Phosphatidylinositol-specific phospholipase C enhances epidermal penetration by Staphylococcus aureus

Staphylococcus aureus (S. aureus) commonly colonizes the human skin and nostrils. However, it is also associated with a wide variety of diseases. S. aureus is frequently isolated from the skin of patients with atopic dermatitis (AD), and is linked to increased disease severity. S. aureus impairs the skin barrier and triggers inflammation through the secretion of various virulence factors. S. aureus secretes phosphatidylinositol-specific phospholipase C (PI-PLC), which hydrolyses phosphatidylinositol and cleaves glycosylphosphatidylinositol-anchored proteins. However, the role of S. aureus PI-PLC in the pathogenesis of skin diseases, including AD, remains unclear. In this study, we sought to determine the role of S. aureus PI-PLC in the pathogenesis of skin diseases. PI-PLC was observed to enhance the invasion and persistence of S. aureus in keratinocytes. Besides, PI-PLC promoted the penetration of S. aureus through the epidermal barrier in a mouse model of AD and the human organotypic epidermal equivalent. Furthermore, the loss of PI-PLC attenuated epidermal hyperplasia and the infiltration of Gr-1+ cells and CD4+ cells induced by S. aureus infection in the mouse model of AD. Collectively, these results indicate that PI-PLC eases the entry of S. aureus into the dermis and aggravates acanthosis and immune cell infiltration in infected skin.


Results
Generation of PI-PLC knockout and complemented strains. We first examined whether PI-PLC is expressed and secreted by the S. aureus strains that colonize the skin in patients with AD. Since an antibody against PI-PLC is not commercially available, a monoclonal antibody against PI-PLC was generated by immunization with a recombinant PI-PLC protein. Western blot analysis with the generated antibody revealed that, even though the levels of PI-PLC secretion varied among different clinical isolates, PI-PLC was detected in the supernatants of all isolates cultured in the tryptic soy broth (TSB) culture medium ( Supplementary Fig. S1a). To evaluate the role of PI-PLC in the pathogenesis of AD, a PI-PLC knockout S. aureus (Δplc) and complemented strain were constructed. In the present study, the methicillin-susceptible S. aureus strain JCM 2874 was used as the parent strain. The Δplc strain was generated by inserting a 0.9-kb group II intron into the plc gene (Supplementary Fig. S1b). The insertion of introns and the disruption of the plc gene were verified using PCR (Supplementary Fig. S1c). Western blot analysis revealed that PI-PLC was detected in the culture supernatant of the wild-type strain, whereas it was not detected in that of the Δplc strain ( Supplementary Fig. S1d). In addition to the Δplc strain, the complemented strain was generated by reintroducing the plc gene into the Δplc strain (designated as Δplc :: plc). Since a previous study demonstrated the necessity of His-80 in the activity of S. aureus PI-PLC 25 , another strain was constructed for the complemented strain with mutated plc gene that encoded His-80-mutated catalytically inactive mutants of PI-PLC (designated as Δplc :: plc-MT). This would help determine the role of PI-PLC activity. Since the anti-PI-PLC monoclonal antibody is able to recognize the catalytically inactive mutants of PI-PLC, the levels of PI-PLC in the culture supernatant of Δplc :: plc and Δplc :: plc-MT strain were examined. Western blot analysis revealed that PI-PLC was present in the culture supernatant of both Δplc :: plc and Δplc :: plc-MT strains at levels similar to those in the culture supernatant of the wild-type strain ( Supplementary Fig. S1d). Next, PI-PLC activity in the culture supernatant of wild-type, Δplc, Δplc :: plc, and Δplc :: plc-MT strains was evaluated using a colorimetric substrate. PI-PLC activity was observed in the culture supernatant of the wild-type and Δplc :: plc strain, whereas the culture supernatant of Δplc and Δplc :: plc-MT strain did not exhibit PI-PLC activity ( Supplementary Fig. S1e). Knockout and complementation of the plc gene did not affect the proliferation of S. aureus in TSB ( Supplementary Fig. S1f). These results indicate that the PI-PLC knockout and complemented strains were successfully constructed, and that the loss of PI-PLC activity did not affect the proliferation of S. aureus under the culture conditions. PI-PLC enhances the persistence and proliferation of S. aureus within keratinocytes. While S. aureus was originally considered an extracellular pathogen 26 , it has been shown to be able to invade and proliferate in mammalian cells [27][28][29][30] . Although antimicrobial peptides (AMPs) secreted from keratinocytes target extracellular