Effect of porcine corneal stromal extract on keratocytes from SMILE‐derived lenticules

Propagating large amounts of human corneal stromal cells (hCSCs) in vitro while maintaining the physiological quality of their phenotypes is necessary for their application in cell therapy. Here, a novel medium to propagate hCSCs obtained from small incision lenticule extraction (SMILE)‐derived lenticules was investigated and the feasibility of intrastromal injection of these hCSCs was assessed. Primary hCSCs were cultured in porcine corneal stroma extract (pCSE) with RIFA medium including ROCK inhibitor Y27632, insulin‐transferrin‐selenium, fibroblast growth factor 2, L‐ascorbate 2‐phosphate and 0.5% FBS (RIFA medium + pCSE). Protein profiling of the pCSE was identified using nanoscale liquid chromatography coupled to tandem mass spectrometry (nano LC‐MS/MS). After subculturing in RIFA medium + pCSE or 10% FBS normal medium (NM), hCSCs at P4 were transplanted into mouse corneal stroma. Compared with NM, ALDH3A1, keratocan and lumican were significantly more expressed in the RIFA medium + pCSE. ALDH3A1 was also more expressed in the RIFA medium + pCSE than in the RIFA medium. Fibronectin and α‐SMA were less expressed in the RIFA medium + pCSE than in the NM. Using Metascape analysis, the pCSE with its anti‐fibrosis, pro‐proliferation and anti‐apoptosis activities, was beneficial for hCSC cultivation. The intrastromally implanted hCSCs in the RIFA medium + pCSE had positive CD34 expression but negative CD45 expression 35 days after injection. We provide a valuable new medium that is advantageous for the proliferation of hCSCs with the properties of physiological keratocytes. Intrastromal injection of hCSCs in RIFA medium + pCSE has the potential for clinical cell therapy.


