Study design
The research objectives of this study were to: (1) adapt our previously published technique to shrink-wrap single cells in order to shrink-wrap small clusters of CE cells into patches of endothelium termed µMonolayers, (2) to investigate whether shrink-wrapped μMonolayers would integrate into endothelium and increase cell density in vitro, and (3) determine if the μMonolayers were capable of engrafting into an existing healthy endothelium in vivo. Primary bovine and rabbit CE cells were used throughout the study using established cell culture methods, with minor modifications in the case of the rabbit cells26,36. Sample sizes for in vitro studies were determined by using the minimum number of samples to be considered statistically significant and time points/endpoints were based on our previously published studies. For the in vitro cell density study, 4 replicates per sample type per time point were used and one full study was completed. Data from the day 3-time point was used to determine if the sample size was sufficient to provide statistical significance. On day 7 one control sample and on day 14 one single cells sample was lost during fixing and staining and therefore n = 3 for those sample types. In vivo studies were designed to be pilot studies and as such, the number of rabbits per study was kept to 2–3 animals per time point and condition to establish repeatability.
ECM scaffold fabrication
The ECM scaffolds were fabricated via previously described surface-initiated assembly techniques with minor modifications24,25. Briefly, 1 cm × 1 cm PDMS stamps designed to have 200 μm square features were fabricated via standard soft lithography techniques. The stamps were sonicated in 50% ethanol for 60 minutes, dried under a stream of nitrogen, and incubated for 60 minutes with a 50:50 mixture of 50 μg/mL COL4 and 50 μg/mL LAM (Fig. 1step 1). Either 50% AlexaFluor 488 labeled COL4 or 50% AlexaFluor 633 labeled LAM (for a final concentration of 25% labeled protein) was used to visualize the pattern transfer. Following incubation, the stamps were rinsed in sterile water, dried under a stream of nitrogen, and brought into conformal contact with PIPAAm (2% high molecular weight, Scientific Polymers) coated 18 or 25 mm glass coverslips for 30 minutes to ensure transfer of the squares (Fig. 1step 2). ECM squares microcontact printed on PDMS coverslips were used as controls. Upon stamp removal, laser scanning confocal microscopy (Nikon AZ100) was used to determine the quality of the transferred ECM squares.
Corneal endothelial cell culture
Bovine CE cells were isolated and cultured as previously described26,27. Briefly, corneas were excised from the whole globe (Pel Freez Biologicals) and incubated endothelial side up in a ceramic 12-well spot plate with 400 μL of TrypLE Express for 20 minutes. The cells were then gently scraped from the cornea using a rubber spatula, centrifuged at 1500 rpm for 5 minutes, resuspended in 5 mL of culture media (low glucose DMEM with 10% FBS, 1% Pen/Strep/AmphB, and 0.5% gentamicin), designated at P0 and cultured in a 50 kPa PDMS coated T-25 flask that was pre-coated with COL426. Fifty whole eyes were received at a time and were used to seed 5 T-25 flasks. Cells were cultured until confluence and split 1:3 until they were used once confluent at P2.
Rabbit corneas were excised from the whole globe (Pel Freez Biologicals), the CE and Descemet’s Membrane were manually stripped with forceps, and then incubated in Dispase (1 U/mL, Stem Cell Technologies) for 1.5 hours at 37 °C to detach the rabbit CE cells (RCECs) from the Descemet’s Membrane. The RCECs were then gently pipetted up and down, diluted in culture media (DMEM/F12, 10% FBS, 0.5% Pen/strep), centrifuged at 1500 rpm for 5 minutes, resuspended in 10 mL of culture media, designated at P0 and cultured on COL4 coated T-25 flasks with the equivalent of 15–25 eyes per flask depending on cell yield. RCECs were cultured until confluence and then split 1:2 and used in all experiments once confluent at P1 or P2.
