Introduction
Inflammatory skin disorders such as atopic dermatitis (AD) and psoriasis are among the most prevalent chronic inflammatory diseases and represent a major public health burden (1, 2). They are characterized by epidermal barrier dysfunction, immune cell infiltration, and recurrent inflammation driven by genetic and environmental interactions. Although traditionally viewed as primarily immune-mediated skin conditions, increasing evidence indicates that epidermal dysfunction serves as an initiating event.
Notably, genetic and functional studies in both AD and psoriasis have identified keratinocyte-intrinsic defects in barrier function and stress responses as important initiators of cutaneous inflammation (3–9).
Keratinocytes are the most abundant cell type in the epidermis and play dual roles in barrier formation and immune surveillance. Through terminal differentiation, they generate the stratum corneum (SC), which forms a physical and biochemical barrier that limits water loss and prevents entry of pathogens and allergens (10). In parallel, keratinocytes act as immune sentinels that respond to microbial, environmental, and tissue-derived signals by secreting cytokines, chemokines, and antimicrobial peptides to shape local inflammatory responses (11, 12).
Disruption of these tightly coordinated processes compromises epidermal homeostasis, resulting in abnormal differentiation, cellular stress, and impaired barrier integrity. Specialized programed forms of keratinocyte cell death, such as cornification, are required for normal barrier formation, whereas dysregulated or stress-induced cell death can disrupt barrier integrity and trigger inflammation through the release of alarmins and danger-associated molecular patterns (DAMPs) that activate innate immune sensing pathways (13–15).
Among proteins implicated in epidermal homeostasis and barrier function, TMEM79 (also known as Mattrin) has emerged as an intriguing yet incompletely characterized regulator. TMEM79 is a five-pass transmembrane protein predominantly expressed in keratinocytes (16, 17), and genetic variants are associated with human AD (17). Genetic studies in flaky-tail mice first demonstrated that the matted allele, a spontaneous loss of function mutation in Tmem79, was responsible for the dermatitis and heightened allergen hypersensitivity observed in these mice (16, 17).
Subsequent work showed that global Tmem79-deficient mice developed barrier dysfunction, epidermal hyperplasia, Th17 immune responses, mast cell accumulation, microbial dysbiosis, and itch behaviors resembling human disease (18–21). Cell-specific studies further showed that TMEM79 function in both keratinocytes and sensory neurons contributed to itch and scratching behaviors (20). Tmem79-deficient mice developed dermatitis in a biphasic manner: An early phase arises at 2–6 weeks of age that occurs independently of adaptive immunity and the skin microbiota, followed by a chronic phase driven by skin microbial dysbiosis and type 17 immune responses (18, 19).
Genetic ablation of adaptive immune cells, Il-17a, or antibiotic treatment markedly attenuates the chronic phase with minimal impact on early disease, indicating that the initial phase is driven by epithelial-intrinsic mechanisms. However, the keratinocyte changes and signals that govern this early phase and enable progression to the chronic type 17 inflammation remain undefined. Defining these mechanisms may reveal therapeutic targets during a critical early-life window that determines progression to chronic inflammation.
TMEM79 has been linked functionally to multiple aspects of epidermal biology, playing roles in lamellar granule (LG) secretion and SC formation (16). It shares distant sequence homology with membrane-associated proteins in eicosanoid and glutathione metabolism (MAPEG) family proteins, which are involved in lipid mediator metabolism and detoxification of reactive electrophiles (17, 20). TMEM79 has been shown to regulate oxidative stress and lipid homeostasis, and its loss is associated with increased sebaceous lipid production (17, 18, 20).
TMEM79 has also been proposed to regulate temperature sensation by acting as a negative regulator of the thermosensitive ion channel TRPV3 (22). In addition, we recently identified TMEM79 as a specific antagonist of Wnt/β-catenin signaling that promotes the degradation of the Wnt receptor Frizzled (23).
Despite these advances, how TMEM79 deficiency initiates early-life dermatitis and licenses subsequent immune responses remain unclear. In this study, we utilized keratinocyte-specific Tmem79 knockout mice and TMEM79-deficient human keratinocytes to show that TMEM79 mediates epidermal differentiation and barrier formation. This keratinocyte-intrinsic defect disrupts epidermal differentiation and barrier formation, is associated with ferroptotic lipid stress, and activates STING-dependent type I interferon responses during the initial phase of dermatitis.
Over time, this pathology drives a type 17-dominated immune response, revealing how keratinocyte-intrinsic defects can disrupt epidermal homeostasis and initiate inflammatory skin disease.
Results
Spatial and temporal expression patterns of Tmem79 in keratinocytes
To begin to define the role of TMEM79 in epidermal homeostasis, we first examined its spatial and age-dependent expression in the skin. Ear skin from wild-type (WT) mice was enzymatically separated into epidermal and dermal sheets (SI Appendix, Figure S1A). Enrichment of epidermal markers (K14 and K10) in epidermal sheets and the dermal marker (Col1a1) in dermal sheets validated the effectiveness of separation (SI Appendix, Figure S1B).
