Staphylococcus aureus causes aberrant epidermal lipid composition and skin barrier dysfunction

2.1 Study subjects

Skin tape strip (STS) samples were obtained from 24 children with AD (6.0 ± 4.4 years) and 16 healthy subjects (7.0 ± 4.5 years) (p = .217). The groups were well matched for gender distribution, with 59.3% of males in the AD group and 50% of males in the healthy subjects group (p = .54). Subjects in the healthy control group had no personal or family history of atopy or skin diseases. AD patients recruited for this study did not use topical corticosteroids, topical calcineurin inhibitors, or topical or oral antibiotics for 1 week prior to enrollment. Patients were not treated with systemic immunosuppressive medications for more than 1 month prior to enrollment into the present study. The diagnosis of AD was based on the criteria defined by Hanifin and Rajka.¹⁷ AD severity was assessed by allergists using the SCORing Atopic Dermatitis (SCORAD), with the score range from 0 to 103.¹⁸ TEWL and skin pH were measured on non-lesional skin from the volar surface of the forearm and lesional skin. All TEWL measurements were done by Tewameter TM300 (Courage & Khazaka, Köln, Germany). In the test room, the relative humidity was kept at 40–50% and temperature was kept at 22–24°C as previously done.¹⁹ This study was approved by the Institutional Review Board at Samsung Medical Center, Pusan National University Hospital, and Inje University-Sanggye Paik Hospital (IRB No. SMC-2018-03-041, H-1808-021-070, and SGPAIK 2018-06-005-001, respectively). Written informed consent was obtained from all parents and/or patients prior to participation in this study.

2.2 Collection of skin tape strip samples

We obtained three different types of skin samples for STS and bacterial swabs: (1) lesional skin from 24 AD patients, (2) non-lesional volar forearm skin from 24 AD patients, and (3) normal skin from 16 healthy control subjects. A total of four consecutive D-Squame® tape discs (22 mm diameter; CuDerm) were applied to the skin as described before.²⁰ Bacterial DNA samples from skin swabs were prepared for polymerase chain reaction (PCR) as described previously.²¹

2.3 Tape strip processing for lipid extraction and protein estimation

Tape strip processing was performed according to the previously published protocol.²² Lipids were dissolved in 0.2 ml methanol for liquid chromatography electrospray ionization tandem mass spectrometry (LC-ESI-MS/MS) analyses.

2.4 Lipid and protein analyses

CERs, lysophosphatidylcholines (LPCs), and sphingomyelins (SMs) from human skin and organotypic 3D skin were identified and quantified using the targeted LC-ESI-MS/MS approach on a Sciex 6500QTRAP mass spectrometer coupled with a Shimadzu Nexera X2 UHPLC system as described previously.¹¹

2.5 Primary human keratinocyte culture

Human primary epidermal keratinocytes (HEKs) derived from neonatal foreskin were grown in serum-free EpiLife cell culture medium (Life Technologies, Grand Island, NY) containing 0.06 mmol/L calcium chloride, 1% human keratinocyte growth supplement S7 (Life Technologies), and 1% of gentamicin/amphotericin.

To investigate the effects of the Staphylococci on the gene expressions of FA elongases (ELOVLs), CER synthase (CERS)2, CERS3, cytochrome P450 family 4 subfamily F polypeptide 22 (CYP4F22), delta 4-desaturase, sphingolipid 2 (DEGS2), and UDP-glucose ceramide glucosyltransferase (UGCG), HEKs were seeded in a 24-well plate with 2 × 10⁵ cells per well and differentiated in the presence of 1.3 mmol/L of CaCl₂ for 3 days. Then, cells were stimulated with three strains of live S. epidermidis (ATCC®-12228™, ATCC®-49461™, and ATCC®-700296™, Manassas, VA, 0.001 or 0.01 CFU/well), three strains of methicillin-sensitive S. aureus (MSSA) (ATCC®-29213™, ATCC®-49230™, and SMC1801-45, Asian Bacterial Bank, Seoul Korea, 0.001 or 0.01 CFU/well), or three strains of MRSA (ATCC®-BAA-1556™, ATCC-700699™, and SMC1803-07, Asian Bacterial Bank, 0.001 or 0.01 CFU/well) for an additional 24 h, prior to RNA collection. Cell culture media were also harvested and stored at −80°C for lactate dehydrogenase (LDH) assays.

To examine Staphylococci-induced cytokine expression, HEKs were differentiated for 3 days, and then, cells were stimulated with three strains of S. epidermidis (same as above, 0.01 CFU/well), MSSA (same as above, 0.01 CFU/well), or MRSA (same as above, 0.01 CFU/well) for 8 or 20 h. Cell culture media were harvested and stored at −80°C for ELISA assays.

