Human Corneal Endothelium Regeneration: Effect of ROCK Inhibitor

We read with great interest the article by Okumura et al., in the April 2013 issue, on Rho-associated protein kinase (ROCK)-inhibitor eye drops' effect on corneal endothelium. ¹ As clinicians, knowing only some very basic concepts of molecular biology and regenerative medical procedures, but very interested in the field of human corneal endothelial regeneration, we dare to make some comments. 

During the last 17 years, in many articles, the group headed by Joyce at Schepens Eye Research Institute has identified various mechanisms causing in vivo human corneal endothelial cells to be arrested in G1-phase of the cell cycle: cell–cell contact inhibition (possibly mediated by the cyclin-dependent kinase inhibitor p27Kip1); deficiency or absence of positive growth factor stimulation; age-related increased expression of G1-phase inhibitors (cyclin-dependent kinase inhibitor p21Cip1, and p16INK4a) that lastly leads the cells to stress-induced premature senescence; and finally, the presence of transforming growth factor-β 2 (TGF-β 2) that suppresses S-phase entry (possibly involving its effects on p27Kip1 and prostaglandin E2 levels). ² However, evidence of mitosis in humans in vivo has been reported since 1982. Treffers studied it in two patients: a wound was created by a central transcorneal freeze, and after removing the cornea from the enucleated eye, using tritiated thymidine showed evidence of both migration and proliferation of endothelial corneal cells. ³ Studies from more than a decade ago by Wollensak and Green ⁴ and, more recently, works by Lagali et al. ⁵ have shown that, in corneal grafts, a partial or total replacement of the donor endothelial corneal cells by recipient endothelium frequently occurs, proving the ability of peripheral recipient endothelial cells to migrate to the central cornea. Although a complete donor cell replacement in the graft by the recipient's endothelial cells is possible without needing cell proliferation (leading to a reduction of the cell density by approximately 50%), in vivo mitotic division, in addition to migration, might contribute to the repopulation of graft endothelium in those cases, as suggested by both groups of authors. 

We also suggested that in vivo corneal endothelial cell proliferation might have been a factor in spontaneous repopulation of the posterior bare stroma with endothelial cells in both eyes of a patient who underwent Descemet's stripping without endothelial replacement, reported by Shah et al. ^6,7 In that patient, the right eye underwent Descemet's stripping endothelial keratoplasty and had graft failure. Later a regraft was performed, but the second posterior corneal lenticule detached. Surgeons decided to remove the donor tissue, but it was not replaced. Finally, the cornea gained transparency, and the patient reached a corrected visual acuity of 20/20. That eye's confocal microscopy image prior to the original surgery showed a very irregular endothelial mosaic pattern with apparently lower cell density than shown in final postoperative images after removing the donor lenticule, which, in our opinion, favors cellular replication in addition to migration. We think descemetorhexis played a role in releasing contact inhibition at least in a group of recipient endothelial cells. 

Studies by Whikehart et al. ⁸ and McGowan and coauthors, ⁹ which reported telomerase activity at the peripheral endothelium in unwounded human tissues and specific stem cell markers in the trabecular meshwork and the transition zone between the trabecular meshwork and the outer edge of the corneal endothelium, suggest that endothelial stem cells reside in the posterior human limbus and respond to corneal wounding by initiating an endothelial repair process, and may also contribute to a normal, slow replacement of corneal endothelial cells. Recently, He et al. published additional evidence that in vivo human corneal endothelial cell proliferation exists. ¹⁰ Cell organization in clusters of two or three layers and radial rows in the extreme periphery of human corneas supports the hypothesis that corneal endothelial cells continuously migrate centripetally from peripherally located stem cells. These authors suggest a model of human corneal endothelium homeostasis: in the periphery of the cornea, cells divide very slowly within a renewal zone and then migrate toward the center but probably desquamate, while the density of the central cells remains stable. They suggest that the cell clusters located in the extreme periphery may be stem cell niches or emerging points for progenitors migrating from deeper niches and that the contact with aqueous humor leads the cells to lose their proliferating capacity, but not their migrating potential. Also recently, Hirata-Tominaga and coauthors (including several of the authors of the article published in the April 2013 issue of IOVS, which is the motive of this letter) found that leucine-rich repeat-containing G-protein–coupled receptor 5 (LGR5), reportedly a marker of multiple tissue stem cells in mice, is expressed in the peripheral region of human corneal endothelial cells and that those LGR5(+) cells show some stem/progenitor cell characteristics. ¹¹ The presence of stem cells supports the hypothesis about the in vivo proliferative activity of human corneal endothelial cells. 

Okumura et al., ¹ in the present study in both a corneal endothelial dysfunction primate model and a human clinical case series of corneal endothelial dysfunction, used a model injuring endothelial cells by transcorneal freezing. Primary effects of that procedure were lysis of the cells affected by the freezing and the release of contact inhibition of at least a fraction of the remaining cells. In the primate model, noncontact specular microscopy revealed that corneal endothelial cell density was significantly higher in the ROCK inhibitor (Y-27632) group compared with the controls at 4 weeks (3000 cells/mm² vs. 1500 cells/mm²), which suggested a positive effect of the substance. 