Quantitative PCR (qPCR) analysis showed that Tmem79 was predominantly expressed in the epidermis (Figure 1A). To identify the cellular origin of this signal, we analyzed a published single-cell RNA-sequencing (scRNA-seq) dataset of mouse adult dorsal skin in telogen (24). Tmem79 expression was enriched in Dkkl1+ suprabasal keratinocytes relative to basal and hair follicle-associated keratinocytes (SI Appendix, Figure S1C and Figure 1B).
To examine age-dependent changes in Tmem79 expression from 2 weeks of age into adulthood, we performed qPCR analysis on WT epidermal sheets collected across developmental stages. Tmem79 expression remained relatively stable between 2 and 7 weeks of age (Figure 1C). To complement these observations made in mice, we cultured primary human keratinocytes under conditions that either maintained a proliferative basal-like state or induced differentiation with high calcium over 1, 3, or 7 days (SI Appendix, Figure S1D).
Progressive keratinocyte differentiation was confirmed by time-dependent enrichment of both early/intermediate (K10 and IVL) and terminal differentiation markers (LOR and FLG) (Figure 1D). TMEM79 expression levels increased concomitantly with this keratinocyte differentiation (Figure 1D). Consistent with these data, scRNA-seq data generated from human skin organoids showed enrichment of TMEM79 in differentiated suprabasal keratinocytes (SI Appendix, Figure S1E) (25).
Overall, these results indicate that TMEM79 is expressed primarily in suprabasal keratinocytes across human and mouse skin and shows stable expression during early postnatal development.
Keratinocyte-specific Tmem79 loss disrupts skin barrier homeostasis
To investigate the specific role of TMEM79 in keratinocytes, we compared global knockout (KO) and K14-Cre keratinocyte-specific conditional knockout (cKO) mice. qPCR analysis of whole-skin and epidermal samples confirmed complete deletion of Tmem79 in whole skin from global KO mice and near-complete deletion in the epidermis of cKO mice, whereas low residual expression persisted in cKO whole skin (SI Appendix, Figure S2A and B), indicating low levels of Tmem79 expression in nonkeratinocyte cell populations in the skin.
To assess the impact of Tmem79 loss on skin structure and barrier function, we performed histological analysis of dorsal skin architecture and transepidermal water loss (TEWL) over time. Epidermal hyperplasia and barrier dysfunction as indicated by higher TEWL measurements were apparent as early as 2 weeks of age in KO mice, with the most pronounced changes observed at 4 weeks (Figure 1E–G). Notably, keratinocyte-specific deletion of Tmem79 recapitulated the inflammatory skin phenotype observed in global KO mice (Figure 1H–J), indicating that pathology is primarily driven by loss of epithelial Tmem79.
Based on these findings, we focused subsequent analyses on two key timepoints: (1) 4 weeks of age, corresponding to peak pathology, and (2) earlier than 2 weeks of age to examine disease initiation. Notably, epidermal hyperplasia spontaneously resolved in global KO mice by 6-7 weeks of age (Figure 1E), whereas it persisted in cKO mice (Figure 1H). In contrast, barrier dysfunction, as measured by TEWL, remained elevated through 16 weeks of age in both mouse lines (Figure 1G and J).
These data indicate that Tmem79 expression in nonkeratinocytes appears to contribute to perturbed epidermal structure rather than its restoration.
Tmem79 loss induces hyperproliferation and impairs epithelial differentiation
To better understand epidermal defects observed in 4-week-old Tmem79 cKO mice, we performed transmission electron microscopy (TEM). Consistent with histological findings, cKO skin showed a marked increase in epidermal thickness (SI Appendix, Figure S2C). Notably, the SC, the outermost layer that forms the primary skin barrier, was significantly expanded (Figure 2A).
Although SC thickening does not necessarily indicate barrier failure, it has been associated with reduced barrier quality and increased TEWL in conditions such as cosmetic dry skin, ageing, and ichthyosis (26–29).
Because the SC represents the terminal stage of epidermal differentiation and TMEM79 is mostly expressed in suprabasal keratinocytes, we next investigated whether differentiation was altered in cKO mice. Immunostaining for Ki67+ cells revealed a significant increase in proliferating cells within the cKO epidermis (Figure 2B). Whereas control skin showed proper spatial organization, with basal keratin 14 (K14) and suprabasal keratin 10 (K10) segregated into distinct layers, cKO mice showed co-expression of these markers, indicative of aberrant differentiation (Figure 2C).
