Safety and efficacy of combination of suberoylamilide hydroxyamic acid and mitomycin C in reducing pro-fibrotic changes in human corneal epithelial cells

Corneal haze post refractive surgery is prevented by mitomycin c (MMC) treatment though it can lead to corneal endothelial damage, persistent epithelial defects and necrosis of cells. Suberanilohydroxamic acid (SAHA) however has been proposed to prevent corneal haze without any adverse effects. For clinical application we have investigated the short and long term outcome of cells exposed to SAHA. Human donor cornea, cultured limbal epithelial cells, corneal rims and lenticules were incubated with SAHA and MMC. The cells/tissue was then analyzed by RT-qPCR, immunofluorescence and western blot for markers of apoptosis and fibrosis. The results reveal that short term exposure of SAHA and SAHA + MMC reduced apoptosis levels and increased αSMA expression compared to those treated with MMC. Epithelial cells derived from cultured corneal rim that were incubated with the MMC, SAHA or MMC + SAHA revealed enhanced apoptosis, reduced levels of CK3/CK12, ∆NP63 and COL4A compared to other treatments. In SAHA treated lenticules TGFβ induced fibrosis was reduced. The results imply that MMC treatment for corneal haze has both short term and long term adverse effects on cells and the cellular properties. However, a combinatorial treatment of SAHA + MMC prevents expression of corneal fibrotic markers without causing any adverse effect on cellular properties.

by modulating cell proliferation and differentiation 7 . Agents limiting keratocytes proliferation and differentiation to myofibroblast can prevent corneal haze formation post PRK.
Mitomycin C derived from Streptomyces species has anti-neoplastic and antibiotic functions exerted by its DNA alkylating property during cell division. It has been approved to be used for preventing haze on corneal surface by inducing apoptosis of keratocytes and stromal cells 8 . It has been elucidated that conditioned medium from MMC treated corneal epithelial cells lead to senescence of corneal epithelial cells and secreted senescence associated secreted proteins that suppressed the collagen deposition implicating a definitive role of corneal epithelial cells in preventing haze in the presence of MMC 9 . It has been shown that prophylactic usage of MMC has more beneficial effects rather than therapeutic effects 10 . The widely accepted prophylactic usage of MMC in ophthalmic clinics for PRK and LASEK are supported by the fact that MMC marginally reduces stromal keratocytes that can be repopulated within a year, none or minimal decrease in endothelial cell density [11][12][13] . However, since several animal studies have revealed decrease in endothelial cell density, further studies investigating the physiological functions are warranted 14 . The dosage and duration of MMC exposure is critical as has been reported in several studies 10 . The debilitating effects of topical application of MMC can be extended persistent epithelial defects, limbal and scleral necrosis, corneal endothelial damage and corneal perforation 15 . In an effort to explore other agents with similar functions as MMC, potential of SAHA has been evaluated.
Studies have reported that histone deacetylase inhibitors are shown to reduce TGFβ1 induced myofibroblast formation along with fibrotic changes using in-vitro model 16 . An analogue of Trichostatin A, vorinostat (Suberoylamilide Hydroxyamic Acid-SAHA) has been approved by United States Food and Drug Administration for medical use in cutaneous T-cell lymphoma 17 . In vitro as well as in vivo animal studies have conclusively demonstrated that SAHA can reduce corneal haze and also short and long term damage to corneal endothelium [18][19][20] . Studies of have shown SAHA to be less toxic at their efficacious dosage compared to MMC 20 . Therefore, it is imperative to establish the individual effects of SAHA on human eyes as well as its combinatorial treatment with MMC.
However, there are no safety and efficacy studies of a combinatorial treatment of SAHA and MMC on human tissues. In order to be able to investigate the applicability of the treatment for clinical trials a targeted study using human samples is warranted. Therefore, our study investigates the effect of SAHA with/without MMC in in vitro primary culture systems as a step towards understanding the safety and efficacy of the combinatorial treatment to prevent corneal haze post PRK.

Concentration of SAHA and MMC based cell viability. Cultured corneal epithelial cells were treated
with different concentrations of SAHA dissolved in DMSO or MMC dissolved in saline and incubated for 24 h. The cell viability was performed with trypan blue dye exclusion analysis. The results revealed that at concentrations higher than 10 μM there was a decrease in the percentage of viable cells in SAHA (Fig. 1A), whereas at concentrations higher than 0.005% there was a decrease in percentage of viable cells for MMC (Fig. 1B). Hence, for all our experiments we have used SAHA at a concentration of 5 μM. There was a significant decrease in the percentage of viable cells at 50 μM (p = 0.0352) and 100 μM (p = 0.0042) compared to untreated controls. Incubation with MMC showed significant decrease in viability at concentrations of 0.01% (p = 0.046) and 0.02% (p = 0.037) compared to controls. Hence, we have used 0.005% MMC and 5 μM of SAHA in this study for treating cells.