Although the authors did not verify that the treatment did not alter the karyotype of endothelial cells, by evaluating the frequency of cells having an abnormal chromosome number (i.e., aneuploid), ¹² they determined histologic phenotypes in the primate corneal endothelial wound model, and they found that the percentages of tight junction protein ZO-1 and Na⁺/K⁺-ATPase–positive cells in the central area were significantly higher in the ROCK-inhibitor treated eyes than those in the control eyes, suggesting that the intervention rapidly enhances functional recovery as well as morphologic recovery. In the series of patients, however, results were less definitive. A point to bear in mind is that pachymetry measurements made with various techniques have a range of intra-individual variability, which should be considered when analyzing changes in corneal thickness. It is also important to know exactly which measurement technique was used, since some techniques, such as ultrasound pachymetry, have proven to be more operator dependent (by the difference in the exact location of the probe and the corneal indentation caused by examiner when applying the probe). ¹³ The patient in case 2 underwent Descemet's stripping automated endothelial keratoplasty apparently before the 6-month follow-up, and data of pachymetry or visual acuity before the corneal graft are not shown. Therefore, it is not possible to assess the effect of treatment, and we consider that case should be excluded from the study. Four patients had diffuse edema, peripheral and central: three with bullous keratopathy (BK) after laser iridotomy (the second cause of BK in Japan) ¹⁴ and one with pseudoexfoliation syndrome (rarely identified as the causative factor in BK). ¹⁵ In three of them, corneal edema did not decrease (pachymetry remained the same or increased), but in one patient (case 5), the central corneal thickness exhibited the greatest decrease of the whole group (almost 180 μm). So, what the authors said, indicating that there was no reduction in pachymetry in eyes with diffuse edema, although true if you look at the average, was clearly inaccurate when looking at every case individually. What happened is that the behavior of the patient in case 5 was significantly different from the other cases in this group, and therefore must be analyzed in a particular way. Therefore, it would have been very important to determine if the noncontact-specular microscopy showed any improvement and if the transparency of the cornea improved in this case, because visual acuity could have been degraded by another concurrent ocular condition, since Table 1 indicates that the patient had cataract. Moreover, although the statement of the authors concerning cases of Fuchs' dystrophy patients (indicating that their pachymetry decreased) is true, the central corneal thickness ranges remained above normal (663 and 687 μm in cases 3 and 4). Pachymetry significantly decreased in one single patient of this group (case 1), who had greatly improved visual acuity (from approximately 20/100 to approximately 20/13). Endothelial cell density after treatment in this case was between 1200 and 1500 cells/mm². These findings suggest very strongly a positive effect of the intervention in this case, but without cell count data, we do not see clear evidence of a positive effect of the treatment in the other two. 

We wonder why, following transcorneal freezing, just 42 drops of ROCK inhibitor over 1 week were given. No complications such as intraocular pressure elevation or systemic complications were detected in relation to the ROCK-inhibitor eye drop application during this short period of time. Would it not be expected that the beneficial effect on the proliferation of endothelial cells would have increased if the time of treatment had been extended? 

Pipparelli et al. ¹⁶ recently (April 2013) reported that ROCK inhibitor is not toxic to human endothelial cells both in vitro and ex vivo, so it sounds reasonable to increase the dose or time of administration. 

Until recently, the mechanism by which ROCK inhibitor promotes proliferation of corneal endothelial cells had not been elucidated, yet at the 2013 ARVO meeting, Numata, Okumura, Kinoshita, and associates showed that in cultivated monkey corneal endothelial cells, the amount of p27 was greatly reduced from 1 hour following treatment of cells with that substance, whereas the control cells maintained high levels of p27 up to 12 hours. The findings of their study demonstrated that ROCK inhibitor activates PI 3-kinase signaling, which subsequently promotes degradation of p27 via Cdc25A pathway, thus leading to cell proliferation (Numata R, et al. IOVS 2013;54:ARVO E-Abstract 1690). They also showed a study indicating that ROCK inhibitor could protect corneal endothelium from the apoptosis (Odajima A, et al. IOVS 2013;54:ARVO E-Abstract 1693). 

However, the issue of the effect of ROCK inhibitor in human corneal endothelial cells is not yet fully resolved. In contrast to what was published by Okumura et al., ¹ another group of researchers from Australia, Switzerland, and France (Pipparelli et al. ¹⁶ ) simultaneously published their results showing it had no effect on the proliferative capacity of human corneal endothelial cells both in vitro and ex vivo (mean donor's age 73 years), as shown by EdU (5-ethynyl-2′-deoxyuridine) incorporation, a sensitive technique used to monitor cell proliferation since it is incorporated during DNA synthesis. This lack of effect on mitosis was also evident by the loss of human endothelial cells during storage and by caspase-3 immunostaining. However, ROCK inhibitor did stimulate changes in cell shape, increase cellular adhesion, and encourage healing of an endothelial wound by cell migration. By the latter capabilities, even if it does not stimulate mitosis of human endothelial corneal cells, ROCK inhibitor might potentially be useful in the clinical management of corneal endothelial dysfunction. 

In sum, we are convinced that pharmacologic advances, such as the use of ROCK inhibitor, are very promising and that along with other approaches, such as cell-injection therapy using cultivated human corneal endothelial cells (Koizumi N, et al. IOVS 2013;54:ARVO E-Abstract 2201), will revolutionize the way we treat endothelial alterations and even the way we perform intraocular surgery and undoubtedly will reduce the need for a high percentage of corneal transplants. Recently, Okumura et al. ¹⁷ also published that in vitro use of SB431542, a selective inhibitor of the TGF-β receptor, counteracted the fibroblastic phenotypes of human corneal endothelial cells to the normal contact-inhibited monolayer, and these polygonal cells maintained endothelial physiologic functions. It would be very interesting to explore the in vivo effect of this TGF-β inhibitor in the G1/S transition of the cell cycle and its possible synergistic effect with the ROCK inhibitor. ¹⁷