This pattern likely reflects abnormal transitional states between basal stem cells and differentiated keratinocytes (30, 31). In addition, the injury- and stress-associated marker K6A, which is normally absent from the interfollicular epidermis (IFE) (32, 33), was ectopically induced in the cKO epidermis (Figure 2D). These findings indicate that Tmem79 deletion promotes stem cell proliferation and disrupted keratinocyte differentiation, likely contributing to the observed epidermal hyperplasia and barrier dysfunction.
By 4 weeks of age, global and cKO mice exhibited clear dorsal skin hair loss (SI Appendix, Figure S2D and E). Because hair follicle abnormalities and hair loss can secondarily influence epidermal architecture (34, 35), we sought to exclude the possibility that hair loss contributes to the observed epidermal hyperplasia. We therefore performed histological analyses of sparsely haired regions (ear and tail) as well as hairless plantar skin to determine whether hyperplasia occurs independently of hair follicles.
H&E staining revealed a marked increase in epidermal thickness in all three regions (Figure 2E and SI Appendix, Figure S2F), ruling out a direct causative role for hair loss in driving epidermal hyperplasia.
Interestingly, within the granular layer, which is characterized by the presence of keratohyalin granules, nuclei in cKO mice remained round rather than elongated (Figure 2E). In normal epidermis, nuclear elongation, condensation, and enucleation accompanying terminal differentiation are required to establish the skin barrier (36, 37). The persistence of rounded nuclei in the granular layer of cKO mice suggests impaired nuclear remodeling and defective terminal differentiation.
At 4 weeks of age, keratinocyte-specific Tmem79 deletion also caused histological features of dermal immune cell infiltration (Figure 1I). To better understand the cellular changes in the skin microenvironment, we performed scRNA-seq of dorsal skin from 4-week-old control and cKO mice (SI Appendix, Figure S3A). This identified 22 distinct cell populations, with Tmem79 cKO skin showing increased proportions of γδ T cells and fibroblasts (SI Appendix, Figure S3B and Figure 2F–G), consistent with immune activation and remodeling of the skin microenvironment.
To understand keratinocyte-intrinsic transcriptional changes, we performed Kyoto Encyclopedia of Genes and Genomes (KEGG) pathway and Gene Ontology (GO) analysis on keratinocyte-specific differentially expressed genes (DEGs) from the scRNA-seq dataset between Tmem79 cKO and control mice. KEGG analysis revealed enrichment of focal adhesion and cell cycle pathways, with the cell cycle pathway supported by increased expression of Mcm2, Mcm3, Mcm5, Mcm6, Mcm7, Cdt1, Cdk1, Cdk4, Ccna2, Bub1, Cdc20, Cdc14a, and Mki67 (Figure 2H). GO analysis showed enrichment of pathways related to actin cytoskeleton organization, epithelial morphogenesis, and cell junction organization (SI Appendix, Figure S3C), consistent with impaired epithelial differentiation and barrier dysfunction.
In contrast, genes involved in immune and defense responses were downregulated in cKO keratinocytes, including Ccl27a, Cxcl10, B2m, H2-K1, H2-D1, Tap1, Tap2, Defb1, Defb6, Pglyrp4, Myd88, Ripk2, and Ly96 (Figure 2I). Consistently, KEGG analysis showed reduced enrichment of pathways related to antigen processing and presentation and virus infection–related pathways (SI Appendix, Figure S3D).
Together, these data indicate that Tmem79 deficiency reprograms keratinocytes toward a hyperproliferative and structurally remodeled state while compromising epithelial immune defense programs.
TMEM79 loss impairs human keratinocyte differentiation and barrier formation
To assess whether the keratinocyte-intrinsic consequences of TMEM79 loss are conserved in a human system, we used N/TERT-1 human keratinocytes, which are capable of full epidermal differentiation and are amenable to genetic editing. We generated three independent TMEM79 knockout and control clones using CRISPR/Cas9. Biallelic disruption was confirmed using Sanger sequencing (SI Appendix, Figure S4) and TMEM79 protein loss by western blot analysis (SI Appendix, Figure S5A).
To define transcriptional programs associated with TMEM79 deficiency and characterize the knockout lines, we performed bulk RNA-seq under two conditions: a proliferative basal-like state (nonconfluent, 0.3 mM CaCl2) and a differentiative state (3 days at confluence, 1.2 mM CaCl2). Upon differentiation, TMEM79-deficient cells downregulated genes associated with keratinocyte differentiation and keratinization, including ZNF750, SPRR1A, SPRR2D, SPRR2E, LCE3D, LCE3E, CDSN, CNFN, ALOXE3, ALOX12B (SI Appendix, Figure S5B–D), indicating intrinsic deficits in differentiation.
We next asked whether impaired differentiation translated into reduced barrier formation. Control and TMEM79-null clones were differentiated in air–liquid interface (ALI) culture to form 3D human epidermal equivalents (HEEs) (38). Transepithelial electrical resistance (TEER) was measured as a quantitative, noninvasive readout of barrier integrity (38, 39).