S. aureus 31 , intracellular S. aureus can evade the attack. It was reported that the secretory lipases of S. aureus enhanced keratinocyte invasion by S. aureus 18 , which suggests that lipid metabolism plays a role in host cell invasion by S. aureus. Therefore, the role of PI-PLC in the invasion and persistence of S. aureus in keratinocytes was evaluated. Monolayers of HaCaT cells or human epidermal keratinocytes were infected with S. aureus, followed by killing of extracellular S. aureus using antibiotics. The number of intracellular S. aureus was measured by the quantification of S. aureus DNA and calculation of the number of relative colony forming units (CFUs). The number of intracellular Δplc strains was observed to be lower than that of the wild-type strain immediately after (0 h) and 24 h after infection (Fig. 1a), which suggests that PI-PLC contributes to the invasion and persistence of S. aureus in HaCaT cells. The number of intracellular Δplc strains in human epidermal keratinocytes was also lower than that of the wild-type strain at 24 h after infection (Fig. 1b). Although the number of relative CFUs in S. aureus-infected HaCaT cell lysates increased regardless of the presence of PI-PLC, the www.nature.com/scientificreports/ growth ratio of the Δplc strain was lower than that of the wild-type strain (Fig. 1c), which suggests that PI-PLC also contributes to intracellular survival and proliferation of S. aureus in HaCaT cells. Next, the role of PI-PLC activity in the survival and proliferation of S. aureus in HaCaT cells and human epidermal keratinocytes was examined by challenging the cells with the Δplc :: plc and Δplc :: plc-MT strains. The number of relative CFUs at 24 h after infection with the Δplc :: plc-MT strain was lower than that after infection with the Δplc :: plc strain in HaCaT cells and human epidermal keratinocytes (Fig. 1d,e). The growth ratio of the Δplc :: plc-MT strain was also lower than that of the Δplc :: plc strain in HaCaT cells (Fig. 1f). These results strongly suggest that PI-PLC promotes the survival and proliferation of S. aureus in keratinocytes in an enzyme activity-dependent manner.
PI-PLC promotes the penetration of S. aureus through the epidermis. We next evaluated the role of PI-PLC in the entry of S. aureus into the dermis in normal mice, since the entry of S. aureus into the dermis and the subsequent aggravation of dermatitis is observed in the skin lesions in patients with AD 32 . The mice were administered an epicutaneous dose of wild-type and Δplc strains by attaching filter paper discs and Finn chambers covered with surgical tapes on the shaved flank skin. To exclude potential confounding variables, the wild-type and Δplc strains were administered to the right and left flanks of the same mice, respectively (Fig. 2a). The entry of S. aureus into the dermis decreased significantly in the skin of normal mice challenged with the Δplc strain compared to that in mice challenged with the wild-type strain (Fig. 2b). In addition, the  www.nature.com/scientificreports/ entry of S. aureus into the dermis decreased in mice infected with Δplc :: plc-MT strain compared to that in mice challenged with the Δplc :: plc strain, which indicates that PI-PLC activity contributes to the entry of S. aureus into the dermis (Fig. 2c). Since the loss of PI-PLC activity led to the inhibition of S. aureus entry into the dermis, we hypothesized that PI-PLC may contribute to epidermal penetration by S. aureus. To assess this, the invasion and penetration potential of the wild-type and Δplc strains in the human organotypic epidermal equivalent was evaluated (Fig. 2d). The findings of immunofluorescence experiments revealed that while the wild-type strain penetrated the epidermal equivalent, the Δplc strain did not enter the epidermal equivalent and was only detected on the surface (Fig. 2e). The number of S. aureus penetrating the human organotypic epidermal equivalent was estimated by spreading the culture medium below the epidermal equivalent onto the agar plates. While colonies were formed when the medium below the epidermal equivalent challenged with the wild-type strain was cultured, no colonies were observed in case of the Δplc strain-challenged epidermal equivalent (Fig. 2f). To evaluate the role of PI-PLC activity in epidermal invasion and penetration by S. aureus, the human organotypic epidermal equivalent was challenged with Δplc :: plc and Δplc :: plc-MT strains. While the Δplc :: plc strain invaded and penetrated the organotypic human epidermal equivalent, the Δplc :: plc-MT strain did not (Fig. 2g,h). These results strongly suggest that PI-PLC contributes to epidermal invasion and penetration by S. aureus in an enzyme activity-dependent manner.