Shrink-wrapping CE Cell µMonolayers in ECM
Patterned coverslips were secured with vacuum grease to the bottom of 35 mm Petri dishes, which were placed on a dry block set to 52 °C. This resulted in the coverslips reaching (within 30 min) and holding at ~40 °C. Bovine CE cells were released from the culture flask with TrypLE Express, centrifuged, and resuspended at a density of 150,000 cells/mL in 15 mL centrifuge tubes. The tubes were placed in a dry block set at 45 °C for ~5 minutes, or until the cell solution just reached ~40 °C and 2 mL of cell suspension was added to each 35 mm dish before it was immediately placed in an incubator (37 °C, 5% CO2). Cells were cultured for 24 hours to allow them to form μMonolayers on the 200 μm ECM squares. Samples were then removed from the incubator, rinsed twice in 37 °C media to remove non-adherent single cells, 2 mL of fresh warm media was added, and the sample was allowed to cool to room temperature. Once the temperature decreased to <32 °C the PIPAAm dissolved and released the µMonolayers. The release process was recorded using a Photometrics CoolSnap camera. Following the release, the µMonolayers were collected via centrifugation at 1500 rpm for 5 minutes before use in further experiments. CE cells seeded onto PDMS coverslips were used as a control.
Immunostaining of shrink-wrapped CE cell μMonolayers
Shrink-wrapped bovine µMonolayers resuspended in PBS containing Ca2+ and Mg2+ (PBS++) were injected through a 28-gauge needle onto a glass coverslip and allowed to settle for ~15 minutes before fixation for 15 minutes in 4% paraformaldehyde in PBS++. Samples were gently washed two times with PBS++ and incubated with 1:100 dilution of DAPI, 1:100 dilution of mouse anti-ZO-1 antibody (Life Technologies), and 3:200 dilution of AlexaFluor 488. Samples were rinsed two times for 5 minutes with PBS++ and incubated with 1:100 dilution of AlexaFluor 555 goat anti-mouse secondary antibody for 2 hours. Samples were rinsed two times for 5 minutes with PBS++, mounted on glass slides with Pro-Long Gold Antifade (Life Technologies), and then imaged on a Zeiss LSM700 confocal microscope.
Viability of shrink-wrapped CE cells μMonolayers post injection
After centrifugation, bovine shrink-wrapped µMonolayers or single cells were resuspended in 200 µL of growth media, drawn up into a 28 G needle, injected into a petri dish, and incubated with 2 µM calcein AM and 4 µM EthD-1 (Live/Dead Viability/Cytotoxicity Kit, Life Technologies) in PBS++ for 30 minutes at 37 °C. After 30 minutes, samples were imaged on a Zeiss LSM700 confocal; 5 images per sample and 3 samples per type were used. The number of live and dead cells was counted manually using ImageJ’s multi-point tool. The number of live cells was divided by the number of total cells to determine the percent viability of both the shrink-wrapped cells and enzymatically released cells. The data were compared using a Student’s t test in SigmaPlot. The same methods were used to test the viability of the cells through a 34 G needle to test the smallest needle that could be used.
Seeding of shrink-wrapped CE cell μMonolayers and single CE cells on stromal mimics
Compressed collagen type I films were prepared as previously described to mimic the structure of the underlying stroma27. Briefly, a 6 mg/mL collagen type I gel solution was prepared per the manufacturer’s instructions and pipetted into 9 mm diameter silicone ring molds on top of glass coverslips. The gels were placed into a humid incubator (37 °C, 5% CO2) for 3 hours to compress under their own weight. The gels were then dried completely in a biohood followed by rehydration in PBS, forming thin collagen type I stromal mimic of approximately 10 µm in thickness (reduced from an original 1 mm in thickness as cast). Shrink-wrapped bovine CE cell µMonolayers were seeded onto the films at a 1:1 ratio of the stamped coverslip to collagen type I film. As a control, bovine CE cells that were cultured in the flasks and enzymatically released using TrypLE Express into a single-cell suspension were seeded onto collagen type I films. The number of control single cells seeded was equal to the number of cells seeded in the μMonolayers, assuming each of the 200 μm ECM squares used for shrink-wrapping was completely covered in cells. The average number of cells on the 200 μm ECM squares was 30 cells, so with 1600 squares per stamp, we seeded ~48,000 cells per sample. Therefore, 50,000 cells per sample were seeded for the controls. At 6 and 24 hours, samples were removed from the culture and fixed and stained for the nucleus, ZO-1 (tight junction protein), and F-actin. Briefly, samples were rinsed two times in PBS++, fixed in 4% paraformaldehyde in PBS++ with 0.05% Triton-X 100 for 15 minutes. Samples were rinsed two times for 5 minutes with PBS++ and incubated with five drops of NucBlue (Life Technologies) for 10 minutes. Samples were rinsed once with PBS++ and incubated with 1:100 dilution of mouse anti-ZO-1 antibody (Life Technologies) and 3:200 dilution of AlexaFluor 488 or 633 phalloidin for 2 hours. Samples were rinsed three times for 5 minutes with PBS++ and incubated with 1:100 dilution of AlexaFluor 555 goat anti-mouse secondary antibody for 2 hours. Samples were rinsed three times for 5 minutes with PBS++, mounted on glass slides using Pro-Long Gold Antifade, and imaged on a Zeiss LSM700 confocal microscope.