Effect of SAHA and MMC on corneal explants. Corneal buttons were incubated with SAHA and MMC to evaluate their effects on expression of pro-fibrotic marker αSMA. The results revealed that MMC upregulated αSMA expression both at protein ( Fig. 2A-F) as well as mRNA levels (Fig. 2G). Corneal buttons incubated with MMC revealed significant increase of αSMA mRNA levels both in 1 week (p = 0.009) as well as 1 month (p = 0.024) incubation compared to 1 week control corneal buttons. There was observed a significant reduc-  www.nature.com/scientificreports/ tion in the αSMA mRNA levels in corneal buttons incubated with SAHA for 1 week (p = 0.0002) and 1 month (p = 0.001) compared to those incubated with MMC for 1 week (Fig. 2G). Excised corneal lenticules were treated with SAHA, TGFβ individually as well as in combination. The mRNA expression revealed significant decrease in αSMA levels (p = 0.0022) by SAHA treatment with or without pro-fibrotic stimulation (Fig. 3A). However, no difference was detected in the mRNA levels of IL6 (Fig. 3B).

Status of apoptosis levels in presence of SAHA and MMC. Cultured limbal epithelial cells were
treated with SAHA and MMC alone or in combination (study group 1). The cell viability was ascertained using Trypan blue vital staining. The percentage of viable cells was similar in control cultures and those that were exposed to SAHA and MMC + SAHA. However, there was a significant decrease in the percentage of viable cells in cultures treated with MMC compared to untreated cultures (p = 0.0034) and those incubated with SAHA (p = 0.0106) (Fig. 4A). Furthermore, to determine the effect of treatment on the human eye, cadaveric corneal rims were incubated with SAHA and MMC alone or in combination (study group 2). Cell viability assay showed results similar to those obtained by incubating the reagents in cultured limbal epithelial cells. Corneal rims incubated with SAHA and MMC + SAHA did not show any difference compared to the untreated corneal rims. However, there was a significant decrease in the percentage of viable cells in rims incubated with MMC alone (p = 0.0038) compared to untreated rims (Fig. 4C). Gene expression analysis revealed significant upregulation of the ratio of Bax and Bcl2 levels in study group 1 (p = 0.028) as well as study group 2 (p = 0.0022) exposed to MMC compared to control cultures, respectively (Fig. 4B,D). There was no significant difference in the ratio of the mRNA levels of Bax and Bcl2 in cells obtained from cultures and corneal rims treated with SAHA and MMC + SAHA in comparison to untreated cultures. Furthermore, corneal rims of study group 2were stained for BCL2 protein expression. The results revealed a significant decrease in the mean fluorescent intensity of BCL2 positive cells obtained from corneal rims exposed to MMC compared to those in controls, SAHA and MMC + SAHA (p < 0.0001) (Fig. 4E,F).

Status of fibrotic gene expression.
Corneal rims from study group 2 were analyzed for expression of fibrotic markers. Corneal rims incubated with SAHA (p = 0.0004) and MMC + SAHA (p = 0.0452) showed significant decrease in the mRNA levels of αSMA compared to those incubated with MMC. The rims treated with MMC showed elevated expression of αSMA compared to controls. SAHA incubated corneal rims showed significant (p = 0.0452) decrease in the mRNA levels of αSMA compared to controls (Fig. 5A). There was a significant decrease in the mRNA levels of TGFβ in the presence of SAHA (p = 0.0255), MMC and MMC + SAHA (p = 0.0138) compared to control rims (Fig. 5B). Similarly, the mRNA levels of Lox and Coll4A showed decreased mRNA expression levels in cells of study group 2 compared to controls (Fig. 5C,D). Significant reduction in the mRNA levels of Lox was noted in rims incubated with SAHA (p = 0.0022) compared to controls (Fig. 5C). Similarly Coll4A mRNA levels were significantly reduced in rims incubated with SAHA (p = 0.0004), MMC (p = 0.0453) and MMC + SAHA (p = 0.0453) compared to controls (Fig. 5D). Furthermore, immunofluorescence staining was performed to corroborate the results obtained by mRNA analysis using samples of study group 2. Quantification of the mean fluorescent intensity of the images revealed that cells obtained from corneal rims incubated with SAHA (p < 0.0001) and MMC + SAHA (p = 0.0002) showed significant lower expression of αSMA staining positivity compared to controls. However, cells obtained from corneal rims treated with MMC showed no change in the levels