Gene Signature–Based Development of ELISA Assays for Reproducible Qualification of Cultured Human Corneal Endothelial Cells

Abstract

Purpose: To develop a method to qualify the function of cultured human corneal endothelial cells (cHCECs) applicable for clinical settings.

Methods: The diversified gene and microRNA (miRNA) signatures in HCECs from a variety of tissue donors were confirmed by three-dimensional (3D) gene human miRNA profiling. These were compared with those of more than 20 cHCECs distinct in their cell morphology or culture lots. Candidate genes were selected after quantitative (q)RT-PCR validation, and gene products were assayed by ELISA. After three additional screening steps, final candidate cytokines for qualification were selected.

Results: Gene and miRNA signatures among distinct cHCEC lots were greatly diversified compared with those among fresh tissues from different age donors. By comparing more than 20 lots of cultures, 32 candidate genes were assigned to be seemingly linked to distinct cHCEC morphologic features. The validation of candidate genes by qRT-PCR revealed the genes, either upregulated or downregulated, corresponding to morphologic variances in cHCECs (e.g., epithelial-mesenchymal transition or cell senescence). Further adding the ELISA results by Bio-Plex Human Cytokine 27-Plex Panel, 11 candidate cytokines suitable to qualify cHCEC function were selected. In consideration of the presence of these cytokines in the anterior chamber, IL-8, tissue inhibitors of metalloproteinases 1 (TIMP-1), monocyte chemotactic protein-1 (MCP-1), and platelet-derived growth factor–BB (PDGF-BB) were ultimately selected and applied in practice for the qualification of cHCECs actually used in our clinical cell-injection studies.

Conclusions: The specified cytokines properly discriminating the functional features of cHCECs indicates a correlation between profiling signatures and cell morphology.

To date, most researchers conceptualize cultured human corneal endothelial cells (cHCECs) only from the aspect that they are derived from corneal endothelium tissue, and disregard details pertaining to the refinement of the biochemical features. In fact, cHCECs contain subpopulations (SPs) that are heterogeneous in their morphology and in their surface markers.^1–3 Studies, either directly or indirectly, including those from Joyce et al.,^4–11 have shown that heterogeneity is present in HCEC cultures. Of particular and striking interest is the finding by Miyai et al.¹ of the presence of frequent chromosomal aneuploidy in cHCECs. Cultures HCECs have an inclination toward cell state transition (CST) into a senescence phenotype, epithelial-mesenchymal transition (EMT), and fibroblastic cell morphology. These findings indicate the necessity of qualifying the features of heterogeneous cHCECs by means of biochemical terms in order to distinguish vulnerable transformed cHCECs² and CST (i.e., in order to clarify in detail the fine and distinct characteristics among cHCECs). 

It is well known that the proliferative potential of HCECs is limited,^12–14 and that severe damage to the corneal endothelium resulting from pathologic conditions leads to corneal endothelial dysfunction and to the loss of corneal transparency.^15–17 Hence, corneal transplantation is the only available therapeutic pathway in such cases. Of note, Koizumi et al.¹⁸ developed a new corneal-cell transplantation method involving the intraocular injection of substrate-free cHCECs, and the efficacy of those cells was confirmed in a cell-transplantation study using a cynomolgus-monkey corneal endothelial dysfunction model. In order to make this innovative therapy applicable in the clinical setting, it is absolutely vital that quality-controlled cHCECs can be reproducibly obtained, thus illustrating the obvious need of a reliable method to qualify the cell cultures. 

Trials pertaining to the in vitro expansion of cHCECs without CST and karyotype aneuploidy have been hampered by the lack of a scientifically reliable and semiquantitative methods to evaluate the cellular features of cHCECs. 

In consideration of the controversy that surrounds the heterogeneity of cHCECs, we first investigated the discrepancy of gene signatures between fresh corneal endothelium tissues and cHCECs. Then, we compared the gene signatures among morphologically distinct cHCECs, and those appearing to have almost no apparent morphologic distinction. 

In this study, we present our novel findings an establishment of a qualification method to monitor protein products in cHCECs, and describe a quality-control method to ensure the functional characteristics of cHCECs that are indispensable for clinical application. 