In vitro integration of shrink-wrapped μMonolayers vs single bovine CE cells
To mimic a low-density CE, 25,000 P5 bovine cells were seeded onto the collagen type I stromal mimics, as described above, until confluent to form the low-density monolayers. Shrink-wrapped bovine µMonolayers and single bovine CE cells were prepared as above, labeled with CellTracker Green (Life Technologies) for 30 minutes, centrifuged, diluted to the equivalent of 50,000 cells/sample, and injected onto the low-density monolayers. Low-density monolayers with no cells injected on top served as controls. Samples were rinsed 3 hours post injection to mimic the in vivo procedures and new media was added. The media was changed every two days thereafter. Samples were fixed and stained on days 3, 7, and 14 as described above. A Zeiss LSM700 confocal was used to image 10 random spots on each sample and the cell density was manually counted using the multi-point selection tool in ImageJ to count cell nuclei. The locations on each sample were chosen randomly using the joystick positioner on the Zeiss confocal without looking at the sample under the eye piece to ensure a random location within each sample. The number of nuclei was divided by the image area to obtain the cells/mm2 per image. The cell density for each sample was determined by averaging the cell densities of each image and the average cell density of each sample type was determined by averaging the cell density of the 3–4 samples. The data were compared using a one-way ANOVA on ranks with Tukey’s test (day 3) or one-way ANOVA (days 7 and 14) with Tukey’s test in SigmaPlot. To examine the outgrowth of the shrink-wrapped µMonolayers over time, confocal images centered around an individually shrink-wrapped µMonolayer (day 3 n = 33, day 7 n = 37, day 14 n = 40) were collected and the CellTracker channel was converted into a binary black and white image. The binary images for each sample type were then converted into one Z-stack and analyzed via the Heat Map for Z-stacks plugin (relative without log10) for ImageJ to determine the average pixel density of CellTracker. To determine the percentage of injected single cells or shrink-wrapped cells that were engrafted into the existing monolayers at each time point, the average number of cells on the control samples was subtracted from the total number of cells on the single-cell or shrink-wrapped samples to get the total number of cells that engrafted using each technique. This difference in the number of cells was then divided by the number of seeded cells (~50,000) and multiplied by 100 to achieve a percentage. The results were compared at each time point for the two sample types using a student’s t test p < 0.05.
Live imaging of in vitro integration of shrink-wrapped bovine CE cells
For live imaging, the bovine monolayer on the collagen type I stromal mimic was first incubated for 30 minutes with CellTracker Orange to differentiate between the existing monolayer and injected cells, which were labeled with CellTracker Green as described above. HEPES buffered Opti-MEM I Reduced Serum Media (Life Technologies) with 10% FBS, 1% Pen/Strep was added to the monolayer and shrink-wrapped bovine µMonolayers that were prepared as described above were injected through a 30 G needle on top of the sample. The sample was placed on the Zeiss LSM700 confocal equipped with a temperature chamber set to 37 °C for 30 minutes to allow for the cells to settle. Using the Definite Focus system, a time-lapse series of one z-stack was obtained every hour for 48 hours. Videos from the time-lapse images were created using the Imaris Software.
