Sphingosine 1-phosphate receptor 2 is central to maintaining epidermal barrier homeostasis

Abstract

The outer layer of the epidermis composes the skin barrier, a sophisticated filter constituted by layers of corneocytes in a lipid matrix. The matrix lipids, especially the ceramide-generated sphingosine 1-phosphate (S1P), are the messengers that the skin barrier uses to communicate with the basal layer of epidermis where replicating keratinocytes are located. S1P is a bioactive sphingolipid mediator involved in various cellular functions via S1P receptor (S1PR) 1–5, expressed by keratinocytes. We discovered that the S1pr2 absence is linked to an impairment in skin barrier function. Although S1pr2^−/− mouse skin have no difference in its phenotype and barrier function compared to wildtype (wt), after tape stripping, S1pr2^−/− mice showed significantly higher transepidermal water loss (TEWL) and required another 24 more hours to normalize their TEWL levels. Moreover, after epicutaneous S. aureus application, impaired S1pr2^−/− mouse epidermal barrier function allowed deeper bacterial penetration and denser neutrophil infiltration in the dermis. Microarray and RNA sequence of S1pr2^−/− mouse epidermis linked the barrier dysfunction with a decrease in filaggrin 2 and tight junction components. In conclusion, S1pr2^−/− mice have compromised skin barrier function and increased bacteria permeability, making them a suitable model for diseases that present similar characteristics, such as atopic dermatitis.

INTRODUCTION

The outer layer of human epidermis gives rise to a very specialized physical barrier constituted of corneocytes (the outermost layer of skin cells) in a lipid matrix. The lipid matrix is composed of ceramides, cholesterol and fatty acids (Elias, 1983, Li et al., 2020). This barrier is not just a physical protection to the inner layer, but it functions like a rheostat to convey signals from the surface to the base membrane using bioactive sphingolipids as messengers (Coant et al., 2017). Sphingosine 1-phosphate (S1P), one of those bioactive lipid mediators, is generated from ceramide by the consecutive actions of ceramidase and sphingosine kinase. S1P can potently regulate a variety of cell activities including cell proliferation via binding to and activating high-affinity G protein-coupled receptors which are named S1P receptor (S1PR) 1–5 (Coant et al., 2017, Obinata and Hla, 2019). S1PRs, including S1PR2, have diverse functions and have been implicated in many organ-system pathologies such as cell proliferation, differentiation, migration, and degranulation in endothelial, epithelial, nervous and immune systems including some kinds of cancer cells (Adada et al., 2013, Jolly et al., 2004, Nema et al., 2016, Wang et al., 2012). On the other hand, in the intestinal epithelium, S1P controls its barrier function by up-regulation of occludin (OCLN) expression (Paszti-Gere et al., 2016) and zonula occludens 1 (ZO1) expression via S1PR2 mediated PI3K/AKT signaling pathway (Chen et al., 2018), and it has already been shown that S1P induces keratinocyte differentiation by induction of the intracellular calcium concentration elevation (Japtok et al., 2014, Lichte et al., 2008), or inhibition of insulin-mediated keratinocyte proliferation through S1PR2 (Schuppel et al., 2008). However, the specific activity of S1P on S1PR2 in respect to epidermal barrier formation has not been investigated. Previously, we have reported that activation of S1PR1 and 2 during bacterial invasion controls proinflammatory cytokine synthesis in human keratinocytes (Igawa et al., 2019). We hypothesized that S1P and S1PR2 are involved in not only proinflammatory cytokine production but also barrier function in the epidermis as a response to various external stimuli, such as mechanical stress or bacterial invasion.

In this study, we have shown how S1P and S1PR2 contribute to maintaining epidermal barrier homeostasis both in vivo and in vitro. Our study started with the observation that S1pr2^−/− mouse skin is fragile and characterized by a compromised skin barrier function. We demonstrated that while S1pr2^−/− mouse skin phenotype and epidermal barrier function appeared normal, their various epidermal barrier-related protein expressions were significantly decreased, especially under bacterial invasion and mechanical stress. Here, we address how S1P stimulation increases these barrier-related protein mRNA expressions via S1PR2 in normal human epidermal keratinocytes (NHEKs), while the loss of cellular capacity to sense S1P through their S1PR2 brings alterations similar to those observed in skin disorders related to epidermal barrier dysfunction such as atopic dermatitis (AD).