Materials and Methods

HCEC Donors

The human tissue used in this study was handled in accordance with the tenets set forth in the Declaration of Helsinki. Human corneal endothelial cells were obtained from human cadaver corneas and were cultured before performing karyotyping analysis. Human donor corneas were obtained from SightLife, Inc. (Seattle, WA, USA). Informed written consent for eye donation for research was obtained from the next of kin of all deceased donors. All tissues were recovered under the tenets of the Uniform Anatomical Gift Act (UAGA) of the particular state in which the donor consent was obtained and the tissue was recovered. 

All donor corneas were preserved in Optisol GS (Chiron Vision, Inc., Irvine, CA, USA) and imported via international air transport for research purposes. Donor information accompanying the donor corneas showed that they were all considered healthy and absent of any corneal disease, and that all donors had no past history of chromosomal abnormality. 

Cell Cultures of HCECs

Unless otherwise stated, the HCECs were cultured according to the published protocols, with some modifications.¹¹ Briefly, the Descemet's membranes with the CECs were stripped from donor corneas and digested at 37°C with 1 mg/mL collagenase A (Roche Applied Science, Penzberg, Germany) for 2 hours. The HCECs obtained from a single donor cornea were seeded in one well of a Type-I collagen-coated six-well plate (Corning, Inc., Corning, NY, USA). The culture medium was prepared according to published protocols. Briefly, basal medium was prepared with OptiMEM I (Life Technologies Corporation, Carlsbad, CA, USA), 8% fetal bovine serum (FBS), 5 ng/mL epidermal growth factor (EGF; Life Technologies), 20 μg/mL ascorbic acid (Sigma-Aldrich Corp., St. Louis, MO, USA), 200 mg/L calcium chloride (Sigma-Aldrich Corp.), 0.08% chondroitin sulfate (Wako Pure Chemical Industries, Ltd., Osaka, Japan), and 50 μg/mL gentamicin. Mesenchymal stem cell (MSC)-conditioned medium was prepared as previously described.¹⁸ The HCECs were cultured using MSC-conditioned medium at 37°C in a humidified atmosphere containing 5% CO₂, and the culture medium was changed twice per week. The HCECs were passaged at ratios of 1:3 using 10× TrypLE Select (Life Technologies) at 37°C for 12 minutes when they reached confluence. The HCECs at passages two through five were used for all experiments. 

Phase Contrast Microscopy

Phase contrast microscopy images ware taken by use of an inverted microscope system (CKX41; Olympus Corporation, Tokyo, Japan). 

Total RNA Extraction

Corneal endothelial tissues and epithelial tissues were stripped from donor corneas and then stored in QIAzol Lysis Reagent (QIAGEN, Hilden, Germany) at −80°C until total RNA extraction. Cultured HCECs and the supernatant of cHCECs were lysed by use of QIAzol Lysis Reagent and then stored at −80°C until total RNA extraction. Total RNA was extracted using the miRNeasy Mini kit (QIAGEN). The quality of the purified total RNA was analyzed by use of a Bioanalyzer 2100 (Agilent Technologies, Santa Clara, CA, USA). 

MicroRNA (miRNA) Expression Profiling

For miRNA expression profiling, 3D-Gene Human miRNA Oligo Chips (miRBase version 17-21; Toray Industries, Inc., Tokyo, Japan) were used. One chip was 250 to 500 ng of total RNA derived from both tissue and cell samples labeled with Hy5 using miRCURY LNA microRNA Power Labeling Kit (Exiqon A/S, Vedbaek, Denmark), while the other chip was all of the miRNA derived from 400 uL supernatant labeled with Hy5. The labeled miRNA were individually hybridized onto the surface of the miRNA chips, and then incubated at 32°C for 16 hours. The washed and dried miRNA chips in an ozone-free environment were scanned using 3D-Gene Scanner 3000 (Toray Industries) and analyzed using 3D-Gene Extraction Software (Toray Industries). 

Messenger RNA (mRNA) Expression Profiling

For mRNA expression profiling, 3D-Gene Human miRNA Oligo Chips 25K (version 2.1; Toray Industries) were used. Total RNA of 200 to 500 ng were amplified by use of an Amino Allyl MessageAmp II aRNA Amplification Kit (Ambion, Inc., Foster City, CA, USA). The antisense (a)RNA were labeled with Cy5 using Amersham Cy5 Mono-Reactive Dye (GE Healthcare, Little Chalfont, Buckinghamshire, UK). The labeled aRNA were individually hybridized onto the Human Oligo Chip surface, and then incubated at 37°C for 16 hours. The chips were washed and dried, and then scanned in an ozone-free environment by use of a 3D-Gene Scanner 3000 (Toray Industries) and analyzed using 3D-Gene Extraction Software (Toray Industries). 