| INTRODUC TI ON
The corneal stroma is mainly composed of collagen and corneal stromal cells (CSCs). The physiological CSCs (also called keratocytes) present the characteristics of quiescent and dendritic cells.
They usually express the cluster of differentiation 34 (CD34), stromal crystallins (ALDH3A1) and keratan sulphate proteoglycans (KSPGs), including lumican, keratocan and mimecan, which contribute to corneal transparency. 1 Keratocytes synthesize collagens, other extracellular matrix (ECM) elements, and various enzymes to degrade old matrix proteins to maintain stromal matrix homeostasis. 2 Keratocytes can be transformed into motile and contractile fibroblasts or myofibroblasts largely due to the activation of the transforming growth factor-β (TGF-β) system during stromal damage and the healing process. 3 Myofibroblasts express high levels of alpha-smooth muscle actin (α-SMA), biglycan and the extra domain A (EDA) of cellular fibronectin, which usually causes corneal scarring. 4 Corneal disease is one of the major leading causes of blindness globally. 5 Corneal transplantation is the definitive treatment for patients with severe corneal lesions. 6 However, the clinically successful application of corneal transplantation is limited due to the lack of donated corneal tissue and post-transplant complications. 7,8 Thus, developing alternatives is imperative.
Cell therapy using healthy CSCs to replace diseased stromal cells is an available approach. Propagating large amounts of human CSCs in vitro while maintaining the physiological quality of their phenotypes is necessary for their application in cell therapy. However, when cultured in serum-containing medium, CSCs easily transform and can lose important cell properties after a limited scale of cell expansion in vitro. 9 Culture substrate/scaffold conditions, cell densities and other elements also influence keratocyte-myofibroblast transdifferentiation in in vitro cultures. 10,11 Although serum-free cultures have been reported to be effective for the maintenance of the phenotype and physiological properties of keratocytes, the subcultivation of a large amount of CSCs remains challenging. 12 Various improved culture techniques have been reported, including spheroid cultures, 13 media supplements, 2 derivation from human stem cells 14 and culturing on special materials. 10,11 In terms of medium supplements, extracts from cells or tissues have also been applied. Several studies demonstrated that extracts obtained from mammalian and embryonic stem cells (ESCs) had beneficial effects on cellular reprogramming. 15,16 Amniotic membrane extract (AME) has also been successfully used as an eye drop in clinical applications to treat dry eye and chemical burns. 17 Yam et al 2 used cocktail medium supplemented with soluble AME, ROCK inhibitor Y-27632, and insulin-like growth factor-1 (IGF-1) to propagate and maintain CSCs. They reported that this type of medium supplementation promoted the keratocyte features and prevented keratocyte-myofibroblast transdifferentiation.
Proteins > 3 kDa in soluble AME enhanced keratocyte growth and inhibited cell fibrosis.
Small incision lenticule extraction (SMILE) is a corneal refractive surgical procedure used to correct myopia or other refractive errors.
With increasing numbers of patients undergoing SMILE, there are many extracted pieces of intrastromal lenticule, which are usually discarded. SMILE-derived lenticules were recently successfully used in both preclinical animal studies and human clinical applications. Clinical studies have demonstrated that corneal intrastromal lenticules can be successfully reutilized to treat high hyperopia and presbyopia, 18,19 thin corneas due to recurrent pterygium, 20 keratoconus, 21,22 corneal ulcer or perforations, 23 and other corneal diseases. 24 Our prior research demonstrated that SMILE-derived lenticules were beneficial scaffolds for the reconstruction of retinal pigment epithelial (RPE) sheets. Lenticules displayed biocompatibility after subretinal implantation. 25 Therefore, discarded corneal lenticules from SMILE can be reused as repair biomaterials or scaffolds for tissue engineering. Fresh lenticules can also be used as sources to acquire human CSCs (hCSCs) in vitro although no reports have been documented in the literature. Mimura et al reported that both the peripheral and central regions of rabbit corneal stroma contained a significant number of precursors using spherical cultivation, but the peripheral stroma had more precursors with a stronger proliferative capacity than cells from the central stroma. 26 Therefore, improving the quantity and quality of expanded and activated cells for cell therapy when using primary cells from SMILE-derived lenticules of central stroma rather than the peripheral region is necessary.
This study aimed to develop an effective approach for the proliferation of hCSCs with preferable cell viability and physiological properties. We used soluble porcine corneal stromal extract (pCSE) with low-serum RIFA medium primarily including ROCK inhibitor Y27632, insulin-transferrin-selenium (ITS), fibroblast growth factor 2 (FGF2), L-ascorbate 2-phosphate and 0.5% foetal bovine serum (FBS) to culture primary hCSCs obtained from SMILE-derived lenticules. This study provides insights into the probable mechanisms of pCSE's effect on hCSCs by analysing the protein composition using nano LC-MS/MS. We also assessed the applicability of injecting hCSCs into mouse corneas and explored the effects. We determined whether using pCSE with low-serum RIFA medium not only promoted hCSC proliferation, but also maintained the physiological function of human keratocytes. We also assessed the potential of using hCSCs from SMILE-derived lenticules for corneal stromal cell therapy.

| Ethics statement
Porcine ocular tissues were obtained from a slaughterhouse and the pigs were certified by the Animal Quarantine Bureau of China.

| Preparation of soluble porcine corneal stromal extract (pCSE)
The pCSE was prepared with reference to published methods reported by Yam GH et al 2 Briefly, porcine corneas (n = 8-10, either gender) were isolated from fresh porcine eyes. After they were rinsed with sterile saline containing gentamicin and the epithelial and endothelial layers were removed, the tissues were used to prepare pCSE. The stromal layers were devitalized in DMEM/F12-glycerol (50:50 vol/vol) at −80°C. After they were rinsed with phosphatebuffered saline (PBS), the corneal stromal layers were drip-dried, weighed, and ground under the air phase of liquid nitrogen. The homogenate was suspended in ice-cold sterile PBS (5 ml per gram of tissue). The subsequent steps were performed at 4°C. The suspension was rotated at 300 rpm (THZ-C-1; Taichang, Changsha, China) for 48 h and centrifuged at 15,000 g for 20 min to remove insoluble debris. The supernatant was further filtered using a 0.22 μm membrane filter. Sterile pCSE was collected and stored at −80°C for further use. The total protein concentration was determined via a BCA assay (SolarBio, Beijing, China).

