Priming human adipose‐derived mesenchymal stem cells for corneal surface regeneration

Abstract Limbal stem cells (LSC) maintain the transparency of the corneal epithelium. Chemical burns lead the loss of LSC inducing an up‐regulation of pro‐inflammatory and pro‐angiogenic factors, triggering corneal neovascularization and blindness. Adipose tissue‐derived mesenchymal stem cells (AT‐MSC) have shown promise in animal models to treat LSC deficiency (LSCD), but there are not studies showing their efficacy when primed with different media before transplantation. We cultured AT‐MSC with standard medium and media used to culture LSC for clinical application. We demonstrated that different media changed the AT‐MSC paracrine secretion showing different paracrine effector functions in an in vivo model of chemical burn and in response to a novel in vitro model of corneal inflammation by alkali induction. Treatment of LSCD with AT‐MSC changed the angiogenic and inflammatory cytokine profile of mice corneas. AT‐MSC cultured with the medium that improved their cytokine secretion, enhanced the anti‐angiogenic and anti‐inflammatory profile of the treated corneas. Those corneas also presented better outcome in terms of corneal transparency, neovascularization and histologic reconstruction. Priming human AT‐MSC with LSC specific medium can potentiate their ability to improve corneal wound healing, decrease neovascularization and inflammation modulating paracrine effector functions in an in vivo optimized rat model of LSCD.

| 5125 NIETO-NICOLAU ET AL. vascularization and opacity [4,7,8]. Chemical burns represent the most common cause of LSCD. This kind of chemical aggression to the eye surface represents 7.7%-18.0% of all ocular traumas [9], and their frequency has augmented by the use of corrosive cleaners in the house field. All in all, LSCD affects approximately 10 million people worldwide [10].
Cell therapy with LSC for the treatment of unilateral LSCD is a well-consolidated treatment [11]. In this treatment, LSC of the healthy eye are expanded ex vivo and transplanted on a biocompatible scaffold onto the injured eye of the patient [12]. Nevertheless, the treatment of bilateral LSCD supposes a more challenging situation. It relies on allogeneic LSC transplantation, requiring systemic immunosuppression [13] that can lead to adverse side effects [14].
This entails the need for alternative treatments.
In this framework, somatic stem cells are a promising tool. Human adipose tissue-derived mesenchymal stem cells (AT-MSC) can be easily obtained from liposuction aspirates and have been applied in tissue-engineering applications for wound healing repair [15] and other cell based therapies [16]. AT-MSC exert a paracrine action that benefit the regenerative processes through trophic factors secretion [15,17] suppressing, the inflammation and immune reaction signalling [18]. In fact, their immunomodulatory properties are explored for the treatment of immune disorders [19] and the therapeutic effects of their secretome are being increasingly studied on several diseases [17]. Moreover, in vitro and in vivo approaches have shown that AT-MSC cells have potential plasticity to differentiate into several lineages and to integrate in the regenerated tissue [20,21].
Recently, MSC have demonstrated to be effective and safe in human clinical application in comparison with allogeneic LSC transplantation in cases of LSCD [22]. In addition, AT-MSC demonstrated their ability to improve total LSCD in rabbit models [23]. With this encouraging outcome in mind, there is a need to not only study the mechanisms through AT-MSC could exert their pleiotropic action, but also to optimize culture conditions to increase their therapeutic potential. Our leading aim was to evaluate the therapeutic effectiveness of AT-MSC primed with different LSC media in the regeneration, inflammatory and angiogenic cytokines profile of the ocular surface of an optimized rat model for total LSCD. also approved the experimental procedure.

| AT-MSC culture characterization
Cultures were fixed and blocked with 10% foetal calf serum (FCS) in 100 mmol/L PBS for 10 minutes and then incubated with conjugated antibodies for 30 minutes at room temperature. Mouse monoclonal immunoglobulin (IgG) isotype was used as a negative control.

| AT-MSC adipogenic and osteogenic differentiation
Cells (10 4 cells/cm 2 ) were plated and cultured in an AT-MSC medium (DMEM, 10% FCS, 2 mmol/L L-glutamine, 10 mmol/L HEPES and antibiotics) for 24 hours. The medium was changed to adipogenic or osteogenic medium, as described [24,27,28] and maintained for 4 weeks, changing the medium every 48 hours. All media and supplements were supplied by Invitrogen (Invitrogen, Carslbad, CA, USA), and other reagents were supplied by Sigma-Aldrich (Munich, Germany). Adipogenic differentiation was confirmed by Oil red staining, and osteogenic differentiation was confirmed by alkaline phosphatase histochemistry.