Normalized Data Processing

The digitalized fluorescent signals provided by the above described software were regarded as the raw data. All of the normalized data were globally normalized per microarray, such that the median of the signal intensity was adjusted to 25. 

Quantitative Real-Time Polymerase Chain Reaction (qRT-PCR)

Total RNA was extracted from cHCECs by use of the miRNeasy Mini Kit (QIAGEN). The complimentary (c)DNA was synthesized by use of the High Capacity cDNA Reverse Transcription Kit with RNase Inhibitor (Applied Biosystems, Foster City, CA, USA). Polymerase chain reaction was performed via TaqMan Fast Advanced Master Mix (Applied Biosystems) and TaqMan Gene Expression Assays (inventoried) (Applied Biosystems) under the following conditions: activation of enzyme at 95°C for 20 seconds, 40 cycles of denature at 95°C for 1 second, and annealing/elongation at 60°C for 20 seconds. The StepOnePlus Real-Time PCR system (Applied Biosystems) was used for PCR amplification and analysis. The levels of gene expression were normalized to that of 18S rRNA. Results were presented as 2^−ΔΔCt (relative units of expression). 

PCR Array

Total RNA was extracted from the cHCECs by use of the miRNeasy Mini kit (QIAGEN). Complimentary DNA synthesis was performed with 100 ng total RNA for a 96-well plate format by use of an RT² First Strand kit (QIAGEN). Expression of endothelial mRNAs was investigated using the RT² Profiler PCR-Array Human Extracellular Matrix and Adhesion Molecules, Human p53 Signaling Pathway, Human Fibrosis, Human Cellular Senescence, and Human EMT (all from QIAGEN) according to the manufacturer's recommendations, and then analyzed using RT² Profiler PCR Array Data Analysis Tool version 3.5 (QIAGEN). 

Integrated Analysis of Cytokines by Bio-Plex

The culture supernatants of cHCECs were harvested after 4-days cultivation and then immediately frozen and stored at −80C until analysis. The cytokine levels of the culture supernatants were analyzed by Luminex Corporation (Austin, TX, USA) xMap Technology (Bio-Plex 200; Bio-Rad Laboratories, Inc., Hercules, CA, USA) with the Bio-Plex Human 27-Plex Panel Kit (Bio-Rad Laboratories) according to the manufacturer's instructions. The measured cytokines were as follows: IL-1β, IL-1ra, IL-2, IL-4, IL-5, IL-6, IL-7, IL-8, IL-9, IL-10, IL-12p70, IL-13, IL-15, IL-17A, basic fibroblast growth factor (b-FGF), eotaxin, granulocyte colony-stimulating factor (G-CSF), granulocyte macrophage colony-stimulating factor (GM-CSF), IFN-γ, interferon-induced protein-10 (IP-10), monocyte chemotactic protein-1 (MCP-1), macrophage inflammatory protein-1α (MIP-1α), MIP-1β, platelet-derived growth factor-BB (PDGF-BB), regulated upon activation normal T cell expressed and secreted (RANTES), TNF-α, and VEGF. Standard curves for each cytokine (in duplicate) were generated using the reference cytokine concentrations supplied in the kit and then used to calculate the cytokine concentrations in the culture supernatants. 

Enzyme-Linked Immunosorbent Assay (ELISA)

Cell culture supernatant was harvested from cHCECs and centrifuged at 1580g at room temperature for 10 minutes to remove detached cells. Culture supernatant was collected and filtered through 0.220-μm filters (Millex-GV; EMD Millipore Corporation, Temecula, CA, USA). ELISA was performed by use of a Clusterin (Human) Competitive ELISA Kits (AdipoGen Corporation, San Diego, CA, USA), Procollagen type I C-peptide (PIP) EIA kit (TAKARA Bio, Inc., Kusatsu, Shiga, Japan), Human CCL2/MCP-1, Human TGF-β1, Human TIMP-1, and Human HGF (R&D Systems, Inc., Minneapolis, MN, USA), BMP2 Human Elisa Kit, PDGF-BB Human ELISA Kit, IL6 Human ELISA Kit, and Human IL8 ELISA Kit (Abcam plc., Cambridge, UK), Human Osteopontin Assay Kit and Human Osteopontin N-Half Assay Kit (Immuno-Biological Laboratories Co., Ltd., Fujioka, Gunma, Japan), Enzyme-Linked Immunosorbent Assay Kit For Early Growth Response Protein 1 (USCN Life Science, Inc., Wuhan, China), Human IP-10 ELISA Kit, Human IL-1RA ELISA Kit, Human IFN-γ ELISA Kit, and TNF-α Human ELISA Kit (Thermo Fisher Scientific, Waltham, MA, USA), Human Collagen Type IIIα1 (COL3A1) ELISA kit, and Human Collagen Type VIIIα2 chain (COL8A2) ELISA kit (CUSABIO, Wuhan, China). 

