Human platelet lysate as a replacement for fetal bovine serum in human corneal stromal keratocyte and fibroblast culture

Abstract The isolation and propagation of primary human corneal stromal keratocytes (CSK) are crucial for cellular research and corneal tissue engineering. However, this delicate cell type easily transforms into stromal fibroblasts (SF) and scar inducing myofibroblasts (Myo‐SF). Current protocols mainly rely on xenogeneic fetal bovine serum (FBS). Human platelet lysate (hPL) could be a viable, potentially autologous, alternative. We found high cell survival with both supplements in CSK and SF. Cell numbers and Ki67+ ratios increased with higher fractions of hPL and FBS in CSK and SF. We detected a loss in CSK marker expression (Col8A2, ALDH3A1 and LUM) with increasing fractions of FBS and hPL in CSK and SF. The expression of the Myo‐SF marker SMA increased with higher amounts of FBS but decreased with incremental hPL substitution in both cell types, implying an antifibrotic effect of hPL. Immunohistochemistry confirmed the RT‐PCR findings. bFGF and HGF were only found in hPL and could be responsible for suppressing the Myo‐SF conversion. Considering all findings, we propose 0.5% hPL as a suitable substitution in CSK culture, as this xeno‐free component efficiently preserved CSK characteristics, with non‐inferiority in terms of cell viability, cell number and proliferation in comparison to the established 0.5% FBS protocol.


| INTRODUC TI ON
The ex vivo cultivation of primary human corneal stromal keratocytes (CSK) and stromal fibroblasts (SF) is crucial for corneal research and treatment. [1][2][3][4] Corneal transplantation is the current treatment modality of choice for patients with advanced corneal disease to revive visual ability. 1,5 However, many factors restrict the long-term success including limited graft survival, allogeneic graft rejection, the need for immunosuppressants, high associated costs and most importantly the global donor material shortage. 1,6 Recent years have seen increasing interest in understanding corneal disease and the implication of cells of the corneal stroma. 3,7 In addition, efforts have been undertaken in the search for tissueengineered alternatives to human donor corneal transplants. 2,8 Corneal cell therapy has seen great advances, particularly for the epithelial and endothelial layers. 1,6,9,10 Nevertheless, the targeted delivery of cells into the cornea to treat stromal disease, which could possibly replace corneal transplantation for some indications in the future, is still in its beginnings. 1,4,11,12 When carefully inspecting the published literature on experiments with human cells of the corneal stroma, it is easily recognized that in the majority of studies SF were used for experiments, 13,14 while very few studies verified the true CSK character of cells before use. 3,15,16 The isolation and propagation of CSK are challenging, as this delicate cell type easily transforms into scar inducing SF and α-smooth muscle actin (SMA) expressing myofibroblasts (Myo-SF). 17 CSK differ fundamentally from SF, for example in terms of phenotype, gene expression, transparency, extracellular matrix (ECM) remodelling and neuroregulatory capabilities. 3,15,16 Therefore, expanding human CSK, SF and Myo-SF ex vivo, while maintaining their unique phenotype, is imperative and extremely desirable for cell research, understanding corneal disease and wound healing as well as their possible future application in tissue-engineering and cell therapy. [1][2][3][4] Yam et al. 17 recently introduced a protocol to safely propagate CSK ex vivo. In their protocol, primary human CSK are "activated" by culturing them with very low fetal bovine serum additive (0.5% FBS), which allows expansion for 6-8 passages ex vivo. When the activated CSK are then returned to serum-free culture, CSK characteristics become reinforced. 17 Concerns have been raised regarding the safety of FBS-based culture media. Bovine antigens, for instance, accumulate intracellularly; hence, cells expanded in FBS containing medium can lead to anaphylactic reactions if administered repeatedly. [18][19][20][21] The ingredients of FBS are not precisely defined, and there is a high lot-to-lot variation. 22 Fetal bovine serum can contain high endotoxin levels, potentially increasing the production of proinflammatory and profibrogenic cytokines in cultured cells. 22,23 Additionally, the bleeding procedure of bovine fetuses, necessary for FBS production, is of animal welfare concern. 22 Therefore, protocols to culture cells for clinical applications should-according to Good Manufacturing Practice-avoid the usage of animal sera. 24 Over the last decade, different preparations of human blood products have been tested regarding their suitability as xeno-free cell culture additives to replace FBS, among them plasma rich in growth factors (PRGF), platelet-rich plasma (PRP) and human platelet lysate (hPL). 25 To date, there are no standardized protocols, which entails heterogeneity in terms of nomenclature, manufacturing and content. 25 However, the production of all these products involves the separation of blood components from platelets and plasma by centrifugation as well as releasing a wide range of growth factors from platelets by cell activation and/or lysis steps. Platelets contain more than 1,100 different proteins, among them transforming growth factor β (TGFβ), platelet-derived growth factor (PDGF), insulin-like growth factor 1 (IGF-1), vascular endothelial growth factor (VEGF), fibroblast growth factor (FGF) and others, which are known to be involved in tissue regeneration. 26,27 Nevertheless, some limitations remain. Depending on the protocol, studies have shown that PRP contained higher amounts of leukocytes than PRGF, 28 which is known to have a negative effect on cell viability. 25 Previous protocols used bovine thrombin to activate platelets during PRP production, which incurs the risk of immunogenic reactions. Activation by calcium chloride is a viable alternative. 26 hPL on the other hand is usually generated by a freeze-thaw procedure of platelet concentrates, which is fast and effective and lyses all corpuscular elements. This leads to a very high growth factor content and low leukocyte concentrations in hPL. 22,25,29 Plasma rich in growth factor and PRP are usually prepared on site using specialized kits. 7 Pooled allogenic hPL, used in this study, is commercially available, allowing improved growth factor control and consistency. 25,30 In addition, hPL is habitually stored frozen and easily used for consecutive applications. 25 It should be noted that several groups freeze their PRP or PRGF before addition to culture medium, which then closely resembles hPL. PRGF, PRP and hPL can all be used in autologous settings to further reduce risks of contamination or immune reactions. 22 Previous research has shown promising effects of hPL on cells of the eye. hPL enhanced the proliferation of human mesenchymal stem cells (hMSC) and conjunctival fibroblasts. 31,32 In a clinical trial, hPL eye drops led to the uncomplicated healing of various corneal lesions. 33 In summary, there is significant evidence promoting a beneficial effect of hPL vs. FBS for human CSK and SF culture, which we investigated in this study.