RESULTS

S1pr2^−/− mouse epidermal barrier function is impaired by mechanical stresses.

To investigate whether S1pr2^−/− is involved in the epidermal barrier function, we decided to induce a mechanical stress using a sequential tape stripping followed by time-dependent transepidermal water loss (TEWL) evaluations. Before tape stripping, shaved back skin of S1pr2^−/− mice showed slightly higher TEWL than wt mice with no significance (Figure 1a). After tape stripping, S1pr2^−/− mice displayed significantly higher TEWL and required another 24 more hours to recover their TEWL levels than wt mice (Figure 1a). To rule out the possible damaging effects of hair removal, we measured TEWL on the ear skin of wt and S1pr2^−/− mice. We found that S1pr2^−/− mice has significantly higher TEWL than wt mice even without tape stripping or hair removal (Figure 1b). We also investigated their phenotype with regards to histology and ultrastructure. S1pr2^−/− mice showed no apparent skin phenotypical, histological or ultrastructural difference from wt mice (Supplementary Figure S1). This result suggests that S1pr2^−/− mouse epidermal barrier function is slightly impaired at baseline when there are no barrier disruptions, but once the barrier has been broken by a mechanical stress, they have a more severe reaction to the stress and an impaired ability to recover their barrier function.

Figure 1. S1pr2^−/− mouse epidermal barrier function was more severely impaired than wt after tape stripping and reduced multiple junctional protein gene expression underlay this impaired barrier function.

(a) The time-dependent change of TEWL measured from wt and S1pr2^−/− mice before and after sequential tape stripping. (b) TEWL of wt and S1pr2^−/− mouse ear skin without any treatments. (c, d) Zo1, Cldn1, Ocln and Cdsn transcriptions in wt and S1pr2^−/− mouse epidermis before and three hours after tape stripping (c) and the ratio of downregulation in these mRNA expressions after tape stripping (d). Each data is shown by a comparison with the wt mRNA expression before tape stripping. TEWL: transepidermal water loss; Zo1: Zonula occludens-1; Cldn1: Claudin-1, Ocln: occludin; Cdsn: corneodesmosine; wt: wild type.

Reduced multiple junctional protein gene expression underlies impaired S1pr2^−/− mouse epidermal barrier function

Since previous reports on S1P and S1PR2 function on the intestinal epithelia focused on tight junctions (TJs) (Chen et al., 2018, Chen et al., 2017), we proceed to verify whether the correlation between S1P signaling and TJs was also true in skin epithelia. To determine if junction proteins were responsible for the impaired barrier function of S1pr2^−/− mice, we used RT² Profiler PCR Arrays® to analyze the epidermis of wt and S1pr2^−/− mice before and after tape stripping. Based on the RT² Profiler PCR Arrays® data, we found that the S1pr2^−/− mouse epidermis showed lower expression of multiple junctional protein genes including TJ protein genes (Supplementary Figure S2). Already at baseline, without tape stripping, S1pr2^−/− mouse epidermis showed lower expression of the TJ protein Zo1, Ocln, claudin 1 (Cldn1) and of a corneodesmosome related protein, corneodesmosin (Cdsn), than wt mice (Figure 1c). After tape stripping, S1pr2^−/− mice showed further downregulated Zo1, Cldn1, and Cdsn expressions than wt mice (Figure 1c and d). We also analyzed S1pr1–5 mRNA expressions of wt and S1pr2^−/− mouse epidermis before and three hours after tape stripping. We could not detect S1pr3 mRNA expression in either wt or S1pr2^−/− mouse epidermis and confirmed that after tape stripping, S1pr1 and 4 mRNA expressions were increased, but S1pr2 deficiency does not alter their transcription levels (Supplementary Figure S3). These data suggest that S1pr2^−/− mice have a subclinical epidermal barrier dysfunction that becomes clinically evident after mechanical disruption.

S1pr2^−/− mice show deeper S. aureus penetration and more massive neutrophil infiltration in the dermis after epicutaneous bacterial application.

To explore whether epidermal barrier dysfunction will compromise skin resistance against bacterial invasion, Staphylococcus aureus (S. aureus) was applied on the back skin of wt and S1pr2^−/− mice. After epicutaneous S. aureus application, S1pr2^−/− mice showed more pustular and erosive skin lesions than wt mice (Figure 2a). Histological analysis revealed that S1pr2^−/− mice showed more massive neutrophil infiltration in the dermis (Supplementary Figure S4). Consistent with histological data, immunofluorescent staining also confirmed deeper S. aureus penetration and more massive neutrophil marker, Ly-6G, staining in the dermis of the S1pr2^−/− mice compared to the wt (Figure 2b).

