GPER activation protects against epithelial barrier disruption by Staphylococcusaureus α-toxin

Sex bias in innate defense against Staphylococcus aureus skin and soft tissue infection (SSTI) is dependent on both estrogen production by the host and S. aureus secretion of the virulence factor, α-hemolysin (Hla). The impact of estrogen signaling on the immune system is most often studied in terms of the nuclear estrogen receptors ERα and ERβ. However, the potential contribution of the G protein-coupled estrogen receptor (GPER) to innate defense against infectious disease, particularly with respect to skin infection, has not been addressed. Using a murine model of SSTI, we found that GPER activation with the highly selective agonist G-1 limits S. aureus SSTI and Hla-mediated pathogenesis, effects that were absent in GPER knockout mice. Specifically, G-1 reduced Hla-mediated skin lesion formation and pro-inflammatory cytokine production, while increasing bacterial clearance. In vitro, G-1 reduced surface expression of the Hla receptor, ADAM10, in a human keratinocyte cell line and increased resistance to Hla-mediated permeability barrier disruption. This novel role for GPER activation in skin innate defense against infectious disease suggests that G-1 may have clinical utility in patients with epithelial permeability barrier dysfunction or who are otherwise at increased risk of S. aureus infection, including those with atopic dermatitis or cancer.

Scientific RepoRts | (2019) 9:1343 | https://doi.org/10.1038/s41598-018-37951-3 activation to skin immunity, particularly with respect to innate defense against bacterial infection, has not been addressed. Therefore, given the role of S. aureus Hla in SSTI and disruption of epithelial cell junctions, we hypothesized that G-1-mediated activation of GPER would limit Hla-mediated epithelial permeability barrier disruption and reduce S. aureus pathogenesis.
To test this hypothesis, we used a murine model of SSTI 9 to test whether G-1 limits S. aureus SSTI and Hla-mediated pathogenesis in a GPER-dependent manner. Specifically, G-1 treatment reduces Hla-mediated skin lesion formation and production of pro-inflammatory cytokines in vivo. Consistent with its ability to support BBB integrity following GCI 35 , G-1 treatment of a human keratinocyte cell line increased intercellular junction integrity in the face of Hla-mediated permeability barrier disruption. Furthermore, G-1 reduced keratinocyte surface expression of the Hla receptor, ADAM10, as well as E-cadherin cleavage with Hla-challenge, a mechanism that may contribute to the overall increase in permeability barrier integrity. Together, these studies clearly demonstrate a novel role for GPER activation in skin innate defense against S. aureus infection and the important virulence factor, Hla, as well as the potential of G-1 as an HDT to limit infectious disease.

Results
GpeR activation reduces pathogenesis in a mouse model of S. aureus sstI. GPER activation has a variety of effects on innate immune function, including modulation of macrophage cytokine production and neutrophil function [30][31][32] , as well as reversing stroke-induced immunosuppression 33 . To determine whether GPER activation would support innate immune defense against infectious disease, we evaluated the effects of GPER activation on the outcomes of S. aureus infection using a well characterized murine model of SSTI 9 . Male mice were treated with the GPER-selective agonist G-1 34,40 or vehicle control prior to subcutaneous (SQ) infection with the community-acquired MRSA isolate LAC 41 (Fig. 1a). Over the course of a three-day infection, G-1-treated mice showed significantly reduced lesion area (neutrophil-filled abscesses with subsequent dermonecrosis) (p < 0.001) and weight loss (p < 0.05) (a general measure of morbidity) compared to vehicle-treated controls (Fig. 1b,c). On day 3 post-infection (typically the peak of lesion formation 42 ), G-1-treated males also had reduced bacterial burden compared to control-treated mice (Fig. 1d). Consistent with reduced lesion area, bacterial burden, and the demonstrated anti-inflammatory effects of G-1 30 , G-1-treated mice also had lower local levels of the inflammatory cytokines IL-1β, TNFα, IL-6 and CXCL1 (Fig. 1e). As expected, given lower levels of the neutrophil-recruiting chemokine CXCL1, local levels of myeloperoxidase (MPO), often used as a surrogate marker for neutrophil presence 43 , were reduced in G-1-treated mice (Fig. 1f) suggesting a potential association between reduced lesion size with G-1-treatment and reduced neutrophil accumulation. In contrast to reduced levels of pro-inflammatory cytokines, levels of the anti-inflammatory cytokine IL-10 did not significantly differ between groups (p = 0.0884). This indicates that while G-1 reduces inflammation, it is not a general suppressor of cytokine production (Fig. 1e).
