Derivation of Corneal Keratocyte-Like Cells from Human Induced Pluripotent Stem Cells

Corneal diseases such as keratoconus represent a relatively common disorder in the human population. However, treatment is restricted to corneal transplantation, which only occurs in the most advanced cases. Cell based therapies may offer an alternative approach given that the eye is amenable to such treatments and corneal diseases like keratoconus have been associated specifically with the death of corneal keratocytes. The ability to generate corneal keratocytes in vitro may enable a cell-based therapy to treat patients with keratoconus. Human induced pluripotent stem cells (hiPSCs) offer an abundant supply of cells from which any cell in the body can be derived. In the present study, hiPSCs were successfully differentiated into neural crest cells (NCCs), the embryonic precursor to keratocytes, and then cultured on cadaveric corneal tissue to promote keratocyte differentiation. The hiPSC-derived NCCs were found to migrate into the corneal stroma where they acquired a keratocyte-like morphology and an expression profile similar to corneal keratocytes in vivo. These results indicate that hiPSCs can be used to generate corneal keratocytes in vitro and lay the foundation for using these cells in cornea cell-based therapies.


Introduction
The cornea represents the transparent anterior-most portion of the eye. It acts to protect the underlying iris, pupil and anterior chamber as well as providing two-thirds of the refractive power of the eye. A healthy cornea has a central thickness of about 490 to 620 μm, 90% of which consists of the stromal layer. The corneal stroma is composed of highly organised collagen fibrils which arrange into lamellae that run parallel to the corneal surface [1]. The corneal stroma is populated by a small number of non-myelinated nerve bundles and trafficking immune cells, but its main cellular occupant is the corneal keratocyte.

Tissue Culture
Tissue samples and cell lines were prepared and used in accordance with Health and Disability Ethics Committee approvals assigned by the Ministry of Health New Zealand (ref# NTX/07/ 08/080/AM04). All cultures were grown in a 37°C, 5% CO 2 humidified incubator. hiPSCs were grown on Geltrex (ThermoFisher) in Modified Tennelle's Special Recipe 1 (MTESR1, StemCell Technologies). All cells were dissociated and passaged to a new well/ plate or into a single cell suspension using Accutase (StemCell Technologies). The iPSC line used was generated using modified RNAs encoding the Yamanaka factors and was first described by Warren et al, (2010) [19].

NCC production
We followed the protocols outlined in Chambers et al (2013), and Lee et al (2010) [19,20]. To begin, a confluent 10 cm cell culture dish was passaged to Geltrex coated 6-well plates (NCCs were not grown on mitomycin treated mouse embryonic fibroblasts). The number of cells was estimated using a haemocytometer and transferred to a 6-well plate at a concentration of 10, 000-20, 000 cells/ cm 2 . 10 μM Y-27632 was used in the media for the first 24 hours after passaging only. Due to growth, cells were split 1:2 once during each protocol, typically between day 6 and day 8, using Accutase (StemCell Technologies). Upon completion of NCC derivation, cells were frozen back for storage in liquid nitrogen (in 10% DMSO/ 90% BSA) or used for downstream applications. We repeated this protocol a total of eleven times on different passages of the iPSCs and consistently yielded >80% NCC conversion.

3D culturing
The cells of one well of a 6-well plate (~2 x 10 5 cells) were dissociated and made into a singlecell suspension in a 15 ml Falcon tube using Accutase. These cells were then pelleted at 700 g for 5 mins. The media was aspirated and carefully replaced with Keratocyte Differentiation Media (Advanced DMEM (Life Technologies), 10 ng/ml FGF2, 0.1 mM ascorbic acid-2-phosphate). Pelleted cells were left to grow for 21 days with media changes every second day.

Limbal rim assay
Corneo-scleral Limbal rims were freeze thawed (20 mins at -80°C, 20 mins in a 37°C water bath x10) to degrade any endogenous protein epitopes. Rims were cut into slices using a scalpel and each slice was placed in one well of a 12-well plate. 1.5 ml of KDM was added to each well. hiPSC-derived NCCs were dissociated from one well of a 6-well plate and placed in a 15 ml Falcon tube. Cells were then pelleted at 700g for 5 minutes. Media was aspirated from the 15 ml Falcon tube and the cells were re-suspended in 8 ml of KDM. Cell number was determined using a haemocytometer and~2 x 10 5 cells were seeded directly onto the sclera region of each limbal rim slice using a Hamilton syringe. We seeded cells on the sclera as we found the conical shape of the cornea caused the unattached cells to accumulate at the region of the limbal rim. The following day, live/ dead cell staining was performed using LIVE/DEAD Viability/Cytotoxicity Kit for mammalian cells (Molecular Probes), to detect if cells had appropriately attached to the rim.