After 10 days of ALI culture, TMEM79-deficient HEEs exhibited a 53% reduction in TEER relative to controls (SI Appendix, Figure S5E). However, EdU incorporation assays showed no differences in proliferation between TMEM79-null and control N/TERT-1 cells (SI Appendix, Figure S5F). These data indicate that TMEM79 deficiency impairs human keratinocyte differentiation and barrier formation without increasing proliferation.
We previously showed that TMEM79 antagonizes Wnt signaling by promoting Frizzled degradation (23). Consistent with this function, TMEM79 loss in human keratinocytes significantly increased cytosolic active β-catenin levels (SI Appendix, Figure S6A) and expression of the Wnt target gene AXIN2 (SI Appendix, Figure S6B), which we observed across independent clones. These findings suggest that Wnt/β-catenin dysregulation may contribute to the differentiation defects associated with TMEM79 deficiency.
TMEM79 deficiency causes decreased viability and ferroptotic stress in keratinocytes
Although TMEM79-null N/TERT-1 clones displayed similar proliferative capacity (SI Appendix, Figure S5F), they consistently yielded fewer cells upon passaging, suggesting a defect in cell survival. Transcriptional analysis of control and TMEM79 KO keratinocytes under basal-like conditions showed upregulation of genes related to cell death pathways (Figure 3A and B), including RBCK1, PNMA1, PDCD4, PHLDA1, EGR1, TXNIP, FOS, and PTGS2 (Figure 3C). Consistent with this, the cell counting kit-8 (CCK8) assay and clonogenic potential assays confirmed reduced viability and colony-forming potential in TMEM79-deficient clones (Figure 3D and E).
Increased keratinocyte cell death was also observed in the epidermis of Tmem79 cKO mice by TEM and terminal deoxynucleotidyl transferase dUTP nick end labeling (TUNEL) staining (SI Appendix, Figure S7A and B). To quantify cell death in the human keratinocytes, we measured lactate dehydrogenase (LDH) release as a readout of membrane integrity and found that TMEM79 KO cells released significantly higher LDH levels than controls (Figure 3F).
To define which cell death pathway contributes to this phenotype, we treated TMEM79 KO cells with pharmacological inhibitors that block apoptosis (Z-VAD-FMK), necroptosis (Necrostatin-1s), or ferroptosis (Ferrostatin-1 or Fer-1). Only Fer-1 significantly improved viability in TMEM79-null cells, implicating ferroptosis in keratinocyte death (Figure 3G). We next asked whether ferroptosis inhibition restores functional viability.
We evaluated the ferroptosis inhibitors Fer-1 and Liproxstatin-1 (Lip-1) in viability and clonogenic assays. Fer-1 rescued cell viability in TMEM79-deficient keratinocytes in a dose-dependent manner (Figure 3H). Both inhibitors enhanced colony-forming capacity and reduced LDH release from TMEM79-null cells to near-control levels (Figure 3I and J).
These data identify ferroptosis as a major contributor to cell death caused by TMEM79 deficiency. Because ferroptosis is driven by iron-dependent lipid peroxidation (40), we next measured lipid oxidation using the BODIPY-C11 probe, which shifts fluorescence from red to green upon oxidation (41). TMEM79-deficient keratinocytes exhibited markedly increased oxidized BODIPY-C11 fluorescence relative to controls, reaching levels comparable to direct induction of ferroptosis in control cells with the GPX4 inhibitor RSL3 (Figure 3K).
Treatment with Fer-1 or Lip-1 significantly reduced oxidized BODIPY-C11 levels in TMEM79-null cells, supporting ferroptosis-associated lipid peroxidation as a driver of cell death caused by TMEM79 deficiency (Figure 3L).
To determine whether lipid peroxidation was also present in vivo, we evaluated epidermal sheets from 4-week-old Tmem79 cKO and control mice using oxidized BODIPY-C11. Tmem79 cKO epidermis displayed increased lipid peroxidation relative to controls (Figure 3M). Consistent with this observation, keratinocytes from Tmem79 cKO mice showed increased expression of genes associated with ferroptosis and lipid metabolism, including Fbxl5, Ftl1, Slc39a14, Pla2g7, Acsl3, Lpin2, and Lclat1 (SI Appendix, Figure S7C).
Therefore, ferroptotic stress is a major component of the keratinocyte response to TMEM79 loss in both human cells and the mouse model.
Type 17 immunity drives epidermal hyperplasia
Having established that TMEM79 loss impairs keratinocyte differentiation and barrier formation, we next asked whether epithelial dysfunction leads to immune responses that amplify skin pathology. Whole-skin scRNA-seq analysis showed a marked increase in γδ T cells at 4 weeks of age, the peak of disease in Tmem79 cKO mice (Figure 2F and G). To define this immune response in greater depth, we sorted CD45+ immune cells and performed scRNA-seq to profile immune populations that are underrepresented in whole-skin datasets.