To investigate whether keratinocyte-derived cytokines inhibit ELOVL3 and ELOVL4, HEKs were Ca²⁺ differentiated and stimulated with various concentrations of S. aureus-induced cytokines such as tumor necrosis factor (TNF)-α, interleukin (IL)-1β, IL-36γ, IL-6, and IL-33 for 24 h. In addition, HEKs were differentiated and preincubated with isotype control antibodies or neutralizing antibodies against TNF-α (0.1 μg/ml; R&D Systems), IL-1β (10 ng/ml; R&D Systems), IL-36γ (1.0 μg/ml; R&D Systems), IL-6 (0.1 μg/ml; R&D Systems), and IL-33 (1.0 μg/ml; R&D Systems) separately or in combination for 3 h, followed by stimulation with MSSA (0.01 CFU/cell) or MRSA (0.01 CFU/cell) for an additional 20 h.

2.6 Real-time RT-PCR

RNeasy Mini Kits (Qiagen) were used for RNA extraction from cultured HEKs. RNA was reverse-transcribed into cDNA using the SuperScript VILO MasterMix (Invitrogen). Real-time reverse-transcribed polymerase chain reaction (RT-PCR) was performed using an ABI Prism 7300 sequence detector (Applied Biosystems). Primers and probes were purchased from Applied Biosystems. Amplification reactions were performed in MicroAmp optical plates (Applied Biosystems) as previously described.²³ Relative expression levels were calculated by the relative standard curve method as outlined in the manufacturer's technical bulletin.

2.7 Organotypic skin model

A mixture of fibroblasts (7.5 × 10⁵/culture insert; CCL-92™, ATCC) and rat tail collagen type I (10 mg/ 2 ml/culture insert; Corning Life Sciences) were plated on tissue culture inserts (MilliporeSigma) to prepare dermal equivalents. HEKs (2 × 10⁶/culture insert) were then plated on top of the dermal equivalent and cultured for 12 days. On Day 13, the organotypic 3D skins were stimulated with media and live S. aureus (0.01 CFU/cell) for an additional 20 h. The epidermal part of the organotypic skin was isolated and stored at −80°C for lipidomic analysis or submerged in 4% paraformaldehyde to fix the tissue for immunostaining.

2.8 TEWL measurement

Transepidermal water loss (TEWL) in human skin was measured as previously described.¹⁹ The relative humidity was kept at 40%–50%, and the temperature was kept at 22–24°C in this room. TEWL in organotypic skin was measured using a device from GPower Inc. (Seoul, Korea) as previously described.²² To measure TEWL in organotypic skin, culture media was removed from the culture chamber and insert. Then, the organotypic skin sat for 10 minutes to equilibrate. TEWL was measured 3 times from each organotypic skin cultures, and the average values were used.

2.9 Preparation of bacteria

Fifty μL of frozen bacterial stock was added into 3 ml of tryptic soy broth (Millipore Sigma) and incubated at 37°C overnight on a rotary shaker. One hour prior to use, 3 ml of fresh tryptic soy broth media (Millipore Sigma) was added to the culture and then returned to incubation. One ml of bacterial culture was pelleted by centrifugation at 2850g-force for 5 min in an Eppendorf 5424 centrifuge, and then, CFUs/ml were determined by optical density at 595 nm using a spectrophotometer Scientific (Cole-Parmer®). Another 1 ml of bacterial culture was pelleted by centrifugation, resuspended in sterile PBS (1 × 10⁷ cfu/ml) and then diluted to the desired concentration for use.

2.10 Multiplex real-time RT-PCR for Staphylococcus genes

The TaqMan probe-based multiplex real-time RT-PCR assay was performed by the method described in the previous reference.²¹ The multiplex real-time RT-PCR primers and TaqMan probes were used to amplify the Staphylococcus spp. (16S rRNA), S. aureus (nuc), and MRSA (mecA). Multiplex real-time RT-PCR was performed following conditions: 95°C for 3 min and 45 cycles of 95°C for 20 s and 60°C for 40 s in a single real-time RT-PCR assay.

2.11 Measurement of protein levels of TNF-α and IL-1β

Protein expression levels of TNF-α and IL-1β in cell culture supernatant were measured using an ELISA kit (R&D Systems) according to the manufacturer's instructions.

2.12 LDH assay

Culture media from HEKs culture and 3D skin experiments were collected and evaluated to examine cell toxicity of Staphylococci. LDH release was determined using the CytoTox-One Homogeneous Membrane Integrity Assay (Promega) according to the manufacturer's instructions.