of αSMA with respect to control. The αSMA positivity was significantly high in cells obtained from rims incubated with MMC compared to those with SAHA (p < 0.0001) and MMC + SAHA (p < 0.0001) (Fig. 5E,H) Cells obtained from corneal rims incubated with SAHA (p < 0.0001; p < 0.0001), MMC (p < 0.0001; p = 0.0114) and MMC + SAHA (p = 0.0034; p < 0.0001) showed significantly lower mean fluorescent intensity of TGFβ and COLL4A levels compared to controls (Fig. 5F,G,I,J). Additionally, it was noted that TGFβ mean fluorescent intensity was significantly lower in SAHA (p = 0.0002) and MMC (p = 0.0003) incubated rims compared to those incubated with SAHA + MMC (Fig. 5I). Similar results were obtained in the levels of COLL4A mean fluorescent intensity levels. Significant reduction was observed in COLL4A levels in rims incubated with   www.nature.com/scientificreports/ SAHA (p = 0.0057) and MMC + SAHA (p < 0.0001) compared to those incubated with MMC alone (Fig. 5J). The results show that SAHA reduced αSMA expression levels whereas MMC promoted the expression of αSMA. Analysis of gene expression levels of study group 1 samples revealed that MMC + SAHA treatment decreased the expression of fibrotic markers compared to control as well as those incubated with MMC + SAHA. Cultures treated with MMC showed upregulation in mRNA levels of αSMA (p = 0.0453), TGFβ, Coll4A and Lox (p = 0.0022) compared to the controls (Supplementary Fig. S2A-D). In the presence of SAHA, the MMC induced upregulation of mRNA levels was significantly reduced. A significant reduction of αSMA (p = 0.0004), TGFβ p = 0.0022), Lox and Coll4A (p = 0.0022) mRNA levels was found in MMC + SAHA treated cultures compared to those incubated with MMC alone (Supplementary Fig. S2A-D). However, in the presence of MMC, SAHA as well as MMC + SAHA the mRNA levels of proliferative marker Ki67 and Cyclin D1 were reduced compared to control mRNA levels ( Supplementary Fig. S2E). A significant decrease in the mRNA levels of Ki67 (p = 0.0022) and Cyclin D1 (p = 0.0022) was noted in cells incubated with MMC + SAHA compared to controls. A decrease in the mRNA levels of Coll3A1, Decorin and Fibronectin 1 was noted in samples of study group 1 compared to controls. A significant decrease in mRNA levels of Coll3A1 (p = 0.0022) and Decorin (p = 0.0022) was observed in cells treated with SAHA compared to controls. A significant decrease in the levels of Fibronectin (p = 0.0022) was observed in cells treated with MMC compared to controls ( Supplementary Fig. S2F).

Regulation of MDR genes in the presence of SAHA + MMC.
In an attempt to understand the underlying mechanism for these effects of SAHA and MMC treatments, we analysed the gene expression levels of multidrug resistance (MDR) proteins. Cells obtained from study group 1 were analysed for mRNA expression of MDR proteins. In the presence of SAHA, a significant upregulation of Abcg2 (p = 0.0453) and Abcb1 (p = 0.0453) was observed compared to untreated day 14 differentiated limbal epithelial cultures. But no change was observed in expression levels of Abcg2 and Abcb1 in MMC treated cultures compared to untreated cultures. A significant decrease was found in mRNA levels of Abcg2 (p = 0.0004) and Abcb1 (p = 0.0004) in cells incubated with MMC compared to those incubated with SAHA alone. Contrarily, a significant increase in the mRNA levels of Abcg2 (p = 0.0453) and Abcb1 (p = 0.0453) in cells treated with MMC + SAHA compared to MMC (Fig. 6A). A similar result was obtained in the mRNA levels of cells obtained from the study group 2. Corneal rims incubated with SAHA showed significantly elevated mRNA levels of Abcg2 (p = 0.0453) and Abcb1 (p = 0.0453) compared to untreated controls. In the presence of MMC, the mRNA of cells from corneal rims revealed reduced levels of Abcg2 and Abcb1 compared to untreated controls. Corneal rims incubated in MMC + SAHA show significantly elevated gene expression of Abcg2 (p = 0.0453) and Abcb1 (p = 0.0453) compared to those treated with MMC alone. There was also a significant upregulation of Abcg2 (p = 0.0004) and Abcb1 (p = 0.0004) mRNA in rims treated with SAHA compared to those with MMC ( Fig. 6B). Flow cytometry analysis for ABCG2 positive population in study group 1 samples revealed increase in the percentage of positive population in SAHA (p = 0.0310) compared to control. The ABCG2 positivity was significantly high in SAHA treated cultures compared to those incubated with MMC (p = 0.0006) (Fig. 6C,D).