Figure 2. Decreased ZO1 and CDSN expression in S1pr2^−/− mouse skin allowed deeper S. aureus penetration and more massive neutrophil infiltration in the dermis after epicutaneous bacterial application.

(a, b) wt and S1pr2^−/− mouse clinical manifestations (a) and S. aureus (green) and Ly-6G (red) immunofluorescent images (b) 48hr-after TSB (a) or sa113 (a, b) application. (c-f) Western blot analysis (c, d) and mRNA expression (e, f) of ZO1 (c, e) and CDSN (d, f) of wt and S1pr2^−/− mouse whole skin 48 hr after TSB or sa113 application. Each RT-qPCR data is shown by a comparison with the wt mRNA expression incubated with TSB. Scale bars: (b) 20 μm, CDSN: corneodesmosin; ko: S1pr2^−/− mice; TSB: 3% tryptic soy broth; wt: wild-type; ZO1: Zonula occludens-1.

To confirm the role of S1PR2 in controlling the expression of TJ and corneodesmosome proteins ZO1 and CDSN, respectively, we performed Western blot analysis using whole skin extracts derived from wt or S1pr2^−/− mice with or without application of S. aureus. Without S. aureus application, S1pr2^−/− mouse skin expressed less ZO1 compared to wt mouse skin (Figure 2c). After S. aureus application, ZO1 expression was substantially reduced in both wt and S1pr2^−/− mouse skin; notably, in the S1pr2^−/− mouse skin, we could not obtain any visible bands (Figure 2c). These data suggest that S1pr2^−/− mouse skin have a decreased ZO1 expression compared to wt mice without bacterial infection. Furthermore, S1pr2^−/− mouse skin also showed reduced CDSN expression after S. aureus application (Figure 2d). Consistent with the protein data, S1pr2^−/− mice showed significantly lower Zo1 (Figure 2e) and Cdsn (Figure 2f) mRNA levels compared to wt mouse skin without S. aureus application. After S. aureus application, S1pr2^−/− mice showed the lowest Zo1 and Cdsn expression (Figure 2e, f); however, there was no significance in the degree of mRNA reduction due to bacterial treatment in Zo1 and Cdsn compared to TSB control treatment (Supplementary Figure S5). Taken together, these findings also suggest an association between S1pr2 deficiency and impaired skin barrier functions due to decreased TJ and corneodesmosome expressions.

Filaggrin 2 expression was decreased in S1pr2−/− mice

In AD, the metabolism of ceramides is disturbed (Holleran et al., 2006) and reduced S1P level was found in canine AD (Baumer et al., 2011); however, S1P expression in human AD is still controversial (Japtok et al., 2014). At the same time, expression of filaggrin, a gene of great significance for the homeostasis of the skin, is also altered (Cabanillas and Novak, 2016, Palmer et al., 2006). Therefore, to investigate whether other skin barrier components were affected by S1pr2^−/−, we performed RNA sequencing using the epidermis of wt and S1pr2^−/− mice. According to the RNA sequencing data, we actually found that the S1pr2^−/− mouse epidermis showed lower gene expression of filaggrin 2 (Flg2) (Figure 3). FLG2 behaves similarly to filaggrin which undergoes proteolytic modification, suggesting that they may also have a function of providing natural moisturizing factors and its expression is also decreased in AD (Makino et al., 2014, Margolis et al., 2014, Pellerin et al., 2013). To assess the relationship between FLG2 and S1P in vitro, we incubated NHEKs with 1 μM S1P for 30 minutes. S1P treatment increased keratinocyte FLG2 expression significantly (Figure 4a). To evaluate this result in vivo, we investigated the Flg2 expression in the whole skin of wt and S1pr2^−/− mice and confirmed that the S1pr2^−/− mouse skin showed significantly lower Flg2 expression than wt mice (Figure 4b). Moreover, immunofluorescent staining revealed that FLG2 staining was lost in S1pr2^−/− epidermis while wt mouse epidermis showed granular and linear FLG2 staining in the granular layer (Figure 4c). These data suggest that S1P regulates FLG2 expression via S1PR2 and may be involved in the function to provide natural moisturizing factor derived from FLG2.