Statistical Analysis

Student's t-test was used to determine by the statistical significance (P value) of the mean values for 2-sample comparisons, and Dunnett's multiple-comparisons test was used to determine the statistical significance for the comparison of multiple sample sets. Values shown on the graphs represent the mean ± SE. 

Results

Diversification of Genes and miRNA Signatures During Cultures

Human corneal endothelium (Endo) tissues from six elderly donors and five corneal epithelium (EP) samples were analyzed by 3D-Gene for their mRNA and miRNA signatures. The discrepancy of the gene signatures between Endo and EP were evident, whereas the miRNA signatures between these two tissues were not very different, as is the case for mRNA (Fig. 1A). The signatures were nearly similar among the six donors. In addition, the changes were not evident in tissues from four age generations, both in Endo and EP, except for the mRNA signatures in neonatal donors. 

Figure 1

 

Analysis by 3D-Gene Human miRNA Oligo Chips (Human 25K ver. 2.1 and Human miRNA ver. 17; Toray Industries) of the mRNA and miRNA signatures of Endo/ EP donor tissues and cultured human corneal endothelial cells (cHCECs). (A) Fresh Endo elicited completely distinct gene signatures from those of EP from the same donors (correlation coefficient: 0.58–0.62). However, the miRNA profiles of Endo did not differ with those of EP (correlation coefficient 0.95). Fresh tissue subjected to analysis were 8/9 (donor age [DA]: 20 years; donor epithelial cell density [D-ECD, cells/mm²]: 3671/3953), 10/11 (DA: 63 years; D-ECD: 2566/2629), 12/13 (DA: 12 years; D-ECD: 3998/3976), 18 (DA: 7 months; D-ECD: 5070), and 20/21(DA: 3 days; D-ECD: 7427/7514). (B) Gene signatures of cHCECs were distinct from those of fresh Endo, as indicated by the correlation coefficient 0.7 to approximately 0.8, and those of miRNA were highly distinct (0.25∼0.50). The fresh tissue subjected to analysis were as in (A), and cHCECs subjected to analysis were #19 (passage [P] 3; DA: 67 years; D-ECD: 2503), #14 (P3; DA: 66 years; D-ECD: 3859), #18 (P3; DA: 62 years; D-ECD: 2514), #29 (P1; DA: 66 years; D-ECD: 2636), #34 (P1; DA: 65 years; D-ECD: 3692), and #35 (P1; DA: 65 years; D-ECD: 3589). (C) Scatter plots of the Endo, EP, and cHCEC gene and miRNA expression profiles. The values shown are the mean global normalization values. The upper and lower of the 2-fold line are indicated in the scatter plot.

To the contrary, both the mRNA and miRNA signatures of cHCECs were found to be greatly distinct from those in fresh Endo (Fig. 1B), and the signatures were far distant from miRNA in fresh tissue, indicating the role of epigenetic regulation in cultures. 

Varied Expression of CST-Related Genes

Coinciding with the finding that the cultures of HCECs caused the remarkable changes of gene expression, typical cultures clearly different in morphology (Fig. 2A), together with fresh Endo, were used for PCR array using the RT² Profiler PCR-Array for EMT, fibrosis, and cell senescence. Heat map revealed the clear distinction among those three groups. It is of note that two groups of cHCECs exerted the distinct gene expression profiles (Fig. 2A). Almost 50 genes were commonly elevated in two cultured cells. These included Col3A1, FN1, IGFBP3,4,5, ITGA3,5, MMP2, TIMP1, ZEB2, MAP1B, Serpine1, THBS2, TGFbR2, TGFb1, and CD44. Between the two cultures shown in Figure 2A, many up- and downregulated genes were clarified (data not shown), indicating the diversified gene expression among cHCEC cultures even under the same culture protocol. From these screening of the varying genes among cultures we selected 50 candidate genes for further validation by qRT-PCR (32 genes selected by screening and 18 auxiliary genes added considering the functional importance in CST (Table). 