Female mice are innately better protected than males against S. aureus SSTI 26 , so we asked whether G-1 would further limit pathogenesis in female mice. At the male infectious dose of 2 × 10 7 LAC CFU, females develop much smaller lesions than males. Therefore, to assess additional G-1 protection in females, the infectious dose was increased to ~3 × 10 7 CFU. Compared to vehicle-treated controls, G-1-treated female mice showed significant reductions in lesion area (p < 0.01), whereas differences in weight loss and bacterial burden did not reach statistical significance ( Supplementary Fig. S1). G-1-treated female mice also had reduced levels of the inflammatory cytokine TNFα at the site of infection, with no significant reduction in levels of the other cytokines tested or MPO compared to vehicle-treated mice ( Supplementary Fig. S1). Given that G-1 does not directly inhibit bacterial growth ( Supplementary Fig. S2), these results suggest that GPER activation limits the severity of S. aureus SSTI by a host-dependent mechanism. G-1 limits the severity of S. aureus sstI in a GpeR-dependent manner. G-1 is a highly selective GPER ligand with respect to both classical estrogen receptors and other GPCRs 30,34 , suggesting that the benefits of G-1-treatment against S. aureus SSTI should depend on host expression of GPER. To test this, we compared outcomes between male GPER knockout (GPER KO) mice and corresponding wild-type (WT) C57BL/6 controls treated with G-1 or vehicle and infected with LAC. Compared to vehicle-treated controls, G-1-treatment significantly reduced lesion size (p < 0.001) and increased bacterial clearance (p < 0.01) without affecting weight loss in infected WT mice, but had no effect on lesion size or bacterial clearance in GPER KO mice (Fig. 2a-c). Notably, infection outcomes in the absence of G-1 did not significantly differ between WT and GPER KO male (Fig. 2a-c) or female (Fig. 2d-f) mice. This indicates that innate defense in this model is normally GPER-independent and suggests that increased estrogen-dependent, innate protection in female mice 26 is mediated through classical estrogen receptors. However, our findings clearly show that G-1 limits pathogenesis in both males and females and that GPER could be a promising target for HDT. Most importantly, these results demonstrate that the efficacy of G-1 in limiting S. aureus SSTI is host-and GPER-dependent. G-1 limits Hla-mediated pathogenesis in a murine dermonecrosis model. In animal models of S. aureus SSTI, the secreted virulence factor alpha-hemolysin (Hla) drives lesion formation at the site of infection 8,10,11 . Given the host-dependent protective effects of G-1 against S. aureus SSTI, we postulated that G-1-treatment would limit pathogenesis in male mice directly challenged with Hla. Consistent with results in mice infected with S. aureus, G-1 treatment significantly reduced lesion formation, inflammatory cytokine production (IL-1β, TNFα, IL-6 and CXCL1) and MPO levels at the site of subcutaneous Hla-injection compared to vehicle-treated controls (Fig. 3). To corroborate the role of Hla in G-1-mediated protection in the skin, we infected male mice with an isogenic Hla deletion mutant of S. aureus (LACΔhla). In the absence of Hla expression, skin lesion formation is minimal or absent, so outcomes are based largely on bacterial burden and weight loss. Importantly, G-1-treatment did not significantly alter LACΔhla infection outcomes based on day three post-infection bacterial clearance and overall weight loss ( Supplementary Fig. S3). Given that G-1 does not  Fig. S3), these results provide support for a mechanism in which G-1 protects against Hla-mediated pathogenesis by altering the host response.

G-1 limits keratinocyte permeability barrier disruption by Hla.