| Isolation of CSK and SF
Human CSK were isolated from 22 corneas (11 donors) unsuitable for transplantation (age 64.6 ± 14.6 years, male = 6, female = 5) supplied by the Cornea Bank Aachen, following institutional review board approval (EK 291/20). CSK were isolated and cultivated as previously described. 8,15,17 Briefly, corneas were washed with sterile phosphate-buffered saline (PBS, 0.1 M; Merck KGaA), the central button was trephined (8.0 mm diameter) and incubated with dispase II (20 mg/ml; Roche) for 1 h at 37℃. The loosened corneal epithelium and endothelium were removed by gentle scrapping. The remaining stromal tissue was then digested with collagenase I (1.5 µg/ml; Gibco, Life Technologies) in CSK basal medium (Table 1) for 12 h at 37℃. Single cells were then suspended in CSK basal medium with 0.5% FBS (Panbiotech<>, Table 1). Cells were seeded on collagen-I-coated (type I collagen, solution from rat tail, Sigma-Aldrich) culture plates (BD Biosciences). The medium was changed every 3 days.

| Cell culture of CSK and SF
Corneal stromal keratocytes were cultured in CSK basal medium containing 0.5% FBS until passage 3 ( Figure 1). After washing three times with PBS, CSK were incubated for 24 h in serum-free basal medium.
After 24 h, the medium was exchanged for new medium containing the according substitutes (Table 1). In media containing hPL (PL Bioscience), 2 IU/ml heparin (PL Bioscience) was added according to the manufacturer's instructions to avoid gel formation. The 0.5% FBS CSK group served as control. At passage 3, CSK from each cornea were also converted into SF by incubating them for 7 days in SF basal medium substituted with 5% FBS. After washing in PBS for three times and 24 h in serum-free medium, SF were also exposed to the five different media (Table 1). After 3 days of culture in the according substitutes, cells were harvested for further testing.

| Viability and cell number analysis
Cells were seeded at 9000 cells/1.8 cm 2 on collagen-I-coated 4well chamber slides (Nunc Labtek Chamber Slide, Sigma-Aldrich) and incubated in different media (Table 1)

| Immunohistochemistry
Cells were seeded at 10,000 cells/2.0 cm 2 on collagen-I-coated glass cover slips (VWR International). After 3 days of culture in different media (Table 1), cells were fixed with neutral-buffered 4% paraformaldehyde (Sigma-Aldrich). After quenching with icecold 50 mM ammonium chloride (Sigma-Aldrich), samples were washed with PBS containing 0.2% bovine serum albumin (BSA; Sigma-Aldrich) and blocked with 1% bovine serum albumin and Triton X (1 µl/ml; Sigma-Aldrich) followed by incubation with primary antibodies for 2 h at room temperature (CSK markers: rab-

| Real-time polymerase chain reaction
Cells were seeded at 47,000 cells/9.5 cm 2 on collagen-I-coated 6well plates (Corning, New York, USA) and incubated in different media (Table 1)

| Statistical analysis
All data were expressed as mean ± standard deviation (SD).
Statistical analyses were performed with SPSS version 22.0 (IBM).
Mann-Whitney U or Wilcoxon rank-sum tests were used to compare cell viability, cell numbers, proliferation rates, gene ratios and growth factor levels. A p value ≤0.05 was considered statistically significant.