PI-PLC enhances epidermal thickening and immune cell infiltration induced by S. aureus
infection in a mouse model of AD. Since PI-PLC plays a positive role in S. aureus invasion and persistence in keratinocytes and penetration through the epidermis and epidermal equivalent, the effects of PI-PLC on the phenotypes of S. aureus-infected skin were examined. Epicutaneous infection with the wild-type and Δplc strains induced mild thickening of the epidermis and infiltration of CD45 + leukocytes, Gr-1 + granulocytes, and CD4 + T cells. However, the loss of PI-PLC did not affect the severity of epidermal thickening, immune cell infiltration, and expression of pro-inflammatory cytokines ( Fig. 3a,b, and Supplementary Fig. S2a). These results suggest that PI-PLC does not play a significant role in the mild skin inflammation induced by S. aureus infection in normal mice. Given that normal mice had an intact epidermal barrier and a stable immune homeostasis, the number of S. aureus invading the dermis of normal mice may have been considerably low to induce clear inflammatory phenotypes. As AD mouse models have a defective physical and immune barrier, S. aureus was applied to the barrier-deficient skin in the AD mouse model that was generated by the topical application of a vitamin D3 analogue MC903 33 . The number of S. aureus that penetrated the epidermis in the AD mouse model infected with wild-type S. aureus was higher than that in normal mice. The entry of S. aureus into the dermis was inhibited significantly in skin infected with the Δplc strain than in that infected with the wild-type strain (Fig. 2b). Infection by wild-type S. aureus induced significant epidermal thickening and immune cell infiltration (Fig. 4a,b) in the skin of the mouse model of AD. However, the Δplc strain only induced mild thickening of the epidermis and low infiltration of CD45 + leukocytes (Fig. 4a,b). Among leukocytes, infiltration by the Gr-1 + granulocytes and CD4 + T cells was attenuated upon the loss of S. aureus PI-PLC (Fig. 4a,c). Regardless of the attenuated infiltration of immune cells in skin infected with the Δplc strain, the expression of pro-inflammatory cytokines remained uninhibited ( Supplementary Fig. S2b).
Tight junctions are one of the major physical barriers in the epidermis. Since the localization of tight junction proteins was disturbed in the lesioned skin of patients with AD 34,35 , the effect of S. aureus infection on the localization of ZO-1, a tight junction protein, was evaluated. Since the AD model mice only exhibited mild skin inflammation in the absence of S. aureus infection, ZO-1 staining was clearly observed in the upper epidermis in non-infected skin. Upon infection with the wild-type S. aureus, the ZO-1 staining intensity reduced and became diffused (Fig. 4a). In contrast to infection with the wild-type strain, that with the Δplc strain did not affect ZO-1 localization (Fig. 4a). These results suggest that S. aureus PI-PLC is involved in the mislocalization of tight junction proteins in AD skin. Therefore, PI-PLC plays a critical role in S. aureus-induced aggravation of epidermal hyperplasia and immune cell infiltration in a mouse model of AD.
Next, PI-PLC activity was evaluated to determine its role in S. aureus-induced aggravation of dermatitis in a mouse model of AD. The Δplc :: plc and Δplc :: plc-MT strains were applied on the dorsal skin of AD model mice. The number of dermal S. aureus decreased significantly in skin challenged with the Δplc :: plc-MT strains than in that challenged with the Δplc :: plc strain (Fig. 2c). Although both Δplc :: plc and Δplc :: plc-MT strains induced epidermal thickening and infiltration of CD45 + leukocytes, Gr-1 + granulocytes, and CD4 + T cells, immune cell infiltration and epidermal thickening induced by the Δplc :: plc strain was more prominent (Fig. 4d-f). ZO-1 localization was also disturbed in Δplc :: plc strain-infected skin (Fig. 4d). In contrast, the Δplc :: plc-MT strain did not induce the mislocalization of ZO-1 (Fig. 4d). These results strongly suggest that PI-PLC enhances epidermal thickening and immune cell infiltration induced by S. aureus infection in a mouse model of AD in an enzyme activity-dependent manner.