Differentiation status of corneal limbal epithelial cells after treatment. Samples of study group 3
were analyzed in order to evaluate the differentiation potential of the cells in the corneal rim exposed to the different treatments. Phase contrast images revealed decreased number of cultured cells in corneal rims treated with MMC whereas no difference was noted in the cells treated with SAHA, SAHA + MMC and control rims (Supplementary Fig. S1). There was a significantly elevated ratio of Bax to Bcl2 mRNA levels in day 14 differentiated cells obtained from corneal rims treated with MMC (p = 0.0004) compared to control rims. Day 14 differentiated cells obtained from SAHA treated corneal rims, revealed significantly low ratio of Bax to Bcl2 mRNA compared to differentiated cells obtained from untreated control rims (p = 0.0453) and MMC treated rims (p = 0.0004). Day 14 differentiated cells obtained from MMC + SAHA (p = 0.0453) treated corneal rims showed significantly low levels of ratio of Bax to Bcl2 compared to differentiated cells from MMC treated rims (Fig. 7A). The results show that day 14 differentiated cells obtained from corneal rims treated with MMC showed lower mRNA expression of Ck3/Ck12 compared to cultures of SAHA treated corneal rims (p = 0.0022; p = 0.0022). The mRNA levels Ck3 and Ck12 of day 14 limbal epithelial cultures from MMC treated corneal rims were lower than the cells cultured from MMC + SAHA treated corneal rims. mRNA expression levels of Ck3/Ck12 were comparable in corneal rims treated with SAHA alone, SAHA + MMC or control cultures (Fig. 7B). Further, the mRNA expression of corneal epithelial stem/progenitor marker (Abcg2 and ∆Np63) levels of corneal rim treated with MMC were significantly lower than those treated with SAHA (p = 0.0022; p = 0.0022). However, the mRNA expression of Abcg2 and ∆Np63 was higher in day 14 cultured cells obtained from SAHA + MMC treated corneal rims compared day 14 cultured cells obtained from MMC treated corneal rims. The mRNA expression levels of Abcg2 and ∆Np63 were similar in control corneal rims and those treated with MMC + SAHA (Fig. 7C). There was a decrease in the mRNA levels of Ki67 and Cyclin D1 in day 14 differentiated cells obtained from corneal rims treated with SAHA + MMC (p = 0.0022; p = 0.0022) compared to control levels (Fig. 7D). Western blot analysis revealed significantly lower expression of CK3/CK12 in day 14 cultured cells obtained from corneal rims treated with MMC (p = 0.0022) compared to untreated rims. However there was no difference in the expression of CK3/CK12 in day 14 cultured cells obtained from corneal rims incubated with MMC + SAHA and untreated rims. Expression of αSMA was significantly higher in day 14 cultured cells obtained from rims treated with MMC compared to cultured cells obtained from control (p = 0.0453), SAHA (p = 0.0004) and SAHA + MMC (p = 0.0453) corneal rims. No significant difference could be observed in day 14 cultured cells obtained from corneal rims treated with SAHA, MMC + SAHA and controls. There was a significant decrease in the expression of BCL2 in comparison to day 14 cultured cells obtained from rims treated with SAHA (p = 0.0359) and control cultures (p = 0.0254).    (Fig. 7E,F, Supplementary Fig. S3).

Discussion
Corneal haze formation post PRK has been one of the primary complications of this laser induced refractive correction procedure. Photorefractive keratectomy (PRK) and laser assisted subepithelial keratomileusis (LASEK) induce corneal epithelial proliferation, migration, differentiation and hemidesmosome formation in an effort to heal the surgically induced epithelial defects 21,22 . Corneal epithelial change drive haze in post-refractive sugery espite the effect in stromal epithelium in development and propagation of sub-epithelial stromal haze 23 . The wound contraction is driven by components of extra-cellular matrix and differentiation of fibroblasts to myofibroblasts or activated keratocytes 24 . Damage to the basement membrane lead to secretion of abnormal stromal extra-cellular matrix components and deposition of abnormal collagen fibrils by activated keratocytes in subepithelial regions leading to the formation of haze 25 . The anti-scarring effects are dependent on the keratocyte death 6,26 . The proliferation and migration of the activated keratocytes in the stromal region gets initiated 12-14 h post corneal epithelial injury activated by surgery 6 . However, studies have shown that the status of corneal epithelial cells also contributes in triggering the cascade leading to haze resolution 9 . Further the effect of drugs to treat haze on limbus need to be addressed as long term outcome may lead to limbal stem cell deficiency. Hence, in this study we focused on corneal epithelium and limbal epithelial cells of ex vivo human samples. It has been shown that a topical application of MMC can lead to an altered phenotype for the corneal epithelial cells as well as a gene expression pattern, as an indicator for cell senescence 27 . The effect of MMC on corneal epithelial cells, keratocytes and corneal endothelial cells is already well known 28 . Animal studies have shown protective effect of SAHA on endothelial cells 26 . It has been illustrated that senescence associated secretory proteins from MMC treated corneal epithelial cells play pivotal role in mitigating the role of MMC in preventing haze 9 . Usage of MMC in ophthalmic surgery was first done as an adjunct in pterygium excision in 1963 and it first use as a corneal healing modulator was performed after almost after more than 35 years in PRK 29,30 . MMC prevents sub-epithelial haze by inducing senescence in activated keratocytes thereby reducing the abnormal ECM by attenuating the TGFβ1 signalling in activated fibroblasts 8,31 . Epithelial basement membrane proteins such as laminin332, perlecan, nidogen are enhanced by MMC treated corneal epithelial cells 9,32 . These proteins play vital role in preventing haze. The cause of haze is being triggered by the initial surgery induced insult to corneal epithelial cells. However, further studies have revealed that MMC reduces corneal fibrosis by reducing collagen deposition and expression of αSMA and fibronectin by corneal fibroblasts and simultaneous deposition of epithelial basement membrane protein Laminin332 9 .
