Abstract

   Disrupted N^6-methyladenosine (m^6A) modification modulates various
   inflammatory disorders. However, the role of m^6A in regulating
   cutaneous inflammation remains elusive. Here, we reveal that the m^6A
   and its methyltransferase METTL3 are down-regulated in keratinocytes in
   inflammatory skin diseases. Inducible deletion of Mettl3 in murine
   keratinocytes results in spontaneous skin inflammation and increases
   susceptibility to cutaneous inflammation with activation of neutrophil
   recruitment. Therapeutically, restoration of m^6A alleviates the
   disease phenotypes in mice and suppresses inflammation in human biopsy
   specimens. We support a model in which m^6A modification stabilizes the
   mRNA of the lipid-metabolizing enzyme ELOVL6 via the m^6A reader
   IGF2BP3, leading to a rewiring of fatty acid metabolism with a
   reduction in palmitic acid accumulation and, consequently, suppressing
   neutrophil chemotaxis in cutaneous inflammation. Our findings highlight
   a previously unrecognized epithelial-intrinsic m^6A modification–lipid
   metabolism pathway that is essential for maintaining epidermal and
   immune homeostasis and lay the basis for potential therapeutic
   targeting of m^6A modulators to attenuate inflammatory skin diseases.
     __________________________________________________________________

   Disrupted epithelial m^6A modification–lipid metabolism pathway drives
   skin inflammation via enhancing neutrophil chemotaxis.

INTRODUCTION

   RNA modification plays a pivotal role in posttranscriptional regulation
   and notably influences the fate and function of RNA molecules ([86]1).
   Among various forms of RNA modifications identified to date,
   N^6-methyladenosine (m^6A) is the most abundant and prevalent
   modification on mRNA that profoundly affects RNA fate-related
   decisions, such as its splicing, decay, stabilization, and translation
   ([87]2). m^6A modification is catalyzed by methyltransferases (which
   are referred to as writers), removed by demethyltransferases (erasers),
   and recognized by m^6A-binding proteins (readers). Among all the known
   RNA methyltransferases, METTL3 is considered as the only
   S-adenosylmethionine–dependent methyltransferase that has catalytic
   effects on mRNA in mammals ([88]3).

   Numerous studies have elucidated the physiological and pathological
   roles of METTL3-mediated m^6A modification in specific cell types
   ([89]4). Increasing evidence underscores the pivotal and diverse roles
   of m^6A modification and its associated biochemical machinery in
   regulating various types of immune responses. For instance, inhibiting
   METTL3 enhances T cell–mediated cancer killing through activation of a
   cell-intrinsic interferon response ([90]5). In renal tubular epithelial
   cells, the lack of METTL3-mediated TAB3 m^6A modification leads to
   increased renal inflammation ([91]6). Reduced m^6A modification
   resulting from reduced METTL3 expression in monocyte macrophages
   enhances T helper 2 (T[H]2) cell responses and exacerbates allergic
   airway inflammation by promoting M2 macrophage activation in vivo
   ([92]7). Similarly, loss of METTL3-mediated m^6A modification in mast
   cells exacerbates their effector function in response to immunoglobulin
   E (IgE) and antigenic complexes ([93]8). In skin diseases, heterozygous
   knockout (KO) of Mettl3 worsens the severity of psoriasis ([94]9),
   while depletion of Mettl3 in CD4^+ T cells or γδ T cells alleviates
   psoriasis by inhibiting interleukin-17A (IL-17A) production ([95]10,
   [96]11).

   Psoriasis and atopic dermatitis (AD) are two kinds of highly prevalent
   chronic inflammatory skin diseases that share common features of
   epidermal hyperplasia and abundant infiltrates ([97]12–[98]14).
   Keratinocytes, in addition to various immune cells, play an essential
   role in their pathogenesis, particularly in driving the production of
   inflammatory molecules such as cytokines, chemokines, and lipid
   mediators ([99]12). Although aberrant responses of keratinocytes to T
   cell–derived cytokines are an intrinsic common pathological mechanism
   of psoriasis and AD ([100]12), other common mechanisms modulating skin
   inflammation remain unclear. Recent studies have highlighted the
   importance of m^6A modifications in skin epithelial progenitors. Xi et
   al. ([101]15) found that m^6A modification can affect cellular fate
   choices and tissue architecture during skin morphogenesis by enhancing
   translation of key morphogenetic regulators. Deletion of Mettl3 in
   epidermal progenitors results in impaired development and self-renewal,
   affecting hair follicle morphogenesis, cell adhesion, and polarity,
   which is associated with altered expression of histone-modifying
   enzymes ([102]16). Nevertheless, the specific impact of m^6A
   modification in epidermal keratinocytes on regulating cutaneous
   inflammation is not well defined.