| Live/Dead viability assay
The live/dead assay was used to examined hCSC viability of lenticules after suffering femtosecond laser irradiation and was performed according to the manual of the Live/Dead Viability Assay Kit (Beyotime, Shanghai, China). Then, images were captured by an LSM800 confocal microscope (Zeiss, Germany).

| Isolation and culture of human corneal stromal cells
The human corneal stromal cells (hCSCs) were isolated from SMILE-derived lenticules (Changsha Aier Eye Hospital, Changsha, China). A total of 20 stromal lenticules were obtained from donor myopic patients, whose age was ranged from 22 to 32 years
Media were changed every third days and every media had freshly prepared TGFβ1 added to ensure similar TGFβ1 activity throughout the experiment. Cells in different culture conditions were collected at 7 days.

| Quantitative polymerase chain reaction (qPCR)
Total RNA from hCSCs was extracted using the High Pure RNA Isolation Kit (Roche). The cDNA was synthesized using Revert Aid  Table 1 (Table S1). For qPCR experiments, gene expression was analysed by qPCR (Roche) with three replicates per sample. The results of amplification were normalized to GAPDH mRNA transcript. Expression changes in the gene transcripts for each sample were calculated using the 2 -△△Ct method. The results from three independent experiments were statistically analysed.

| Western blotting assay
The total hCSC proteins were extracted and the protein concentrations were detected using a BCA Protein Assay Kit

| 5-ethynyl-2′-deoxyuridine (EdU) labelling assay
The hCSCs at P4 in RIFA medium supplemented with pCSE were seeded on 48-well plates at 1 × 10 4 cells/well and continually cultured in the RIFA medium, RIFA medium supplemented with pCSE or NM for another 2 days. The EdU labelling assay was conducted according to the manual of the EdU labelling/detection kit (Keygen, Jiangsu, China). Samples were then observed and photographed under a fluorescence microscope. The percentage of EdU-positive cells was calculated using ImageJ software, respectively, from six random fields in three random wells.

| Phycoerythrin (PE) Annexin V and 7-Amino-Actinomycin (7-AAD) assay
The hCSCs at P4 in RIFA medium supplemented with pCSE were seeded into 6-well plates at the density of 1 × 10 5 cells/well in the RIFA medium or RIFA medium supplemented with pCSE for 24 h.
To establish an oxidative stress model, 500 μM hydrogen peroxide (H 2 O 2 ) was added to the culture medium for 24 h at 37 ℃, 5% CO 2 incubator. Apoptosis staining was performed using a PE Annexin V Apoptosis Detection Kit I as the manufacturer's instructions and stained cells were instantly measured using flow cytometry (BD FACSCelesta, USA). One hundred thousand events were collected for each sample.

| Nanoscale liquid chromatography coupled to tandem mass spectrometry (Nano LC-MS/MS) assay, enzyme-linked immunosorbent assay (ELISA) and bioinformatic analyses
The three independent pCSE samples were prepared. The nano LC-MS/MS and ELISA assays were subcontracted to Beijing Biotech-Pack Scientific (Beijing, China). The details are described in Supplementary Methods 1-4. The identified proteins' biological processes were analysed using Metascape. 27

| Intrastromal injection of hCSCs
On the transplantation day, the P4 cells in RIFA medium supplemented with pCSE and NM were digested and passed through a cell strainer to obtain single-cell suspensions. After they were washed, the total cell count was measured and living cells were counted using a trypan blue exclusion assay. The hCSCs were labelled with

| Ophthalmic examination and measurements
The injected corneas were divided into three groups: a RIFA medium supplemented with pCSE group, an NM group and a blank group. and CD45, respectively. 4',6-diamidino-2-phenylindole (DAPI) was used to stain the nuclei. Images were obtained via confocal laser scanning microscopy. For whole-mount immunostaining, the corneas were incubated with primary antibody followed by incubation with fluorescence-conjugated secondary antibody. The corneas were then mounted on slides for confocal laser scanning microscopy.