The corneal scarring is induced by excessive wound healing caused by a variety of mechanisms typically involving the TGFβ pathway and its associated proteins. While MMC is clinically used to reduce scarring, its mode of action is to induce apoptosis and inhibition of mitosis in myofibroblast precursors by causing cellular DNA damage 33 . However, in the present study we have used sub-lethal dosage of MMC on in order for us to compare the effects of MMC and SAHA on the differentiated limbal epithelial cells without inflicting any wound. Limbal cells are exquisitely more susceptible to stress compared to corneal fibroblasts and hence clinical application of these drugs needs to be learned.
The effect of MMC treatment on expression of inducible genes can be evaluated if MMC is used at the time of injury but not after wound repair has started. Hartnick et al., found that MMC had no beneficial effect in comparison to placebo in patients undergoing stent or endotracheal tube removal with 1 year follow-up 34 . Similarly our present study is done by treating corneal epithelial cells obtained by differentiating limbal epithelial cells with MMC without any inflicted wound. While this experimental design may be considered a limitation of our study, it should be noted that our aim was to investigate the safety and efficacy of SAHA at the human cornea keeping MMC treatment as a control. MMC is known to have clear negative effects on corneal cells stability and repopulation post refractive surgery, thereby making it important to have alternatives 35,36 . Upregulation of profibrotic genes such as LOX, TGFβ and corresponding decrease in decorin expression demonstrates the pleiotropic effects of MMC in this model of limbal derived epithelial cells.
Decorin, has anti-fibrotic properties and studies have shown its antagonistic functionality towards TGFβ thereby reducing fibrosis 39 . Collagen III, a key component of extracellular matrix along with Collagen I, V, lumican, keratocan and is expressed weakly in cornea in physiological conditions. Its expression increases in a time-dependent manner during wound healing and inflammation 40 . In our experiments cells were cultured in physiological conditions before incubating with MMC, this might have resulted in lower mRNA levels of Col 3A1 gene. It has also been shown that Col 3A1 expression increases in the initial stages of wound healing and eventually replaced with Col 1 41,42 . Col 4A is primarily present in the corneal basement membrane, Bowman's membrane and Descemet's membrane with role in development, maintainence and wound healing process of the cornea 43,44 . Treated with TGFβ cells are known to upregulate the expression of myofibroblast markers along with Col 4A 45 . The derranged gene expression profile could be attributed to the DNA interstrand crosslinks that could not be repaired. It has been shown that terminally differentiated cells such as muscle and nerve cells lack a normal DNA repair mechanism resulting in accumulated DNA damage 46 . A similar effect was observed in our differentiated corneal epithelial cultures treated with MMC.