Figure 2

 

Comparison of gene signatures among Normal and fibroblastic (Fibro), and Fresh tissues by use of the RT² profiler PCR-Array (QIAGEN) in regard to cell senescence, EMT, and fibrosis. (A) Microscopy images of Normal (#14, #18, and #19, P3), Fibro (#29, #34, and #35, P1), and Fresh (Endo tissues of 8, 9, and 12, no image indicated) tissues. Scale bar: 200 μm. The data of the Normal, Fibro, and Fresh Endo tissues are shown in Figure 1. (B) Heat-map hierarchical-clustering showing the comparison of mRNA extracted from fresh tissues and cHCECs to determine gene signatures.

Table.

 

Candidate of 50 Genes by PCR Array and 3D Gene

 

Validation of the Selected Genes by qRT-PCR

The candidate genes were validated via qRT-PCR by comparing their expression in the two cHCECs depicted in Figure 3A, namely #66 and #67, which were morphologically quite contrasting (some of the results are shown in Fig. 3B). Gene expressions of CDH2, TGFb2, and Col8A2 were clearly upregulated in cHCECs without CST (i.e., #66), whereas those of TIMP1, Col3A1, CD44, IL-6, IL-8, and BMP2 were upregulated in cHCECs with CST (i.e., #67). In this comparison, VIM, CD166, CD105, CD24, and MMP4 genes were expressed at comparable levels in both cHCECs. The results are summarized in Figure 3C (the Table does not include the genes expressed at very low levels). 

Figure 3

 

Validation of selected genes by qRT-PCR. (A) Microscopy images of RNA extracted from #66 (P5; DA: 23 years; D-ECD: 3504) and #67 (P5; DA: 67 years; D-ECD: 2717) showing significant morphologic contrast. (B) Bar graphs showing the qRT-PCR validation via relative mRNA amount of 50 candidate genes by RT² profiler PCR-Array and 3D-Gene chip analysis. (C) Summary of the mRNAs expressed distinctly between #66 and #67.

Validation of the Selected Genes Morphologically Graded cHCECs

The above mentioned results clearly indicated that the CST might be more diverse than imagined, even in terms of EMT, senescence, and fibrosis, thus prompting us to investigate the expression levels in more detail. Eleven cHCECs were graded by points from 0 to 10 in terms of the cultured cell morphology (graded by 3 persons; Fig. 4). The expression levels of some of the genes investigated are also illustrated in Figure 4. The first group showed the upregulated expression in inverse correlation with the order of the points, the second group showed a positive correlation, whereas the third group elicited mostly no correlation with the points. Of interest, even in the 10-point cHCECs, there was a distinct expression of CD24, indicating that this gene would be regulated more finely than the other genes shown in Figure 4. Next, validation was performed using cHCECs produced in the cell processing center under GMP (Fig. 5). In this comparison, two cHCECs seemingly distinct in their CST, C9, and C11 elicited a contrasting gene expression profile. For example, CD24 was upregulated in C9 as well as Col4A1 and 4A2, whereas MMP2, TIMP1, IL-6, IL-8, and TGFβ1 were elevated only in C11. CD44, THBS2, Col3A1, and HGF were upregulated in both C9 and C11. 

Figure 4

 

Validation of the selected genes of morphologically-graded cHCECs by qRT-PCR. The criteria for morphologically grading cHCECs were as follows: no abnormal (larger or fibrous) cells in three fields: 10 points, 0% to approximately 20% abnormal cells: 8 to approximately 9 points, 20% to approximately 40%: 6 to approximately 7 points, 40% to approximately 60%: 4 to approximately 5 points, 60% to approximately 80%: 2 to approximately 3 points, 80% to approximately 100%: 0 to approximately 1 points. Microscopy images of RNA extracted from #66 (data shown in Fig. 3), #72 (P3; DA: 4 years; D-ECD: 3818), #74 (P2; DA: 69 years; D-ECD: 2775), #68 (P2; DA: 53 years; D-ECD: 2586), #78 (P2; DA: 8 years; D-ECD: 3984), #73 (P2; DA: 59 years; D-ECD: 2696), #62 (P1; DA: 68 years; D-ECD: 3122), #70 (P2; DA: 66 years; D-ECD: 2612), #75 (P2; DA: 4 months; D-ECD: 5557), #79 (P2; DA: 4 days; D-ECD: 4529), and #85 (P0; DA: 1 day; D-ECD: 4312). (B) Bar graphs illustrating qRT-PCR evaluation of the mRNA expression levels of eight genes.