| Human amniotic membrane (AM) preparation
Human placenta samples were obtained at the time of elective caesarean and prepared at the facilities of the Barcelona Tissue Bank (BTB-BST, Barcelona, Spain). Placenta was cleaned of blood clots with saline solution containing penicillin (50 μg/mL; ICN, Costa Mesa, CA, USA), streptomycin (50 μg/mL; ICN) and amphotericin B (2.5 μg/mL; Bristol Myers Squibb Co, Princeton, NJ, USA). The AM was separated from the chorion by blunt dissection. Approximately 4 × 4 cm of the AM was placed on Millipore filters in a plastic sterile container with DMEM (Invitrogen) and glycerol (ICN) at a 1:1 ratio and preserved at −80°C until use.

| Treatment of AT-MSC cells with specific media
AT-MSC cells were cultured (10 4 cells/cm 2 ) under the following conditions: AT-MSC medium; SHEM medium prepared as described above; and CnT30 medium (CellnTech Advanced Cell System, ZenBio, NC, USA). All culture media were supplemented with 2% FCS. After 7 days, cells were harvested and 10 6 cells/ml were seeded on 4 × 4 cm sections of AM. Implants were observed by phase contrast microscopy to confirm cellular adhesion. Comparison of the expression of progenitor and corneal markers was performed at different time-points by qPCR as described below.

| In vivo rat model for LSCD
Thirty Wistar albino rats (n = 6 per group; weighing 250-350 g) were anaesthetized with intramuscular ketamine hydrochloride (35 mg/ kg; Phoenix Scientific Inc, USA) and xylazine hydrochloride (5 mg/kg; Phoenix Scientific). Study eyes (left eyes) were topically anaesthetized with 1% proparacaine hydrochloride (Bausch & Lomb, Madrid, Spain) and decontaminated with 5% povidone iodine drops for three minutes before procedures. Each cornea was enclosed with a plastic cylinder (3-mm diameter), and several drops of n-heptanol (Sigma) were placed on the centre of the cornea for 2 minutes to injure the corneal epithelium. Mechanical debridement was performed, followed by 360° surgical destruction of a lamellar superficial limbal ring, defined as a 1.5-mm segment on either side of the anatomic junction between the cornea and the conjunctiva [29]. Topical antibiotics (levofloxacin 0.5%) and steroids (betamethasone 0.1%) were applied twice daily. Rats were examined using a handheld light daily and a slit lamp on days 3 and 7. After one week, n-heptanol application with mechanical debridement was repeated before ocular surface treatment as described above.
Animals were classified into five study groups (n = 6), as follows:

| Histopathology
Enucleated eyes were obtained after 30 days of cell implant procedures. Half of each cornea was immediately frozen in liquid N 2 and used for molecular analysis. The other half was fixed in a 4% paraformaldehyde buffered solution and prepared for conventional histopathologic studies. Material was embedded in paraffin blocks, sectioned (6-8 μm) and stained with haematoxylin and eosin for light microscopy examination.

| In vitro model of corneal inflammation by alkali treatment
Human corneal epithelial cells (HCE) were seeded at a density of 3 × 10 5 cells/cm 2 and cultured in DMEM, 10% FCS, 2 mmol/L L-glutamine, 10 mmol/L HEPES and 1% antibiotics medium. After 24 hours, medium was changed by 10 mmol/L, 25 mmol/L or 50 mmol/L NaOH that were dissolved in the same medium. Medium without NaOH was used as control. Treatment was applied for 10 minutes, 3 washes with PBS were carried out and cells were cultured with new medium for 24 hours.
Then, medium was collected for paracrine secretion analysis. Treatment was repeated again for 10 minutes. Supernatants were collected at 24 hours from the second round for paracrine secretion analysis and further experiments with AT-MSC. Cell toxicity assay was carried out.
HCE cells were seeded and alkali-treated as explained above on 96 well plates. WST-1 assay (Abcam, Cambridge, UK) was performed following manufacturer's instructions 72 hours after seeding. Absorbance was read at 450 nm with a reference wavelength of 680 nm.

