Short-Chain Fatty Acids from Cutibacterium acnes Activate Both a Canonical and Epigenetic Inflammatory Response in Human Sebocytes

Acne vulgaris is an inflammatory skin disorder involving hair follicles and associated sebaceous glands. Ranking third in global disease burden of chronic skin conditions, acne is a significant health care problem and a leading source of medical expense (1). It is widely thought that the skin-resident bacterium Cutibacterium acnes [formerly known as Propionibacterium acnes (2) and referred to by that name throughout this article] is a driver of skin inflammation in this disease (3). TLR2 ligands present on P. acnes have been shown to initiate inflammatory gene expression from cells such as keratinocytes and monocytes (4–7). However, P. acnes is an abundant member of the microbiome in healthy skin in which inflammation is not activated by the presence of these TLR ligands (8, 9). Furthermore, despite the existence of this microbe in the majority of sebaceous follicles, only a select few at any time demonstrate the inflammation characteristic of acne. Thus, critical underlying events that regulate the inflammatory response to P. acnes are not completely understood.

Taking these observations into account, we hypothesized that a change in the local microenvironment of individual sebaceous follicles may lead to a change in the metabolic activity of P. acnes and that this local shift in metabolites may be then responsible for promoting the inflammatory responses observed in acne. Supporting this hypothesis, our previous work demonstrated that short-chain fatty acids (SCFAs) produced by P. acnes in hypoxic, lipid-rich conditions (conditions that mimic an occluded, sebaceous follicle) have a robust proinflammatory effect on epidermal keratinocytes (10). Keratinocytes treated with SCFAs showed an enhanced cytokine response to TLR activation, a distinct and striking phenotype of the keratinocyte inflammatory response that is opposite to the well-documented anti-inflammatory effects of SCFAs on cells of myeloid origin (11–13). This proinflammatory response by keratinocytes could be linked to the inhibition of histone deacetylase (HDAC) activity by SCFAs, and the target of SCFAs was associated with unique HDACs (HDAC8 and HDAC9) because specific depletion of HDAC8 or HDAC9 led to an enhanced response to TLR activation in a manner similar to SCFA treatment. These observations suggested that these HDAC enzymes in particular are responsible for repressing excessive levels of proinflammatory gene expression in keratinocytes and that when SCFAs that inhibit HDAC activity are produced by P. acnes, then the epithelial surface breaks tolerance to these commensal microbes.

In addition to keratinocytes, sebocytes that make up the sebaceous glands associated with hair follicles have also been identified as a major contributor of proinflammatory cytokine and antimicrobial peptide expression in response to P. acnes (14–17). In fact, recent studies have identified immunomodulatory roles of human sebocytes through their production of cytokines, chemokines, and lipid mediators (18–20). Historically viewed primarily as lipid producers, sebocytes are now recognized as another immunocompetent cell type that may participate in, and have regulatory effects on, inflammatory and immune processes in the skin (21–23). However, whether HDAC activity is involved in regulating the expression of inflammatory genes in sebocytes, and whether P. acnes may modulate this regulation through the production of SCFAs, is unknown.

In this study, we set out to explore the effects of P. acnes metabolites and HDAC inhibition on the inflammatory response of sebocytes. Through this work, we demonstrate that human sebocytes, like keratinocytes, exhibit an increased response to TLR activation and P. acnes stimulation when HDAC activity is depleted. Furthermore, the same HDAC enzymes that were demonstrated to be important in keratinocytes, HDAC8 and HDAC9, play a role in regulating proinflammatory gene expression in sebocytes because depletion of either of these proteins largely replicated the effects of SCFA treatment. Finally, we show that unlike keratinocytes, a portion of the proinflammatory effects of SCFAs in sebocytes is due to the activation of fatty acid receptors. Together, these observations show a role for lipids, hypoxia, and P. acnes in promoting the development of acne by both epigenetic and canonical inflammatory mechanisms and define the sebocyte as a unique cell type that may participate in the inflammatory response of the skin.