   Given the versatility of the m^6A modification, we aimed to investigate
   its critical role in propagating inflammation in the skin, particularly
   within epidermal keratinocytes. Herein, we investigated the m^6A
   modification and its methyltransferase METTL3 in psoriasis and AD,
   using a genetic ablation approach to delineate the transcriptional and
   fatty acid metabolic rewiring that occurs during cutaneous inflammation
   resembling psoriasis or AD subsequent to depletion of the m^6A
   methyltransferases METTL3, which might elucidate metabolite-facilitated
   activation of the innate immunity. Our findings also underscore an
   m^6A-dependent effect in maintaining epidermal and immune homeostasis
   in the skin, with substantial implications for the treatment of
   psoriasis, AD, and other inflammatory diseases.

RESULTS

METTL3-m^6A pathway is down-regulated in psoriasis and AD

   To underline the dynamic of RNA m^6A in psoriasis, we analyzed the m^6A
   level by performing m^6A enzyme-linked immunosorbent assay (ELISA) and
   immunofluorescence staining. The results revealed a lower number of
   m^6A signals in the lesional skin of patients with psoriasis compared
   to the healthy skin ([103]Fig. 1A and fig. S1A). To identify the
   enzymes responsible for this m^6A reprogramming in skin inflammation,
   we analyzed the expression of m^6A writers and erasers in the lesional
   skin of patients with psoriasis and in the skin of healthy donors using
   publicly available RNA sequencing (RNA-seq) or microarray datasets
   [accession nos. [104]GSE13355 ([105]17) and [106]GSE121212 ([107]18)],
   quantitative polymerase chain reaction (PCR) with reverse transcription
   (RT-qPCR), and immunofluorescence staining. Our results showed a
   significant reduction in the key RNA methyltransferase METTL3 (both
   mRNA and protein levels) in the skin of patients with psoriasis,
   consistent with the decreased m^6A abundance in psoriatic lesions
   ([108]Fig. 1, B and C, and fig. S1B). In addition, the expression of
   METTL3 in the skin lesions of patients with psoriasis was restored
   after receiving biologic therapies, such as infliximab (anti–TNF-α) and
   secukinumab (anti–IL-17A) (fig. S1, C and D), supporting the strong
   association between METTL3-mediated m^6A and psoriasis development.
   Decreases in m^6A abundance and METTL3 expression were also detected in
   the epidermis keratinocytes of AD lesional skin and could be partially
   restored after crisaborole treatment ([109]Fig. 1D and fig. S1, E to
   G), indicating that the reduction in METTL3-mediated m^6A might be
   conservative changes in both psoriasis and AD.

Fig. 1. METTL3-m^6A pathway is down-regulated in keratinocytes of psoriasis
and AD.

   [110]Fig. 1.
   [111]Open in a new tab

   (A) Representative images of m^6A immunofluorescence staining in the
   lesional skin of patients with psoriasis or healthy control. (B and C)
   RT-qPCR of the indicated genes (B) and immunochemistry staining of
   indicated proteins (C) in the lesional skin of patients with psoriasis
   or the healthy skin. Scale bar, 100 μm. (D) Representative images of
   m^6A and METTL3 immunofluorescence staining in the lesional skin of
   patients with AD or healthy control. (E and F) Representative images of
   m^6A (E) and METTL3 (F) immunofluorescence staining in the dorsal skin
   from WT mice topically treated with IMQ for 6 days consecutively. (G
   and H) RT-qPCR analysis of Mettl3 mRNA (G) and immunoblot of METTL3 (H)
   in the epidermis from WT mice topically treated with IMQ. (I)
   Immunoblot of METTL3 in the dorsal skin treated with IMQ for 6 days
   consequently or in the recovery skin (on day 30). (J and K)
   Representative images of m^6A (J) and METTL3 (K) immunofluorescence
   staining in the dorsal skin from WT mice topically treated with MC903
   for 8 days consecutively. (L and M) RT-qPCR analysis of Mettl3 mRNA (L)
   and immunoblot of METTL3 (M) in the epidermis from WT mice topically
   treated with MC903. (N) Immunoblot of METTL3 in ear skin treated with
   MC903 for 12 days consequently or in the recovery skin (on day 30).
   Scale bar, 50 μm [(A), (D), (E), (F), (J) and (K)]; scale bar, 100 μm
   (C). Pso, psoriasis; KRT14, keratin 14; IMQ, imiquimod; AD, atopic
   dermatitis; GAPDH, glyceraldehyde-3-phosphate dehydrogenase. *P < 0.05;
   **P < 0.01; ***P < 0.001. The P values were analyzed by unpaired,
   two-tailed Student’s t test [(B), (G), and (L)]. Data are presented as
   means ± SEM.

