Lactobacillus rhamnosus GG Lysate Increases Re-Epithelialization of Keratinocyte Scratch Assays by Promoting Migration

A limited number of studies have investigated the potential of probiotics to promote wound healing in the digestive tract. The aim of the current investigation was to determine whether probiotic bacteria or their extracts could be beneficial in cutaneous wound healing. A keratinocyte monolayer scratch assay was used to assess re-epithelialization; which comprises keratinocyte proliferation and migration. Primary human keratinocyte monolayers were scratched then exposed to lysates of Lactobacillus (L) rhamnosus GG, L. reuteri, L. plantarum or L. fermentum. Re-epithelialization of treated monolayers was compared to that of untreated controls. Lysates of L. rhamnosus GG and L. reuteri significantly increased the rate of re-epithelialization, with L. rhamnosus GG being the most efficacious. L. reuteri increased keratinocyte proliferation while L. rhamnosus GG lysate significantly increased proliferation and migration. Microarray analysis of L. rhamnosus GG treated scratches showed increased expression of multiple genes including the chemokine CXCL2 and its receptor CXCR2. These are involved in normal wound healing where they stimulate keratinocyte proliferation and/or migration. Increased protein expression of both CXCL2 and CXCR2 were confirmed by ELISA and immunoblotting. These data demonstrate that L. rhamnosus GG lysate accelerates re-epithelialization of keratinocyte scratch assays, potentially via chemokine receptor pairs that induce keratinocyte migration.

where it is severely breached. For example, morbidity and mortality in victims of severe burns are associated with the dehydration and infection that occurs due to the poor skin barrier in these conditions 10 . However, as well as these extreme situations, breaches in the skin barrier can occur in daily life due to wounding and in surgery 11 . In healthy individuals, wounds normally heal in a timely fashion 12 . However, due to underlying conditions such as diabetes, some wounds heal slowly or not at all. In these cases, infection is a potential complication that can then further inhibit wound healing 13,14 . Chronic wounds are therefore a significant pathology. It follows then that treatments that can promote the wound healing process will be of considerable benefit to patients.
In the gut, certain species of probiotic bacteria have been demonstrated to increase barrier repair in in vitro and vivo models 15,16 . For example, healing of acid-induced gastric ulcers in the rat is reportedly accelerated in the presence of L. rhamnosus GG and L. gasseri 16 . Because of their purported effectiveness at promoting the healing of gastric wounds, attention has begun to focus on whether probiotic bacteria can impact upon the healing of cutaneous wounds. A recent study in rabbits showed that patches containing L. fermentum accelerated wound healing 17 . L. plantarum has been shown to improve healing of burns in mice and humans in a limited number of studies and also reportedly promotes healing of leg ulcers in humans [17][18][19][20] . However, in general, the mechanisms underlying these effects have not been fully explored.
We hypothesised that bacterial lysates would represent a safer alternative to the use of live bacteria in a wound situation because although probiotic lactobacilli have a GRAS (generally regarded as safe) status for food, the potential risks of live probiotic entering the bloodstream through breached skin has not been assessed. Furthermore, the use of lysates may be of more utility to potential wound care manufacturers than live bacteria because the logistical requirements of maintaining viability of bacteria within a formulation or wound dressing can be bypassed. In this study therefore, we have compared the ability of lysates made from four different species of lactobacilli to accelerate an important aspect of wound healing, re-epithelialization. To this end the scratch assay a well-established wound model 21 , was used to assess the effect of Lactobacillus lysates on re-epithelialization of human primary keratinocytes, the main cell type present within the epidermal layer of skin.