Figure 5

 

Validation of the selected genes by qRT-PCR of cHCECs produced in a cell processing center under Good Manufacturing Practices (GMP). (A) Microscopy images of the subjected cHCECs from #66 (data shown in Fig. 3), C11 (P2; DA; 26 years; D-ECD: 3322), and C09 (P2; DA: 16 years; D-ECD: 3590). (B) Bar graphs showing the expression levels of miRNAs in the cHCECs of #66, C09, and C11 as evaluated by qRT-PCR. (C) Summary of the mRNAs expressed distinctly among #66, C9, and C11.

Screening of the Cytokines Produced by the Bio-Plex Human Cytokine 27-Plex Panel

Gene expression usually corresponds to products consisting of a single chain, and does not sufficiently reflect the products as a heterodimer. To overcome this issue, and to confirm the possible assays by culture supernatants, we next analyzed the culture supernatants of diverse cHCECs by Bio-Plex Human Cytokine 27-Plex Panel. Three cHCECs (#82, #84, and #88), different in culture passages, were used for the analysis (typical results are illustrated in Fig. 6). Depending on the increase of passage frequency, most of the cytokines secreted showed a periodical decrease or increase. The findings of this investigation showed that IL-6, MCP-1, and IL-8 increased in correlation with the worsening of the culture quality. Conversely, IL-1Ra, IFN-γ, IP-10, PDGF-BB, and MIP-1β in the culture increased in correlation with the improvement of the culture quality. 

Figure 6

 

Bar graphs of cHCEC supernatants of #82 (P0, P1, P2, and P3; DA: 72 years; D-ECD: 3301), #84 (P0, P1, P2, and P3; DA: 75 years; D-ECD: 2598), and #88 (P0, P1, P2, and P3; DA: 10 years; D-ECD: 3879) subjected to analysis by Bio-Plex (Bio-Rad). (A) IL-6, (B) IFN-g, (C) MCP-1 (MCAF), (D) PDGF-bb, and (E) MIP-1b.

ELISA Assays for Reproducible Qualification of cHCECs

As the result of the extensive investigations performed in this study, we have developed a reproducible method for the qualification of cHCECs, and four cytokines, namely TIMP1, MCP-1, IL-8, and PDGF-BB, were selected. The final selection was done after testing the practicability of qualifying 32 lots of cHCECs produced in the Cell Processing Center of Kyoto Prefectural University of Medicine, Kyoto, Japan. The selection was also validated in correspondence with this qualification and that by fluorescence-activated cell sorting (FACS) at the time of cell harvesting on the day of cell-infusion therapy. The standard values for each cytokine were found to be less than 500 ng/mL TIMP1, 500 pg/mL IL-8, 3000 pg/ml MCP-1 (Fig. 7), and greater than 30 pg/mL PDGF-BB. The qualification should be performed 7 days prior to cell-injection therapy. 

Figure 7

 

Qualification of cHCECs by ELISA assay. The quality of cHCECs from C14 (P3, at week 4; DA: 22 years; D-ECD: 2813), C15 (P3, at week 4; DA: 15 years; D-ECD: 3733), C24 (P3, at day 27; DA: 29 years; D-ECD: 3413), C23 (P2, at day 31; DA: 18 years; D-ECD: 3280), and C32 (P2, at day 45; DA: 17 years; D-ECD: 3344) was checked by ELISA assay of secreted TIMP-1, IL-8, PDGF-bb, HGF, and MCP-1.

Discussion

The expansion of cHCECs from donor corneal endothelium could provide a pragmatic tool for cHCEC-infusion therapy.^19–24 Cultured HCECs expanded in an in vitro culture system can be contaminated by cHCECs with CST cells. Cultured cells also tend to be inclined toward karyotype changes.¹ Thus, cHCECs should be carefully monitored in regard to quality with a view toward application in the clinical setting. The morphologic features of cHCECs varied greatly from culture to culture, even under identical culture protocols. To date, one of the most notable obstacles against the application of cHCECs for cell-based therapy is the lack of a reproducible method to validate the quality of cHCECs. 