   Next, we evaluated the m^6A levels in the mouse models induced by
   imiquimod (IMQ, a Toll-like receptor 7 agonist)–induced psoriasis-like
   skin inflammation and calcipotriol (MC903)–induced AD-like skin
   inflammation. In both IMQ-induced and MC903-induced skin inflammation,
   decreases in m^6A modification and METTL3 expression were observed
   primarily within the epidermis rather than the dermis ([112]Fig. 1, E
   to H and J to M, and fig. S1, H, I, and K). Intriguingly, the decrease
   in METTL3 expression was transient as it was restored within 30 days
   after the initial IMQ or MC903 treatment, at which point the skin
   regained homeostasis ([113]Fig. 1, I and N, and fig. S1, J and L).
   These findings suggest that a similar pattern was observed in the mouse
   models of psoriasis and AD as in the lesional skin of patients with
   psoriasis and AD.

   Last, the decrease in METTL3 expression not only was limited to the
   skin inflammation induced by IMQ or MC903 but also occurred in
   inflammation induced by tumor necrosis factor–α (TNF-α) and IL-17A
   (fig. S1, M and N). These data suggest that a reduction in the
   METTL3-m^6A axis may be prevalent in skin inflammation.

Inducible keratinocyte-specific Mettl3-deficient mice develop spontaneous
skin lesions

   It has been previously reported that loss of m^6A in preimplantation
   epithelial cells leads to early implantation lethality ([114]19).
   Notably, mice with decreased m^6A modification in keratinocytes
   [Mettl3^f/f; keratin 14 (KRT14)–Cre mice] have a limited life span, and
   most do not survive beyond postnatal day 6 ([115]15). Their poor
   survival is possibly related to a reduction in tongue filiform papillae
   rather than concomitant alterations in the skin barrier function or
   inflammation ([116]15). Therefore, to investigate the role of epidermal
   m^6A modification in vivo, we generated an inducible,
   keratinocyte-specific Mettl3-ablation mouse model by crossing
   Mettl3^flox/flox mice with KRT14-CreER^T2 mice in which the deletion of
   Mettl3 in keratinocytes was induced following tamoxifen injection,
   hereafter referred to as Mettl3 cKO mice ([117]Fig. 2, A and B). As
   expected, marked reductions in both protein and mRNA levels of METTL3
   as well as in m^6A modification levels were observed in the epidermis
   of Mettl3 cKO mice ([118]Fig. 2, C and D, and fig. S2, A and B). Skin
   lesions exhibiting erythema, scaling, and hair loss developed
   spontaneously in Mettl3 cKO mice 2 weeks after the initial tamoxifen
   injection, mainly at the base of the tail and in the palm area
   ([119]Fig. 2, E to H). Histological examination of the skin lesions
   revealed epidermal hyperplasia, hyperkeratosis, parakeratosis, and
   extensive cellular infiltration in the dermis of Mettl3 cKO mice
   ([120]Fig. 2, G and H). Furthermore, the lesional skin section of
   Mettl3 cKO mice showed a significant increase in the number of
   epidermal cells positive for Ki67^+ (a marker of cell proliferation)
   and a marked decrease in the expression of loricrin (a marker of
   terminal differentiation) in the epidermis ([121]Fig. 2I and fig. S2, C
   and D). These data provide strong evidence that METTL3-mediated m^6A in
   epidermal keratinocytes is crucial for maintaining skin homeostasis in
   adult mice.

Fig. 2. Deletion of Mettl3 in the epidermal keratinocytes of adult mice leads
to spontaneous skin inflammation.