Results
Bacterial lysates accelerate scratch closure in a species dependent manner. Re-epithelialization in keratinocyte monolayers treated with L. rhamnosus GG lysate was significantly accelerated compared to that of untreated monolayers. At 18 h, 95% (P = 0.03, n = 3) of the scratch area was re-epithelialized compared with 75.2% in the control monolayer (Fig. 1a,b). Similar results were found when scratched cells were subjected to treatment with L. reuteri lysate, where 90.4% (P = 0.05, n = 3) of the scratch area was closed at 18 h (Fig. 1a,b). In contrast, L. plantarum lysate did not stimulate monolayer re-epithelialization (70.5%, n = 3). L. fermentum lysates caused a significant reduction in monolayer re-epithelialization rate compared with the control (40%, P = 0.01, n = 3). In keeping with the literature, the positive control, Keratinocyte growth factor (KGF) stimulated re-epithelialization of scratches although it was no more efficacious than the L. rhamnosus GG lysate (Fig. 1).
Bacterial lysates increase keratinocyte migration and proliferation. Since only L. rhamnosus GG and L. reuteri were efficacious in the scratch assay, these two lysates were analyzed further. Re-epithelialization could be due to keratinocyte migration and/or proliferation. Hence, the effect of L. rhamnosus GG and L. reuteri lysates on these keratinocyte activities was tested alongside that of the positive control KGF. In cultures treated with L. rhamnosus GG lysate and L. reuteri lysate there were about 48 × 10 5 and 24 × 10 5 migrated cells respectively after 4 h (P = 0.001, P = 0.004, respectively, n = 3), compared with 7 × 10 5 cells in control cultures (Fig. 2a). Importantly, there was a significant difference between two lysates (P = 0.002, n = 3), with L. rhamnosus GG lysate being the most efficacious (Fig. 2a) in stimulating cell migration. KGF stimulated 34 × 10 5 cells to migrate. However, there was no significant difference between the effects of KGF and L. rhamnosus GG lysate on keratinocyte migration. Incubation of keratinocytes with 100 μ l of L. rhamnosus GG lysate or L. reuteri lysate resulted in significant proliferation as double the number of cells was found in these cultures relative to control cultures (P = 0.02, P = 0.03, respectively, n = 3) at 12 h post-inoculation (Fig. 2b). KGF caused a two-fold increase in the number of cells compared with the untreated control (Fig. 2b). Interestingly, both lysates and KGF afforded equal stimulation of proliferation.
L. rhamnosus GG induces cell migration as the dominant mechanism. Since the L. reuteri and L. rhamnosus GG lysates both stimulated migration and proliferation of keratinocytes, we next asked which of these mechanisms was most dominant in the acceleration of re-epithelialization induced by the two lysates. The rate of re-epithelialization of scratched keratinocyte monolayers was measured in the presence of the well-characterised inhibitor of cell proliferation, Mitomycin C. The data in Fig. 3a demonstrate that in monolayers treated with L. rhamnosus GG lysate and Mitomycin C, 94.5% of the scratch area re-epithelialized after 18h incubation (P = 0.03, n = 3) which was not significantly different to the re-epithelialization of monolayers treated with L. rhamnosus GG alone. However, in the presence of Mitomycin C the L. reuteri lysate-treated monolayer was only 52% re-epithelialized compared to 90% in cells treated with L. reuteri lysate alone (P = 0.01, n = 3, Fig. 3b).
Genome-wide Affymetrix microarray study of the effects of the L. rhamnosus GG lysate on keratinocyte gene expression. Our data suggest that the L. rhamnosus GG lysate increases re-epithelialization by enhancing keratinocyte migration. Therefore, the pathways underlying this effect were explored using microarray analysis of scratched keratinocytes treated vs. untreated with L. rhamnosus GG lysate. The genes identified in the microarray were grouped using IPA according to cellular function. Table 1 considers functions known to be relevant to the wound-healing process 10,22,23 . The groups of genes involved in 'cell movement' , 'migration' and 'proliferation' had the highest P value suggesting modulation of these genes by the bacterial lysate. The group of genes that was most significantly altered in expression was 'cell movement' . However, all the genes contained within this group were also common to the group called 'cell migration' . More than 1760 genes involved in migration, were altered in response to the L. rhamnosus GG lysate. The top 10 genes from the common group between 'migration' and 'cell movement' are presented in Table 2. Table 2 shows that the expression of two genes, CXCL2 and its receptor CXCR2 were highly up-regulated by the L. rhamnosus GG lysate. The microarray was validated by performing quantitative PCR analysis of CXCL2 and CXCR2 mRNA levels in scratched keratinocytes treated with L. rhamnosus GG lysate vs untreated controls (data not shown). Since the L. rhamnosus GG lysate up-regulated the gene expression of CXCL2, and its corresponding receptor, we next investigated whether the proteins corresponding to these genes were also increased by the lysate. The secretion of CXCL2 by scratched keratinocytes in response to L. rhamnosus GG lysate after 12 h and 18 h incubation was measured using ELISA. These showed that increased CXCL2 secretion was stimulated by L. rhamnosus GG lysate-treated keratinocytes, with 6.30 ± 0.4 μ g/ml, compared with 3.10 ± 0.7 μ g/ml in untreated cells after 12h incubation (P = 0.002, n = 3, Fig. 4a). Immunoblot analysis was performed to quantify the changes in the levels