In this study, fibroblastic transformation of cHCECs was easily detected by microscopy observation (Fig. 2). Consequently, a relevant and more difficult qualification will be the discrimination of cHCECs without CST from those in cellular senescence (ref. Fig. 5). 

Epithelial-mesenchymal transition is aberrantly induced in many disease settings. cHCECs have an inclination toward CST into EMT. Through EMT, cell-to-cell adhesion is reduced. Cells undergoing EMT frequently gain stem cell–like properties,²⁵ including that induced by TGF-β. At the beginning of this study, we often experienced this EMT-like CST, yet later, and depending on the improvement of culture protocols, many cultures tended to show senescent-type CST. 

Senescence is regulated in a stimulus-dependent manner by a signaling network involving the tumor suppressors p53 and pRb.²⁶ Cultured HCECs enter senescence, even after a short lifespan, as monitored by SA-β-Gal staining (data not shown). The senescence is characterized by long-term cell-cycle arrest accompanied by altered cellular morphology and physiology.^27,28 However, senescent cells are metabolically active, thereby propagating the senescence process to their neighboring cells, notably via the secretion of a vast number of inflammatory mediators known as senescence-associated secretory phenotype (SASP) mediators.²⁹ The dual role of senescence in the prevention and progression of chronic disease pathogenesis has recently become well known, via a focus on the link between senescence and inflammation.²⁹ The inflammatory mediators IL-6, IL-8, and MCP-1 are also defined as SASP mediators. Thus, monitoring the maximum value for these three mediators is rational, considering their senescence-propagating effect during culture. Regulation of senescence-associated inflammation by targeting these inflammatory mediators is reportedly beneficial in the treatment of aging-related diseases.²⁹ The SASP includes a large number of inflammatory cytokines and chemokines,³⁰ thus providing a strong link between senescence and inflammation.^31,32 The role of PDGF-BB in qualifying a HCEC culture has yet to be clarified, and is an elusive issue that requires further investigation. 

Although many previous works have dealt with the relationship between cHCEC transcripts and their cell-cycle arrest or cellular senescence, none of those studies have correlated their protein products (SASP) in culture supernatant with their functions, especially with CST intrinsic in HCEC cultures.^33–38 

In this present study, we succeeded in providing a candidate of monitoring the cell quality in a noninvasive way. Comprehensive survey of the gene and miRNA signatures of diverse cHCECs made it possible to develop the ELISA assay to discriminate the quality of cHCECs. The selection of SASP was, however, coincidental. The assay using secreted miRNA will be published elsewhere. To the best of our knowledge, our findings are the first to describe the discrimination of cHCECs with or without CST by culture supernatants. 

The qualification assay presented in this study may enable the availability of the highest-quality cHCECs for cell-injection therapy for HCEC tissue damage due to Fuchs' endothelial corneal dystrophy, trauma, or surgical intervention. 

It should be noted that in vivo data to support that the quality control standard established here, which has a direct impact on the outcome of endothelial repair, is critical. The good correlation between the in vivo endothelial density measured by specular microscopy, and the improvement of central corneal thickness and the visual acuity post cHCEC injection therapy with the standard established here, will be published elsewhere (manuscript in preparation). 

Acknowledgments

The authors thank Michio Hagiya. The authors also thank Yuki Hosoda, and Shunsuke Watanabe for constant encouragement, and Yoko Hamuro, Keiko Takada, and Tomoko Fujita for their excellent secretarial assistance. The authors also wish to extend their sincere thanks to John Bush for his thorough and excellent review of the manuscript, and to Satoshi Kondo (Toray Industries, Inc.) for his invaluable assistance with the 3D-Gene analysis. Finally, the author's wish to extend their sincere thanks to Noriko Koizumi and Naoki Okumura for their helpful discussion. 

Supported by the Highway Program for Realization of Regenerative Medicine from Japan Agency for Medical Research and Development, AMED and JSPS KAKENHI Grant Numbers JP26293376. 

Disclosure: M. Ueno, None; K. Asada, None; M. Toda, None; U. Schlötzer-Schrehardt, None; K. Nagata, Toray, Inc. (E); M. Montoya, None; C. Sotozono, None; S. Kinoshita, None; J. Hamuro, None