| Statistical analysis
Values are expressed as the mean ± SD of values obtained from three or more samples. Statistical analysis between two groups was carried out using Student's unpaired t test; comparison among multiple groups was determined by one-way ANOVA P < .05 was considered to be statistically significant.

| Isolation of hCSCs from SMILE-derived lenticules and culture in RIFA medium supplemented with pCSE (RIFA medium + pCSE)
The hCSCs were successfully isolated from SMILE-derived lenticules

| Optimal concentration of pCSE in RIFA medium for the growth of hCSCs
We explored the optimal concentration of pCSE in RIFA medium using a qPCR assay. The hCSCs at passage 4 (P4) were cultured in different concentration groups (0 μg/ml, 0.05 μg/ml, 0.5 μg/ ml, 5 μg/ml and 50 μg/ml) of pCSE in RIFA medium and treated with 10 ng/ml TGFβ1 for 7 days. As demonstrated by light microscopy, the hCSCs expanded considerably under these conditions. The cells had many morphological characteristics, such as slender, spindle and polygonal shapes. Cells had small and stellate shapes only in the 5 μg/ml group. In contrast, the hCSCs cultured in RIFA medium supplemented with pCSE had dendritic or stellate shapes ( Figure 1E). The qPCR analysis indicated that fibrotic genes FN1, ACTA2, COL3A1 and THBS1 were low in the 5 μg/ml pCSE group at all of the concentrations ( Figure 1F). Based on these results, 5 μg/ ml pCSE in RIFA medium is an optimal concentration for hCSC cultures and was used in later studies.

| Characteristics of the hCSCs cultured in lowserum RIFA medium supplemented with pCSE
Immunofluorescence staining, Western blotting and qPCR were used to study the morphology and phenotype of the adherent hCSCs at P4 in RIFA medium, RIFA medium supplemented with pCSE and NM. After attaching for 24 h, the hCSCs at P1-P4 in the RIFA medium supplemented with pCSE showed more physiological morphology. The majority of the adherent hCSCs had dendritic or stellate shapes and formed cellular networks with cell processes extending to connect with neighbouring cells as observed using light microscopy ( Figure 2A). The hCSCs expressed keratocyte markers (ALDH3A1 and lumican) and were weakly stained with fibroblast and myofibroblast markers (fibronectin and α-SMA) via immunofluorescence staining ( Figure 2B). Additionally, F I G U R E 2 Characteristics of adherent hCSCs cultured in the RIFA medium + pCSE and normal medium (NM) groups. (A) The hCSCs at P1-P4 had differing morphologies. (B) Immunofluorescence image of ALDH3A1, lumican, fibronectin and α-SMA of the hCSCs at P4. Scale bars: 50 μm many hCSCs at P3 were obviously positive to ALDH3A1 staining (Fig. S1). In contrast, the hCSCs at P1-P4 in the NM had spindle shapes ( Figure 2A) and were negligibly expressed ALDH3A1 but were strongly positive for lumican, fibronectin, and α-SMA ( Figure 2B). RT-qPCR analysis showed that keratocyte marker related genes, such as ALDH3A1, CD34, KERA and LUM, were significantly greater in the RIFA medium supplemented with pCSE group than those in the NM group (14.08-, 236.1-, 79.1-and 12.57-fold, respectively; P < .05). ALDH3A1 and CD34 were also more upregulated in the RIFA medium supplemented with pCSE group than in the RIFA medium group (8.2-and 13.54-fold, respectively; P < .05) ( Figure 3A). The fibrotic genes (FN1 and THBS1) were significantly lower in the RIFA medium supplemented with pCSE group than in the NM group (0.26-and 0.32-fold, respectively; P < .01). Other fibrotic genes (COL3A1 and ACTA2) were significantly higher in the RIFA medium group than in the RIFA medium supplemented with pCSE (4.60-fold; P < .01) and NM groups (5.94-fold; P < .05) ( Figure 3B). Western blotting confirmed that ALDH3A1, keratocan and lumican were significantly more expressed (7.19-, 2.16-and 1.66-fold, respectively; P < .05) in the RIFA medium supplemented with pCSE group. Compared with the RIFA medium group, ALDH3A1 was also more expressed in the RIFA medium supplemented with pCSE group (3.52-fold; P < .05).
Fibronectin and α-SMA were less expressed (0.15-and 0.14-folds, respectively; P < .01) in the RIFA medium supplemented with pCSE group than in the NM group ( Figure 3C). Taken together, RIFA medium supplemented with 5 μg/ml pCSE can help hCSCs maintain the keratocyte phenotype and prevent hCSC fibrotic transition at least within P4 in vitro subcultures.

