Plasma gelsolin promotes re-epithelialization

Woundhealing disorders characterized by impaired or delayed re-epithelialization are a serious medical problem that is painful and difficult to treat. Gelsolin (GSN), a known actin modulator, supports epithelial cell regeneration and apoptosis. The aim of this study was to estimate the potential of recombinant gelsolin (rhu-pGSN) for ocular surface regeneration to establish a novel therapy for delayed or complicated wound healing. We analyzed the influence of gelsolin on cell proliferation and wound healing in vitro, in vivo/ex vivo and by gene knockdown. Gelsolin is expressed in all tested tissues of the ocular system as shown by molecular analysis. The concentration of GSN is significantly increased in tear fluid samples of patients with dry eye disease. rhu-pGSN induces cell proliferation and faster wound healing in vitro as well as in vivo/ex vivo. TGF-β dependent transcription of SMA is significantly decreased after GSN gene knockdown. Gelsolin is an inherent protein of the ocular system and is secreted into the tear fluid. Our results show a positive effect on corneal cell proliferation and wound healing. Furthermore, GSN regulates the synthesis of SMA in myofibroblasts, which establishes GSN as a key protein of TGF-β dependent cell differentiation.

Wound healing is a complex response to injuries of tissues and cell clusters. The main steps in wound healing are inflammation, tissue formation and tissue remodeling [7][8][9] . Injuries activate inflammatory cells, trigger release of mediators and cause contraction of blood vessels and formation of blood platelet clots. Furthermore, migration of keratinocytes, angiogenesis and, ultimately, wound closure succeeded by tissue remodeling are part of the wound healing cascade 8,[10][11][12] .
A main part of wound healing is a complex interaction between macrophages, granulocytes and fibroblasts 10 . A key factor in the underlying signaling cascades is transforming growth factor-β (TGF-β). TGF-β activates epithelial to mesenchymal conversion, which is important for sufficient wound healing, and also regulates maturation from fibroblasts to myofibroblasts, which contribute to wound contraction 13,14 . The decisive role of myofibroblasts in the process of wound healing is discussed controversially, but the prevailing opinion is that they are necessary 15,16 . TGF-β is an important regulator of epithelial wound healing and leads to an increased gene expression of gelsolin (GSN). GSN, on the other hand, can modulate TGF-β-induced cell differentiation 17 .
GSN is a multifunctional actin-binding protein that is produced by almost every cell type 18 . The GSN gene undergoes post-transcriptional splicing, yielding a cytoplasmatic isoform (cytoplasmatic gelsolin -cGSN) and a secreted plasmatic isoform (plasmatic gelsolin -pGSN) [20]. cGSN is essential to the processes of cellular locomotion and morphogenesis, as it regulates assembly and degradation of intracellular actin filaments by capping, nucleation and severing 19 . It is therefore of special importance for phagocytosis and the proper function of thrombocytes 20,21 . pGSN,on the other hand, is produced mainly by skeletal muscle cells and distributed into the blood 22 . pGSN binds (scavenges) and removes free toxic actin after cell damage through non-proteolytic deterioration. Circulating pGSN seems to be a humoral reserve recruited in case of inflammation or trauma. Other functions of pGSN are elimination of bacterial toxins, e.g. lipopolysaccharide and lipoteichoic acid, and amelioration of immune response by reducing the release of proinflammatory interleukins 18,[23][24][25][26] . Witke et al. showed that skin fibroblasts of GSN null mice demonstrated severe impairments in cell motility based on an incapacity to reorganize their actin skeleton 27 .As such, GSN has tremendous potential to enhance wound repair and might also qualify as a future therapeutic.