Single-cell analysis identified 16 distinct CD45+ immune populations, with cKO mice showing increased proportions of γδ T cells, neutrophils, and mast cells (SI Appendix, Figure S8A and B and Figure 4A). Neutrophil infiltration was confirmed by flow cytometry (SI Appendix, Figure S8C). Immunostaining for the neutrophil marker myeloperoxidase (MPO) showed that neutrophils were predominantly located in the lower dermis in control skin but were positioned closer to the epidermis in cKO skin (SI Appendix, Figure S8D).
This positioning is consistent with increased expression of neutrophil-recruiting chemokines by cKO keratinocytes, such as CXCL1, as detected by both scRNA-seq and qPCR (SI Appendix, Figure S8E and F). Mast cell infiltration in cKO mice was confirmed by toluidine blue staining (SI Appendix, Figure S8G), consistent with a previous report that mast cells are increased and mediate histaminergic itch in Tmem79 KO mice (20).
Flow cytometry confirmed the increase in dermal γδ T cells in cKO skin (Figure 4B). We next performed pseudobulk differential expression analysis to define their transcriptional state. cKO γδ T cells displayed a strongly activated and inflammatory profile. GO analysis of DEGs revealed enrichment of pathways related to leukocyte and lymphocyte activation, supported by increased expression of Cd83, Cd84, Cd226, Slamf1, Tyk2, Stat6, Nfkbiz, Tnfaip3, Cd69, Il2ra, Zap70, Lat, Jak3, Stat5b, and Pdcd1 (Figure 4C).
KEGG analysis further indicated signs of γδ-Th17 cell differentiation, with upregulation of Il17a, Rorc, Rora, Il23r, and Hif1a (Figure 4D and E), along with enrichment of NF-κB and TNF signaling pathways (Figure 4D). In contrast, downregulated DEGs point to a suppression of cytoplasmic translation, ribosome function, and natural killer cell–mediated cytotoxicity, suggesting a shift away from a homeostatic and cytotoxic program (SI Appendix, Figure S8H and I). Together, these findings indicate that loss of Tmem79 in keratinocytes drives γδ T cells toward an activated, IL-17–skewed effector state.
The expansion of γδ T cells in cKO skin is likely not a result of local proliferation, as immunofluorescence showed distinct γδ TCR and Ki67 signals (SI Appendix, Figure S8J). This suggests a potential recruitment of γδ T cells, likely from the lymph nodes, consistent with increased frequencies of γδ T cells in skin-draining lymph nodes of cKO mice (SI Appendix, Figure S8K). Analysis of skin cytokine levels revealed significant elevation of Il-17a, unchanged type 2 cytokines (Il-4, Il-13), and a nonsignificant downward trend in IFN-γ (P = 0.15) (Figure 4F).
Serum IL-17A was also increased (Figure 4G). These findings indicate that 4-week-old cKO mice show a type 17-dominant immune response, consistent with the γδ T cell transcriptional signatures (Figure 4C–E).
To test whether suppressing type 17 immunity could ameliorate TMEM79-associated phenotypes, we generated Tmem79 KO; Il17a KO mice (Figure 4H). These double KO mice exhibited a marked reduction in epidermal hyperplasia and dermal γδ T cell infiltration to levels comparable to control mice (Figure 4I and J). Although skin barrier dysfunction was significantly improved, it was not fully restored to baseline (Figure 4K).
These findings, together with data from TMEM79 KO human keratinocytes (SI Appendix, Figure S5), indicate that early skin barrier dysfunction is largely driven by keratinocyte-intrinsic factors, whereas epidermal hyperplasia is primarily driven by IL-17A dependent immune responses that further exacerbate barrier dysfunction.
Early IFN signaling and keratinocyte changes precede type 17 inflammation
Because IL-17A contributed mainly to peak disease at 4 weeks of age, we next asked whether key earlier epithelial and immune-related programs emerge before overt skin pathology. To define these initiating events, we examined epithelial abnormalities in the epidermis at time points prior to the scaling, hyperplasia and skin barrier dysfunction observed at 2 weeks of age (Figure 5A–D). We performed longitudinal bulk RNA-seq profiling of epidermal sheets isolated from control and cKO mice at P1 (postnatal day 1), P3, P6, and P14 (Figure 5E).
Aside from Tmem79, no DEGs were observed at P1, indicating that cutaneous pathology develops postnatally rather than embryonically. The number of DEGs then increased from 13 at P3 to 136 at P6 and to 113 at P14 (Figure 5E).
At P3, beyond Tmem79, the other 12 DEGs remain poorly characterized in the context of skin biology or pathology (SI Appendix, Figure S9A). At P6 and P14, downregulated genes were significantly enriched for processes related to epidermal development and keratinization (SI Appendix, Figure S9B and C), consistent with the differentiation and barrier defects observed in TMEM79-deficient human keratinocytes and adult cKO skin. In parallel, genes upregulated at P6 were enriched for innate immune response, including early induction of IFN-stimulated genes such as Gvin1, Ddx60, Ifit1, Oasl2, suggesting the initial activation of type I IFN signaling (Figure 5F and G).