See Table I.

The sebocyte cell lines SEB-1 (24) and SZ95 (25) were maintained in Sebomed Basal Medium supplemented with 10% FBS, 1× antibiotic-antimycotic, and recombinant human epidermal growth factor (rhEGF; 5 ng/ml) in incubators maintained at 37°C in 5% CO₂. Cells were passaged before reaching 100% confluence in tissue culture flasks. For experiments, cells were switched to medium containing 1% FBS overnight before stimulation.

BD OptEIA ELISA Kits were used to determine cytokine concentrations in cell supernatants (BD Biosciences). Kits were used according to the manufacturer’s instructions, and results were determined using a 96-well plate reader (Beckman Coulter Diagnostics).

RNA from cultured sebocytes was isolated using the PureLink RNA Mini Kit (Thermo Fisher Scientific) according to the manufacturer’s instructions. RNA concentration was determined using a NanoDrop 2000 spectrophotometer (Thermo Fisher Scientific), and up to 1 μg of RNA was converted into cDNA using the iScript cDNA Synthesis Kit (Bio-Rad Laboratories). Five nanograms of cDNA per sample was used in each subsequent quantitative RT-PCR (RT-qPCR), which was performed in 20 μl reactions using TaqMan Universal PCR Master Mix (Thermo Fisher Scientific) and TaqMan Gene Expression Assays (Thermo Fisher Scientific, assay numbers described in Supplemental Table I). All RT-qPCR were performed on a CFX96 Touch Real-Time PCR Detection System (Bio-Rad Laboratories).

Purified isolated RNA was submitted to the University of California, San Diego (UCSD) Institute for Genomic Medicine core facility for quality analysis, library preparation, and high-throughput sequencing. RNA quality was assessed using the Agilent TapeStation instrument (Agilent). Libraries were constructed using TruSeq Stranded mRNA Library Prep Kits (Illumina) and analyzed on a HiSeq 4000 instrument (Illumina). Raw data were obtained from the UCSD Institute for Genomic Medicine core and analyzed using Partek Flow software (Partek) to determine transcript abundance and identify differentially expressed genes. Data are deposited in the National Center for Biotechnology Information Sequence Read Archive (https://trace.ncbi.nlm.nih.gov/Traces/sra/?study=SRP151073) with the accession number SRP151073. Gene ontology analysis was performed using the Metascape Web site (www.metascape.org) (26).

Total protein was isolated from cultured sebocytes by cell lysis on ice in 1× RIPA buffer (150 mM NaCl, 25 mM Tris-HCl [pH 7.6], 0.1% SDS, 1% sodium deoxycholate, and 1% NP-40) supplemented with 1× protease inhibitor mixture (Roche) followed by incubation in a sonic water bath at 4°C for 30 min. Cellular debris was pelleted by centrifugation at 12,000 × g for 15 min at 4°C. Protein concentration in supernatants was then determined by bicinchoninic acid assay (Pierce Biotechnology/Thermo Fisher Scientific), and protein concentrations were normalized among samples. Equal amounts of protein were then denatured by heating at 100°C for 10 min in 1× Lamelli sample buffer (Bio-Rad Laboratories), separated on 4–20% Mini-Protean TGX gels (Bio-Rad Laboratories), and transferred to PVDF membranes using the Trans-Blot Turbo system (Bio-Rad Laboratories). Membranes were blocked with Odyssey Blocking Buffer (LI-COR Biosciences) for 1 h at room temperature, followed by overnight incubation at 4°C with primary Abs. Membranes were washed three times for 5 min in PBS plus 0.05% Tween-20 and then incubated for 1 h, protected from light, at room temperature with IRDye-labeled species-specific secondary Abs (LI-COR Biosciences). Membranes were subsequently washed with PBS plus 0.05% Tween-20 and PBS, after which membranes were visualized with the Odyssey Infrared Imager (LI-COR Biosciences). Odyssey program software was used to perform densitometry analysis.