   [122]Fig. 2.
   [123]Open in a new tab

   (A and B) Mouse construct (A) and schematic diagram (B) for
   CreER-inducible Mettl3 conditional ablation specific in keratinocytes
   and harvest time in (C) and (D). (C and D) Immunofluorescence staining
   of METTL3 (C) or m^6A (D) in the dorsal skin from Mettl3 cKO
   (KRT14-CreER^T2; Mettl3^fl/fl) or WT (Mettl3^fl/fl) mice. Scale bars,
   50 μm. (E) Schematic diagram for experimental design for inducible
   keratinocyte-specific ablation of Mettl3 observation in (F) to (N). (F)
   Incidence of skin lesions in the tail or palm of Mettl3 cKO and WT mice
   over time (n = 20). (G and H) Dermoscopic images, H&E staining of the
   tail (G) or palm (H) skin lesions, and their quantitation of acanthosis
   and dermal cellular infiltrates on the lesional skin of Mettl3 cKO and
   WT mice skin in the indicated locations. Scale bars, 100 μm. (I)
   Immunofluorescence staining of loricrin in lesional skins of Mettl3 cKO
   mice or WT mice skin in the indicated locations. Scale bars, 50 μm.
   Dashed line indicates the border between the epidermis and dermis. (J
   and K) Top 10 GO terms (J) and KEGG pathways (K) enriched in DEGs
   identified by RNA-seq between the spontaneously inflamed tail skin of
   Mettl3 cKO mice and the normal tail skin of WT mice. (L) Heatmap
   showing expression of selected genes [Fragments Per Kilobase of
   transcript per Million mapped reads (FPKM)] involved in immune
   responses in WT and Mettl3 cKO skin. (M) GSEA of genes in Mettl3 cKO
   skin relative to control skin with DEGs in the human psoriatic skin
   (left) or AD skin (right) ([124]GSE121212) enriched. TAM, tamoxifen;
   AD, atopic dermatitis; ECM, extracellular matrix; NES, normalized
   enrichment score. *P < 0.05; **P < 0.01; ***P < 0.001. The P values
   were analyzed by unpaired, two-tailed Student’s t test [(G) and (H)].
   Data are presented as means ± SEM.

   To seek the molecular insights into the formation of spontaneously
   inflamed skin lesions in Mettl3 cKO mice, we conducted transcriptome
   analysis and compared the results to those in wild-type (WT) mice.
   Through gene ontology (GO) and Kyoto Encyclopedia of Genes and Genomes
   (KEGG) pathway enrichment analysis, we identified that the
   differentially expressed genes (DEGs), selected based on the criteria
   of adjusted P < 0.05 and log[2] fold change (|log[2]FC|) > 2 from
   RNA-seq, were significantly enriched in pathways related to innate
   immune response, proliferation, and activation of T cells,
   cytokine–cytokine receptor interaction, and neutrophil chemotaxis
   ([125]Fig. 2, J and K). Furthermore, several specific inflammatory
   signaling pathways, such as the cytokine-related signaling pathway and
   the nuclear factor κB (NF-κB) signaling pathway, were also enriched
   ([126]Fig. 2K). The heatmap showed increased expression of several
   immune response–related genes in the spontaneously inflamed lesional
   skin of Mettl3 cKO mice compared with the normal skin of WT mice
   ([127]Fig. 2L). Moreover, RT-qPCR analysis confirmed elevated
   expression of multiple inflammatory factors, including Tnf, Il23a,
   Cxcl1, Cxcl2, Cxcl3, Il6, Tslp, Il33, S100a8, and S100a9, in the skin
   lesions of Mettl3 cKO mice (fig. S2E). These factors are crucial for
   T[H]17 (e.g., Il23) and T[H]2 (e.g., Tslp and Il33) immune responses.
   Furthermore, as the phenotypes closely resemble the phenotypes of skin
   lesions from patients with psoriasis or AD, we next performed gene set
   enrichment analysis (GSEA) by comparing transcriptomic data from
   spontaneously inflamed skin lesions of Mettl3 cKO mice with the data
   from the normal skin tissue and the data from the GEO database
   [accession no. [128]GSE121212 ([129]18)] from psoriatic or AD skin
   samples relative to normal skin samples. It was shown that the
   down-regulated transcriptional signatures of skin lesions in Mettl3 cKO
   mice significantly overlapped with the down-regulated DEGs identified
   in psoriatic or AD skin samples relative to normal skin samples
   ([130]Fig. 2M), implying the molecular resemblance between these
   spontaneous mouse skin lesions and both psoriasis and AD skin in human
   patients.

Mettl3 deficiency in epidermal keratinocytes exacerbates both psoriasis-like
and AD-like skin inflammation

   To directly examine the impact of METTL3-mediated m^6A methylation in
   epidermal keratinocytes on psoriasis-like and AD-like skin
   inflammation, we next compared the IMQ-induced or MC903-induced skin
   responses in Mettl3 cKO and WT mice. IMQ elicited more severe
   psoriasis-like skin symptoms in epidermal m^6A-diminished Mettl3 cKO
   mice ([131]Fig. 3, A and B, and fig. S3A). This was evident from the
   higher psoriasis area and severity index (PASI) score and greater
   transepidermal water loss (TEWL) as well as the more prominent
   acanthosis and elevated dermal inflammatory cell infiltration
   ([132]Fig. 3, B to D, and fig. S3B). In addition, compared to WT mice,
   Mettl3-deficient mice showed a more pronounced increase in the
   percentage of epidermal Ki67^+ cells and a marked decrease in the
   expression of the terminal differentiation markers, loricrin and
   filaggrin, in the epidermis after IMQ application (fig. S3, C to E).
   Furthermore, the mRNA expression of psoriasis-related inflammatory
   cytokines/chemokines, such as Il17a, Tnf, Il6, Mmp9, Cxcl1, Cxcl2,
   Cxcl3, Cxcl5, Cxcl15, and Csf3, was elevated in the skin of IMQ-treated
   Mettl3 cKO mice compared with WT mice ([133]Fig. 3E and fig. S3F).