Results
GSN is expressed and synthesized in human and murine ocular tissues and cell lines. Tissue samples from human and murine cornea, conjunctiva, lacrimal gland, efferent tear duct, eyelid, lung, liver and stomach as well as three different human epithelial cell lines, i.e. a human corneal epithelial cell line (HCE), a human conjunctival epithelial cell line (HCjE) and a human meibomian gland epithelial cell line (HMGEC) were tested positive for GSN mRNA (Fig. 1A,B). ß-actin served as loading control for all samples. The sequenced PCR bands were in accordance with the expected sequences from GSN in the gene database (www.ncbi.com).Protein extracts from cornea, conjunctiva, lacrimal gland, efferent tear duct, eyelid, lung, liver, stomach, HCE, HCjE, HMGEC (Fig. 1C,D)and human tear fluid samples (Fig. 2D) were tested for presence of GSN by Western blot analysis. Specific GSN protein bands were detected at 80 kDa and 47 kDa, respectively.
For immunohistochemistry, paraffin-embedded tissue sections from cornea, conjunctiva, eyelid, lacrimal gland, stomach and lung were analyzed. All investigated tissue samples showed positive anti-GSN antibody reactivity ( Fig. 2A,B). Negative control sections were negative for all tissues (data not shown).
Human (Hu) cornea: GSN was detected cytoplasmically in all epithelial cell layers of the corneal epithelium (Fig. 2A1). Mouse (Mo) cornea: GSN was only detected in the basal epithelial cell layer of the corneal epithelium (Fig. 2B1). Hu-conjunctiva: Epithelial cells of the conjunctiva revealed intracytoplasmatic reactivity (red staining) with the anti-GSN antibody, with the exception of the secretion product of intraepithelial goblet cells (Fig. 2A2). Mo-conjunctiva: GSN was visible in cells of the conjunctival epithelium (Fig. 2B2). Hu-meibomian gland: Meibocytes reacted positive with the anti-GSN antibody as did the lining cells of the excretory duct system (Fig. 2A3). Mo-meibomian gland: Murine meibocytes reacted positive with the antibody to GSN as well (Fig. 2B3). Hu-lacrimal gland: GSN reactivity was detected intracytoplasmatically within acinar cells (Fig. 2A4). Mo-lacrimal gland: There was a GSN reactivity in acinar cells (Fig. 2B4). Hu-stomach: Only parietal cells revealed a positive antibody reactivity (Fig. 2A5). Mo-stomach: The antibody to GSN reacted positive in all cells of mouse gastric glands (Fig. 2B5). Hu-lung and Mo-lung: GSN reactivity was visible in type I and type II alveolar epithelial cells (Fig. 2A6,B6).
Immunofluorescence showed presence of GSN in human lacrimal and meibomian glands (Fig. 2C). Green fluorescence indicates positive antibody reactivity, visible in particular in lining cells of the excretory ducts.
All human and murine tissues of the ocular surface and lacrimal system showed higher concentrations of GSN than the respective positive control tissues of liver, stomach and lung.
The pGSN concentration is significantly increased in tear fluid of patients suffering from aqueous-deficient or evaporative DED. Quantification of pGSN in tear fluid of patients suffering from aqueous-deficient dry eye (ADDE) or evaporative dry eye (EDE) revealed a significantly (p ≤ 0.0005) higher pGSN concentration compared to tears from healthy volunteers (Fig. 2G, Table S1).
The influence of rhu-pGSN on cell proliferation. In order to mimic inflammatory conditions similar to bacterial infection or DED, we challenged rhu-pGSN stimulated HCE cells with the bacterial toxin LPS and the inflammatory cytokine TNFα. Flow cytometry analysis based on BrdU incorporation demonstrated increased HCE cell proliferation in cells treated with rhu-pGSN (Fig. 3A,B). Stimulation with 300 µg/ml rhu-pGSN yielded an increased cell proliferation rate (34.6%) compared to treatment with 300 µg/ml BSA (20.7%). LPS and TNFα were used to simulate a bacterial inflammation and an immune response, respectively. An additional treatment with 1 µg/ml lipopolysaccharides (LPS) or 10 ng tumor necrosis factor alpha (TNFα) had an expected diminishing effect on the proliferation rates ( Fig. 3C-F). However, rates under rhu-pGSN stimulation were still markedly higher than under BSA control treatment (LPS: 26.5% versus 18.2%; TNFα: 20.8% versus 13.4%) ( Table 1).