Select genes were depleted from cultured sebocytes using Silencer Select small interfering RNA (siRNA) oligonucleotides (Thermo Fisher Scientific, catalog numbers listed in Supplemental Table I). Briefly, sebocytes at 60% confluence in 6-cm tissue culture dishes were treated with siRNA oligonucleotides diluted to a final concentration of 10 nM in 5 ml of a 9:1 Sebomed Basal Medium:Opti-MEM Medium mixture containing 0.25% RNAimax Lipofectamine reagent (Thermo Fisher Scientific). Cells were treated for 18 h with siRNA, after which media was changed and cells were allowed to recover for 24 h. Cells were then collected and plated into appropriate 12-well, 24-well, or 96-well plates for subsequent analysis. Cell treatments occurred within 120 h of original siRNA application, and RT-qPCR measurement of target transcript abundance was measured simultaneously with additional experiments.

P. acnes strain ATCC6919 was maintained in Difco Reinforced Clostridial Medium (BD Biosciences). For stimulation of sebocytes, P. acnes was inoculated into Sebomed Basal Medium (with 10% FBS and 5 ng/ml rhEGF, without antibiotics) with or without the addition of 2% glycerol. Bacteria were cultured in anaerobic conditions using BD GasPak EZ pouches (BD Biosciences) at 37°C for 7 d. Culture supernatants were then sterile filtered and added at 20% vol to Sebomed Basal Medium and used to stimulate sebocytes.

Measurements of HDAC activity in sebocytes in the presence of SCFAs were performed using the Epigenase HDAC Activity/Inhibition Direct Assay Kit (EpiGentek) according to the manufacturer’s instructions. Briefly, nuclear extracts from SEB-1 and SZ95 sebocyte cells were incubated in 8-well strips with an acetylated histone HDAC substrate in the presence or absence of ranging concentrations of propionate, butyrate, or trichostatin A (TSA) at 37°C for 90 min. Test wells were subsequently incubated with capture Ab (room temperature, 60 min), detection Ab (room temperature, 30 min), and detection solution (room temperature, 10 min) with washing of wells between incubation steps. Stop solution was added to test wells after 10 min of incubation with detection solution, and the absorbance of wells in the test plate at 450 nm was measured in a plate reader (Beckman Coulter Diagnostics). Inhibition of HDAC activity by propionate, butyrate, or TSA was calculated using the following formula as directed by the manufacturer:

Sebaceous glands in the skin are exposed to resident microbes and their products and have been previously shown to produce inflammatory cytokines (14). To test the hypothesis that metabolic products of C. acnes (P. acnes) could promote cytokine release from sebocytes, we first evaluated the release of IL-6 and IL-8 from a cultured sebocyte cell line (SEB-1) after exposure to supernatants of P. acnes (ATCC6919) that were grown under hypoxic conditions for 7 d in the presence or absence of glycerol. It has previously been demonstrated that glycerol, a naturally occurring metabolite on human skin that arises from the catalysis of sebaceous gland–derived triglycerides (27), is used by P. acnes under anaerobic conditions and results in the generation of SCFAs (28). The conditioned supernatant from these bacteria was then sterile filtered and added to sebocytes. Exposure of sebocytes to supernatant from P. acnes grown without glycerol led to a modest release of IL-6 and IL-8, but when exposed to bacterial supernatants from P. acnes grown with glycerol, a significant increase in these cytokines was seen (Fig. 1A). Similarly, measurement of mRNA by quantitative PCR for the inflammatory cytokines IL-8, TNF-α, CXCL10, and thymic stromal lymphopoietin (TSLP) demonstrated that sebocytes exposed to bacterial culture supernatant with glycerol had higher levels of gene expression than with exposure to supernatant from cultures grown without glycerol (Fig. 1B). These results did not appear to be a result of increased bacterial load in the glycerol-containing cultures as P. acnes growth and total abundance was similar in both culture conditions (Fig. 1C).