2.13 Immunofluorescent staining

Organotypic skin sections were blocked with Super Block (Scyteck Laboratories). Slides were then stained with antibodies against ELOVL3 (Abcam; cat# ab253016, 1:500 dilution) and ELOVL4 (Abcam; cat# ab224608, 1:200 dilution). The slides were visualized with the fluorescent microscopy (Leica), and the levels of mean fluorescent intensity (MFI) were measured with Slidebook 6.0 (Intelligent Imaging Innovations).

2.14 Statistical analysis

Data were analyzed using SPSS for Windows (version 27.0; SPSS) and Graphpad Prism (version 9; Graphpad Software). Categorical data are presented as numbers and percentages. Continuous data are presented as median and interquartile ranges. Categorical variables between groups were compared using the chi-square test and Fisher's exact test. In comparisons of the non-lesional and lesional skins with the control group, data were analyzed by the Kruskal–Wallis test with the post hoc test. Statistical differences between the two groups were determined using a Mann–Whitney U-test. A p value < .05 was considered to be significant.

For details about lipidomics, organotypic skin model, LDH assay, TEWL measurement, RT-PCR, immunofluorescent staining, and bacterial DNA preparation for PCR from skin swabs, refer to the supplemental method file.

3.1 Increased TEWL and altered lipid profiles in the skin of AD children colonized with MRSA

We have evaluated SCORAD, TEWL, and epidermal lipid profiles of pediatric AD patients and normal healthy controls. We also performed PCRs to evaluate S. aureus colonization using skin swab samples, and subjects were then subgrouped based on S. aureus methicillin sensitivity into MSSA and MRSA groups (Table S1). Importantly, 43.4% of S. aureus-positive swabs from AD lesional skin were found to be MRSA positive, while 8.3% swabs from AD non-lesional skin and 6.3% swabs from normal healthy skin were MRSA positive (all p < .001) (Figure S1A).

AD severity (SCORAD score) was significantly (p < .01) higher in AD patients colonized with MRSA (median 31.6, IQR 28.8–33.7) than in the no S. aureus group (median 21.9, IQR 21.2–22.5), while no difference in SCORAD score was noted between the MSSA group (median 23.7, IQR 23.6–25.0) and the no S. aureus group (Figure S1B) in AD patients. Lesional skin TEWL of MRSA group was significantly higher as compared to no S. aureus group (p < .01) and MSSA group (p <.01) (Figure 1A). Additionally, non-lesional skin TEWL of the MRSA group was significantly (p < .05) higher compared to the no S. aureus group (Figure 1A). These data suggest that AD patients colonized with MRSA have more severe AD symptoms and skin barrier disruption than AD patients colonized with MSSA.

We have examined whether these changes in MRSA-colonized skin can be further corroborated by changes in epidermal barrier lipid profiles. Lipid molecular species with palmitic and stearic fatty acids (C16–C18) were globally increased in the skin samples from patients colonized with S. aureus (Figure 1). Nonhydroxy fatty acid C18-sphingosine ceramides (NS-CERs) N-acylated with C16–C22 FAs (Figure 1B), SM N-acylated with C16 & C17 FAs (Figure 1C), and LPCs with C16–C18 FAs (Figure 1D) were significantly increased in the skin samples colonized with MRSA compared to skin colonized with MSSA or without S. aureus colonization (p < .05 for all corresponding comparisons). Additionally, SM N-acylated with palmitic acid (C16) was significantly (p < .01) increased in the skin colonized with MSSA compared to that of skin without S. aureus colonization (Figure 1C). In contrast, very long chain fatty acid (VLCFAs) containing NS-CERs (FA carbon chain length C26–C30, Figure 1B), SMs (C26, Figure 1C), and LPCs (C24 and C28, Figure 1D) were decreased in the skin samples colonized with MRSA compared to skin colonized with MSSA or without S. aureus colonization. Additionally, SMs (with C20 and C22 FAs) and LPCs (with C26 and C28 FAs) were significantly (p < .05) decreased in the skin colonized with MSSA compared to skin without S. aureus colonization (Figure 1C,D). SM with C24:1 FA was significantly increased in the skin samples colonized with MRSA compared to those colonized with MSSA or without S. aureus colonization (Figure 1C). SM (with C16 FA) and LPCs (with C18 FA) were significantly (p < .01 for all) increased in the skin colonized with MSSA compared to skin without S. aureus colonization (Figure 1C,D). These data demonstrate that S. aureus colonization of skin is associated with aberrant lipid profiles, with more pronounced changes toward lipids with shorter-chain FAs (C16-C18) observed in the MRSA-colonized skin.

