After differentiation, the phenotype of hPSC-CEnCs used in the experiments were analyzed with immunocytochemistry. For that, the cells were fixed with 4% paraformaldehyde (PFA, Sigma-Aldrich) for 15 min. Next, the cells were permeabilized for 10 min with 0.1% Triton X-100 (Sigma-Aldrich) followed by blocking with 3% bovine serum albumin (BSA) for 1 h. Then the cells were first incubated with 1:400 ZO-1 (#61–7300, Thermo Fisher) 1:200 Na+/K+-ATPase (#ab7671, Abcam, #55,187–1-AP, Proteintech, IL, USA), 1:400 CD166 (#559,260, BD Biosciences), STEM121 1:100 (#Y40410, Takara Bio Inc., Japan), 1:400 Ki67 (#AB9260, Millipore,F MA, USA), 1:400 Keratocan (sc-33243, Santa-Cruz, Dallas, TX, USA), 1:400 PAX6 (HPA030775, Sigma-Aldrich) primary antibodies and 1:200 phalloidin Alexa Fluor 647 (Thermo Fisher) overnight at 4 °C. The cells were next treated with 1:800 Donkey anti-Rabbit IgG Secondary Antibody, Alexa Fluor 488, 1:800 Donkey anti-Mouse IgG Secondary Antibody, Alexa Fluor 568 (both from Thermo Fisher) and 1:800 Donkey anti-Goat IgG Secondary Antibody, Alexa Fluor 647 (Abcam) according to the host of the primary antibody for 1 h at room temperature. The nuclei were counterstained with 1:1000 Hoechst 33,342 (Thermo Fisher) with secondary antibody incubation or with 4’,6-diamidine-2’-phenylindole dihydrochloride in mounting medium (DAPI; Vector Laboratories, Peterborough, UK).

The flat mount technique was employed for rat corneas. Briefly, three cuts were made with scalpel blade no. 10 around the cornea approximately 1/3 of the diameter of the cornea before mounting. Vectashield Antifade Mounting Medium (Vector Laboratories) was used for mounting. The images of stained cells were captured using a fluorescence microscope (Olympus IX51; Olympus, Tokyo, Japan) or confocal microscope (Zeiss LSM 800, Carl Zeiss AG, Germany) and prepared using image editing software (Adobe Photoshop CC 2021; Adobe Systems) or Zen 3.0 (Blue edition) (Carl Zeiss AG).

Immunohistochemistry of the cells on fibrin membrane, ex vivo rat and porcine corneas

The hPSC-CEnCs bioprinted on fibrin membrane, injected with bioink or seeded without bioink on ex vivo rat and porcine corneas were fixed with 4% PFA for 3 h and put into 30% sucrose overnight at + 4 °C. Then, the samples were put into Tissue-Tek® O.C.T. Compound (Sakura Finetek, USA) and kept overnight at + 4 °C. For snap freezing, samples were put vertically in cryomolds filled with O.C.T. Molds were then submerged into isopentane in decanter glass which was surrounded by liquid nitrogen. Frozen samples were stored at -80 °C. The frozen sample blocks were cut into 8 µm thick sections with cryostat (MEV + , SLEE medical GmbH, Germany). The sections on glass slides were stained for the evaluation of the tissue integration with hematoxylin and eosin (H&E) using standard protocols. Next, the sections were permeabilized with 0.1% Triton X-100. Sections were treated with blocking buffer consisting of 5% BSA for 1 h in moisture chamber at room temperature. Then the sections were first incubated with 1:200 ZO-1, 1:100 Na+/K+-ATPase 1:200 CD166 and anti-human cytoplasm STEM121 1:80 (for porcine and rat ex vivo sections) primary antibodies in blocking buffer in moisture chamber overnight at 4 °C. After 20 min DPBS wash, the sections were incubated in 1:400 Donkey anti-Rabbit IgG Secondary Antibody, Alexa Fluor 488, 1:400 Donkey anti-Mouse IgG Secondary Antibody, Alexa Fluor 568 and 1:500 Hoechst 33,342 in blocking buffer for 1 h in moisture chamber at + 37 °C. Sections were washed for 20 min in DPBS and mounted with Prolong Gold Antifade Mountant (Thermo Fisher). The images of mounted cells/tissue were captured using using a fluorescence microscope (Olympus IX51; Olympus, Tokyo, Japan) or confocal microscope (Zeiss LSM 800, Carl Zeiss AG, Germany) and prepared using image editing software (Adobe Photoshop CC 2021; Adobe Systems) and Zen 3.0 (Blue edition) (Carl Zeiss AG).