| Improving hCSC proliferation and antiapoptosis using low-serum RIFA medium supplemented with pCSE
To investigate the functions of pCSE in terms of hCSC proliferation, cytoactivity and anti-apoptosis, EdU, CCK-8 and H 2 O 2 -induced apoptosis assays were conducted. After treatment with RIFA medium, RIFA medium supplemented with pCSE or NM for 2 d, the results of EdU assay demonstrated that, at the same seeding density, the

F I G U R E 3
The maintenance of the hCSC phenotype after treatment with RIFA medium + pCSE. (A) RT-qPCR analysis showed that the keratocyte marker-related genes ALDH3A1, CD34, KERA and LUM were significantly greater in the RIFA medium + pCSE group than in the NM group. ALDH3A1 and CD34 were also up-regulated in the RIFA medium + pCSE group compared with the RIFA medium group. (B) The fibrotic genes (FN1 and THBS1) were significantly lower in the RIFA medium + pCSE group than in the NM group. Other fibrotic genes (COL3A1 and ACTA2) were significantly higher in the RIFA medium group than in the RIFA medium + pCSE and NM groups, respectively. (C) Western blotting confirmed that ALDH3A1, keratocan and lumican were significantly more expressed in the RIFA medium + pCSE group than in the NM group. Compared with the RIFA medium group, ALDH3A1 was also more expressed in the RIFA medium + pCSE group. Fibronectin and α-SMA were less expressed in the RIFA medium + pCSE group than the NM group (*P < .05; **P < .01; ***P < .001; ****P < .0001) hCSCs at P4 cultured in the NM had more EdU-positive cells than those in the RIFA medium and RIFA medium supplemented with pCSE on day 2 ( Figure 4A). The percentage of EdU-positive hCSC nuclei in the RIFA medium, the RIFA medium supplemented with pCSE and the NM were 19.43 ± 0.47%, 28.88 ± 0.84% and 35.92 ± 1.08%, respectively ( Figure 4B). There was a significant increase in the cell density in the RIFA medium supplemented with pCSE (208 ± 6.37 cells/ mm 2 ) compared with the RIFA medium (143.1 ± 2.53 cells/mm 2 ) after 2 days of culture. The cell density in NM (253.1 ± 4.32 cells/mm 2 ) was also significantly higher than the other two groups ( Figure 4C). The CCK-8 was used to assess cytoactivity of hCSCs cultured in different media, in which the results showed that the hCSC cytoactivity of NM group was significantly increased compared with other two groups, and that for RIFA medium supplemented with pCSE group was more prominent than in the RIFA medium group from day 2 (P < .01). The cell growth curve was further assessed for cytoactivity of hCSCs in different media ( Figure 4D). Besides, following exposure to 500 μM H 2 O 2 for 24 h, the apoptotic rate of the hCSCs in RIFA medium supplemented with pCSE (1.87 ± 0.27%) was lower than in the RIFA medium (10.03 ± 0.73%) (P < .01) ( Figure 4E), indicating that the pCSE reduced the apoptosis of hCSCs when exposed to H 2 O 2 . Therefore, pCSE not only increases hCSC proliferation and cytoactivity, but also prevents hCSC apoptosis in vitro.