Discussion
PI-PLC plays a critical role in the invasion and penetration of mouse epidermis and human organotypic epidermal equivalent by S. aureus . PI-PLC enhanced epidermal penetration by S. aureus and aggravated epidermal hyperplasia and immune cell infiltration only in the mouse model of AD, and not in normal mice. Since AD lesions are characterized by defective antimicrobial or physical barriers 36,37 , PI-PLC may facilitate epidermal penetration and skin inflammation by S. aureus under such conditions. Epidermal penetration by S. aureus is enhanced in the skin lesions of patients with AD, leading to the exacerbation of AD 32 . Since PI-PLC supports the entry of S. aureus into the dermis, PI-PLC may act as a promising target in AD treatment.  18 , which suggests that lipid metabolism plays a role in host cell invasion. PI-PLC hydrolyses GPI and removes GPI-anchored proteins from the cell surface. Since the nanoclustering of GPI-anchored proteins was reported to regulate the functions of integrin 38 , the PI-PLC-mediated shedding of GPI-anchored proteins may affect the function and clustering of α5β1 integrin and FnBP-mediated invasion by S. aureus. The GPI-anchored proteins CD55 and CD59 present on keratinocytes are suggested to play a role in the alleviation of AD 39 . Since S. aureus PI-PLC was observed to cleave CD55 and CD59 from the surface of human umbilical vein endothelial cells and mouse pneumocytes 25 , the PI-PLC-mediated shedding of CD55 and CD59 from keratinocytes might enhance epidermal penetration by S. aureus, epidermal hyperplasia, and immune cell infiltration in S. aureus-infected skin in AD model mice.  www.nature.com/scientificreports/ Although keratinocyte-derived AMPs kill extracellular S. aureus in the epidermis and inhibit epidermal penetration by S. aureus, AMPs are less effective in combating intracellular S. aureus. Since PI-PLC contributes to the intracellular survival and proliferation of S. aureus in HaCaT cells, PI-PLC may protect S. aureus from AMPs and enhance epidermal penetration by supporting the intracellular persistence of S. aureus. Besides AMPs, S. aureus is able to evade antibiotic eradication by invasion and persistent colonization in cells such as keratinocytes 40 . Therefore, PI-PLC inhibition may enhance the efficiency of antibiotic treatment. Another study reported that the loss of PI-PLC reduced the viability of S. aureus in human blood and PMNs 24 . Given that PI-PLC supports the intracellular persistence of S. aureus in phagocytes and keratinocytes, PI-PLC may act as a therapeutic target for skin diseases such as AD as well as for other infectious diseases.
PI-PLC in Listeria monocytogenes is known to contribute to phagosomal escape, possibly by hydrolyzing PI in the phagosome membrane 41,42 ; therefore, S. aureus PI-PLC may enhance its intracellular persistence by supporting endosomal escape in keratinocytes via mechanisms similar to those followed by Listeria monocytogenes PI-PLC..
The Δplc :: plc strain exhibited lower relative CFU and growth ratio compared to the wild-type strain. Although the exact reason for the lower relative CFU and growth ratio of the Δplc :: plc strain was not clear, chloramphenicol selection for the generation of transformants harboring the plc gene may affect the character of the S. aureus strain. Nonetheless, the relative CFU and growth ratio of the Δplc :: plc strain were higher than those of the Δplc :: plc-MT strain, which indicates that the enzyme activity of PI-PLC plays a positive role in the persistence and proliferation of S. aureus in keratinocytes.
PI-PLC secreted by intracellular S. aureus may disturb PI metabolism in keratinocytes. Exogenous PI metabolism by S. aureus PI-PLC also affects the concentration of the phosphorylated forms of PI (PIPs). Since PIPs metabolism plays a crucial role in maintenance of skin barrier integrity 43 , PI-PLC may damage the epidermal barrier by affecting the concentration and metabolism of PIPs in keratinocytes.
Stratum corneum and tight junctions are physical barriers of the epidermis 44,45 that prevent the penetration of S. aureus through the epidermis. Since S. aureus infection was observed to disturb the localization of the tight junction protein ZO-1 in a PI-PLC activity-dependent manner, PI-PLC may enhance epidermal penetration by S. aureus by disturbing the tight junction barrier.
GPI-anchored proteins are located in cholesterol/sphingolipid-rich membrane domains, known as lipid rafts. Lipid rafts serve as platforms for signal transduction proteins, including GPI-anchored proteins. Since keratinocytes with damaged lipid rafts exhibit gene expression patterns similar to those observed in AD biopsies, lipid raft dysfunction is suggested to be related to AD 46 . Given that PI-PLC hydrolyses and removes GPI-anchored proteins from the cell surface, PI-PLC may aggravate AD by perturbing lipid raft-mediated signal transduction. Detailed analysis of the changes in composition and localization that occur in phospholipids and GPI-anchored proteins as a result of S. aureus infection will help clarify the mechanisms underlying PI-PLC-mediated epidermal penetration by S. aureus.