Discussion
Non-healing wounds such as diabetic foot ulcers are a significant cause of morbidity and can lead to mortality 24 . Over-use of antibiotics has given rise to many antibiotic-resistant infections that are frequently associated with poor outcomes in wound healing, especially following surgery 25 . Currently, there is an unmet, clinical need for new therapies in the treatment of wounds. Since probiotics have been demonstrated to promote wound healing in the gut 15 , the aim of this study was to examine the effects of enteric probiotic lysates on keratinocyte migration and proliferation that are important aspects of the wound healing response.
Lysates of L. rhamnous GG and L. reuteri promoted the re-epithelialization of keratinocyte monolayers. In contrast, lysates from the other lactobacilli tested either had no effect on re-epithelialization, (L. plantarum) or inhibited re-epithelialization (L. fermentum). This latter result is probably related to our previous observation that L. fermentum reduces keratinocyte viability 6,8 . A previous study examined the effects of cell-free supernatants of L. plantarum on the wound healing process especially, the proliferation and migration of human keratinocytes in vitro 26 . The study demonstrated that the L. plantarum cell free supernatant promoted re-epithelialization in keratinocyte cultures. It was hypothesized that L. plantarum produces a specific substance, Plantaricin A that stimulates keratinocyte migration and proliferation in a scratch assay 26 . The authors investigated the mechanism underlying improved migration and proliferation and showed increased expression of transforming growth factor (TGF-β 1), vascular endothelial growth factor (VEGF) and fibroblast growth factor (FGF-7). In the present study, the effect of spent culture fluid of L. plantarum on the stimulation of keratinocyte re-epithelialization was not measured. Therefore, it is not possible to exclude the likelihood that the spent culture fluid from L. plantarum may have been efficacious in our assay. What is clear is that lysates of both L. reuteri and L. rhamnosus GG are more efficacious than the L. plantarum lysate in promoting keratinocyte proliferation/migration. The enhanced re-epithelialization induced by L. rhamnosus GG and L. reuteri was probably due to increased keratinocyte migration and proliferation although the two lysate were not equally effective in promoting cell migration. L. rhamnosus GG was much more effective at stimulating migration than L. reuteri. Indeed, migration is probably the dominant mechanism promoted by L. rhamnosus GG because in the presence of Mitomycin C, scratches were still able to re-epithelialize at a significantly faster rate than untreated scratches. Conversely, the dominant mechanism for the L. reuteri lysate was probably stimulation of proliferation, because Mitomycin C completely negated the stimulatory effects of the L. reuteri lysate. This suggests that lysates from different lactobacilli have species-specific effects on keratinocyte functions and is in agreement with previous observations [6][7][8] .
The mechanism underlying the increase in re-epithelialization induced by L. rhamnosus GG potentially involves increased expression of the chemokine, CXCL2 and its receptor, CXCR2. The expression of these genes was greatly enhanced at both the mRNA and protein levels as demonstrated by microarray, q-PCR, ELISA and immunoblotting data. CXCL2 is known to have several roles in the wound-healing process such as the stimulation of keratinocyte migration, proliferation and adhesion 27,28 . Indeed, it has been suggested that CXCL2 may be a chemoattractant for human keratinocytes 27,28 . Moreover, a number of vitro studies have shown that CXCL2 stimulates the production of additional chemokines, such as CXCL12 and CXCL10 also act as chemoattractants for keratinocytes in the wound-healing process 26,29,30 . The CXCL2 receptor, CXCR2 is also known to be important to the wound healing process. Mice lacking CXCR2 have delayed wound healing and this was related to a delay in cell recruitment to the wound site and a delay in re-epithelialization 26,29,30 . Based on these findings and the present data, we hypothesize that L. rhamnosus GG lysate increases re-epithelialization potentially via the interaction of CXCR2 and CXCL2 that accelerates keratinocyte migration, although other factors could also be involved.
A study conducted in 2011 by Polk 31 reported that material derived from L. rhamnosus GG regulated intestinal epithelial cell survival and growth in injury via stimulating epidermal growth factor (EGF) and its receptor 31,32 . The microarray data in the present study did not highlight a role for EGF nor its receptor alpha chain EGFR-α since the gene levels of neither of these were significantly altered in cells treated with L. rhamnosus GG lysate. However, some of the main genes in the EGF signaling cascade, such as EGFR-β , MEK1, STAT3 and ADAM17, were significantly up regulated in response to L. rhamnosus GG lysate. CXCL2 and CXCR2 were both significantly upregulated and higher than the expression of genes associated with the EGF pathway. Polk's study 32 also identified specific proteins from L. rhamnosus GG spent medium fluid that promoted cell growth in cultured human and mouse colon epithelial cells. In the current study, the spent culture fluid was tested but found not to be as efficacious as the lysate ( Figure S1). It is possible that this is due to a concentration effect i.e. the effective concentration is not sufficiently high in the spent culture fluid for an effect on re-epithelialization to be observed. Conversely,  Table 2. L. rhamnosus GG modulates the expression of genes involved in cell migration. The top 10 genes modulated by the lysate that are involved specifically with cell migration, their corresponding P values and fold changes. The change in gene expression in cells treated with lysate were considered significant when P-value was < 0.05 and the fold change was greater than two compared with control (n = 3). the L. rhamnosus GG lysate may be working via different receptors and ligands in keratinocytes than has been reported for enterocytes [29][30][31] .
In conclusion, L. rhamnosus GG lysate is a potential new wound healing treatment. We have demonstrated previously 6 that this lysate also has protective effects against S. aureus toxicity towards keratinocytes. Thus, it is possible that L. rhamnosus GG lysate may have utility in the treatment of infected, non-healing wounds.