Hla is a major contributor to the pathogenesis of S. aureus infections involving epithelial cells, including SSTI and pneumonia (PNA) 8,[10][11][12][13][44][45][46] . During these infections, Hla disrupts host permeability barriers to facilitate invasive infection [8][9][10][11][12][13][14] . Recently, GPER activation was reported to limit disruption of BBB integrity in a rodent model of GCI 35 . Furthermore, GPER is expressed in numerous types of skin cells including melanocytes, dermal fibroblasts and the most prominent skin cell type, keratinocytes [36][37][38][39] . Given our in vivo data showing that G-1 limits skin damage caused by Hla, and the role of S. aureus Hla in disrupting epithelial cell junctions, we hypothesized that activation of GPER with G-1 would limit Hla-mediated epithelial permeability barrier disruption. One measure of permeability barrier integrity is resistance of cell monolayers to passage of an electric current (electrical cell-substrate impedance sensing (ECIS)) 47 . As keratinocytes are the major skin cell type, we used the HaCaT human keratinocyte cell line 48 to determine whether G-1 limits Hla-mediated disruption of epithelial barrier integrity. GPER expression in HaCaT cells was first verified by immunofluorescent staining with and without siRNA knockdown of GPER, after which we demonstrated that G-1 did not affect cell growth or viability during culture ( Supplementary Fig. S4). Next, HaCaT cells were grown to confluence in the presence of vehicle or G-1 and changes in transepithelial electrical resistance (TER) were measured by ECIS. After reaching stable resistance, monolayers were exposed to Hla or the inactivate Hla mutant, Hla H35A 49,50 . As previously reported, there was no significant reduction in TER between control cells and those challenged with Hla H35A 10 regardless of G-1-treatment ( Fig. 4a and data not shown). In contrast, whereas Hla rapidly reduced keratinocyte barrier integrity (decreased TER) in vehicle-treated cells, cells grown in the presence of G-1 were significantly more resistant (p < 0.01) to permeability barrier disruption, with an average 33% increase (p = 0.0079) in TER compared to vehicle control (Fig. 4a,b). The G-1 mediated reduction in permeability barrier disruption was reversed in the presence of the GPER-antagonist, G15 40 (Fig. 4c), consistent with the requirement for GPER for G-1 efficacy in vivo (Fig. 2a-c). Notably, G-1 did not alter HaCaT cell numbers ( Supplementary Fig. S4), suggesting that increased barrier integrity is independent of potential G-1-mediated effects on cell growth or viability.
Hla disrupts epithelial barriers by binding its cell surface receptor, ADAM10 10,14,51 , which in turn cleaves E-cadherin. Given that E-cadherin is a component of the adherens junctions responsible for epithelial permeability barrier integrity 10,51 , and that G-1 limited keratinocyte permeability barrier disruption by Hla (Fig. 4a,b) and reduced skin damage (lesion area) in Hla-injected mice (Fig. 3a), we predicted that G-1 would reduce E-cadherin cleavage following Hla challenge. To test this, we used immunoblotting to measure full-length E-cadherin (FL) and the cleaved C-terminal fragment (CTF) from HaCaT cell monolayers grown in the presence of vehicle or G-1 and exposed to Hla (Fig. 4d,e). In the absence of Hla, G-1 significantly increased expression of FL E-cadherin (Fig. 4e, left) while decreasing baseline cleavage (CTF) (Fig. 4e, right). As expected, following an eight-hour culture with Hla, FL E-cadherin was significantly decreased (Fig. 4e, left) and cleavage (CTF) was increased independent of treatment ( Fig. 4e, right). However, whereas FL E-cadherin levels were equivalent between vehicle-and G-1-treated cells in the presence of Hla (Fig. 4e, left), G-1-treatment reduced E-cadherin cleavage (CTF) (Fig. 4e, right). Given reduced Hla-mediated E-cadherin cleavage with G-1, we asked whether G-1 altered expression of E-cadherin or the Hla receptor ADAM10. Although transcription of CDH1, which encodes E-cadherin, was not altered by G-1 alone, CDH1 transcription was increased in G-1-treated keratinocytes exposed to Hla (Fig. 4f). Also, whereas G-1 alone increased ADAM10 transcription approximately 15% (p < 0.05) (Fig. 4g), HaCaT cells treated with G-1 displayed 23% less ADAM10 (p < 0.0001) on the cell surface compared to control-treated cells (Fig. 4h). Together, these findings suggest that G-1 may contribute to transcriptional regulation of CDH1 when Hla is present, as well as post-transcriptional regulation of ADAM10 expression or trafficking to the keratinocyte cell surface. This in turn may contribute to maintenance of epithelial permeability integrity in the face of Hla-challenge (Fig. 4a,b).