| Multiplex analysis
Supernatants were collected, centrifuged at 13 000 g for 5 minutes and stored at −80°C until analysis. Concentrations of MMP-2, VEGF-A (isoform A), MCP-1, IL-6 and TGFβ were analysed using Luminex multiplex immunobead assays from Millipore according to the manufacturer's protocols. The reactions were detected with a Luminex 100 IS 2.3 system (Luminex, Austin, TX).

| mRNA isolation and reverse transcription
Total RNA from rat corneas was extracted using phenol/chloroform purification with TriPure Isolation Reagent ® (Roche Diagnostics GmbH, Mannheim, Germany), following the manufacturer's instructions. RNA was purified using an RNeasy MinElute Cleanup Kit (Qiagen, Madrid, Spain), following the manufacturer's instructions. The concentration was measured using Tecan infinite m200 pro (Tecan, Männedorf, Switzerland) and adjusted to 100 µg/µL. Then, 8 μL was reverse transcribed using SuperScript ® III Reverse Transcriptase (Thermo Fisher Scientific, Invitrogen) in a final volume of 20 μL, following the manufacturer's recommendations. Total RNA from AT-MSC, LSC and CO were extracted using PureLink RNA Mini Kit (Ambion, Invitrogen) following the manufacturer's recommendations. One μg was reverse transcribed as explained above.

| Quantitative real-time polymerase chain reaction
One μL of cDNA was used for qPCR in a final volume of 18 µL with SYBR ® Green PCR Master Mix (Thermo Fisher Scientific, Invitrogen).  (Table S2, supplemental material).

| Polymerase Chain Reaction analysis and sequencing
To analyse the presence of human CK12 in the rat corneas, we used a specific primer for human CK12. Reactions were carried out as commented above using Taq Platinum PCR SuperMix F I G U R E 1 Schematic representation of the experimental design. A, In vivo assay. Upper timeline represents the LSCD generation. Corneas were damaged with n-heptanol followed by mechanical debridement and surgical destruction of the limbal ring. Rats were examined on days (d) 3 and 7. The procedure was repeated at d7. Transplantation was performed at d14 from the first damage. At d44, corneas were analysed by qPCR for inflammatory markers and by histology. Bottom timeline represents the culture of AT-MSC under different conditions (standard medium, SHEM medium and CnT30 medium). AT-MSC seeding on amniotic membrane (AM) was performed at d7. Corneal and limbal markers were assayed by qPCR. B, In vitro assay. Upper timeline shows the culture of AT-MSC with the different media. Inflammatory markers were analysed by multiplex assay at d1, 3, 5 and 7, and by qPCR at d7. At d7 of culture, media were changed by conditioned medium (CM) of alkali-treated HCE. AT-MSC were cultured for 72 hours (h) with this medium. Multiplex assay was performed after 48 and 72 h from the medium change. Bottom timeline exemplifies the in vitro inflammation model by alkali treatment. NaOH treatments were applied after 24h from HCE seeding. A second round of NaOH treatment was applied after 48h from the seeding. Media were recovered at 48 and 72 h and analysed by multiplex assay for inflammatory markers. CM of alkali-treated HCE was recovered at 72 h to culture AT-MSC cells that were grown previously in the 3 different culture media (Thermo Fisher Scientific, Invitrogen) following manufacturer's instructions. Products were electrophoresed in 2% agarose gel for 1 hour. Ethidium bromide was used to visualize the PCR bands.
PCR products were purified using QIAquick PCR Purification Kit (Q iagen, Madrid, Spain), following the manufacturer's instructions.
The product was sequenced using the ABI Prism 3730 Genetic Analyzer (ABI Applied Biosystems, Life Technologies). Finally, sequences were aligned with mRNA using the ClustalW software (http://www.ebi.ac.uk/Tools/ clust alw2).

| Statistical analysis
Data are reported as mean ± standard error of the mean (MD ± SE).
Statistics were analysed using GraphPad Prism software (version 4; GraphPad Software Inc, CA, USA). Statistical significance was determined by one-way analysis of variance with Bonferroni's post hoc analysis. p-values ≤ 0.05 were considered statistically significant.