| Proteomic identification of pCSE
We used the nano LC-MS/MS assay and bioinformatic analysis to identify the proteomics of the pCSE. The nano LC-MS/MS assay showed that there were approximately 740 kinds of pCSE proteins as listed in Table S2. The biological processes of the identified proteins were analysed by Metascape as represented by the gene symbols. Significant biological processes potentially affected by pCSE proteins were listed in Table 1, which included extracellular matrix organization, developmental growth, fibroblast proliferation and regulation of the apoptotic signalling pathway. The 10 most notable biological processes were illustrated (Table S3), including exocytosis regulation, supramolecular fibre organization, extracellular structure organization, negative regulation of proteolysis, negative regulation of endopeptidase activity and wound healing. Fig. S2A showed the significant biological processes of the identified pCSE specimen proteins and the protein count of each biological process. The 10 most notable biological processes in GO enrichment and the protein count of each biological process demonstrated the pCSE's characteristics (Fig. S2B). The ELISA assay verified the concentration of decorin (90.65 ± 14.1 ng/ml), insulin-like growth factor 2 (84.56 ± 11.7 ng/ml) and clusterin (196 ± 46.61ng/ml) (Fig. S3).
In summary, the pCSE with its anti-fibrosis, pro-proliferation and antiapoptosis activities, is beneficial for hCSC cultivation.

F I G U R E 4
The effect on hCSC proliferation and apoptosis of RIFA medium + pCSE. (A-C) The percentage of EdU-positive cells and cell density in the RIFA medium group, RIFA medium + pCSE group and NM group. (D) The hCSC cytoactivity in the RIFA medium group, RIFA medium + pCSE group and NM group during 7 d via the CCK-8 assay. (E) Compared to the RIFA medium group, RIFA medium + pCSE significantly reduced the apoptosis rate of the hCSCs (**P < .01; ***P < .001; ****P < .0001). Scale bars: 100 μm for A

| Intrastromal injection of hCSCs into normal mouse corneas
The injected corneas were divided into three groups: a RIFA me-  Figure 6A). In addition, corneal whole-mount staining showed persistence of the PKH26-labelled hCSCs (red) in the central corneal region. A few PKH26-labelled hCSCs were distributed in the corneal TA B L E 1 Significant biological processes potentially affected by porcine corneal stromal extract (pCSE) proteins using Metascape analysis Extracellular matrix organization  −39.485  A2M, AEBP1, AGT, ANXA2, BGN, SERPINH1,  COL1A1, COL5A1, COL5A2, COL6A2, COL6A3,  COL11A1, COL17A1, COMP, CTSL, DCN,  FBLN1, FGA, FGG, FMOD, FN1, HSPG2, LOX,  LUM, NID1, PLG, SERPINF2, SPARC, TGFBI,  THBS1, TIMP1 periphery. Compared to the RIFA medium supplemented with pCSE group, gradual cell migration to the corneal periphery was obvious in the NM group ( Figure 6B). Confocal microscopic imaging of the partially thick corneal tissue showed that the hCSCs in the NM group also tended to vertically migrate, while the hCSCs in the RIFA medium supplemented with pCSE seldom migrated in the z-position ( Figure 6C). Human nuclear antigen (HNA)-positive cells were observed both in the RIFA medium supplemented with pCSE and the NM group corneas using corneal whole-mount staining ( Figure 6D), indicating that mouse corneal transparency was not significantly altered by the long-term presence of hCSCs in the RIFA medium supplemented with pCSE.