Discussion
The skin permeability barrier provides protection against transcutaneous water loss, invasion by microbial pathogens and access of environmental toxins to underlying sensitive tissues 4,5 . However, bacteria have had countless generations to evolve powerful tools to disrupt this barrier and cause SSTIs resulting in annual treatment costs of billions of dollars 6 . As the most common cause of SSTI, S. aureus secretes Hla to disrupt epithelial barriers and facilitate invasive infection. Specifically, Hla binds ADAM10 on host cells, resulting in cleavage of the cell junction protein E-cadherin and loss of permeability barrier integrity [8][9][10][11][12][13][14][15][16] . Here we show that G-1, the highly selective ligand of the non-classical estrogen receptor GPER, limits the severity of S. aureus SSTI and production of pro-inflammatory cytokines in a murine challenge model. The effects of G-1 are dependent on S. aureus expression of Hla, a finding supported by reduced skin pathogenesis (lesion formation) in G-1-treated mice compared to controls following direct Hla challenge. Not surprisingly, G-1 efficacy is dependent upon host expression of GPER, as protection against S. aureus SSTI is lost in GPER KO mice. Furthermore, G-1 reduces keratinocyte surface expression of the Hla receptor ADAM10 and limits Hla-mediated disruption of epithelial barrier integrity in vitro. Therefore, along with supporting endothelial permeability barrier integrity following ischemic injury 35 , our findings show that G-1 promotes epithelial barrier integrity and host innate defense against a major bacterial toxin. Given that Hla-mediated epithelial injury controls infection outcome 17 , and that G-1 lacks the feminizing effects seen with estrogen treatment 52 , this work demonstrates the potential efficacy of G-1 as an HDT to promote skin innate defense and reduce the burden of S. aureus SSTI. The role of estrogen signaling in immune regulation has historically been studied in terms of the classical estrogen receptors ERα and ERβ 27 . In contrast, although GPER is expressed by a variety of immune cells, including monocytes, macrophages, neutrophils and lymphocytes 31,32,53-60 , the impact of GPER signaling on immune function is in the early stages of investigation. Interestingly, GPER activation can result in both pro-and anti-inflammatory responses. For example, GPER signaling can modulate neutrophil function 31,32 , regulate both pro-and anti-inflammatory cytokine production 30,56 and promote regulatory T-cell responses 55 . GPER has also been shown to provide protection in a mouse model of multiple sclerosis 30 , to contribute to monocyte-dependent skin inflammation in response to serum from lupus patients, and to reverse peripheral immunosuppression in an ovariectomized mouse model of stroke 33 . GPER is also expressed in skin cells [36][37][38][39] , where it contributes to cytoskeletal organization 37,61 and melanin synthesis 36 . Here we demonstrate the contribution of GPER activation to host innate defense against S. aureus skin infection and to increasing epithelial barrier integrity in the face of Hla challenge. These findings not only significantly expand our understanding of GPER signaling in innate immune defense, but also demonstrate a novel role for G-1 in the maintenance of barrier integrity in human keratinocytes.