Quantification of the measured impedance or simplified as cell proliferation under rhu-pGSN stimulation was also analyzed by Electric Cell-Substrate Impedance Sensing (ECIS ® ) (Fig. 3G). Again, HCE cells were used as a model system for the ocular surface. There was a significant difference in cell proliferation rates between the control group (BSA) and both rhu-pGSN stimulated groups (30 µg/ml and 300 µg/ml) after 96 hours and 144 hours of stimulation. HCE cells treated with 300 µg/ml rhu-pGSN showed an even higher cell proliferation rate than cells treated with only 30 µg/ml rhu-pGSN (Fig. 3G).
rhu-pGSN increases the wound closure rate in vitro and in a combined corneal in vivo/ex vivo defect model. The cell culture-based wound healing assay revealed increased gap closure of HCE cells in the presence of 300 μg/ml rhu-pGSN compared to no treatment or treatment with 300 µg/ml BSA (Fig. 3H,I). After 24 hours incubation time, the remaining wound area was significantly (p < 0.001) narrowed with a more than 2.5 fold increase in wound area restoration (Supplement Statistic PDF). Positive effects of rhu-pGSN with regard to wound healing were also visible in a corneal wound healing mouse model (in vivo/ex vivo cornea defect model) (Fig. 3J). Red coloration indicates the wound surface. To compare the corneal wound healing process between placebo-treated and rhu-pGSN groups, we performed Kaplan-Meier analysis using digitized areas of the remaining corneal defect area over time. No significant reduction of the wounded area over time was observed in the untreated group as well as after treatment with 3 µg/ml rhu-pGSN. Treatment with30 µg/ml rhu-pGSN revealed a statistically not significant tendency to enhanced wound surface closure. However, mice corneae treated with 300 µg/ml rhu-pGSN showed complete restitutio ad integrum after To determine morphological effects of rhu-pGSN and/or recombinant human TGF-β (rhuTGF-β) stimulation, we analyzed human primary corneal fibroblasts stimulated with either one substance or both substances. Control fibroblasts appeared longer, thinner and showed no intracellular granulation compared to stimulated fibroblasts (Fig. 4A). Immunofluorescence revealed subjectively more immunoreactivity to smooth muscle actin  (SMA) (green staining) in fibroblasts stimulated with 3 ng/ml rhuTGF-β and 300 µg/ml rhu-pGSN compared to unstimulated corneal fibroblasts (Fig. 4B).
Analysis of SMA by Western blotting after protein isolation detected a singular band at the expected molecular weight (about 40 kDa) in all analyzed samples of stimulated (3 ng/ml rhuTGF-β or 300 µg/ml rhu-pGSN or both in combination) and unstimulated human primary corneal fibroblasts (Fig. 4C). Semi-quantitative densitometry of Western blot analysis revealed significantly (p < 0.0001) higher synthesis of SMA in cells cultured with 300 µg/ ml rhu-pGSN or a combination of 3 ng/ml rhuTGF-β and 300 µg/ml rhu-pGSN, each in comparison to control (Fig. 4D). The latter had comparable effects to 3 ng/ml rhuTGF-β alone.

Amount of smooth muscle actin mRNA is decreased after GSN gene knockdown of both
isoforms. Values were normalized for fibroblasts that were treated with 3 ng/ml rhuTGF-β (control). The amount of SMA mRNA was 75% lower in cultivated human corneal fibroblasts treated with 50 nmol GSN-siRNA compared to control (Fig. 4E). This decrease was statistically significant (p < 0.01). In contrast, at the protein level, semi-quantitative densitometry of Western blot analyses (data not shown) after gene knockdown did not show a significant change in SMA synthesis (Fig. 4F).

Discussion
The goals of our present work were to study the expression patterns of pGSN in tissues from either human or mice ocular surface, to quantify pGSN levels in tear fluid of healthy human subjects and dry eye patients and to investigate the significance of pGSN for corneal wound healing.
Protein size variations for GSN were also described in other journals 28 . The protein band at 80 kDa is specific for the plasmatic form of GSN 18 . The 47 kDa band derives from split of Gelsolin through caspase-3 29 .Interestingly, all cell types related to the ocular surface or participating in tear film production or transport revealed expression and production of pGSN, i.e. corneal and conjunctival epithelial cells (with the exception of goblet cells), acinar epithelial cells of the lacrimal gland and meibocytes as well as epithelial cells of the nasolacrimal duct. This implies multiple origins for the high pGSN concentrations found in tear fluid.