Cryosectioning
3D cultured NCCs or limbal rim slices seeded with NCCs were washed twice in PBS and fixed in 4% PFA overnight at 4°C. The tissues were washed twice in PBS and then placed in 1% low melting point agarose containing 5% sucrose and 0.9% agar (made up in water) in cryomoulds. After one hour the cryoblocks were removed from the cryomoulds and placed in 30% Sucrose/ PBS. Cryoblocks were incubated overnight at 4°C then 30% Sucrose/PBS was replaced and left for another night at 4°C. Cryoblocks were flash frozen on dry ice and stored at -80°C prior to sectioning. 14 μm transverse sections were cut in a cryostat machine (set to -25°C) and transferred to a gelatine-coated microscope slide.

Quantitative Real-Time PCR
Cultures of hiPSCs and NCCs in 6-well plates were washed twice in PBS and then exposed to 1 ml of Trizol for 10 mins. Trizol extracted RNA was then transferred to a 1.5 ml eppendorff tube and stored at -80°C. 3D-cultured NCCs and limbal rim slices were similarly treated with 1 ml of Trizol and transferred to a flat-bottomed 2 ml eppendorff tube. The tissue was then homogenised and stored at -80°C. RNA was purified from Trizol using a PureLink TM RNA Mini Kit (Ambion) and cDNA was prepared using a PrimeScript TM RT Master Mix (Takara). Each reaction of real-time PCR used 8 μl of a master mix containing SYBR1 Premix Ex Taq (Tli RnaseH Plus), ROX plus (Takara) and 10 μM primer mix, and 2 μl of diluted cDNA. Primers for SYBR assays were synthesized using the Harvard PrimerBank as a reference. All amplifications were performed in triplicate, with HPRT1 used for normalization of RNA content. A negative control with no cDNA was also used in each qPCR. Relative expression was determined using the comparative C T method (2 -ΔΔCT ).

Differentiation of hiPSCs into NCCs
As corneal keratocytes are derived from cephalic NCCs during embryogenesis [2,3] it was reasoned that a 2-step protocol could be developed whereby hiPSCs were first induced to differentiate into NCCs, of which a number of protocols have been developed [20][21][22][23][24][25], and then matured into keratocytes in the second step. Extensive qPCR analysis of NCC and pluripotency gene markers was undertaken to confirm the NCC identity of the hiPSC-derived NCCs. All neural crest markers tested were more highly expressed in the hiPSC-derived NCCs than in hiPSCs (Fig 2). In particular, PAX3, SOX10, ZIC1, MSX2 and TFAP2A were >100-fold more highly expressed in the hiPSC-derived NCC population than in hiPSCs. The pluripotency genes SOX2, OCT3/4 and NANOG were all reduced by at least 10-fold in hiPSC-derived NCCs compared to hiPSCs (Fig 2). KLF4 expression did persist at a relatively unchanged level, which is consistent with KLF4 being highly expressed in NCCs [26] and their derivatives (such as smooth muscle cells [27]), and with KLF4 being the most highly expressed transcription factor in the mouse cornea [28]. In addition to analysing gene expression, we also undertook protein analysis ( Fig 3A). We performed antibody labelling for HNK1, AP2a and Vimentin to test for NCC markers. All hiPSC-derived NCCs were positive for Vimentin and AP2a, and approximately 20% were positive for HNK1 ( Fig 3A). These results, in addition to cell morphology observations, loss of SOX2 antibody staining and AP2a + and p27 NTR+ staining, support a conclusion whereby NCCs were successfully derived from hiPSCs.