At present MMC remains the primary drug of choice to manage the burden of adverse effects. Reports have shown that prophylactic usage of MMC resulted in poor differentiation of epithelial cells, slow wound healing due to its effect on keratocytes in the stroma, endothelial damage, corneal perforation, limbal and scleral necrosis 37,47,48 . Since, MMC acts on dividing cells, its effects can have long term ramifications 49 . The primary side effect of MMC treatment is endothelial cells damage though there has been no case of endothelial decompensation in any patient 50 www.nature.com/scientificreports/ a decrease in endothelial cell density 33,52 . This would prompt an investigation for endothelial cell physiological function on a long term study with MMC. An intact and healthy corneal epithelial cell layer would protect the corneal endothelial cell layer 53 , hence the study was attempted to investigate the role of SAHA in protecting MMC treated corneal epithelial cells. SAHA has been investigated as a plausible alternative to MMC application to avoid haze post refractive surgery. Animal experiments have provided safety and efficacy of SAHA in order to avoid post-refractive laser surgery [18][19][20] . In continuation with earlier studies, we have evaluated the safety and efficacy of a combinatorial treatment of SAHA and MMC on primary donor derived human corneal cells and tissues. Our data shows significant decrease in the cell viability at 25 μm concentration of SAHA (Fig. 1). Hence we used 5 μm concentration of SAHA in subsequent experiments. Gronkiewicz et al., have used 2.5 μm of SAHA on canine corneal fibroblast cultures to study its role in corneal fibrosis 54 . Buss et al., have shown that αSMA expression is significantly higher in cultured equine corneal myofibroblasts compared corneal keratocytes and fibroblasts 55 . Interestingly, in our experiments on human donor derived tissues, the expression of αSMA post refractive surgical tissue ablation was significantly reduced by SAHA in comparison to MMC or control treatments. Though MMC has been shown to prevent haze formation successfully, it can induce DNA (CpG-rich promoter, exon, and gene upstream regions) damage leading to abnormal proliferation and gene expression patterns 37 26 . In the presence of TGFβ stimulation on corneal lenticules, SAHA prevented the mRNA expression of αSMA, as described in other systems 63 .
Cell viability assays revealed that MMC treated 14-day limbal epithelial cultures as well as corneal rims had significantly lower percentage of viable cells in agreement with other studies which have shown MMC induced cell apoptosis 70 . This observation is further supported by the enhanced ratio of mRNA levels of Bax/Bcl2. We as well as others have a shown that the ratio of Bax/Bcl2 determines the cell susceptibility to apoptosis [71][72][73][74] . However, in the presence of SAHA, the MMC induced apoptosis of cells were prevented. Woo et al., showed higher TUNEL-positivity implicating higher apoptosis in rat cornea on MMC treatment was significantly reduced when low dosage of MMC treatment was combined with SAHA 68 . Kim et al., also showed that a low dose of MMC along with SAHA prevented apoptosis of conjunctival epithelial cells as well as tenon capsule fibroblasts in rabbits 75 . Cells obtained from corneal rims exposed to MMC showed significantly lower BCL2 positivity compared to controls. Thus, the data demonstrates that in the presence of SAHA, deleterious effect of MMC is mitigated.
MMC, SAHA and their combination treatment decreased the expression of corneal fibrotic markers (TGFβ, LOX and COL4A) as shown by the mRNA as well as immunofluorescence results. Similarly, Anumanthan et al., have shown using rabbit models of corneal haze that MMC and SAHA are effective in preventing corneal haze post photorefractive keratectomy 20 . Woo et al. have shown using rat models and cultured human corneal epithelial cells that though MMC and SAHA prevent proliferation of corneal myfibroblasts, a combination of low dosage of MMC with SAHA is better as MMC at high dosage is toxic to cells 68 . Conjunctival epithelial cells cultured from rabbits were used to study subconjunctival fibroblast in the presence of MMC and SAHA. Expression of proliferative markers was also reduced in the presence of MMC, SAHA and a combination of MMC and SAHA. Studies have reported the role of HDAC inhibitors in preventing cellular proliferation 76 . Additionally, the MMC is known to prevent proliferation of keratocytes 20,68 . Our findings are in agreement with the known mechanism that suppression of myofibroblast proliferation would prevent fibrosis formation 68 . Hence, a combination of MMC + SAHA might have better clinical applicability in blocking corneal haze post PRK.
SAHA is a known modulator of MDR group of proteins 77 . Hence, our results revealed that in the presence of SAHA and MMC + SAHA there was an upregulation of the mRNA/protein expression of Abcg2 and Abcb1 both in cells obtained from corneal rims as well as in cultured differentiated limbal epithelial cells. It has been shown that through the activation of MDR proteins effluxes MMC thereby reducing the harmful effects of MMC on cells 78,79 . It has been shown that Abcb1 and Abcg2 play crucial role in imparting efflux of MMC 80,81 .