This pattern became more pronounced at P14, with the emergence of a robust type I IFN response signature characterized by genes such as Ifi202b, Oasl2, Usp18, Ifit1, Aim2, Ddx60, and Xaf1, which we further validated by qPCR (Figure 5H–J). Together, this temporal profiling indicates that Tmem79 loss induces early epidermal differentiation defects together with a type I IFN response, preceding the later type 17 inflammatory response.
To define the cellular basis of these early changes, we performed scRNA-seq on dissociated skin collected at P6, an early time point corresponding to the emergence of numerous DEGs in cKO epidermal sheets (Figure 5E). We observed an increased proportion of proliferating interfollicular epidermal (IFE) cells in cKO skin (SI Appendix, Figure S10A and B). Differential expression analysis across keratinocytes from cKO and control skin revealed a strong induction of cell cycle and mitosis-related genes like Top2a, Mki67, and Wee1, consistent with heightened proliferative activity (SI Appendix, Figure S10C).
Inflammatory and stress–response genes such as Tnfaip3 and Nod2 were also upregulated, reflecting an inflammatory-activated state (SI Appendix, Figure S10C). In contrast, genes involved in epidermal differentiation (Krt10, Krt77, Ivl) and regulators of epidermal homeostasis and barrier integrity (Zfp36, Igfbp3) were significantly downregulated in P6 keratinocytes (SI Appendix, Figure S10C). The increased proliferative activity and impaired epidermal differentiation in cKO IFE are consistent with our earlier observations during peak disease (Figure 2).
Langerhans cells are dispensable for TMEM79-associated skin pathology
Beyond keratinocytes, P6 cKO skin showed a nonsignificant trend toward increased Langerhans cells (LCs) and dendritic epidermal T cells (DETCs), which becomes more pronounced at peak disease at 4 weeks of age (SI Appendix, Figure S10B and S10D). We found that the inflammatory genes upregulated in P6 cKO epidermal sheets (Figure 5G) were predominantly expressed by LCs and not DETCs nor keratinocytes (SI Appendix, Figure S10E). These genes (Lyz2, C1qc, C1qb, Cybb, and C1qa) are associated with phagocytic activity, indicating potentially enhanced LC activation (42–44).
LCs can migrate to draining lymph nodes, where they present antigens and promote immune activation, including type 17 responses (45, 46). DETCs showed elevated expression of Tnf, Egr1, Il13 and Ifng in P6 cKO skin (SI Appendix, Figure S10F). However, previous work showed that DETC-deficient Tcrd KO mice did not impact initial-phase dermatitis (<7 weeks) (19).
We therefore investigated whether LC activation may serve as an upstream trigger of skin pathology or the immune response in Tmem79 cKO mice. To test this, we crossed Tmem79 cKO mice with Langerin-diphtheria toxin A (DTA) mice to ablate LCs (SI Appendix, Figure S10G and H). LC depletion did not affect epidermal hyperplasia, barrier dysfunction, or the type 17 immune response in cKO mice (SI Appendix, Figure S10I–K).
These results indicate that while we see changes in LCs, their activation is not a direct driver of the pathological features of Tmem79 cKO mice. Instead, they support a model in which immune activation arises as a consequence of keratinocyte-intrinsic changes and epithelial stress.
STING-dependent signaling promotes type 17 inflammation and skin pathology
The progressive emergence of a type I IFN gene signature in the epidermis from P6 to P14 (Figure 5F–J) suggested that this pathway may contribute to the later development of TMEM79-associated skin pathologies. To test this, we crossed Tmem79 cKO mice with Ifnar1 KO mice, which lack functional type I IFN receptors (Figure 6A). Tmem79 cKO; Ifnar1 KO mice exhibited reduced epidermal hyperplasia, improved barrier integrity, and a diminished type 17 immune response, as evidenced by decreased γδ T cell infiltration and lower Il17a expression (Figure 6B and C and SI Appendix, Figure S11A and B).
These findings support a role for early type I IFN signaling in driving TMEM79-associated skin and immune pathologies.
Type I IFNs can be induced downstream of multiple nucleic acid sensors, including the cGAS/STING pathway, which detects intracellular DNA, and the RIG-I/MAVS pathway, which senses double-stranded RNA (47–49). To identify the key molecular sensors upstream of type I IFN induction in Tmem79 cKO skin, we crossed Tmem79 cKO mice with Sting1 KO or Mavs KO mice. Tmem79 cKO; Sting1 KO mice exhibited reduced epidermal hyperplasia, improved barrier function, and diminished type 17 responses (Figure 6D–F and SI Appendix, Figure S11C and D).