Figure 3. Filaggrin 2 expression was decreased in S1pr2^−/− mice.

Heatmap clustering showing the difference of epidermal gene expressions in wt and S1pr2^−/− mouse epidermis. Flg2: filaggrin 2; wt: wild type.

Figure 4. Filaggrin 2 expression is regulated via S1P-S1PR2 signaling.

(a) FLG2 mRNA expression levels in NHEKs incubated with PBS or 1μM S1P for 30 minutes. (b) Flg2 mRNA expression levels in wt and S1pr2^−/− mouse whole skin without any barrier disruption. Each RT-qPCR data is shown by a comparison with the PBS control (a) or wt skin (b) mRNA expression. (c) Flg2 immunofluorescence images of wt and S1pr2^−/− mouse skin. Lower panels are magnified images of the different fields from upper panels. Scale bars: (c) upper panels 100 μm, lower panels 50 μm. Flg2: filaggrin 2; NHEK: normal human epidermal keratinocyte; PBS: phosphate-buffered saline; S1P: sphingosine 1-phosphate; S1PR: S1P receptor; wt: wildtype.

S1P stimulation increases barrier-related protein mRNA expression via S1PR2 in NHEKs.

To further analyze the effect of S1P-S1PR2 axis on skin barrier functions in vitro, we measured barrier-related protein mRNA expressions in NHEKs with or without S1P stimulation. We confirmed that 10μM S1P treatment significantly increased keratinocyte ZO1, CDSN, FLG2 and keratin10 (KRT10) expression, whereas pre-treatment of 10μM S1PR2 antagonist, JTE013, in NHEKs prevented the S1P induced ZO1, CDSN and FLG2 expressions but not KRT10 (Figure 5a–d). These data suggest that S1P directly increases the expression of epidermal barrier-related proteins in keratinocytes via S1PR2. Our results further highlight the role of S1PR2 in the regulation of the skin barrier function.

Figure 5. The S1P stimulation significantly increases epidermal barrier related protein expressions via S1PR2.

mRNA expression levels of (a) Zo1 (b) Cdsn (c) Flg2 and (d) Krt10 measured in NHEKs incubated with PBS or 10μM S1P for 30 minutes. Before treatment, S1PR2 was inhibited by incubating NHEKs with 10μM JTE013 for 2 hours. Each data is shown by a comparison with mRNA expression of NHEK treated with PBS control. Cdsn: corneodesmosin; Flg2: filaggrin 2; Krt10: keratin 10; NHEK: normal human epidermal keratinocyte; PBS: phosphate-buffered saline; S1P: sphingosine-1-phosphate; S1PR: S1P receptor; Zo1: Zonula occludens 1.

DISCUSSION

Maintenance of epidermal barrier homeostasis is important not only to prevent our body from external stimuli but also to avoid water loss from our body (Eyerich et al., 2018, Jonca et al., 2011). A variety of epidermal barrier gene-related genetic disorders, such as peeling skin syndromes, SAM syndrome, and Netherton syndrome, indicate how dysfunction of epidermal barrier affects body homeostasis (Chavanas et al., 2000, Ishida-Yamamoto and Igawa, 2014, Mohamad et al., 2018, Oji et al., 2010, Samuelov et al., 2013). These diseases are also known to have AD-like symptoms such as multiple allergies and increased serum IgE levels (Ishida-Yamamoto and Igawa, 2014). Since FLG gene mutation has been associated with AD symptoms (Palmer et al., 2006), a compromised epidermal barrier function is now recognized to be a key factor in AD development (Weidinger et al., 2018). The discovery of the association between AD, FLG and abnormal barrier function has led to the discovery of many other gene mutation and/or malformations as a genetic background of AD especially in TJ-related genes (Fortugno et al., 2012, Margolis et al., 2014, Saunders et al., 2013, Yuki et al., 2016).

S1P is widely known as a powerful modulator of homeostasis and pathogenesis in multiple organ systems (Obinata and Hla, 2019). The S1P-S1PR signaling system has been repeatedly shown to play a crucial role in regulating cell survival, proliferation, migration, phenotype and various inflammatory processes (Nema et al., 2016). S1P also induce keratinocyte differentiation by increasing its intracellular calcium concentration (Allende et al., 2013, Lichte et al., 2008) and is produced from both sphingomyelin and ceramides when the stratum corneum and/or epidermal keratinocytes are affected by external stimuli (Coant et al., 2017). These previous data made us consider the hypothesis that S1P-S1PR signals also control epidermal barrier functions. In this study, we demonstrated the crucial role of S1P-S1PR2 signals in regulating epidermal barrier-related protein expression and its function.