Mitomycin C belongs to a class of antitumor antibiotics that needs to undergo bioreductive alkylation by enzyme systems such as DT-diaphorase, NADPH-cytochorme P-450 reductase, NADPH-cytochorme C reductase, xanthine oxidase and falvoprotein transhydrogenase 82 82,83 . The alkylating groups of MMC bind to two nitrogen atoms forming interstrand DNA links thereby crosslinking the DNA strands and inhibiting cell proliferation 84 . MMC has been primarily used in cancer treatment primarily because of its antiproliferative role 82 . Intracellular accumulation of MMC results in high number of DNA crosslinking leading DNA damage. This eventually prevents normal DNA replication causing cell death [85][86][87][88] . It has been shown that low dosage of MMC activated Abcb1 gene and protein 89,90 . It has been shown that COX-2 inhibitors prevent induction of MDR proteins 86 . Coadministration of COX2 inhibitor, celecoxib with MMC in COX2-deficient urinary bladder cell line UMUC-3, resulted in increased intracellular MMC levels 86 . The cytosolic concentration of MMC has been reported in the literature [85][86][87][88] . Wilson et al. using colon cancer cell line showed that resistance to MMC was induced by decrease in DNA crosslink formation 91 . Resistance to MMC is mediated partly by blocking the alkylation of MMC and also by efflux mechanism of multi-drug resistance proteins 92,93 . Dorr et al. using a mouse leukaemia cell line, L1210 demonstrated that resistance to MMC is imparted by expression of P-glycoprotein on the membrane and decreased accumulation of intracellular MMC 94 . In similar most likely, our results here support the efflux mechanism of imparting MMC resistance, since SAHA induces expression of P-glycoprotein as well as BCRP, multi-drug resistant proteins. The safety of SAHA and MMC treatments were evaluated in a longer timescale by culturing the donor corneal rim derived primary cells. These cells were cultured in corneal differentiation medium to evaluate the effect of treatment on cell differentiation and proliferation capacity. The corneal rim harbours limbal epithelial as well as transient amplifying cells which are essential for the health of the corneal surface 95 . Results revealed that a combinatorial treatment with SAHA and MMC prevented any compromise of the differentiation potential towards corneal lineage cells when compared to MMC alone. Moreover, the rims exposed to MMC + SAHA did not show any adverse effects on corneal stem/progenitor marker population when compared with MMC treatment.
These findings implicate that a combination treatment of MMC + SAHA or possibly SAHA alone, could effectively prevent corneal haze formation post PRK surgery in human eyes by reducing fibrosis without excessive cell death or compromised corneal cell differentiation. Additionally, our in-vitro data suggests that the combination treatment of MMC + SAHA might also be safe in the long term with respect to maintaining cellular health. Further clinical trials with a combinatorial treatment on post PRK complications are warranted to establish this treatment as a standard of care to prevent corneal haze.

Materials and methods
Human corneoscleral rim collection and limbal primary culture. The current study was approved by the Narayana Nethralaya institutional Review Board. Samples were collected as per Narayana Nethralaya ethics committee (Bangalore, Karnataka, India) and adhered to the tenets of the Declaration of Helsinki guidelines. Informed written consent was obtained from all subjects prior to sample collection. The tissue used for this study was sourced from the Shankar Anand Singh Eye Bank (Bangalore, Karnataka, India). The residual corneoscleral rim post corneal transplantation was used for the limbal primary culture. Tissue samples were collected from subjects within the age group of 25-65 years and both genders were included. We used 50 corneoscleral rims in this study. Limbal primary culture was carried out with modification of a previously described protocol 95 . Briefly, the excessive sclera and peripheral cornea were trimmed from the corneal rim. The enzymatically treated Dispase II (2 mg/ml in DMEM; Sigma-Aldrich Missouri, USA) for 30 min at 37 °C, 5% CO 2 ) corneoscleral rim was then chopped into small pieces, placed on de-epithelialized human amniotic membrane (dHAM) and cultured for obtaining corneal epithelial cells using growth medium. The growth medium contained Dulbecco's Modified Eagle's Medium (DMEM)/Ham's F12 nutrient mix (v/v, 1:1), human recombinant EGF (10 ng/ml), human recombinant insulin (5 μg/ml) (Gibco, Grand Island, New York, USA), penicillin (100 U/ml), streptomycin (100 μg/ml) and amphotericin B (2.5 μg/ml) (HiMedia, Mumbai, Maharashtra, India) along with 10% fetal bovine serum (Gibco, Grand Island, New York, USA). The protocol followed for limbal cells culture was the explant culture technique.
Study group 1. Cultured limbal epithelial cells differentiated to corneal lineage after 14 days of culture are used. These cells are treated 24 h with 5 µM SAHA, 0.005% MMC, 0.005% MMC + 5 µM SAHA or mock control. These cells are then collected for further analysis for mRNA or protein.

Study group 2.
In this group corneoscleral rims were treated with 5 µM SAHA, 0.005% MMC, 0.005% MMC + 5 µM SAHA or mock control for 24 h. At the end of incubation, the peripheral cornea and the sclera tissue was trimmed from these corneoscleral rims. Finally, the limbal epithelial cells were scarped and collected for mRNA and protein analysis 95 .