3.2 S. aureus increases TEWL and causes abnormal lipid profiles in organotypic skin cultures

We have modeled the effects of MSSA and MRSA exposure on skin barrier function and lipid profiles using organotypic skin cultures.

TEWL was not affected in organotypic skin cultures exposed to MSSA but was significantly increased in organotypic skin cultures after MRSA inoculation (p < .05 as compared to cultures exposed to MSSA; p < .001, as compared to cultures grown in media alone) (Figure 2A). These data suggest that MRSA, not MSSA, may cause skin barrier dysfunction.

An increase in the amounts of C14 and C16 FAs in C18-sphingosine containing-CERs (Figure 2B) and an increased in LPC with stearic acid (18:0) (Figure 2D) were found in organotypic cultures stimulated by three different MSSA strains. In addition, some minor changes in N-22:0 SM and 18:1-LPC were observed in MRSA-stimulated cultures (Figure 2C,D). More pronounced changes were observed in organotypic cultures stimulated with three different MRSA strains. Namely, C14 and C16 FAs of N(С18S)-CERs (Figure 2B, p < .01, p < .05, respectively), N-16:0-SM (Figure 2C, p < .01), and LPCs with 14:0- and 18:0-fatty acids (Figure 2D, p < .01) were significantly increased in organotypic skin cultures stimulated with MRSA compared to the skin cultures stimulated with media alone. In contrast, the amounts of VLCFAs in N(C18S)-CERs (Figure 2B, C24 and C26) and LPCs with C18:1 FA (Figure 2D) were significantly decreased in organotypic skin cultures stimulated with MRSA compared to those stimulated with media alone. These data strongly support that MRSA exposure alters production of epidermal lipids with shorter chain fatty acids (C16–C18) and VLCFAs (C24–C28), resulting in accumulation of lipid species with C16–C18 FAs, which is likely detrimental to the skin barrier. Exposure to MSSA strains resulted in an increase in C16–C18 FA lipid species only, with no inhibition of C24–C28 FA lipid species production.

3.3 S. aureus modulates expression of fatty acid elongases

In mammalian cells, FA elongation proceeds through repetition of the FA elongation cycle whereby two carbons are added to the carboxyl end in each cycle. This reaction is catalyzed by FA ELOVLs; they constitute the ELOVL family of proteins, which includes seven isozymes (ELOVL1-7).^24-26 We investigated whether Staphylococci exposure modulates ELOVL production in keratinocytes. In these experiments, we have compared ELOVL expression in keratinocyte cultures that were stimulated with 0.001 and 0.01 CFU of S. aureus (three different strains of MSSA and MRSA were tested, respectively) as compared to S. epidermidis, a commensal strain of Staphylococci which often colonizes human skin (three different S. epidermidis strains were used).

As depicted in Figure 3A, gene expression of ELOVL5 was significantly (p < .05) increased in HEKs stimulated with S. epidermidis, but other ELOVLs were not affected by S. epidermidis. In contrast, MSSA significantly (p < .01) inhibited ELOVL3 (Figure 3B) expression by keratinocytes. Furthermore, MRSA significantly inhibited both ELOVL3 (p < .001) and ELOVL4 (p < .05) (Figure 3C). Strain-dependent differences in MSSA and MRSA inhibition of ELOVLs were noted (Figure S2A,B).

Staining intensities of ELOVL3 (p < 001) and ELOVL4 (p < .01) were significantly decreased in organotypic skin cultures stimulated with MRSA compared to skin cultures stimulated with media alone (Figure 3D). However, 0.001 cfu/cell of all Staphylococci (S. epidermidis, MSSA, MRSA) did not affect gene expression of ELOVLs (Figure S3A–C). Other enzymes²⁶ that are involved in the synthesis of CERs and VLCFAs (including CERS2, CERS3, CYP4F22, DEGS2, and UGCG) were not affected by MRSA (Figure S3D).

Additionally, UV light-killed S. aureus and Escherichia coli lipopolysaccharide (LPS) did not inhibit ELOVL3 and ELOVL4 (Figure S4). Moreover, the effects of toxins produced by MSSA and MRSA including Panton-Valentine leucocidin (PVL) and phenol-soluble modulin alpha3 (PSMα3) were examined. PVL did not inhibit ELOVL3 and ELOVL4, but 10 μg/mlof PSMα3 significantly (p < .01) inhibited gene expression of ELOVL3 only (Figure S5).