Culturing hiPSC-derived NCCs in 3D produces a corneal keratocyte phenotype
For the second step of our 2-step protocol, hiPSC-derived NCCs were further differentiated into cells with a keratocyte-like phenotype. Given that the corneal keratocyte exists within a collagenous matrix in the cornea stroma, we hypothesized that culturing NCCs in a 3D matrix may promote their conversion to a corneal keratocyte-like cell. To do this, a cell suspension of hiPSC-derived NCCs was pelleted and then cultured in a defined keratocyte differentiation media (see schematic in Fig 3 and ref [29]). After 21 days, this 3D culture had grown and did not dissociate upon agitation. These aggregates were fixed and cryo-sectioned for analysis by immuno-staining. Labelling for HNK1 and AP2a showed these markers were reduced after 3D-culture but for small patches of AP2a-positive staining ( Fig 3B). Vimentin labelling persisted after 3D-culture, which is consistent with this protein being a label of corneal keratocytes as well as NCCs (Fig 3B). Keratocan and ABCB5 are specific markers of corneal keratocytes and were not detectable by protein analysis in hiPSC-derived NCCs ( Fig 3A). Strikingly, all cells in the 3D-culture were positive for Keratocan and extracellularly deposited Keratocan could be observed (Fig 3B). ABCB5 staining was detected in approximately 30% of 3D-cultured cells (Fig 3B). These results suggest that hiPSC-derived NCCs cultured in 3D lose their neural crest identity and convert to a corneal keratocyte-like cell type.
To better analyse the potential fate change of the hiPSC-derived NCCs to a corneal keratocyte phenotype, we performed extensive qPCR analysis on the 3D-cultures. Relative to hiPSCderived NCCs, there was a general down-regulation of NCC genes after the 3D-culture. PAX3, SOX10, CX32, FOXD3 and MSX1, and to a lesser extent TFAP2A, SNAIL, ID2 and MSX2 had reduced levels of gene expression in 3D-cultured NCCs relative to hiPSC-derived NCCs ( Fig  4A). A few neural crest genes increased their levels of expression when cultured in 3D, such as ZIC1 and SLUG, but most of the assayed neural crest genes had reduced expression in hiPSCderived NCCs cultured in 3D for 21 days (Fig 4A). Analysis of genes that are highly expressed in corneal keratocytes showed their expression levels increase after 3D-culture (Fig 4A). In particular, BMP3, CDH5, B3GnT7, PtDGS, AQP1 and KLF4 were >10-fold more highly expressed   after 3D culturing (Fig 4A). ALDH1A1, ALDH3A1, KERATOCAN and CHST6 were also increased. These results suggest that 3D culture of NCCs promotes conversion of these cells towards a keratocyte-like fate.
Subsequently, these in vitro derived corneal keratocyte-like cells were compared to in vivo human corneal keratocytes obtained from cadaveric corneas. RNA acquired from corneal buttons, stripped of their endothelial and epithelial layers, leaving only the stromal layer of the cornea, was used as a source of corneal keratocytes. qPCR analysis showed keratocyte markers ALDH3A1, KERATOCAN, PtDGS, AQP1 and KLF4 were expressed at much lower levels in 3D-cultured NCCs compared to human corneal keratocytes (Fig 4B). However, other corneal keratocyte markers were expressed at similar levels in both cell populations, including ALDH1A1, BMP3, B3GnT7 and CHST6 (Fig 4B). Neural crest genes ZIC1, SLUG and MSX2 were more highly expressed in 3D-cultured NCCs relative to human corneal keratocytes whereas SNAIL, FOXD3 and MSX1 were more lowly expressed (Fig 4B). Despite this, a number of genes were expressed at similar levels in 3D-cultured NCCs relative to human corneal keratocytes, including PAX3, SOX10 and ZIC5 (Fig 4B).
In conclusion, the 3D culture of hiPSC-derived NCCs promotes both the expression of corneal keratocyte-specific genes and also the deposition of Keratocan in extracellular matrix suggesting that these cells can be induced to adopt a keratocyte fate in vitro.