F I G U R E 1
Experimental flow of this study, with the aim to find a xeno-free alternative to fetal bovine serum (FBS) for the culture of primary human corneal stromal keratocytes (CSK) and stromal fibroblasts (SF). CSK were isolated from 22 donor corneas. CSK have a dendritic phenotype and express characteristic markers, for example aldehyde dehydrogenase family 3 member A1 (ALDH3A1+), keratocan (KERA+) or lumican (LUM+). CSK were expanded in 0.5% FBS for three passages. At passage 3, 50% of cells were incubated for 7 days in 5% FBS, to generate SF. SF are crucial for corneal wound healing. They have a spindle-cell morphology and lose CSK marker expression (ALDH3A1−, KERA−, LUM−). After keeping these CSK and SF for 24 h in serum-free media, they were then exposed to media containing different levels of FBS or human platelet lysate (hPL) for 3 days. The cells were then analysed for viability, proliferation, and marker expression.

| Viability and cell morphology analysis
Viability analysis of CSK via FDA/PI staining showed high viability rates in all groups and containment of typical dendritic morphology in 0.5% FBS, 0.5% and 2% hPL (Figure 2A). 5% FBS CSK showed a fibroblastic appearance, and 10% hPL led to a spider web-like arrangement of CSK. Viability rates of CSK 0.5% hPL (99.56 ± 0.31%,  Figure 3A, Table S2).
Stromal fibroblasts showed high viability in all tested groups without significant differences ( Figures 2B and 3A, Table S3). 10% hPL SF also showed a spider web-like arrangement of cells.
No SMA expression was detected in any CSK group ( Figure 4A, second row).
ALDH3A1 was highly expressed in CSK 0.5% FBS and in 0.5% hPL.
Low expression was seen in 2% hPL. In CSK 5% FBS and 10% hPL no expression of ALDH3A1 was detected ( Figure 4A, third row).
Immunohistochemistry staining of SF showed a weak expression of LUM in SF 0.5% FBS, SF 0.5% hPL and SF 2% hPL. No expression was seen in SF 5% FBS and SF 10% hPL ( Figure 4B, top row). ALDH3A1 was weakly expressed in SF 0.5% FBS and SF 0.5% hPL. No expression was seen in SF 5% FBS, SF 2% hPL and SF 10% hPL ( Figure 4B, third row).

SF induction
Corneal stromal keratocytes were transformed to SF by culturing them for 7 days in 5% FBS medium. The transformation was verified by comparing the relative gene expression between CSK 0.5% FBS and the SF. SF showed a significant decrease in the expression of CSK markers ALDH3A1 (0.37 ± 0.11, p = 0.020, Figure 5,

| ELISA
Via ELISA a bFGF content of 0.067 ± 0.017 ng/ml was found in our hPL solutions. In our FBS solutions bFGF was not detectable ( Table 2).
The HGF content of our hPL solutions was 1.074 ± 0.050 ng/ml.
HGF was not detectable in our FBS solutions. The amount of TGF-β1 was significantly higher in our hPL (1.861 ± 0.231 ng/ml, p < 0.001) compared to our FBS solutions (0.015 ± 0.010 ng/ml).

| DISCUSS ION
In this study, we investigated the suitability of hPL as a replacement for FBS for the xeno-free culture of primary human CSK and SF. We