Unexpectedly, the concentrations of pGSN in ocular surface tissues were markedly higher compared to lung, stomach or liver samples, which we used as positive controls. As reported previously, the main sources of pGSN in various vertebrate species are skeletal muscle and smooth muscle myocytes 30 . Abundance of pGSN in the tissues belonging to the ocular surface in comparison to relatively low levels of pGSN in parenchymal organs therefore might indicate an important function of pGSN at the ocular surface and lacrimal apparatus.
To further elucidate this functional significance, samples of human tear fluid from healthy subjects and patients suffering from different forms of dry eye disease (DED) were analyzed regarding their pGSN concentration. Interestingly, pGSN levels of physiological tear fluid were 17-34-fold lower than the aforementioned tissues of the ocular surface. However, tear samples from DED patients revealed a significantly increased concentration of pGSN in comparison to tear fluid from healthy volunteers, while patients with the aqueous-deficient form of DED (ADDE) in particular revealed highly significantly increased pGSN concentrations. DED is generally characterized by varying amounts of inflammation and tissue damage 31 . Both events are associated with the release of actin and other intracellular components, for example Galectin-3, into the extracellular space and hence into the tear film 32,33 .
Actin, being the most abundant intracellular protein in mammalian cells, has vital roles in cellular integrity, structure and motility, but becomes toxic and develops serious detrimental effects if liberated into the extracellular space secondarily to cell damage 32 .
pGSN is a major player of the extracellular actin-scavenger system, being responsible for clearing free actin from the circulation via the liver and preventing systemic actin toxicity 32 . It also has been shown to be of great importance in elimination of bacterial toxins and regulation of possibly harmful immune responses 23,34,35 . The markedly alleviated pGSN concentration in the tear fluid of DED patients could, hypothetically, be a direct response to an abundance of free actin in DED. In this context it seems that pGSN plays a more important role in the aqueous form of DED than in the evaporative form. This subset of patients is more prone to tear hyperosmolarity, a key driver in ocular surface inflammation, already in the earlier stages of the disease 36 . Hyperosmolarity triggers the release of matrix metalloproteinases and inflammatory cytokines like IL-1β and TNFα, subsequently causing apoptotic death of ocular surface epithelial cells 37 . Further experiments are needed to elucidate whether this is the mechanism of action behind the markedly elevated pGSN concentration in DED, especially ADDE. pGSN significantly increased the proliferation rate of human corneal epithelial cells in vitro (FACS and ECIS ® ). These findings are consistent with the results of Zhang et al., who were able to demonstrate that pGSN has a proliferation-stimulating effect on mesangial cells via the TGF-ß1 pathway 38,39 . The ocular surface responds to inflammatory events such as bacterial, viral or fungal infections or DED with an upregulation of inflammatory mediators (e.g. IL-1β and TNFα) 40 . As expected, the proliferative effect of rhu-pGSN was partly reduced. Such antiproliferative effects have been well documented for both LPS and TNFα 40,41 . LPS in particular has been reported to decrease GSN levels actively 42 . However, HCE cells treated with rhu-pGSN instead of BSA retained higher proliferation rates even under LPS or TNFα stimulation. This finding To investigate further whether rhu-pGSN has the potential to promote wound healing at the ocular surface, the human HCE cell line was studied in vitro, and mice bulbs were studied in vivo/ex vivo. Our results revealed that a concentration of 300 µg/ml rhu-pGSN significantly increased wound healing in both models. rhu-pGSN has been connected to beneficial effects in various disease models and hypogelsolinaemia, a condition describing lack of pGSN in the circulation, has been associated with poor outcome in numerous conditions 43,44 . However, the value of pGSN for wound healing has not yet been investigated in detail. In this context, our findings are consistent with a report by Rodriguez et al., who first described positive immunoreactivity for gelsolin in a case of corneal wound healing in the context of epikeratoplasty 45 . As the corneal stroma is not affected by the "defect inducing procedure" no fibrotic effects result or were observed during or after bathing the eyeballs in rhu-pGSN containing medium.