| Corneal epithelium showed improved regeneration with SHEM-cultured AT-MSC
We evaluated the efficiency of AT-MSC cells cultured with different media to regenerate the damaged ocular surface on an in vivo model of total LSCD in rats following the schematic representation depicted in Figure 1. As the use of NaOH has been reported to induce stromal damage along with severe corneal ulcers, hyphema and hypopyon [31], which could mislead the interpretation of the outcome, we used n-heptanol accompanied by epithelial debridement and excision of the superficial limbal ring for LSCD. Our approach is generally used in the generation of LSCD because allows specifically a complete LSC niche removal and corneal epithelium denudation without further stromal damage [23,29].
Non-treated damaged corneas showed severe neovascularization and opacification (Figure 2A  with AT-MSC ( Figure 3B). Sequencing of the PCR products obtained from the rat corneas confirmed the specificity (approximately 98%) of the amplicons for human CK12 mRNA ( Figure 3C), indicating the presence of AT-MSC on the rat corneas 30 days after implantation.

| Treatment with 25mM NaOH generated inflammation without cytotoxicity in an in vitro corneal alkali injury model
We  Figure 4B), which could be explained by the short exposure to NaOH. However, treatment with higher concentrations of NaOH, such as 50mM, generated obvious cell death at the first round ( Figure   S1, supplemental material). However, after the double 'burn' treatment, a morphologic change could be observed on HCE ( Figure 4C).
Cells appeared more flattened and bigger. Because 25mM NaOH treatment raised more the secretion of inflammatory cytokines, supernatants of 25mM NaOH treated HCE were collected for further experiments with AT-MSC.

| Specific media change the paracrine secretion of AT-MSC and prime different paracrine responses to alkali-treated HCE conditioned medium
We  Figure 5A).
Analysis of mRNA levels at 7 days of culture with specific media (SHEM or CnT30) or control medium supported the Luminex assay data, further indicating that TNFα was lower in SHEM cell culture conditions and that the mRNA expression level of IL-10 did not differ between treatments ( Figure 5B).

| AT-MSC cultured with LSC specific media did not transdifferentiate to corneal cells
As AT-MSC were cultured on specific LSC media, we aimed to study  Figure 6A). CK3 also declined with culture time in all situations, as well as ΔNp63α did ( Figure 6A). E-cadherin incremented in standard medium culture at 7 days and 14 days, but showed a reduction with LSC specific media ( Figure 6A).
ITGβ1 was elevated in SHEM at 5 days in comparison with control AT-MSC at day 0. However, the expression decreased with CnT30 medium at 5 days ( Figure 6A). LSC expressed higher levels of ABCG2, whereas corneal epithelial cells showed an evident reduction in this expression. In comparison with AT-MSC at day 0, the highest enhancement in the expression of this marker was found in SHEM at 5 days and at 7 days. However, ABCG2 was down-regulated in SHEM at 14 days demonstrating a trend to decrease during culture ( Figure 6A).
AT-MSC showed a typical spindle shaped morphology at 24 hours in all treatments. During culture, cells growing with CnT30 medium showed a smaller polygonal morphology, whereas SHEM-cultured cells showed an elongated phenotype. Neither treatment changed the morphology to epithelial phenotypes ( Figure 6B).

| AT-MSC cells accomplished mesenchymal stem cells criteria characterization
FACS analysis showed that AT-MSC cells were positive for CD105, CD90, CD73 and negative for CD34, CD14, CD45 and HLA-DR markers ( Figure S2, supplemental material) which confirm their mesenchymal characteristics according to the International Society for Cell Therapy (ISCT) [32]. To ensure AT-MSC cell plasticity and their ability for multilineage differentiation, cells were induced to differentiate into adipogenic and osteogenic lineages. Intracellular lipid accumulation was confirmed using Oil Red O staining and osteogenic differentiation was detected by Alizarin Red staining ( Figure S2).