P. acnes cultured with glycerol results in the generation of SCFAs, and the SCFAs propionate, butyrate, and valerate have been shown to inhibit HDAC activity and increase histone acetylation (29–31). Increased histone acetylation can result in variable effects on cytokine expression depending on cell type (10, 32). We therefore next assessed if the crude bacterial supernatant resulted in a change in histone acetylation. Global histone H3 acetylation in sebocytes was increased by propionate but was not affected by stimulation with the TLR2/6 ligand macrophage-activating lipopeptide-2 (MALP-2) (Fig. 1D). Whereas the addition of supernatant from P. acnes grown without glycerol led to a modest increase in global H3 acetylation, supernatant from the glycerol-containing cultures led to a significantly higher increase in H3 acetylation to levels comparable to treatment with propionate (Fig. 1D).

To determine if SCFAs alone could recapitulate the responses observed by sebocytes to the crude P. acnes supernatant, sebocytes were next exposed to a TLR2/6 ligand (MALP-2) in the presence or absence of three purified SCFAs: propionate, butyrate, and valerate. Consistent with our previous findings in human keratinocytes, we saw that SCFAs significantly enhanced the TLR-mediated transcriptional induction of various proinflammatory cytokines, such as IL-1β, IL-8, TNF-α, and CXCL10 from SEB-1 sebocytes (Fig. 2A–D). This enhanced cytokine response in the presence of SCFAs was also seen with an independently derived sebocyte cell line (SZ95) (Fig. 2E–H) (25). Taken together with our previous findings that P. acnes can generate SCFAs under anaerobic, lipid-rich conditions, these results suggest that production of SCFAs by P. acnes increases inflammatory cytokine expression induced by its TLR ligands from sebocytes.

To further assess if SCFAs lead to increased inflammatory response by inhibition of HDACs, we compared responsiveness to propionate to two well-established chemical HDAC inhibitors (TSA and suberoylanilide hydroxamic acid [SAHA]). We observed that both SCFAs and non-SCFA HDAC inhibitors all increased TLR-mediated production of IL-6 and IL-8 from SEB-1 sebocytes (Fig. 3A, 3B). Consistent with their known ability to inhibit HDAC activity, these molecules led to increased global levels of histone acetylation in sebocytes as demonstrated by Western blot analysis of multiple lysine residues of histone H3 (H3K9ac, H3K27ac) (Fig. 3C). Furthermore, we observed an increase in H3K9 acetylation throughout the promoter region of the IL-6 gene when sebocytes were treated with propionate (Fig. 3D), suggesting that the increased TLR-mediated induction of this and other inflammatory genes by HDAC inhibitors may be due to changes in chromatin accessibility. To verify that the observed increase in histone acetylation in sebocytes treated with SCFAs was a direct result of inhibition of HDAC activity, HDAC activity was measured from nuclear extracts of SEB-1 and SZ95 cells in the presence of SCFAs. Both propionate and butyrate were seen to directly inhibit HDAC activity to a comparable degree as the well-established HDAC inhibitor TSA (Supplemental Fig. 1).

Although HDACs are responsible for the removal of acetylation marks in the genome, another class of enzymes, histone acetyltransferases (HATs), are responsible for the addition of these marks. Interestingly, when we stimulated sebocytes with anacardic acid, a chemical known to inhibit the action of HAT enzymes, we observed a decrease in TLR-mediated expression of IL-6 and IL-8 and a complete elimination of the proinflammatory effects of treatment with butyrate (Fig. 3E, 3F). Together, these data demonstrate that sebocytes exhibit a proinflammatory phenotype in response to HDAC inhibition, a finding that contrasts to the anti-inflammatory effects of HDAC inhibitors on bone marrow–derived immune cells like macrophages and dendritic cells.