Ex vivo integration of shrink-wrapped rabbit CE cells
Whole rabbit eyes were placed cornea up in a 12-well plate and shrink-wrapped rabbit CE µMonolayers were prepared as described above. Two samples of µMonolayers per ex vivo eye were prepared and resuspended in 100 µL of DMEM/F12. A 30-G insulin syringe was used to draw up the full 100 µL, the needle was inserted into the center of the cornea until it was visible in the anterior chamber and 50 µL of the suspension was injected. This resulted in the equivalent of 50,000 cells injected into the anterior chamber. The needle was held in place for a few seconds to ensure the media and cells did not come back out of the injection site. The injection was viewed under a stereomicroscope and the pink color of the media filling the anterior chamber was visible, indicating successful injection. The whole eyes were flipped and incubated cornea down for 3 hours at 37 °C, 5% CO2 in a humidified incubator. No media was added directly to the eyes, instead, media was included in two empty wells to maintain hydration throughout the study (Supplementary Fig. 3). After 3 hours, the whole eye was placed in 2% paraformaldehyde (PBS++) at 4 °C for 24 hours. After 24 hours the eye was rinsed in PBS and the cornea was excised and rinsed three times for 5 minutes. The cornea was then incubated CE facing down on 1 mL of PBS++ containing 2 drops of NucBlue (Life Technologies), 2:100 dilution of mouse anti-ZO-1 antibody (Life Technologies), and 3:200 dilution of AlexaFluor 488 Phalloidin (Life Technologies) for 2 hours at room temperature. Corneas were then rinsed three times for 5 minutes in PBS followed by a 2-hour incubation in 1 mL PBS++ with 2:100 dilution of AlexaFluor 555 goat anti-mouse secondary antibody for 2 hours and stored in PBS before imaging on the Zeiss LSM700 confocal.
In vivo injection and integration of shrink-wrapped CE cells
All experimental procedures were reviewed and approved by the University of Pittsburgh Institutional Animal Care and Use Committee (IACUC) and carried out according to the ethical guidelines of the Association for Research in Vision and Ophthalmology Resolution on the Use of Animals in Ophthalmic and Vision Research. For both in vivo experiments, shrink-wrapped rabbit µMonolayers were prepared as described above with one minor modification: cells were labeled with Vybrant DiO 1 day prior to seeding onto the ECM nano-scaffolds by incubating cells in 1 mL of media with 5 µL of Vybrant DiO for 30 minutes followed by three, 10-minute rinses with fresh media. An excess number of µMonolayer samples were prepared to ensure there was enough volume for injection. The shrink-wrapped µMonolayers were released as described above and after centrifugation at 1500 rpm for 5 min, the shrink-wrapped µMonolayers were resuspended in DMEM/F12 at the equivalent of 100,000 cells per 50 μL injection volume (2 stamped samples per 50 μL).