Using two in vivo mouse models, we have demonstrated that S1pr2^−/− mouse skin has subclinical barrier impairment. After barrier disturbance via tape stripping, S1pr2^−/− mice displayed significantly higher TEWL and required a longer time to normalize their TEWL compared to wt (Figure 1a), which indicates barrier dysfunction. After S. aureus epicutaneous application, S1pr2^−/− mice also experienced more pustular and erosive lesions as well as increased inflammation and bacterial penetration in the dermis compared to wt. These data suggest that S1PR2 is central to maintaining epidermal homeostasis and integrity. Furthermore, since S1pr2^−/− mice have reduced expression levels of TJ and barrier-related proteins such as ZO1, FLG2 and CDSN, S1PR2 deficiency may contribute to a dysregulation of vital TJs, corneodesmosomes and cornified envelopes needed for front-line innate defense against infection (Alberola et al., 2019, Ishida-Yamamoto and Igawa, 2015). Consistent with this observation, in endothelial and intestinal epithelial cells S1PR1 and 2 control TJ formation. (Chen et al., 2018, Lee et al., 2006, Li et al., 2015, Paszti-Gere et al., 2016). Epidermal TJs are dynamic structures formed by adhesive and scaffolding proteins that play a role in maintaining cell polarity, regulating paracellular movement between different cell layers of the stratum granulosum as well as forming and maintaining epithelial and endothelial barriers (Aijaz et al., 2006, Kuo et al., 2013, Niessen, 2007, Shin et al., 2006). In addition, TJ proteins are involved in cell differentiation and proliferation, cell signaling and vesicle transport (Matter et al., 2005, Schneeberger and Lynch, 2004) and permeability barrier in epidermal keratinocytes (Yuki et al., 2007). When TJ proteins are impaired, the uptake of antigens is considerably increased, thus may result in increased levels of exogenous pathogens or particles entering the skin and triggering inflammatory response like in AD (Kezic et al., 2014, Yoshida et al., 2013). This explains why S1pr2^−/− mouse skin displays increased inflammation and deeper S. aureus penetration consistent with dysregulation of junctional and barrier proteins. However, S1pr2^−/− mice show no clinical or histological manifestations at baseline, possibly due to compensation by upregulation of other small proline rich proteins (Sppr1b), gap junction beta-2 protein (Gjb2), gap junction beta-6 protein (Gjb6) shown in RNA sequence and RT² Profiler PCR Arrays® (Figure 3 and Supplementary Figure S3). In addition, higher TEWL observed in S1pr2^−/− mice could be a result of impaired stratum corneum (SC) integrity/cohesion and altered lipid composition related to CDSN and FLG2. FLG2 is protein of the filaggrin family expressed in the granular layer of the epidermis and localized in keratohyalin granules (Wu et al., 2009), and is diffusely present in the stratum corneum (Makino et al., 2014). Like filaggrin, FLG2 was suggested to also have hydration and photoprotection properties in the epidermis (Pendaries et al., 2015). Previous studies have shown that FLG2 levels were significantly decreased in tissue samples from patients with skin diseases such as ichthyosis vulgaris, AD and psoriasis vulgaris compared to normal skin samples (Makino et al., 2014). Our immunofluorescent staining confirmed that FLG2 staining was lost in S1pr2^−/− epidermis while wt mouse epidermis showed FLG2 staining in the granular layer. Interestingly, filaggrin deficiencies were shown to influence TJ proteins (Nakai et al., 2012) and FLG2 deficiency reduce CDSN expression and cause peeling skin syndrome type A (Bolling et al., 2018, Mohamad et al., 2018). A recent paper discovered that filaggrin mutations are responsible for an altered phase-separation dynamic between keratohyalin granules and cytoplasm, showing how associated skin barrier disorders can be exacerbated by environmental extremes (Quiroz et al., 2020). Hence, FLG2 deficiency could result in decreased SC hydration and barrier dysfunction, leading to an increase in TEWL as previously discussed.