Transendothelial electrical resistance measurement

Transendothelial electrical resistance (TEER) of the hPSC-CEnCs was measured with Millicell electrical resistance system volt-ohm meter (Merck Millipore, Darmstad, Germany). TEER values were measured from hPSC-CEnCs cultured for 6 days on 24 well plate hanging cell culture inserts with pore size 1.0 µm (Sarstedt, Germany). TEER values were obtained from four different layouts: bioink with cells (n = 4), seeded cells without bioink (n = 4), blank empty insert (n = 4) and bioink without cells (n = 3). All samples were treated with identical medium changes. Measured TEER values (Ωcm2) were multiplied by the surface area of the insert (0.3 cm2). Average of the values with each layout were calculated and the TEER values of blank empty insert was subtracted from the result. The statistical data analysis of TEER measurements was determined with non-parametric Mann–Whitney U test (IBM SPSS Statistics software). A P-value ⩽ 0.05 was considered statistically significant.

Ca2+ imaging and cell size analysis

For Ca2+ imaging, the hPSC-CEnCs injected with bioink (n = 2) or seeded (100 000 cells cm−2) without bioink (n = 2) were cultured for 6 days on sterile 13 mm diameter glass coverslips (VWR, PA, USA). The ATP-induced Ca2 + responses was measured as described in [29,30,31]. Briefly, the cells were loaded with permeable Ca2+ sensitive fluorescent dye fluo-4-acetoxymethyl ester (1 mM, fluo-4 AM; Thermo Fischer Scientific) diluted in Elliot buffer solution (pH 7.4, osm 330 mOsm) for 30 min RT. After loading, the cells were washed with Elliot buffer for 10 min RT. During imaging, the cells were perfused with pre-warmed (approx. 37 °C) Elliot solution alone or Elliot containing 100 µM ATP (Sigma-Aldrich). Nikon Eclipse FN1 upright fluorescence microscope with water immersion 25 × objective (NA = 1.10) was used. The cells were imaged for 10 min which included 5 min of baseline imaging, 2 min of ATP stimulus and 3 min of additional imaging. All steps in the Ca2+ imaging were performed protected from light. Data analysis was performed with ImageJ and MATLAB (R2021a) as described previously [31]. Responsive cells were calculated from three randomly selected region of interests (ROI, 200 × 200 pixels) from each replicate to reach total of 6 analyzed ROIs per sample type.

In addition, cell numbers from the ROIs during the Ca2+ imaging were manually counted and used to compare the representative cell size of bioink and seeded samples. For additional quantification, cells injected with bioink and seeded without bioink were cultured on cell culture plastics for 6 days and imaged with Contrast light microscope Nikon Eclipse TE2000-S with a DS-Fi1 camera (Nikon Corp. Tokyo, Japan). For quantification, 14–30 ROIs (500 × 500 pixels) were chosen from the CEnC-like cell areas and the cells were counted manually. Furthermore, the average cell sizes in the cell cultures were measured using NucleoCounter® NC-200™ (Chemometec, Allerod, Denmark). This involved 27 individual hPSC-CEnCs in bioink and 7 seeded hPSC-CEnCs (both from Regea08/017 line), as well as 190 hPSC-CEnCs in bioink and 89 seeded hPSC-CEnCs (both from the WT001.TAU.bB2 line). All statistics were performed with Mann–Whitney U with GraphPad Prism (version 5.02) to compare statistical significances. A p-value of ⩽ 0.05 was considered statistically significant.

RNA extraction and real-time qPCR

Total RNA was extracted from hPSC-CEnCs injected with bioink and seeded without bioink on cell culture plastics (d6) using TRI reagent (Sigma-Aldrich). RNA concentration of each sample was determined using NanoDrop-1000 spectrophotometer (NanoDrop Technologies, Wilmington, DE, USA). RNA was purified from endogenous DNA using Dnase I (Thermo Fisher Scientific). From each RNA sample 400 ng were used to synthesize cDNA using the High-Capacity cDNA RT kit (Applied Biosystems, Foster City, CA, USA). The resulting cDNA samples were analyzed with qPCR using sequence-specific TaqMan Gene Expression Assays (Thermo Fisher) for ATCAM (CD166, Hs00977641_m1). All samples were run as triplicate reactions with the QuantStudio 12 K Flex Real-Time PCR System (Applied Biosystems). Results were analyzed with the QuantStudio 12 K Flex Software (Applied Biosystems) and Microsoft Excel. Based on the cycle threshold (CT) values given by the software, the relative quantification of each gene was calculated by applying the -2ΔΔCt method [33]. Results were normalized to GAPDH (Hs99999905_m1), with the hPSC-CEnCs injected with bioink as the calibrator to determine the relative quantities of gene expression in each sample.