For the first experiment, control single cells were prepared as described above and resuspended in DMEM/F12 at a density of 100,000 cells in a 50 μL injection volume. Six female New Zealand white rabbits with healthy intact CEs weighing approximately 2.5 kg were used for this study. Rabbits were anesthetized with Ketamine (40 mg/kg) and Xylazine (4 mg/kg) intramuscular injection followed by isoflurane inhalation to keep rabbits under sedation for 3 hours. One rabbit did not survive the anesthetization. Rabbits #1 & 2 were injected in the right eye with 50 µL (~100,000 cells) of the single-cell suspension. Rabbits #3, 4, and 5 were injected with 50 µL of the shrink-wrapped μMonolayer suspension into the right eye using a 30 G needle attached to a 500 µL syringe. A tunnel in the corneal stroma was made for the injecting which prevented cell leakage after injection. Immediately after injection, each rabbit was placed on its side with the injected eye facing down for 3 hours to ensure attachment of the cells. On day 7, rabbits were anesthetized with an intramuscular injection of ketamine (40 mg/kg) and xylazine (4 mg/kg) and then euthanized with Euthasol solution (1 mg per 4 lbs) containing (390 mg/mL Sodium Pentobarbitol, 50 mg/mL Phenytoin Sodium) through an ear vein injection. Photographic images were obtained via the Google Pixel 2 camera (Supplementary Fig. 4) to document eye clarity and the endothelium was viewed using a Nidek Confoscan 3 (Supplementary Fig. 5). Eyes were then immediately enucleated and intravitreally injected with 100 µL 2% paraformaldehyde in PBS++. The whole eye was then immersed in 2% paraformaldehyde in PBS++ and fixed at 4 °C for 24 hours. After 24 hours the eye was rinsed in PBS and the cornea was excised and rinsed 3x’s for 5 minutes. The cornea was then incubated CE facing down on 1 mL of PBS++ containing two drops of NucBlue (Life Technologies) and 2:100 dilution of mouse anti-ZO-1 antibody (Life Technologies) for 2 hours at room temperature. Corneas were then rinsed three times for 5 minutes in PBS followed by 2-hour incubation in 1 mL PBS++ with 2:100 dilution of AlexaFluor 555 goat anti-mouse secondary antibody for 2 hours and stored in PBS before imaging on the Zeiss LSM700 confocal or a Nikon FN1 base with an A1R HD MP Confocal module. To quantify the density of the integrated cells, the green cells were manually traced with the freehand selection tool in ImageJ, and the “Measure” function was used to determine the area. The multi-point selection tool was used to determine the number of nuclei within that area and the density was determined by dividing the number of nuclei by the area. The rectangle tool was used to select control areas, areas of non-green (i.e., non-DiO-labeled) cells, and elsewhere in the image of a similar area. The area and cell number were determined the same way as for the DiO-labeled areas and the cell density was calculated by dividing the number of nuclei by the area. The number of control areas per image was matched to the number of DiO-labeled areas (Supplementary Fig. 6). Data from the injected eye of both rabbits were pooled together and the cell density of the areas with green cells were statistically compared to those without using SigmaPlot for a student t test.
For the second experiment, only shrink-wrapped μMonolayers were used to determine if cells would integrate and remain in the CE long-term. Six female New Zealand white rabbits with healthy intact CEs weighing approximately 2.5 kg were used for this study. Because these were initial proof-of-concept studies, we wanted to eliminate any possible sex differences and included only female rabbits. Rabbits were anesthetized as described above and injected with 50 µL of the shrink-wrapped μMonolayer suspension into the right eye. Immediately after injection, each rabbit was placed on its side with the injected eye facing down for 3 hours to ensure attachment of the cells. One rabbit went into tachycardia right at the end of the 3 hours and was revived, however, it suffered brain damage as a result of the amount of time without oxygen. That rabbit was sacrificed 24 hours post injection and the eye was removed and processed as described above and used to determine if the injection processed had been successful (data not shown). At 14 days (2 rabbits) or 28 days (3 rabbits) post injection, rabbits were sacrificed (as described above) and photographic images were obtained via the Google Pixel 2 camera (Supplementary Fig. 4) to document eye clarity and the endothelium was viewed using a Nidek Confoscan 3 (Supplementary Fig. 5). Following imaging, eyes were immediately enucleated and fixed, stained, and stored as described above, before imaging on the Zeiss LSM700 confocal or a Nikon FN1 base with an A1R HD MP Confocal module. To quantify the cell density, images with DiO-labeled cells present were taken (n = 10+ images) and images were taken far away in areas where there were no green cells (n = 5 images) in each cornea. Because the DiO-labeled may have faded over time, in this case, the cell density of the entire image was counted using the number of nuclei (counted manually via the multi-point tool with only full nuclei being counted) divided by the area of the image. To statistically compare the data, for each rabbit, the density of the images with green cells was compared to the density of the images with no green cells in SigmaPlot using the student t test.
Reporting summary
Further information on research design is available in the Nature Research Reporting Summary linked to this article.