By contrast, loss of Mavs did not alter these phenotypes in Tmem79 cKO mice (Figure 6G–I and SI Appendix, Figure S11E).
Together, these findings indicate that a STING-mediated, MAVS-independent mechanism drives a type I IFN response during early life and underlies the development of persistent epidermal hyperplasia, barrier dysfunction, and type 17 immune activation observed in adult Tmem79 cKO mice.
Overall, these findings indicate that Tmem79 cKO mice spontaneously develop a STING-dependent response that promotes a subsequent type 17 immune response. TMEM79 deficiency in human keratinocytes captures several epithelial-intrinsic features of Tmem79 cKO skin, including impaired differentiation, defective barrier formation, reduced cell viability, and ferroptotic lipid peroxidation. However, TMEM79-deficient human keratinocytes did not exhibit increased proliferation, and we did not detect a type I IFN signature or STING activation in vitro.
Thus, while TMEM79 loss is sufficient to induce epithelial stress responses in human keratinocytes, epidermal hyperplasia and inflammatory activation in vivo likely require additional tissue-level cues.
Discussion
Our study identifies TMEM79 as a keratinocyte-intrinsic regulator that integrates epidermal differentiation, barrier formation, and epithelial stress responses. By combining keratinocyte-specific genetic ablation in mice with human keratinocyte models, we show that loss of TMEM79 disrupts epidermal differentiation and barrier formation, induces ferroptotic lipid peroxidation, and is associated in vivo with early STING-dependent type I interferon signaling. This epithelial-intrinsic stress program precedes the emergence of barrier dysfunction and a type 17 immune response, which in turn drives epidermal hyperplasia (SI Appendix, Figure S12).
Together, our findings indicate that keratinocyte-intrinsic defects in barrier formation and lipid stress are associated with innate immune activation and progression of inflammatory skin disease.
Previous studies have demonstrated that mutations or global deletion of Tmem79 lead to spontaneous dermatitis (16, 17, 19). Work from the Julius laboratory dissected tissue-specific contributions of TMEM79, showing that keratinocyte-specific loss was sufficient to induce skin inflammation, as evidenced by increased CD45+ immune cell infiltration and mast cell accumulation in the dermis (20). Work from the Schwartz group further demonstrated that K14-Cre; Tmem79fl/fl mice develop both spontaneous skin and lung inflammation (21).
Consistent with these findings, our keratinocyte-specific deletion experiments show that epidermal loss of TMEM79 is sufficient to recapitulate major pathological features observed in global knockout mice, including epidermal hyperplasia, barrier dysfunction, and type 17 immune activation. Notably, epidermal hyperplasia persists in keratinocyte-specific knockout mice even as it partially resolves in global knockouts, suggesting that Tmem79 in nonepithelial compartments contributes to disease progression. Collectively, these studies support an epithelial origin for TMEM79-associated pathology, but the keratinocyte-intrinsic mechanisms that initiate epithelial stress and license downstream immune activation remained unresolved.
Our study addresses this gap by defining epithelial differentiation defects, Wnt/β-catenin dysregulation, ferroptotic lipid stress, and STING-dependent type I IFN signaling as mechanisms linking TMEM79 deficiency to inflammatory skin pathology. In vivo, Tmem79 cKO keratinocytes displayed hyperproliferation, aberrant K14/K10 organization and impaired terminal differentiation. However, TMEM79-deficient human keratinocytes exhibited no differences in proliferation, indicating that hyperproliferation is not a cell-intrinsic consequence of TMEM79 loss.
Instead, our data suggest that TMEM79 plays a primary role in keratinocyte differentiation and barrier formation, whereas epidermal hyperplasia is driven by the inflammatory tissue environment. This interpretation is supported by the marked reduction in epidermal thickness in Tmem79 KO; Il17a KO mice, consistent with prior studies showing that IL-17 acts on keratinocytes to induce proliferative and inflammatory programs in psoriasis (50). The barrier defect is evident both in vivo, where Tmem79 cKO mice showed increased TEWL, and in vitro, where TMEM79-deficient HEEs show reduced TEER.
TMEM79 expression also increased during keratinocyte differentiation, supporting a role in late-stage epidermal maturation rather than basal stem cell maintenance.
Temporal transcriptomic profiling of Tmem79 cKO mice revealed increased type I IFN signaling in the epidermis at early postnatal ages. Genetic ablation of Ifnar1 in Tmem79 cKO mice substantially rescued epidermal hyperplasia, improved barrier function, and reduced type 17 immune responses. Type I IFN signaling has been extensively implicated in human inflammatory skin diseases, including psoriasis and cutaneous lupus, where IFN-inducible gene signatures are strongly upregulated in lesional skin and contribute to pathogenic immune activation (51–55).
Together, these findings position type I IFN signaling as a critical molecular bridge linking epithelial stress to the development of chronic inflammation.