ECM-related
By day 14 after injection, both the RIFA medium group and RIFA medium supplemented with pCSE group recovered to be transparent (Fig. S4A). Immunostaining of the mouse cornea sections showed that positive CD34 and negative CD45 both in the RIFA medium group and RIFA medium supplemented with pCSE group (Fig. S4B). By day 35 after injection, immunostaining demonstrated that the expression of human CD34 was positive in the RIFA medium supplemented with pCSE group and negligible in the NM and blank groups ( Figure 6E). We also observed positive CD45 expression cells localized around the injected hCSCs in the NM group corneas and negligible in the RIFA medium supplemented with pCSE and blank groups ( Figure 6F). Collectively, these results demonstrate the feasibility of intrastromal injection of hCSCs in the RIFA medium supplemented with pCSE, which did not increase the mouse MCT 35 days after injection. Similar to native keratocytes, these cells increase the expression of the appropriate markers without causing corneal haze and host inflammatory response in vivo.

| D ISCUSS I ON
There are presently several sources of hCSCs, including somatic phospho-ascorbic acid to successfully maintain dendritic morphology in human keratocytes. 39 Our study was consistent with the reported findings without down-regulating the expression of KSPGs in the propagated hCSCs.
Characterizing the ECM, which is the backbone of the cornea, is key for elucidating the differences between corneal stroma cells in vivo and ex vivo. 30  whereas CD34cells are activated stromal cells and stromal fibroblasts do not fully revert to the quiescent condition. 45 Yam et al 46 reported that post-natal periodontal ligament cells injected into porcine corneas expressed CD34 after corneal organ culture for 7 days. In our study, CD34 exhibited negative expression in vitro (data not shown) but positive in vivo, indicating that the injected cells transitioned to the quiescent condition in the native physiologic condition.
We obtained a substantial amount of hCSCs via significant serum supplementation (5%-10%). However, after intrastromal injection, these fibroblast characteristic cells did not produce KSPGs and stromal crystallins but induced corneal haze, host inflammatory response and corneal neovascularization (CNV). 43 Prior research reported that the expression of CD45 (a leucocyte marker) was negligible in corneal stromal keratocyte-injected rat corneas, even in fibroblast-injected corneas, at 2 and 4 weeks post-injection. 43 In contrast, we did not find CD45 + cells in the corneas in the RIFA medium supplemented with pCSE group but they were present in the NM group corneas by day 35 after injection. CNV was negligible in all of the operated corneas. These results indicated there were no obvious adverse reactions after the intrastromal injection of hCSCs in the low-serum RIFA medium supplemented with pCSE.
The limitation of this study is that hCSCs were injected into normal mouse corneal stroma rather than the corneal injury model.
And the short-term animal experiments between RIFA medium and RIFA medium supplemented with pCSE group were not obviously different. The long-term animal studies will be our future work.
Additionally, according to our nano LC-MS/MS results, there were approximately 740 proteins involved in pCSE. Thus, further research into the links and mechanisms by which proteins and CSCs propagate is needed. In the future, we will devote more time to observing injected hCSC interactions with cornea-resident cells. These are crucial factors for clinical applications.
In conclusion, this study on the potential of hCSCs suggested that soluble pCSE with low-serum RIFA medium improved the proliferation of hCSCs from SMILE-derived lenticules with the properties of physiological keratocytes. The pCSE with its anti-fibrosis, pro-proliferation, and anti-apoptosis activities, was beneficial for hCSC cultivation. The intrastromal injection of hCSCs into normal mouse corneas demonstrated that the hCSCs in the low-serum RIFA medium supplemented with pCSE led to a more rapid recovery of corneal thickness and transparency, less migration, positive CD34 expression and negative CD45 expression in corneas compared with implanted hCSCs cultured in the NM. Therefore, using pCSE with low-serum RIFA medium not only promoted hCSC proliferation, but also maintained the physiological function of human keratocytes. The hCSCs from the SMILE-derived lenticules in the pCSE with low-serum RIFA medium demonstrated good potential for corneal stromal cell therapy.

ACK N OWLED G M ENTS
This study was supported by the National Natural Science

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