As SCFAs and other chemical HDAC inhibitors used in these experiments are broad acting and inhibit the activity of numerous HDAC enzymes, we next sought to determine which individual HDACs may be important in the regulation of inflammatory gene expression in sebocytes. First, we characterized the expression of the 11 classical HDAC family members in sebocytes and compared them with expression levels from epidermal keratinocytes (Fig. 4A). HDAC8 and HDAC9 were partially depleted using siRNA-mediated gene silencing (Fig. 4B, Supplemental Fig. 2A–D). This resulted in a significant amplification of MALP-2–mediated induction of IL-1β, IL-6, IL-8, and CXCL10 (Fig. 4C–F). Expression of these cytokines was not increased in HDAC-depleted sebocytes without TLR activation. Furthermore, protein levels of IL-6 and IL-8 as measured by ELISA were also increased in sebocytes with partial silencing of HDAC8 and HDAC9 and treated with MALP-2 (Fig. 4G, 4H). To validate that these findings were indeed specific to depletion of HDAC8 and HDAC9 and not an off-target effect of the silencing oligonucleotides used, we tested additional siRNA oligonucleotides specific for HDAC8 and HDAC9. Consistent with our initial results, partial depletion of HDAC8 and HDAC9 using alternate oligonucleotides significantly enhanced the TLR-mediated induction of IL-6 and IL-8 from SEB-1 sebocytes (Supplemental Fig. 3A–C). These data indicate that HDAC8 and HDAC9 influence inflammatory cytokine expression in sebocytes and suggest that the activity of SCFAs on sebocytes is also mediated by the inhibition of HDACs.

Having identified HDAC8 and HDAC9 as being important in restraining the sebocyte inflammatory response, we sought to use RNA sequencing to evaluate the global transcriptomic effects of depletion of these enzymes. In sebocytes treated with scrambled control siRNA, stimulation with MALP-2 led to an upregulation of 379 genes and a downregulation of 321 genes (Fig. 5A). Gene ontology analysis of the set of genes upregulated by TLR2 activation in these cells identified enrichment of immune- and inflammation-related pathways, such as inflammatory response, cytokine-mediated signaling pathway, and the TNF-signaling pathway, among others (Fig. 5B). Consistent with our hypothesis that HDAC8 and HDAC9 work to restrain inflammatory gene expression in sebocytes, cells depleted of HDAC8 and HDAC9 and stimulated by MALP-2 resulted in the induction of over twice as many genes (790 versus 379, Fig. 5C). Furthermore, a more significant enrichment of pathways involved in inflammatory responses was observed, such as cytokine-mediated signaling pathway, regulation of cell adhesion, and leukocyte migration (Fig. 5D).

Next, we compared the global inflammatory response of sebocytes with or without functional HDAC8 and HDAC9 to the inflammatory signature of an acne lesion. A total of 2382 genes were upregulated in an inflamed acne lesion compared with healthy skin from the same donor. Comparison of this dataset with the list of genes induced by MALP-2 alone in sebocytes (treated with control siRNA) showed a relatively minor overlap of 157 genes (Fig. 5E). However, when an acne lesion was compared with sebocytes lacking HDAC8 and HDAC9 and stimulated by MALP-2, almost twice as many genes overlapped, with 277 genes common between the samples (Fig. 5F). These data demonstrate that when the activity of HDAC8 and HDAC9 are impaired in sebocytes, activation of TLR 2 leads to the induction of many more inflammatory genes that are detected in clinical acne samples.