Taken together, we have demonstrated that S1pr2^−/− skin displays barrier impairments associated with a downregulation of FLG2, CDSN and TJ-related proteins (Figure 1–4). Consistently, our in vitro data indicated that treatment of S1P directly increased barrier-related mRNA expression levels in keratinocytes, while chemical block with S1PR2 antagonist, JTE013, significantly reduced them (Figure 5). In the present study, we did not confirm that S1P directly increases FLG2 expression in keratinocytes after differentiation but only on undifferentiated keratinocytes. This is a limitation to our study that in the future could possibly be answered by studying keratinocytes’ FLG2 response to S1P in a 3D skin model. Our data implicate S1P and S1PR2 as a modulator of skin barrier function and a possible pharmaceutical target for skin disorders based on impaired epidermal barrier function, such as AD. Since S1P regulates FLG2 via S1PR2 and may be involved in providing natural moisturizing factors derived from FLG2, treatment with S1P might be able to alleviate dry skin and proinflammatory responses in AD. According to our RNA sequencing data, S1pr2^−/− mouse epidermis showed increased IL19, 24 and 33 expressions (Figure 3), so we can also speculate that a S1PR2 deficiency in the epidermis will affect Th2 response in the dermis, as observed in AD patients. Moreover, increased proinflammatory cytokines will further downregulate filaggrin and filaggrin 2 (Pellerin et al., 2013), creating a vicious cycle.

Since S1PR2 is present in other immune cell types such as macrophages and Th cells, to be able to assess the repercussion of epithelial deficiency of S1PR2 and to rule out the effects of system-wide S1PR2 deficiency, we plan to assess skin barrier function using Cre-loxP mice with conditional S1pr2 deletion in the epidermis. We also should consider the possible compensation mechanism of S1pr1, 3–5 in the absence of S1pr2. Moreover, to further establish the relationship between S1PR2 deficiency and barrier impairment, a thorough look at the morphology of tight junction network using confocal microscopy is needed.

In conclusion, our study indicates that S1P-S1PR2 signaling is critical in epidermal barrier function, suggesting their involvement in barrier dysfunctional phenotype of S1pr2^−/− mice. To detect whether S1pr2^−/− mice may have a weakened cytotoxic response against S. aureus infection and inability to lyse invasive pathogens and a propensity to develop a Th2 phenotype, further investigation into their dermal cytokine and antimicrobial profiles is needed. This will be the focus of our future investigation.

MATERIALS AND METHODS

Mice

BALB/c wildtype and S1pr2^−/− mice were provided by Dr. Jerold Chun (Herr et al., 2016, Ishii et al., 2002). S1PR2 genotyping were performed by PCR using tail genomic DNA (Ishii et al., 2002). Three S1pr2^−/− mice and three of their littermate controls were used for each experiment. We confirmed that in the epidermis, S1pr2 deficiency does not affect S1pr1, 3–5 expressions before and after tape stripping (Supplementary Figure S3). All animal protocols were reviewed and approved by the University of California San Diego (approval number: S10288).

Tape stripping and transepidermal water loss evaluation

To evaluate response to skin barrier disruption in vivo, transepidermal water loss (TEWL) was measured before and after tape stripping on wt and S1pr2^−/− mouse back skin. After 24 hours from hair removal, we tape stripped mouse back skin ten times using D-squame® D100 (Clinical & Derm, Dallas, TX). TEWL was measured at 2, 6, 24, 48 and 72 hours after tape stripping using Tewameter TM300 Courage+ (Khazaka, Clologne, Germany). We also measured TEWL on ear skin without any treatment in both wt and S1pr2^−/− mice. To investigate epidermal barrier function-related gene expressions before and after tape stripping, we isolated mouse epidermis by treating wt and S1pr2^−/− mouse back skin samples with Dispase II (Sigma-Aldrich, St. Louis, MO) at 4°C overnight and these samples were analyzed by real-time quantitative RT-PCR (RT-qPCR).

Epicutaneous S. aureus application on the mouse back skin

According to the previous report (Nakatsuji et al., 2016), we applied S. aureus (sa113, ATCC35556, ATCC, Manassas, VA) on wt and S1pr2^−/− mouse back skin. After their hair removal, we applied 10 mm-sized agar discs containing TSB as a control or 1×10⁷ CFU S. aureus on their tape stripped back skin and the entire dorsal skin was then covered with Tegaderm (3M, Maplewood, MN) for 48 hrs. After disc removal, each whole skin samples were analyzed histologically, RNA was extracted for RT-qPCR, and protein was extracted for western blot analysis from each sample.