Culturing hiPSC-derived NCCs on limbal rims promotes keratocyte-like fate
Having demonstrated that a 3D-culture environment can induce keratocyte differentiation, it was next examined whether native corneal tissue would be more effective at inducing a keratocyte phenotype. To achieve this aim, hiPSC-derived NCCs were exposed to cadaveric corneal tissue. This technique used a donated, cadaveric, peripheral corneo-scleral rim (see Fig 5A), which was put through ten freeze-thaw cycles to deplete the cornea of endogenous proteins. Interestingly, before freeze-thaw cycling was performed, the limbal rims contained cells positive for AP2a and Keratocan (S3 Fig). AP2a is a marker of cephalic neural crest cells, therefore this result suggests corneal keratocytes still retain a marker of neural crest identity even in adult tissue.
To test if exposure to corneal tissue promoted conversion to a keratocyte-like fate, we seeded hiPSC-derived NCCs onto the sclera of a corneal rim slice (Fig 5A). These cells attached to the corneal slice and calcein AM live-cell staining showed they were still alive by day 4 (see S4 Fig for the dead cell stain) and had migrated to the corneal region of the rim (Fig 5B). Calcein AM imaging also showed that these cells displayed a NCC-like morphology and their movement towards the cornea indicated an intrinsic migratory capacity (Fig 5B). At day 7, NCCs on the surface of the cornea could no longer be observed (Fig 5B). Ethidium staining for dead cells showed no positive labelling (data not shown), but higher magnifications of the Calcein AM live cell stain revealed that the NCCs had migrated into the stroma (Fig 5B). In transverse sections performed on day 5 after seeding, derived NCCs can be observed on top of the cornea and at the lateral edges where these cells appear to be migrating into the stroma between the collagen fibrils (S5A Fig). Sectioning through the sclera region of the rim showed now cells reside within this region of the tissue (S5B Fig). Taken together, our results suggest that derived NCCs migrate over the sclera to the corneal region of the rim, where they then migrate round the dissected lateral edges and into the exposed collagen fibrils.
We next wished to observe if hiPSC-derived NCCs cultured on limbal rims acquire a corneal keratocyte-like identity. In transverse sections of control non-seeded corneal rim slices no DAPI, AP2a or Keratocan staining was observed, indicating successful depletion of any endogenous signal (Fig 6A, top panels). DAPI stained nuclei in cells positive for AP2a and Keratocan were observed in limbal rims seeded with hiPSC-derived NCCs (Fig 6A, bottom panels). Higher magnification showed that these cells had migrated in-between the collagen fibrils that constitute the corneal stroma (Fig 6B) and had a similar morphology to human corneal  keratocytes found in vivo (see S3 Fig and [30]). In addition to Keratocan and AP2a staining, we also observed iPSC-derived NCCs seeded on to corneal rims were positively labelled with ABCB5, Vimentin and HNK1 (Fig 6B). These results suggest that seeding of hiPSC-derived NCCs onto corneal rims promotes expression of corneal keratocyte markers, such as Vimentin, Keratocan and ABCB5, but also maintains NCC-specific marker labelling (HNK1 and AP2a). qPCR analysis of hiPSC-derived NCCs cultured on limbal rim slices for 21 days showed NCC markers such as PAX3, ZIC1, SOX10, SNAIL, TWIST, ZIC5, FOXD3, MSX1 and MSX2 were similar to levels of expression found in human corneal keratocytes (Fig 7). Only TFAP2A, CX32, P75NTR and ID2 had levels of expression that were disproportionately higher in the rim-cultured NCCs relative to human corneal keratocytes (Fig 7). Levels of expression for many keratocyte markers were also similar between these two cell populations, especially ALDH1A1, BMP3, B3GnT7, CHST6, PtDGS, AQP1 and KLF4 (Fig 7). ALDH3A1 and KERA-TOCAN were less highly expressed in hiPSC-derived NCCs cultured on limbal rims than in endogenous corneal keratocytes, and CDH5 was more highly expressed (Fig 7). Despite these exceptions, the overall gene expression profile of rim-cultured NCCs was more similar than 3D-cultured NCCs to in vivo corneal keratocytes (Fig 7).
Taken together, our findings demonstrate that a two-step protocol involving the conversion of hiPSCs to NCCs using the Chambers et al (2013) protocol followed by culture on limbal rims is sufficient to induce formation of keratocyte-like cells in vitro.