Finally, in addition to the capacity to inhibit HDAC activity, SCFAs also have the potential to interact with cell surface G protein–coupled receptors (GPCRs). In particular, SCFAs are known ligands of GPR41 (also known as free fatty acid receptor [FFAR] 3), GPR43 (also known as FFAR2), and GPR109a (also known as hydroxycarboxylic acid receptor 2 [HCAR2]) (33, 34). We next assessed if there was a direct response to SCFAs by sebocytes and sought to distinguish this response from the amplification of TLR signaling by the epigenetic mechanism of histone acetylation. Prior observations in keratinocytes showed no detectable response to SCFAs alone, which may be explained by the extremely low expression levels of the three relevant GPCRs by this cell type (10). In contrast, sebocytes express FFAR2, FFAR3, and HCAR2 at relatively high levels, several hundred-fold higher than keratinocytes (Fig. 6A). Furthermore, SZ95 sebocytes treated with SCFAs in the absence of TLR2 ligands resulted in increased expression of the proinflammatory mediator TSLP, an effect that was lost when FFARs were depleted via siRNA-mediated knockdown (Fig. 6B). When FFAR2, FFAR3, or HCAR2 were depleted individually from sebocytes, the SCFA-mediated increase in MALP-2–induced cytokine secretion was partially decreased (Fig. 6C). Meanwhile, when cells were depleted of all three receptors in combination, the proinflammatory effect of SCFAs was almost completely abolished (Fig. 6C). Importantly, this did not occur to the same extent when cells were activated by the non-SCFA HDAC inhibitor TSA, demonstrating that these cells still respond appropriately to inhibition of HDAC activity (Fig. 6D). These data demonstrated that sebocytes, compared with other cell types in the skin, exhibit a unique response to SCFAs, which promote cytokine expression by both activation of GPCRs and inhibition of HDAC activity.

These data show for the first time, to our knowledge, how metabolic products of skin-resident bacteria can induce inflammatory gene expression from sebocytes. Although it has been known that sebocytes can respond to P. acnes through activation of TLR 2, the question of why these bacteria only drive an inflammatory response in the skin in certain settings has remained unanswered. We have previously shown that when cultured in hypoxic conditions and supplied a lipid substrate (an environment reminiscent of a plugged, sebaceous follicle) P. acnes can generate high levels of SCFAs through anaerobic fermentation. Interestingly, these molecules were shown to amplify TLR-driven cytokine responses from epidermal keratinocytes through their inhibition of HDAC function. In this study, we now demonstrate that cytokine expression in sebocytes is also regulated through this epigenetic mechanism and also show that sebocytes have a unique capacity to enhance cytokine expression by direct activation of surface-bound FFARs. Thus, sebocytes in the skin are uniquely sensitive to a change in metabolites by P. acnes and respond to SCFAs by two independent mechanisms.

The proinflammatory effects of HDAC inhibition in both sebocytes and keratinocytes, combined with the similar pattern of expression of the 11 classical HDAC family members observed in the two cell types, suggested that the same HDAC enzymes might control inflammatory gene expression in both keratinocytes and sebocytes. Indeed, specific depletion of HDAC8 or HDAC9, which we had previously demonstrated to control cytokine expression in keratinocytes, resulted in an amplified response to TLR activation in sebocytes. When examined at a global level by RNA sequencing, the number of genes upregulated in response to TLR2/6 activation was nearly doubled when HDAC8 and HDAC9 activity was inhibited. Furthermore, the set of genes induced in sebocytes depleted of HDAC8 and HDAC9 more closely resembled the transcriptome of whole tissue samples from acne, suggesting that inhibition of HDAC8 and HDAC9 in the sebaceous gland by P. acnes–derived SCFAs is modeling aspects of the pathophysiology of the disease. Although the overlap with the sebocyte transcriptome still represents a relatively small portion of the genes identified from a clinical acne lesion, one must consider that the acne biopsy represents a heterogeneous mixture of various cell types from a physiological setting, whereas the sebocyte data represents a single cell type grown in culture. More precise measurements of the sebocyte contribution to an acne lesion (for example, through single cell sequencing of sebocytes from clinical specimens) could further elucidate the contribution of this cell type, and the importance of sebocyte HDAC activity, to acne inflammation in vivo.