Primary normal human epidermal keratinocytes

Undifferentiated NHEKs (Thermo Fisher, Waltham, MA) were cultured in EpiLife^™ Medium with 60 μM calcium (Thermo Fisher), seeded sufficiently for cell treatments. Subconfluently conditioned NHEKs were used for the experiments according to the previous report (Igawa et al., 2019).

S1P treatment and chemical S1PR2 block in NHEKs

To block S1PR2, NHEKs were incubated with 10μmol/L JTE013 (Cayman, Ann Arbor, MI) at 37°C for 2 hours prior to S1P treatment. After that, NHEKs were incubated with PBS and 1 or 10μmol/L S1P (TOCRIS, Minneapolis, MN) at 37°C for 30 minutes.

Histology and Immunofluorescence staining

Formalin fixed and paraffin embedded mouse skin samples were sectioned and hematoxylin-eosin stained at UCSD Comprehensive Moores Cancer Center Biorepository and Tissue and Technology Shared Resource. Skin sections were immunostained, as described previously (Wang et al., 2017), with the primary and secondary antibodies listed in Supplementary Table S1. Fluorescence images were obtained with a fluorescence microscope (Olympus BX51).

Quantitative real time reverse transcriptase-PCR

Total mouse RNA from whole mouse skin samples were isolated using Direct-zol Miniprep kit (Zymo Research, Irvine, CA). Total RNA from mouse epidermis and NHEKs were isolated by Quick RNA Miniprep kit (Zymo Research). cDNA conversion from RNA and real time RT-qPCR were performed according to the previous report (Igawa et al., 2019). We listed the probes used for real time RT-qPCR in Supplementary Tables S2 and S3 (Gandy et al., 2013, Li et al., 2012, Pendaries et al., 2014, Ryu et al., 2018, Vivinus-Nebot et al., 2014) online. The expression of target genes was normalized to GAPDH expression and analyzed by the 2^{− ΔΔCt} method.

Western blot analysis

According to the previous report (Takahashi et al., 2018), the mouse skin samples were lysed in RIPA Lysis and Extraction Buffer (Thermo Fisher) containing a protease inhibitor cocktail (Complete ultra mini, Sigma-Aldrich). Protein concentrations were determined with the BCA protein assay kit (Thermo Fisher). Proteins (10–20 μg/lane) were separated on a 4–20% precast polyacrylamide gel (Bio-Rad, Hercules, CA), transferred to PVDF membrane (Bio-Rad), followed by immunoblotting using indicated primary antibodies followed by fluorescent secondary antibodies (Supplementary Table S1) (LICOR Biosciences, Lincoln, NE) and imaging using fluorescent Odyssey System (LICOR Biosciences). Experiments were performed in triplicates and all results showed the same trend.

RNA sequencing

Purified RNA samples from wt and S1pr2^−/− mice were submitted to the UCSD Institute for Genomic Medicine core facility for library preparation and high-throughput next-generation sequencing, according to the previous report (Liggins et al., 2019). Subsequent analysis was conducted by Center for Computational Biology and Bioinformatics at UCSD.

Statistical Analysis

In all in vitro experiments, all samples were performed in triplicates, and values are expressed as mean ± standard deviation. Mann-Whitney test were applied to analyze the differences between two groups. One- or two-way analysis of variance (ANOVA) and Tukey tests were applied to analyze the differences among more than two groups. P < 0.05 was considered significant.

¹ Abbreviations:

AD

atopic dermatitis

CDSN

corneodesmosin

CLDN1

claudin 1

FLG2

filaggrin 2

KRT10

keratin10

NHEK

normal human epidermal keratinocyte

OCLN

occluding

PBS

phosphate buffered saline

RT-PCR

Reverse transcriptase-PCR

S1P

sphingosine 1-phosphate

S1PR

sphingosine 1-phosphate receptor

TEM

transmission electron microscope

TEWL

transepidermal water loss

TJ

tight junction

wt

wild type

TSB

3% tryptic soy broth, ZO1, zona occludens 1

Data availability statement

Datasets related to this article can be found at https://data.mendeley.com/datasets/csrs6vw8ps/1, hosted at Mendeley. The RNA sequence data is uploaded to the Sequence Read Archive, hosted by NCBI (accession number PRJNA658157).