Study group 3. In this last group corneoscleral rims were treated with 5 µM SAHA, 0.005% MMC, 0.005% MMC + 5 µM SAHA or mock control for 24 h. At the end of incubation, limbal epithelial cells were isolated from the rims. Cells obtained by the rims were cultured on de-epithelialized human amniotic membrane and differentiated to corneal lineage by growing them for 14 days. At the end of 14 days cells were collected for mRNA and protein analysis 95 . Human lenticule and treatment. Lenticules (n = 4) obtained intraoperatively during SMILE surgery from each subject undergoing refractive correction surgery were included in the study. Each lenticule was cut into four equal pieces and were subjected to 48 h TGFβ (10 ng/ml), SAHA (5 µM), combination of SAHA (5 µM) + TGFβ ( www.nature.com/scientificreports/ each lenticule followed by cDNA conversion and amplified using real time PCR for αsmooth muscle actin (profibrotic marker) and interleukin 6 (pro-inflammatory marker) expression levels. Actin served as housekeeping control.
Human donor cornea collection and treatment. Donor human corneas unsuitable for transplantation were subjected to excimer laser ablation. Corneal tissues (n = 4) were treated with MMC (0.005% per ml) and SAHA (5 µM) for 1 week and 1 month to study laser induced wound and differential healing response. Post 1 month treatment, corneal button was cut into two halves, one part subjected for gene expression analysis and other half was used for making corneal paraffin blocks. After 1 week treatment, total RNA was isolated from the treated corneal tissues and were analysed for α-smooth muscle actin.  CAC TCC TCC ACC TTT GAC  TGT TGC TGT AGC CAA ATT CGTT   bax  NM_138761  TTG CTT CAG GGT TTC ATC CA  AGA CAC TCG CTC AGC TTC TTG   bcl2  NM_ 000633  TGG CCA GGG TCA GAG TTA AA  TGG CCT CTC TTG CGG   . Cells were stained for FACS analysis. Briefly, cells were trypsinized and fixed with 4% paraformaldehyde before being permeabilized with 0.1% Triton X (Qualigens, Mumbai, India). The permeablized cells were stained with antibody for ABCG2. Unstained cells and cells stained with secondary antibodies alone were used as controls. The fluorescence emitted by cells in FL2 channel was recorded and analyzed using BD CellQuest Pro software, (BD Biosciences, CA, USA) 96 . Primary and secondary antibodies are listed in Table 2.

Cell viability count-Trypan blue.
Immunofluorescence. Scrapped cells obtained from treated corneal rims were fixed using 2% paraformaldehyde (Sigma Aldrich, MO, USA) for 10 min, permeabilized with 0.1% Triton X 100 (Thermo Fischer Scientific, Mumbai, India) in 1× PBS for 15 min, blocked with 1% bovine serum albumin/phosphate-buffered saline (Himedia, Mumbai, India) at room temperature for 1 h. The blocked samples were incubated with primary antibody overnight at 4 °C. After a brief rinse in 1× PBST (0.02% Tween 20 (MP Biomedicals, CA, USA), cells were incubated with secondary antibodies for 1 h. The slides were finally mounted using Vector shield containing 2-(4-amidinophenyl)-1H-indole-6-carboxamidine (DAPI) aqueous mounting medium (Vector laboratories, CA, USA). The fluorescent images were documented using ProgRes Capture Pro 2.5 software on fluorescent microscope (Olympus BX41). Fluorescence intensity was quantified using Image J 1.48 version software (http:// image j.nih.gov/ij/; provided in the public domain by the National Institutes of Health, Bethesda, MD, USA) 74 . Primary and secondary antibodies are listed in Table 2.
Immunofluorescence of corneal tissues. 4 µm sections of MMC and SAHA treated corneal tissues were deparaffinised, rehydrated, and heated in antigen retrieval citrate buffer (pH 6) solution. Sections were incubated with blocking buffer (3%BSA in 1 × PBS) for 1 h at room temperature. Slides were then incubated overnight with mouse monoclonal anti α-smooth muscle actin (Abcam, ab7817; dilution 1:500) followed by secondary antibody Anti-mouse Cy3 (715-165-150, Jackson ImmunoResearch, dilution 1:2000). Further 4,6-diamidino-2-phenylindole (DAPI) staining, slides were mounted in Fluoroshield (Sigma) and stored at 4 °C in the dark. Microphotographs were obtained on Olympus CKX53 inverted microscope. Primary and secondary antibodies are listed in Table 2. www.nature.com/scientificreports/ Statistical analysis. All the experiments were performed in triplicate and results of three independent experiments were used for statistical analysis. Data are represented as the mean ± SD and were analyzed with one-way ANOVA and Mann-Whitney U test for multiple group comparison using statistical software GraphPad PRISM Ver 6.01. Significance value denoted, p* < 0.05, ** < 0.01, *** < 0.001, **** < 0.0001.