Finally, the potential of pGSN to induce differentiation of fibroblasts into myofibroblasts was tested in vitro. These experiments with fibroblasts were performed to demonstrated, that gelsolin is not only acting on epithelial cells, but also can have effects in deeper corneal layers such as the stroma. In the early phase of wound healing, differentiation of residential fibroblasts into myofibroblasts via TGF-β is an important step in wound closure and a prerequisite for definite healing 46 . It's know that TGF-β regulates GSN expression 17 . Our study revealed that rhu-pGSN significantly increased the expression of SMA in cultured fibroblasts, which is a morphological marker for differentiation into myofibroblasts 47 . In line with this, gene knockdown of GSN demonstrated a decreased expression of SMA mRNA. These findings indicate that pGSN not only promotes wound healing through TGF-β-associated cascades by affecting epithelial cells, but also by fibroblast transformation. In contrast to the measured SMA mRNA regulation the SMA protein level was not changed. The presumable reason for this is the short incubation time with siRNA for only 24 hours. Our results in human corneal fibroblasts indicate that SMA is increased after treatment with rhu-pGSN suggesting that the regulation of the TGF-β cascade could promote scar formation. Further experiments are needed to elucidate this effect further.
In conclusion, our study demonstrates that pGSN is an important molecular player at the ocular surface, as it is present in the tear film of healthy subjects (at least human and mice) and increased in the tear film of patients suffering from DED. Considering the well-known functions of pGSN in the circulation, a significant role for the regulation of inflammatory conditions of the ocular surface, besides its importance as actin-scavenger, may be postulated. Furthermore, our study shows the beneficial effects of rhu-pGSN for corneal wound healing. These findings may be extrapolated to the wound healing process in general, but further evidence in other, exemplary skin models of wound healing is necessary to confirm our results. Our study identifies rhu-pGSN as a promising target for new pharmacological approaches, aiming to enhance impaired or delayed wound healing in difficult clinical scenarios like diabetes or burn patients.

Methods Summary
All tissue samples were obtained from cadavers donated to the Department of Anatomy, Friedrich-Alexander-University Erlangen-Nürnberg, Germany. All animals used in this study were treated in accordance with the Association for Research in Vision and Ophthalmology, Resolution on Use of Animals in Ophthalmic and Vision Research and the recommendations of the National Institute of Health Guide for the Care and Use of Laboratory Animals. RNA preparation, cDNA synthesis as well as RT-PCR analysis were performed using the respective primers (Table 3S) as previously described by Schicht et al. 2014 (cf. also supplemental part) 48 . Quantitative real-time RT-PCR using cultivated as well as stimulated ocular surface cell lines, i.e. a human corneal epithelial cell line (HCE), a human conjunctival epithelial cell line (HCjE) and a human meibomian gland epithelial cell line (HMGEC), was performed to quantify the respective genes as previously described by Hampel et al. 2013 (cf. also supplemental part) 49 . Western blot, ELISA and immunohistochemical localization of the protein was performed as described previously (cf. also supplemental part) 48 . Immunofluorescence analysis and investigation of proliferation enhancing properties using fluorescence-activated cell sorting (FACS) with a bromodeoxyuridine (BrdU) assay were performed as described in Mauris et al. 2015 (cf. also in supplemental part) 50 . Electric Cell-Substrate Impedance Sensing (ECIS ® ) for cell proliferation analysis was performed as described by Hampel et al. 2015 (cf. also supplemental part) 51 . A migration assay (scratch assay) with HCE cells was performed as described in Dieckow et al. 2016 (cf. also in supplemental part) 52 . The corneal wound healing mouse model has been previously described (cf. also supplemental part) 49,53 . Measurement of smooth muscle actin expression depending on interaction between GSN and TGF-β was performed by gene knockdown. Extraction and cultivation of human primary corneal fibroblasts was performed as described by Hampel et al. 49,54 . Stimulation and gene knockdown with human primary corneal fibroblasts is described in the supplemental part.