Hla plays a major role in the pathogenesis of S. aureus infection in numerous animal models, particularly models of SSTI and PNA 8,10,11,14-16 . Its importance is further evidenced by ongoing efforts to develop prophylactic and therapeutic strategies targeting Hla expression or function 8,[18][19][20][21][22][23][24][25] . For example, recombinant Hla toxoid was recently shown to be a safe and immunogenic candidate vaccine antigen in healthy adults in Phase 1-2 clinical trials (NCT01011335) 25 . In addition, Suvratoxumab (formerly known as MEDI4893), an anti-Hla monoclonal antibody, has successfully completed Phase 1 safety trials (NCT01769417) 62 and is currently in a Phase 2 safety and efficacy trial (NCT02296320) 63 to prevent or limit S. aureus pneumonia in mechanically ventilated adults. Along with these approaches, the use of G-1, which also protects against Hla-mediated pathogenesis, could provide an additional therapeutic benefit to patients. Interestingly, combination treatment with the Hla neutralizing antibody MEDI4893, and either vancomycin or linezolid, two clinically important antibiotics, improved outcomes in mouse models of S. aureus SSTI 18 and PNA 19,64 . Whether G-1 will have similar adjunctive efficacy with antibiotic therapy against S. aureus infection is an important point for future investigation. Further studies are also required to determine the mechanism by which G-1 increases keratinocyte transcription of the gene encoding the Hla receptor ADAM10 while also reducing its surface expression (Fig. 4f,g). Although speculative, this mechanism may involve G-1-mediated effects on post-translational regulation of ADAM10, including suppression of ADAM10 trafficking to the cell surface. In any case, the use of G-1 as an HDT to promote skin innate defense may prove a valuable component of a multifaceted approach to reduce the burden of S. aureus infection.
The ability of G-1 to promote epithelial barrier integrity and to limit S. aureus infection supports its potential clinical utility in patient populations at increased risk for infection. For example, S. aureus colonization is frequently associated with atopic dermatitis and psoriasis, skin diseases that feature epidermal barrier dysfunction [65][66][67][68] , and S. aureus may actually contribute to skin inflammation in these patients 69 . Recurrent S. aureus SSTI is also a hallmark of autosomal dominant hyper IgE syndrome (AD-HIES) 70 and it has recently been shown that the impaired epithelial response to infection in these patients results from overproduction of the pro-inflammatory cytokine TNFα 71 . TNFα also contributes to the severity of atopic dermatitis, though TNF blockade by anti-TNF biologicals has thus far failed to improve outcomes in these patients 72 . Here we show that whereas G-1 provides greater overall improvement in S. aureus skin infection outcomes in the more susceptible male 26 versus relatively resistant female mouse population, G-1-treatment significantly reduces lesion size as well as TNFα production in S. aureus infected mice of both sexes. This suggests that G-1, which reduces but does not completely prevent TNFα production and signaling, may be efficacious in limiting the severity of S. aureus skin infection in highly susceptible AD-HIES and atopic dermatitis patients.
Here we used a short-term skin infection model to capture the innate immune response to S. aureus infection in advance of the development of any adaptive immunity. Therefore, our findings indicate that G-1 enhances innate immune defense against infection. This suggests the potential of G-1-treatment for limiting infection in groups with increased risk of S. aureus infection due to impaired innate immunity such as cancer patients undergoing chemotherapy or surgery [73][74][75][76] . Given that what begins as an SSTI can lead to S. aureus pneumonia, sepsis or other life threatening infection 77 , it will be essential to experimentally determine the efficacy of G-1 in limiting S. aureus skin infection in disease models of patients with increased susceptibility to S. aureus. Furthermore, since a single dosing regimen was utilized for the current studies, additional investigations will be needed to determine the optimal dosing strategy for efficacy against S. aureus SSTI in immunocompetent mice as well as disease models of susceptible patient populations. Overall, developing an HDT that limits infection in diverse patient populations could positively influence clinical practice and improve human health.