Discussion
In this study, we have shown hiPSC-derived NCCs adopt corneal keratocyte-like characteristics when cultured in three dimensions or seeded onto cadaveric corneal tissue. Of these two culture methods, we found seeding hiPSC-derived NCCs onto cadaveric human limbal rims was Expression analysis shows limbal rim seeded NCCs adopt an expression profile that is similar to human corneal keratocytes. qPCR analysis indicated hiPSC-derived NCCs cultured on limbal rim slices (blue bars) adopted an expression profile that was more analogous to human corneal keratocytes than hiPSC-derived NCCs grown in 3D cultures (red bars). Importantly, the expression of a number of keratocyte markers was much higher when derived NCCs were cultured on limbal rim slices, such as ALDH3A1, KERATOCAN, PtGDS and AQP1. doi:10.1371/journal.pone.0165464.g007 Derivation of Corneal Keratocytes from Human iPSCs most successful, with these cells adopting a similar morphology and molecular marker profile to native corneal keratocytes after just 21 days of culture.
Other stem cell sources have been utilised for differentiating corneal keratocyte-like cells [31][32][33][34], but our study represents the first attempt to generate corneal keratocytes from hiPSCs. A recent report used a similar approach as we present here, by first converting H1 hESCs to NCCs, and then further differentiating these cells in 3D cultures to corneal keratocytes [35]. We believe our protocol is more amenable to downstream applications (including translation to human therapies) as we use a feeder-free system to culture hiPSCs and generate NCCs, whereas the protocol described by Hertsenberg and Funderburgh (2016) uses the PA6 mouse embryonic fibroblast line as feeder cells [35]. In addition, we find seeding corneo-scleral limbal rims with hiPSC-derived NCCs better simulates the in vivo environment corneal keratocytes inhabit, and greatly improved the conversion of derived NCCs to a corneal keratocyte-like fate compared to 3D cultures. NCCs seeded onto limbal rims expressed markers such as KERATOCAN, PTGDS, AQP1 and KLF4 at between 10-and 100-fold higher levels compared to 3D-cultured NCCs. ALDH3A1, which encodes an aldehyde dehydrogenase important in maintaining the transparency of the corneal stroma [36], was expressed 1000-fold higher in NCCs seeded to limbal rims compared to NCCs cultured in 3D. These differences in expression levels between 3D-cultured and corneal rim seeded hiPSC-derived NCCs are important as they highlight that limbal rim seeded cells are more comparable to native human corneal keratocytes. Our results therefore support a model whereby direct seeding of NCCs to cadaveric corneal tissue effectively promotes corneal keratocyte differentiation. This likely indicates that the corneal stroma contains keratocyte differentiation stimuli despite the removal of endogenous cells. It also suggests that the use of NCCs as a cell-based therapy to repopulate corneal tissue may be a viable clinical option for corneal dystrophies such as keratoconus and should be explored further.
Our study concentrated on looking at the endpoint of NCC differentiation. Follow up studies may benefit from understanding how derived NCCs progressively differentiate into keratocytes and if methods can be modified to improve this process. During embryogenesis, NCC migration and differentiation is closely linked to the extracellular matrix [37]. In vitro studies have shown the diameter of matrix fibres acts as a guidance cue and can define neuronal and neural crest differentiation [38,39]. Given the corneo-scleral rims used here offer a matrix that corneal keratocytes occupy in vivo, it is possible that their topographical makeup (such as orientation and diameter of the collagen fibrils) better enables the acquisition of a corneal keratocyte-like fate in our derived NCCs.
Our antibody staining for AP2a showed native corneal keratocytes still express this marker of cranial neural crest, suggesting they retain some level of neural crest fate even in adulthood. Whilst it is difficult to ascertain the developmental time point to which our NCCs correspond relative to in vivo embryogenesis, we did find that cells positive for Keratocan are observed after just 21 days. These results support previous findings that corneal keratocytes maintain a level of fate plasticity that is closely matched to the stem cell-like characteristics of NCCs [40]. Thus, our approach of primarily deriving NCCs and then exposing these cells to corneal tissue represents a simple pathway for corneal keratocyte differentiation. Future studies will determine if hiPSC-derived NCCs transplanted into live corneal tissue can aid in recovery from injury or disease. For any clinical or tissue engineering application, it is important to ensure that the cells can be used safely. Whilst we have not analysed cell proliferation in the current study, we did observe extensive proliferation of hiPSCs cultured on limbal rims (data not shown). This same rate of division was not observed when hiPSC-derived NCCs were seeded onto limbal rims, suggesting these cells have a lower proliferative capacity.
In summary, we have established a novel protocol for converting hiPSCs to keratocyte cells. These cells may be of use in future clinical applications, such as cell-based therapies that could treat causes of corneal blindness, such as inherited corneal dystrophies and keratoconus.