Collectively, the findings of this study indicate that PI-PLC promotes epidermal penetration by S. aureus and enhances S. aureus-induced epidermal hyperplasia and immune cell infiltration. Given that PI-PLC enhances epidermal penetration by S. aureus, it might induce systemic infection by inducing the entry of S. aureus through the skin surface. In this study, we used JCM 2874 as the parent strain. Future studies are necessary to clarify the role of PI-PLC in pathogenesis by a virulent strain such as MRSA USA300 and the exact mechanisms by which PI-PLC enhances epidermal penetration and intracellular persistence of S. aureus.

Methods
Bacterial strains, media, and growth conditions. S. aureus strains were cultured under aeration in TSB (BD, Franklin Lakes, NJ, USA) or tryptic soy agar (TSA). Clinical isolates were collected from patients with AD admitted in a hospital in Tokyo, Japan, from 2010 to 2011. All isolates were obtained from pus samples. All methods used in this study were performed in accordance with the relevant guidelines and regulations in this research field, and the study protocol was approved by the Tokyo University of Pharmacy and Life Sciences Ethics Committee (#12-08). Informed consent was not required from the patients because the study did not involve clinical interactions. To assess growth in TSB, overnight cultures of S. aureus JCM 2874 (ATCC 29213) were diluted in TSB (1:200). The strain was then cultured at 37 °C under shaking conditions, and absorbance was measured at 600 nm. The PI-PLC knockout Δplc strain was constructed from JCM 2874 by the insertion of a group II intron into plc using the primer design software and plasmid system provided with the TargeTron Gene Knockout System (Sigma-Aldrich, St. Louis, MO, USA) according to the manufacturer's instructions. Intron insertion was confirmed by performing PCR with two plc-specific primers: 5′-ATG AGT GGT TGG TAT CAT TC-3′ and 5′-CAC TTA CGA TAT CAT CAT ATCC-3′ ( Supplementary Fig. S1b, S1c, S3a).
Cloning of plc. plc was amplified by PCR and the products were ligated into the polylinker segment of pTZN10 and cloned into Escherichia coli DH5α 47 . The pTZN10 plasmids carrying plc were introduced into S. aureus RN4220 and then into JCM 2874Δplc by electroporation 48 . The transformants carrying the plc gene were directly selected using TSA supplemented with 10 µg/mL chloramphenicol. The inserts of the plasmid were confirmed by DNA sequencing 47 .
Western blot analysis. The culture supernatants of various S. aureus strains were subjected to SDS-PAGE and subsequently transferred onto PVDF membranes. The membranes were blocked with 10% skim milk and probed with an anti-PI-PLC antibody, followed by incubation with an HRP-conjugated secondary antibody (Dako, Glostrup, Denmark). Images were recorded using C-DiGit (LI-COR Biosciences, Lincoln, NE, USA) or  Fig. S3b-c). The culture supernatants were obtained by culturing various strains of S. aureus overnight in TSB at 37 °C.
PI-PLC activity assay. The activity of PI-PLC was determined using the artificial substrate 5-bromo- In vitro invasion and penetration assays in human organotypic epidermal culture. NHEK (KURABO, Osaka, Japan) were cultured in HuMedia-KG2 (KURABO) supplemented with insulin, bovine pituitary extract, epidermal growth factor, hydrocortisone, kanamycin, and amphotericin B. Cells from the second passage were used for the experiments. The cells were seeded onto cell culture inserts (Millipore, Billerica, MA, USA) and cultured overnight in an assay medium (Japan Tissue Engineering, Aichi, Japan). Next, the cultures were raised to the air-liquid interface and cultured in the assay medium for 6 days to form a multi-layered epidermis. were shaved at least 24 h before infection. JCM2874 wild-type, Δplc, Δplc :: plc, and Δplc :: plc-MT strains were cultured in TSB, washed with PBS, and resuspended in PBS. After careful disinfection of the skin surfaces, filter paper discs and Finn chambers (SmartPractice, Phoenix, AZ, USA) containing each S. aureus strain (1 × 10 8 CFU in 30 µL) were applied to the skin. Next, the filter paper discs and Finn chambers were covered with surgical tapes. After 96 h, the filter paper discs and Finn chambers were removed and the skin sample was harvested for the experiments. All animal studies were approved by the animal experiments review board of Tokyo University of Pharmacy and Life Sciences. All animal experiments were performed in accordance with relevant guidelines and regulations.
RNA extraction and real-time RT-PCR. Total  Statistical analysis. Results are expressed in terms of mean ± SEM. Statistical analyses were performed using a two-sided Welch's t-test. For data on S. aureus infection in mouse skin, statistical analyses were performed using a paired t-test. Tukey-Kramer method was used to adjust for multiple comparison. p-value < 0.05 was used to determine the statistical significance.