Of note, keratinocyte cultures exposed to Staphylococci remained viable with minimal cytotoxicity observed in both organotypic skin cultures (Figure S6A) and HEKs (Figure S6B).

3.4 S. aureus induces expression of keratinocyte-derived cytokines

To further understand S. aureus-mediated inhibition of ELOVLs, HEKs were differentiated and stimulated with S. epidermidis, MSSA, or MRSA for 8 and 20 h. S. epidermidis exposure did not affect HEK cytokine profile (Figure S7). MSSA significantly induced gene expressions of TNF-α and IL-1β after 8 h of stimulation (Figure 4A). At the same time, a significantly increased gene expression of TNF-α (p < .0001), IL-1β (p < .0001), IL-36γ (p < .001), IL-6 (p < .0001), IL-33 (p < .0001), IL-1α (p < .01), IL-8 (p < .001), and thymic stromal lymphopoietin (p < .001) were observed in HEKs stimulated with MRSA for 8 h (Figure 4B). Gene expression of TNF-α (p < .0001) and IL-1β (p < .0001) remained upregulated in HEKs stimulated with MSSA and MRSA for 20 h (Figure 4A,B). In addition, IL-36γ (p < .0001), IL-6 (p < .05), and IL-33 (p < .01) remained upregulated in HEKs stimulated with MRSA for 20 h (Figure 4B). Strain-dependent variabilities in TNF-α, IL-1β, and IL-33 mRNA induction were noted (Figure S8A–C).

TNF-α protein levels were confirmed to be increased only in keratinocyte cultures stimulated with MRSA (Figure 4C) (p < .01, as compared to untreated cells). Protein levels of IL-1β were significantly increased in culture supernatants from MSSA and MRSA stimulated HEKs (p < .05 and p < .01 compared to untreated keratinocytes) (Figure 4D).

3.5 Keratinocyte-derived cytokines inhibit fatty acid elongase expression

To examine whether S. aureus-induced, keratinocyte-derived cytokines inhibit ELOVL3 and ELOVL4, HEKs were Ca²⁺ differentiated and stimulated with various concentrations of S. aureus-induced cytokines for 24 h. Gene expression of ELOVL3 was significantly inhibited in HEKs treated with TNF-α (Figure 5A), IL-1β (Figure 5B), IL-6 (Figure 5C), IL-33 (Figure 5D), and IL-36γ (Figure S9A) compared to cells treated with media alone. In contrast, ELOVL4 was only significantly inhibited in HEKs treated with IL-33 (Figure 5D). IL-1α (Figure S9B) and IL-8 (Figure S9C) did not inhibit ELOVL3 and ELOVL4.

3.6 Neutralization of keratinocyte-derived cytokines attenuates S. aureus-mediated inhibition of fatty acid elongase expression

We have investigated whether inhibition of MRSA-induced cytokines will interfere with ELOVL3 and ELOVL4 expression by keratinocytes. HEKs were differentiated and preincubated with isotype control antibodies or neutralizing antibodies against TNF-α (0.1 μg/ml), IL-1β (10 ng/ml), IL-36γ (1.0 μg/ml), IL-6 (0.1 μg/ml), and IL-33 (1.0 μg/ml) separately or in combination for 3 h, followed by stimulation with MRSA (0.01 CFU/cell) or MSSA (0.01 CFU/cell) for an additional 20 h. ELOVL3 gene expression was restored in HEKs stimulated with MRSA and individually added neutralizing antibodies against TNF-α (p < .01), IL-1β (p < .05), IL-6 (p < .05), or IL-33 (p < .05) compared to cells stimulated with a combination of MRSA and isotype antibodies (Figure 6A). Gene expression of ELOVL3 was fully restored (p < .05) in HEKs stimulated with MRSA combined with all neutralizing antibodies (against TNF-α, IL-1β, IL-6, and IL-33) (Figure 6B).

Gene expression of ELOVL4 was only restored in HEKs stimulated with the combination of MRSA and anti-IL-33 neutralizing antibody compared to cells stimulated with the combination of MRSA and an isotype antibody (Figure 6C). MRSA-mediated inhibition of ELOVL4 was not significantly altered by neutralizing antibodies, including anti-TNF-α, anti-IL-1β, anti-IL-36γ, or anti-IL-6 antibodies (Figure 6C). Additionally, MSSA-mediated inhibition of ELOVL3 expression was significantly attenuated by a combination of TNF-α and IL-1β neutralizing antibodies (Figure S10A,B). However, ELOVL4 gene expression was not inhibited by MSSA (Figure S10C).