Despite the commonalities in regard to HDAC8 and HDAC9, the impact of SCFAs on the inflammatory response of sebocytes and keratinocytes is not completely identical. As this work shows, SCFAs have a unique ability to induce cytokine expression from sebocytes through the activation of FFARs, a finding that was not observed in prior work with keratinocytes. Some proinflammatory mediators, such as TSLP, were induced from sebocytes by treatment with SCFAs alone, even in the absence of TLR ligands. These effects may be due to the high expression of these FFARs in sebocytes, which were detected at several hundred-fold higher levels than in keratinocytes. These findings are interesting when taken together with previous studies demonstrating that free fatty acids present in sebum, such as lauric, palmitic, and oleic acids, can induce antimicrobial peptide expression from sebocytes (35). Together, these findings illustrate how the local follicular microenvironment, consisting of fatty acids from the host, resident microbes, and their metabolic products, can influence inflammatory and antimicrobial processes in the skin. Future studies to characterize the fatty acid profiles of healthy and diseased skin and investigate how different mixtures of fatty acids may further inform the relative role of these metabolites in skin inflammation. Additionally, evaluating the generation of SCFAs and other bioactive metabolites by P. acnes cultured with lipid sources that more closely resemble human sebum, as opposed to simply glycerol, may shed interesting light onto the impacts of microbial metabolism on skin inflammation.

The effects of P. acnes–derived SCFAs on inflammatory gene expression from sebocytes through both epigenetic and receptor-dependent mechanisms described in this study invite the question of whether distinct strains of P. acnes have a differential capacity to generate these molecules. Whereas P. acnes proliferation and overgrowth in the follicle was once considered a driving factor behind the inflammation seen in acne, recent work has demonstrated that the overall abundance of microbes in healthy and affected follicles is actually quite comparable (8). Thus, the current view has shifted toward the idea that distinct strains of P. acnes are unique in their abilities to induce cytokine expression from sebocytes, keratinocytes, and cells of the innate immune system. Our results support speculation that SCFAs from P. acnes will drive cytokine expression prior to rupture of the follicle and may influence both the local pilosebaceous unit as well as surrounding skin. Indeed, characterizations of the P. acnes populations isolated from the skin of acne patients and healthy controls have identified “acne-associated” and “healthy-associated” strains of P. acnes, which exert distinct effects on the immune and inflammatory responses of the skin (9, 36, 37). However, whether the ability to produce SCFAs, or the ratios of SCFAs produced by these different bacteria, contribute to the host response has yet to be investigated.

Overall, this study expands our knowledge into the regulation of inflammatory gene expression in the skin, an organ that must strictly control its capacity to respond to microbes. Although this work was conducted using immortalized cell lines, the observation of the same trends in response to HDAC inhibition, specific depletion of HDAC8 and HDAC9, and depletion of FFARs in two independently derived cell lines supports that these responses are not restricted to a cell line but more likely reflect the sebocyte itself. Replicating these experiments in primary human sebocytes isolated from skin samples would help support the notion that these phenomena may occur in vivo, but even these culture-based studies are not a completely accurate representation of the physiological setting of the pilosebaceous unit. In the future, genetic mouse models can be used to generate tissue-specific deletions of HDAC8 and HDAC9 (from sebocytes, epidermal keratinocytes, or both) to gain a more in-depth understanding of the role of these enzymes in regulating skin inflammation and immunity. However, the lack of appropriate animal models for inflammatory acne vulgaris makes it difficult to specifically link our findings to acne rather than skin inflammation in a broader sense.

In conclusion, this work helps to promote a model of skin inflammation in which changes in the local microenvironment of the pilosebaceous unit drives metabolic shifts in resident skin microbes, resulting in the production of bioactive metabolites that promote the shift from the cutaneous immune system tolerating P. acnes colonization to one that is inflammatory. This shift may be the fundamental event in the pathophysiology of acne vulgaris.

We thank Dr. Diane Thiboutot from Pennsylvania State University Dermatology for providing the SEB-1 sebocyte cell line for use in these experiments. We also thank Dr. Yun “Larry” Tong from the UCSD Department of Dermatology for the acquisition of human samples used in immunohistochemistry and RNA sequencing.

R.L.G. is a consultant and has equity interest in MatriSys Bioscience and Sente Inc. C.C.Z. owns an international patent on the SZ95 sebaceous gland cell line (WO2000046353). The other authors have no financial conflicts of interest.