Materials and Methods
Bacterial cell culture. Lactobacillus rhamnosus GG (ATCC 53103) Lactobacillus reuteri (ATCC55730), Lactobacillus plantarum (UCC118) (ATCC10241), and Lactobacillus fermentum (ATCC14932) were purchased LGC Ltd, Ireland, UK. These were grown routinely in Wilkins-Chalgren Broth or Agar (Oxoid, Basingstoke, UK) at 37 °C incubated in an anaerobic cabinet (atmosphere, 10:10:80, H 2 -CO 2 -N 2 ). Lysates of all these species were prepared as previously described 6 . Primary keratinocyte culture and scratch assay. Primary human epidermal keratinocytes were cultured as described in 7 . For scratch assays, keratinocytes were plated on 24 well tissue culture plates and allowed to reach confluence. At this point a scratch was introduced into the monolayer using a sterile p200 pipette tip. The cells were then washed in phosphate buffered saline (0.01M PBS, pH = 7.4) to remove the debris and the medium was replaced with either fresh medium (control samples) or medium containing 100 μ l bacterial lysate (0.93 mg/μ l). In some experiments, keratinocyte growth factor (KGF, 0.5 mg/ml) was included as a positive control.
In other experiments, 0.5 mg/ml Mitomycin C from Streptomyces caespitosus was added before the addition of the lysate to inhibit cell proliferation. Scratches were then monitored and documented at 0,6,12,18 and 24 h post scratch by staining cells with crystal violet solution. Images were captured under the Keyence microscope (x5 magnification, Keyence, Osaka, Japan) and analysed using IMAGE J-64 software (http://imagej.nih.gov). This enabled determination of the percentage of scratch area (gap) with respect to the percentage of starting scratch area at time zero. For each image, the area between one side of the scratch and the other was measured. Quantification of scratch closure was performed by comparing the area of the scratch at set time points according to the equation: Keratinocyte proliferation assay. The proliferation assay (CellTiter 96 ® AQueous) was performed according to the manufacturer's instructions (Promega, Madison, USA). Cells were seeded at an initial density of 2.5 × 10 5 cells per well in 96-well plates. After 24 h incubation, cells were exposed to 100 μ l of probiotic lysates or to the positive control, KGF at 0. Bioinformatics analysis was performed with dChip (V2005) (www.dchip.org). The model-based analysis of oligonucleotide arrays and gene expression analysis were carried out using Bioconductor as described in 33 . Filtering for probe sets with a P ≤ 0.05 created a list of differentially expressed genes. P value refers to the alteration in gene expression in treated vs untreated cells. Data was analysed using the Ingenuity Knowledge Base program (IPA, http://www.ingenuity.com/products/ipa) with the aim of identifying the genes related to the wound-healing process. The microarray data was submitted to the array express repository, ID= E-MTAB-3485.
Analysis of chemokine production. A scratch assay was performed and the cells treated with or without lysate as described above. At 12 h or 18 h after scratching, the spent cell culture fluid yielded from the plates was removed. The CXCL2 level was determined by ELISA (Invitrogen, Life Technologies Ltd, Paisley, UK), in adherence to the manufacturer's instructions. The optical density of wells was determined using a Labtek LT400 microplate reader (Labtech International Ltd, UK) set to read absorbance at 450 nm. The experiment was carried out on three separate occasions with triplicate samples within each individual experiment.

SDS-PAGE and
Immunoblotting. This was performed as described previously 8 using primary antibodies CXCR2/IL-8 RB (2.5 μ g in1 ml of PBS) overnight at 4 °C. Membranes were washed for 3 × 5 minutes with TBS/Tween containing 5% skimmed milk solution and then incubated with secondary antibody (HRP-conjugated goat anti-mouse IgG) at 1:5000 dilutions for 1h at room temperature. Then the immunoblotting assay was performed as described in 8 .

Statistical analyses.
All experiments were performed three times with triplicate samples within each individual experiment. For experiments comparing two or more treatments, a two-way ANOVA with post hoc Tukey test were utilised to analyse the main effects of, and interactions between, multiple factors. Results were considered significant if P ≤ 0.05, and all analysis were performed using the SPSS (IBM SPSS Statistics version 16.0) program.