Although our findings suggest the potential utility of G-1 to limit pathogenesis during S. aureus skin infection, it could have much broader clinical utility. For example, since many innate defense mechanisms are effective against a variety of bacterial pathogens, therapies aimed at GPER activation may also prove efficacious for treating a variety of infections. In addition, given that a monoclonal antibody targeting Hla has shown synergy with antibiotic therapy in animal models of S. aureus infection 18,19,64 , G-1 may likewise have adjunctive efficacy. Furthermore, the ability of G-1 to promote endothelial 35 and epithelial barrier integrity may prove useful in treating AD, psoriasis, or other diseases involving dysfunctional permeability barriers. Finally, regardless of its therapeutic potential, G-1 may provide a powerful tool for identifying other GPER-dependent host targets to improve epithelial barrier integrity and promote host innate immune defense.

Methods
Reagents and cell culture. G-1 was synthesized as previously described 34 . G-1 was dissolved in absolute ethanol to make a 1 mg ml −1 stock and stored at −20 °C until use. HaCaT cells were generously provided by Dr. Laurie Hudson (University of New Mexico Health Sciences Center, Albuquerque, NM, USA). Prior to use as described below, HaCaT cells were cultured at 37 °C, 5% CO 2 in Hyclone TM Dulbecco's Modified Eagle Medium with low glucose (DMEM low), sodium pyruvate and without phenol red (GE Healthcare, Pittsburgh, PA, USA), plus L-glutamine, 1% Hyclone TM Minimal Essential Media with Non-Essential Amino Acids (MEM NEAA) and 10% FBS (Gibco ® , ThermoFisher Scientific, Grand Island, NY, USA).
Bacterial strains and growth conditions. The MRSA USA300 isolate LAC 41  For infection, bacteria were cultured in trypticase soy broth (TSB) at 37 °C to early exponential phase as described previously 78 . Stocks were prepared in TSB with 10% glycerol and maintained at −80 °C for no more than two weeks prior to use. The number of CFU per ml of frozen stock was determined by plating ten-fold serial dilutions onto trypticase soy agar containing 5% sheep blood  Immunofluorescent staining and flow cytometry. HaCaTs were seeded in 24-well plates (4 × 10 5 cells in one ml in media A with G-1/Veh) and replaced as needed until confluent. Media was changed to media B + G-1/Veh 24 hours prior to flow analysis. HaCaTs were trypsinized (0.25% Trypsin-EDTA, Gibco) and cells pooled from either G-1-or Veh-treated wells. Pooled cells were subsequently washed and exchanged into flow buffer (PBS, 2% FBS, and 0.1% NaN 3 ) then diluted to equal cell concentrations (~3 × 10 6 cells ml −1 ). Fc inhibitor antibody (14-9161-71, ThermoFisher) was added at 20 µl ml −1 and cells incubated at room temperature (RT) for 20 min. Cells were then stained on ice in the dark for 30 min with either isotype control (12-4714-81, ThermoFisher) or anti-ADAM-10 (352703, BioLegend, San Diego, CA) antibody at 10 µg ml −1 . Cells were washed three times with flow buffer and 20,000 events were recorded on an Accuri C6 flow cytometer (BD Biosciences, San Diego, CA) with isotype corrected mean fluorescence normalized to Veh-treated controls.
Western blot analysis of e-cadherin cleavage. Throughout this assay, all media included either 100 nM G-1 or vehicle control (G-1/Veh) as described above and cells were maintained at 37 °C, 5% CO 2 . HaCaTs were seeded in 24-well plates (4 × 10 5 cells in one ml of media A + G-1/Veh) and media was replaced every 48 hours until cells were confluent. Cells were then grown for 24 hours in media B + G-1/Veh. After 24 hours, cells were treated for eight hours with 0.5 µg ml −1 Hla (or PBS) in media B + G-1/Veh. To collect both cleaved E-cadherin (CTF), which is released intracellularly, and full-length E-cadherin (FL), which is cell-associated, an equal volume of 2X RIPA lysis buffer 81 (Triton X-100 was substituted for Nonidet P-40) was added to each well and the plate incubated on ice for five min. Equal volumes of lysate were resolved on 4-12% Bis-Tris Plus gels in MES buffer (ThermoFisher) and transferred to 0.45 µm nitrocellulose membranes (Bio-Rad, Hercules, CA

Data Availability
The datasets generated during the current study are available from the corresponding author on reasonable request.