Background The risk of Severe Acute Respiratory Syndrome Coronavirus 2 (SARS-CoV-2) transmission through corneal graft is an ongoing debate and leads to strict restrictions in corneas procurement, leading to a major decrease in eye banking activity. The aims of this study are to specifically assess the capacity of human cornea to be infected by SARS-CoV-2 and promote its replication ex vivo, and to evaluate the real-life risk of corneal contamination by detecting SARS-CoV-2 RNA in corneas retrieved in donors diagnosed with Coronavirus Disease 2019 (COVID-19) and nonaffected donors. Methods and findings To assess the capacity of human cornea to be infected by SARS-CoV-2, the expression pattern of SARS-CoV-2 receptor angiotensin-converting enzyme 2 (ACE-2) and activators TMPRSS2 and Cathepsins B and L in ocular surface tissues from nonaffected donors was explored by immunohistochemistry (n = 10 corneas, 78 ± 11 years, 40% female) and qPCR (n = 5 corneas, 80 ± 12 years, 40% female). Additionally, 5 freshly excised corneas (80 ± 12 years, 40% female) were infected ex vivo with highly concentrated SARS-CoV-2 solution (106 median tissue culture infectious dose (TCID50)/mL). Viral RNA was extracted from tissues and culture media and quantified by reverse transcription quantitative PCR (RT-qPCR) (viral RNA copies) 30 minutes (H0) and 24 hours (H24) after infection. To assess the risk of corneal contamination by SARS-CoV-2, viral RNA was tested by RT-qPCR (Ct value) in both corneas and organ culture media from 14 donors diagnosed with COVID-19 (74 ± 10 years, 29% female) and 26 healthy donors (79 ± 13 years, 57% female), and in organ culture media only from 133 consecutive nonaffected donors from 2 eye banks (73 ± 13 years, 29% female). The expression of receptor and activators was variable among samples at both protein and mRNA level. Based on immunohistochemistry findings, ACE-2 was localized mainly in the most superficial epithelial cells of peripheral cornea, limbus, and conjunctiva, whereas TMPRSS2 was mostly expressed in all layers of bulbar conjunctiva. A significant increase in total and positive strands of IP4 RNA sequence (RdRp viral gene) was observed from 30 minutes to 24 hours postinfection in central cornea (1.1 × 108 [95% CI: 6.4 × 107 to 2.4 × 108] to 3.0 × 109 [1.4 × 109 to 5.3 × 109], p = 0.0039 and 2.2 × 107 [1.4 × 107 to 3.6 × 107] to 5.1 × 107 [2.9 × 107 to 7.5 × 107], p = 0.0117, respectively) and in corneoscleral rim (4.5 × 109 [2.7 × 109 to 9.6 × 109] to 3.9 × 1010 [2.6 × 1010 to 4.4 × 1010], p = 0.0039 and 3.1 × 108 [1.2 × 108 to 5.3 × 108] to 7.8 × 108 [3.9 × 108 to 9.9 × 108], p = 0.0391, respectively). Viral RNA copies in ex vivo corneas were highly variable from one donor to another. Finally, viral RNA was detected in 3 out of 28 corneas (11%) from donors diagnosed with COVID-19. All samples from the 159 nonaffected donors were negative for SARS-CoV-2 RNA. The main limitation of this study relates to the limited sample size, due to limited access to donors diagnosed with COVID-19 and concomitant decrease in the procurement corneas from nonaffected donors. Conclusions In this study, we observed the expression of SARS-CoV-2 receptors and activators at the human ocular surface and a variable increase in viral RNA copies 24 hours after experimental infection of freshly excised human corneas. We also found viral RNA only in a very limited percentage of donors with positive nasopharyngeal PCR. The low rate of positivity in donors diagnosed with COVID-19 calls into question the utility of donor selection algorithms. Trial registration Agence de la Biomédecine, PFS-20-011 https://www.agence-biomedecine.fr/.

No abstract

Corneal endothelial diseases are the leading cause of corneal transplantation. The global shortage of donor corneas has resulted in the investigation of alternative methods, such as cell therapy and tissue-engineered endothelial keratoplasty (TEEK), using primary cultures of human corneal endothelial cells (hCECs). The main challenge is optimizing the hCEC culture process to increase the endothelial cell density (ECD) and overall yield while preventing endothelial–mesenchymal transition (EndMT). Fetal bovine serum (FBS) is necessary for hCEC expansion but contains TGF-βs, which have been shown to be detrimental to hCECs. Therefore, we investigated various TGF-β signaling pathways using inhibitors to improve hCEC culture. Initially, we confirmed that TGF-β1, 2, and 3 induced EndMT on confluent hCECs without FBS. Using this TGF-β-induced EndMT model, we validated NCAM as a reliable biomarker to assess EndMT. We then demonstrated that, in a culture medium containing 8% FBS for hCEC expansion, TGF-β1 and 3, but not 2, significantly reduced the ECD and caused EndMT. TGF-β receptor inhibition had an anti-EndMT effect. Inhibition of the ROCK pathway, notably that of the P38 MAPK pathway, increased the ECD, while inhibition of the ERK pathway decreased the ECD. In conclusion, the presence of TGF-β1 and 3 in 8% FBS leads to a reduction in ECD and induces EndMT. The use of SB431542 or LY2109761 may prevent EndMT, while Y27632 or Ripasudil, and SB203580 or SB202190, can increase the ECD.

PurposeThe HEC (Hoechst‐Ethidium‐Calcein) labeling, which we described in 2011, is a simple destructive test unanimously adopted to determine the pan‐endothelial viable endothelial cell density (vECD) using image analysis (e.g. using CorneaJ, ImageJ dedicated plugin). Sometimes, the endothelial surface folded by stromal edema remains difficult to analyze. We present here the HECplus technique, which, with an optimization of image acquisition and analysis, further increases the accuracy of vECD determination, particularly in endothelial folds.Methods10 pairs of corneas stored in organoculture at 31°C were analyzed. The endothelial side was incubated with 4 µM Calcein‐AM, 1 µM Ethidium and 11 µM Hoescht 33342 for 45 min at RT. Fluorescence was acquired in 3 channels first with a macroscope, in 3D with an image stacking of the entire graft then with a microscope after flat mounting. A 2 mm plastic pellet emitting constant fluorescence at 510 nm was imaged at the same time to normalize the images. Only one sharp image was reconstructed in ImageJ from the stacks of images. The final standardized images were then analyzed by CorneaJ (HEC group) and then manually by delimiting all viable surfaces on a graphic tablet (HECplus group)ResultsThe vECD was significantly higher by 8.2 ± 1.9% in the HECplus group versus HEC (2059 ± 127 versus 1989 ± 156 cells/mm2). Endothelial cells died preferentially in the top of the folds and were better preserved in the bottom of the folds. While HEC wrongly considered some folds as dead, HECplus allowed the error to be corrected. The folding pattern was similar between the 2 corneas of the same pair but differed from one donor to another. The normalization of images for calcein showed overall differences in the level of fluorescence between corneas, suggesting differences in metabolic activity.ConclusionsWe recommend applying the HECplus protocol when a very high degree of accuracy is required, for example when comparing storage media.