F I G U R E 3 (A) Percentage of viable cells, (B) cell numbers and (C)
percentage of Ki67+ corneal stromal keratocytes (CSK) and stromal fibroblasts (SF) after incubation in media containing 0.5% fetal bovine serum (FBS), 5% FBS, 0.5% human platelet lysate (hPL), 2% hPL and 10% hPL for 3 days. Differences were compared to the reference of 0.5% FBS CSK for CSK (*p ≤ 0.05, **p ≤ 0.01, ***p ≤ 0.001) and 0. Nevertheless, Col8 is also expressed in the central corneal stroma, 38 and Col8A2 by CSK. 39 We previously demonstrated, that Col8A2 expression is higher in CSK than in SF, and can be used to differentiate between the two. 17 The conversion of CSK to SF is usually described by a loss of typical CSK markers, and the expression of unspecific fibroblastic cell markers such as Thy-1 membrane glycoprotein (Thy-1/CD90), fibronectin or tenascin C. 11,14,36,[40][41][42] Stimulated by TGF-β1, corneal SF can further transform to SMA expressing Myo-SF, one of the most important processes in corneal fibrosis (scarring). 43,44 The biological effects of TGF-β1 in the cornea have been shown to follow SMAD-dependent as well as SMAD-independent signalling pathways depending upon cellular responses and microenvironments. 45 The complex process of corneal scarring to date is still not fully understood. 45 In this study, we confirmed Myo-SF induction via elevation of SMA expression and loss of the CSK markers LUM, ALDH3A1 and Col8A2.  17 In agreement, we also found higher CSK Ki67+ rates in 5% FBS (35.29 ± 5.95%) compared to 0.5% FBS (11.18 ± 6.08%) after 3 days of culture. Ki67 is a nuclear proliferation factor expressed at all stages of the cell cycle except G0. 48 Therefore, Ki67+ fractions are generally higher than the MI; however, the difference varies depending on cell type. [49][50][51] Liliensiek et al. 52 53  The risk of CSK transformation with higher proliferation rates was previously shown by Jester et al. 55,56 and attributed to TGF-β1 induced changes in gene expression patterns.
Different approaches have been investigated to ex vivo, pharmacologically impede Myo-SF conversion through, for example proliferator-activated receptor gamma (PPARγ) ligands that downregulate TGF-ß1 induced ß-catenin signalling through p38 MAPK inhibition in corneal fibroblasts. 57,58 However, the only known method to preserve CSK characteristics during culture ex vivo to date is the usage of very low growth factor substitution.
This approach was initially suggested by Yam et al. applying a two media culture protocol switching between 0.5% FBS for propagation and 0% FBS for CSK stabilization. 17 His results agree with our findings and in addition to FBS also apply for hPL, as we were able to demonstrate.
Interestingly, we detected that the expression of the Myo-SF marker SMA increased with higher amounts of FBS but decreased with incremental hPL substitution in both cell types. We conducted a literature research paired with ELISAs (Table 2) and found differences in the content of crucial cytokines in both supplements as a possible explanation.
TGF-β1 is known to transform primary CSK and immortalized corneal fibroblasts into Myo-SF. 59,60 Interestingly, we found a higher amount of TGF-β1 in our hPL solutions than in our FBS solutions, which agrees with the literature (Table 2) and would imply incremental SMA expression with higher hPL substitution.
A possible explanation for these contradictory findings is the higher amount of HGF and bFGF in hPL compared to FBS, described in the literature and detected by us via ELISA (Table 2). HGF is known to counteract the TGFβ signalling pathway, via Smad7 activation and Smad2 inhibition. 61 Therefore, HGF has been shown to impede Myo-SF conversion, even in the presence of TGF-β1, in human SF after 24 h of incubation. 60

Maltseva et al. induced a Myo-SF phenotype in primary rabbit
CSK via incubation with 0.25-1 ng/ml TGF-β1 and found that 20 ng/ ml bFGF plus 5 µg/ml heparin promoted the SF phenotype by reversing the SMA expression of Myo-SF after 3 days of incubation. 62 Jester et al. incubated primary rabbit CSK with 10 ng/ml bFGF and found a fibroblast like phenotype but negative SMA immunohistochemistry after 7 days of culture. 59 Anitua et al. 63   Considering all findings, we found that primary human CSK and SF can be cultured with xeno-free hPL. We propose 0.5% hPL as a suitable substitution in CSK culture, as this xeno-free component efficiently preserved CSK characteristics, with non-inferiority in terms of cell viability, cell number and proliferation in comparison to the established 0.5% FBS protocol. Unfortunately, the higher proliferation rates with incremental hPL substitution came at the price of CSK marker diminution and therefore seem unsuitable for the culture of this delicate cell type. hPL contains the antifibrotic HGF and bFGF, potentially suppressing Myo-SF conversion, which could be useful in its future application in corneal cell research and treatment but requires further investigation. Aachen for supplying human corneas unsuitable for transplantation for this study, as well as Antje Schiefer, Anna Dobias, and Anne

ACK N OWLED G EM ENTS
Freialdenhoven for assisting in the laboratory work. Open Access funding enabled and organized by Projekt DEAL.

CO N FLI C T S O F I NTE R E S T
The authors have no relevant financial or non-financial interests to disclose.

DATA AVA I L A B I L I T Y S TAT E M E N T
All data can be requested form the corresponding author.