| D ISCUSS I ON
Currently, different stem cells sources are investigated for the treatment of bilateral LSCD. A recent proof of concept clinical trial with MSC has showed promising results for the treatment of this challenging pathologic condition [22]. As far as we know, this is the first report showing that priming human AT-MSC with LSC specific culture medium can potentiate their therapeutic potential to promote corneal wound repair, restore transparency, decrease inflammation and modulate paracrine effector functions in an in vivo rat model of total LSCD.
We optimized an experimental model of total LSCD in rats. The resulting clinical features, evaluated using previous defined standards [33], are compatible with the clinical and histopathologic findings in LSCD in humans [34,35], other rat LSCD models [31,36,37] and in other species, such as rabbits [23,29]. The limbic deficiency was achieved in all damaged corneas, faster than in other models [23,29,36]. Damaged corneas showed loss of transparency, corneal neovascularization and stromal cell infiltration, indicatives of an inflammatory state that correlated by an increased expression of pro-angiogenic and inflammatory cytokines such as VEGF, MMP-2, MCP-1, IL-6, TNFα and TGFβ, consistent with previous data in murine alkali burned corneas [4,38]. Although these cytokines are implied in corneal wound healing [4,39,40], the specific contribution of each cytokine to corneal re-epithelization and homeostasis restoration needs to be well-defined.
Amniotic membrane is used in clinical application to overcome the effects of inflammatory events in LSCD [5]. However, our data showed that amniotic membrane alone did not suffice to avoid corneal opacification, neovascularization, epithelial and stromal disorganization, agreeing with other investigations in animal models [23,31]. In fact, amniotic membrane only down-regulated the expression of MMP-2 and IL-6.
Our results agree with previous results showing that MSC play an advantageous role during wound healing mitigating the adverse effects of inflammation and angiogenesis [23,31,41,42] increasing re-epithelialization [49,50]. This could explain the better corneal regeneration, transparency and histologic reconstruction in comparison with the other treatments. Some studies indicated that MSC mediate their anti-inflammatory effects increasing the secretion of IL-10 as well as blocking the secretion of TNFα by dendritic cells [51], which also support our results.
So, the composition of SHEM medium enhanced the wound healing properties of AT-MSC cells and produced better outcome in vivo. Interestingly, among other trophic factors, EGF was a component of SHEM media which enhances the wound healing therapeutic potential of MSC through paracrine secretion [52]. In fact, MSC preconditioning with growth factors or cytokines constitutes a reliable approach to enhance their therapeutic properties [53], through activation of trophic signalling pathways [54]. Despite the complex mechanisms governing the immunomodulation in AT-MSC [18,19], we demonstrated that SHEM-treated AT-MSC lowered the levels of IL-6, TGFβ, TNFα and MMP-2, whereas CnT30 up-regulated the levels of TGFβ and VEGF and did not change the levels of TNFα.
Overall, SHEM-treated AT-MSC showed a better anti-inflammatory profile than AT-MSC cultured either with CnT30 or standard medium. This could influence the therapeutic effect of AT-MSC as the increase in IL-6 expression hinder the therapeutic activity of MSC [55]. Further studies would determine whether upstream regulators of immunoregulation, such as nuclear factor-kappa B (NF-κB) and its regulator Rap 1 [56,57], could be implicated in the paracrine effector functions mediated by AT-MSC treated with different mediums.
We aimed to demonstrate that AT-MSC cultured with different media react differently when exposed to the same inflammatory stimuli from an alkali-treated corneal cell line. In vitro alkali-treated HCE presented increased expression of IL-6, MMP-2, MCP-1 and TGFβ, indicating an inflammatory state that did not generated cy- AT-MSC were detected in all corneas regardless of the media used to culture AT-MSC. However, the best outcome were found with SHEM-treated AT-MSC that showed an improved anti-inflammatory and anti-angiogenic paracrine profile. This indicated that the therapeutic effects of AT-MSC were mediated mainly through paracrine effector functions than by a transdifferentiation mechanism on the corneal surface. However, more studies would elucidate whether SHEM could prime AT-MSC to exert therapeutic effects through other mechanisms than paracrine regulation, such as mesenchymal mitochondrial transfer [66] that have demonstrated protective roles in corneal wound healing [67].
We demonstrate that culture media can potentiate the therapeutic potential of AT-MSC to promote corneal wound repair, restore transparency and decrease inflammation on chemical burn corneas modulating paracrine effector functions. The use of LSC specific medium to culture AT-MSC is a feasible option for cell therapy because this medium is already used in LSC ex vivo culture before transplantation.

CO N FLI C T S O F I NTE R E S T
The authors declare no potential conflicts of interest.