Mechanistically, we identify STING as the dominant innate sensor driving skin inflammation in Tmem79 cKO mice. Although induction of type I IFN responses is a major outcome of STING activation, STING can also promote proinflammatory cytokine production through type I IFN–independent mechanisms, including NF-κB signaling (48, 56). The diminished type 17 immunity in Tmem79 cKO; Sting1 KO mice is more pronounced, nearly comparable to control mice, than in Tmem79 cKO; Ifnar1 KO mice, suggesting that STING functions through both type I IFN–dependent and type I IFN–independent proinflammatory mechanisms.
Genetic deletion of Sting1, but not the RNA-sensing adaptor Mavs, attenuated epidermal pathology and immune activation, indicating that cytosolic DNA sensing rather than MAVS-dependent RNA sensing is the dominant innate sensing pathway in this model. Although the precise source of immunostimulatory DNA remains to be defined, our data establish a STING-mediated, MAVS-independent pathway as the central driver of early IFN responses. Consistent with this framework, the cGAS–STING pathway has emerged as a critical mediator of cutaneous inflammation in experimental models of lupus and psoriasis (57–59).
How TMEM79 loss engages STING activation remains unclear. One consequence of TMEM79 deficiency we found in vitro is ferroptotic lipid stress. Ferroptosis is a regulated form of cell death driven by iron-dependent lipid peroxidation (40, 60), and has recently been implicated in inflammatory diseases (61, 62).
In our study, human TMEM79-null keratinocytes exhibit increased lipid peroxidation and reduced viability, both reversed by ferroptosis inhibitors, while Tmem79 cKO mice display increased lipid droplet accumulation (Figure S2C), enrichment of ferroptosis-associated pathways in keratinocytes, and elevated lipid peroxidation. These findings are mechanistically consistent with prior studies implicating TMEM79 in lipid biology and oxidative stress regulation (17, 18, 20). Ferroptotic lipid peroxidation has been shown in other contexts to activate cGAS–STING and induce type I IFN signaling, which in turn can reinforce ferroptotic susceptibility, establishing a bidirectional loop (63–66).
While our findings are consistent with this framework, they do not establish the hierarchy or causal relationship of ferroptosis and STING activation in vivo. Moreover, TMEM79-deficient keratinocytes do not autonomously mount a clear STING/type I IFN response in vitro. Ferroptotic stress alone may be insufficient to activate the full STING-dependent inflammatory program outside the tissue environment.
Instead, TMEM79 loss may create a permissive epithelial stress state that, together with barrier disruption, accumulated cell damage, microbial or environmental exposure, and immune cell interactions, engages STING-dependent type I IFN signaling in vivo.
Together, our findings support a model in which keratinocyte-intrinsic TMEM79 deficiency compromises keratinocyte differentiation and barrier integrity, induces ferroptotic lipid stress, and creates a tissue environment permissive for STING-dependent type I IFN signaling. This early barrier defect and inflammatory milieu, together with previously reported skin microbial dysbiosis (19), facilitates type 17 inflammation, which drives subsequent epithelial hyperplasia and chronic skin pathology. A limitation of the present study is that the causal relationship between ferroptosis and STING activation could not be directly tested at early postnatal time points.
Nonetheless, our findings indicate that TMEM79-dependent epithelial homeostasis and innate immune sensing are determinants of inflammatory skin disease, and suggest that restoring TMEM79 function or epithelial differentiation, or targeting downstream ferroptotic lipid peroxidation and STING activation, may help prevent the transition from epithelial stress to chronic immune dysregulation. Given the emerging links among barrier dysfunction, ferroptosis, and innate immunity across epithelial tissues, this TMEM79-associated epithelial stress pathway may represent a generalizable mechanism by which epithelial barrier defects evolve into chronic inflammatory disease.
Materials and methods
All materials and methods used in this study are detailed in the SI Appendix, Materials and Methods, including mice, quantitative RT-PCR analysis, histology and immunostaining, TEWL, epidermal and dermal sheet separation, preparation of single-cell suspensions from skin tissue, scRNA-seq, data processing, quality control, and annotation, differential gene expression analysis, bulk RNA-seq analysis of mouse epidermal sheets, flow cytometry, TEM, enzyme-linked immunosorbent assay (ELISA), cell culture and chemicals, generation of N/TERT-1 TMEM79 knockout lines, 2D differentiation of N/TERT-1 cells, TEER assay, transcriptomic analysis of human keratinocytes, LDH cytotoxicity assay, CCK-8 assay, lipid peroxidation assay (BODIPY-C11), clonogenic assay, EdU proliferation assay, and statistics. All animal experiments were approved by the Institutional Animal Care and Use Committee (IACUC) of Boston Children's Hospital (Protocol No. 00002200) and conducted in accordance with institutional guidelines and regulations.
Supplementary Material