Fig. 3. Mettl3 deficiency in epidermal keratinocytes exacerbates skin
inflammation by enhancing neutrophil infiltration.

   [134]Fig. 3.
   [135]Open in a new tab

   (A) Schematics of IMQ treatment strategy. (B) Representative images and
   PASI score of the skin after treated with IMQ for 6 days (left) or for
   the indicated time (right). (C and D) H&E staining images (C) and their
   quantification (D) of the skin treated with IMQ for 6 days. (E) Heatmap
   showing the expression of indicated genes in the skin after 3 days of
   IMQ treatment. (F) Schematics of MC903 treatment strategy. (G)
   Representative images and EASI score of the skin after 6 days (left) or
   for the indicated time (right). (H and I) H&E staining images (H) and
   their quantification (I) of the skin. (J) Heatmap showing the
   expression of indicated genes in the skin by performing RT-qPCR. (K)
   Flow cytometry of CD45^+ cells and neutrophils in the skin. (L and M)
   Immunofluorescence staining of Ly6G and CD11b in the skin after IMQ (L)
   or with MC903 (M) treatment. (N and O) Heatmap showing the expression
   of indicated genes in the epidermis after IMQ treatment for 3 days (N)
   or in the 6-day MC903-treated mouse skin (O). (P) Schematics of
   neutrophil depletion strategy. (Q) Representative images of the skin
   after IMQ treatment for 6 days (left) or for the indicated time
   (right). (R and S) H&E staining images (R) and their quantifications
   (S) of the skin. (T and U) Immunofluorescence staining of METTL3 (T)
   and m^6A (U) in the lesional skin from patients with generalized
   pustular psoriasis (GPP) or the healthy skin. Scale bars, 100 μm [(C),
   (H), (L), (M), and (R)] and 50 μm [(T) and (U)]. Ab, antibody. *P <
   0.05; **P < 0.01; ***P < 0.001. The P values were determined by two-way
   ANOVA [(B) and (Q)] or unpaired, two-tailed Student’s t test [(D), (I),
   (K), and (S)]. Data are presented as means ± SEM.

   Similar to the above results, we found a protective role of epidermally
   expressed Mettl3 against the MC903-induced AD-like skin inflammation.
   MC903 treatment of the ear skin in Mettl3 cKO mice led to an increase
   in erythema and scaling that was accompanied by a significant increase
   in skin thickness, acanthosis, and dermal cell infiltration compared to
   WT mice (fig. S3, G to K). In addition, MC903-treated Mettl3 cKO mice
   also developed skin lesions in areas that were not in direct contact
   with MC903 (e.g., eyes, face, and tail root): They exhibited hair loss
   and skin erythema covered with scaling (fig. S3, L and M). Similarly,
   MC903 treatment of the dorsal skin resulted in more severe skin
   phenotypes, including more pronounced keratinocyte terminal
   differentiation defects, in Mettl3 cKO mice than in the WT mice
   ([136]Fig. 3, F to I, and fig. S3, N to P). In addition, the expression
   of the AD-related inflammatory factors (Tslp, Il33, and Il6), but not
   Il4 and Il13, was elevated in the skin of MC903-treated Mettl3 cKO mice
   ([137]Fig. 3J and fig. S3Q). Notably, our results indicate no
   discernible disparity in TEWL between Mettl3 cKO mice and WT mice 9
   days after tamoxifen injection treatment (fig. S3, B and N, day 0),
   consistent with previous evidence indicating that mice with specific
   Mettl3 knockdown in keratinocytes do not exhibit barrier impairment
   ([138]15).

   Collectively, our data demonstrate through IMQ-induced and
   MC903-induced mouse models with Mettl3 deficiency that decreases in
   METTL3 and m^6A modification of RNA in epidermal keratinocytes are also
   critically important for maintaining skin homeostasis and suppressing
   skin inflammation in response to psoriasis-associated or AD-associated
   stimuli.

Disrupted m^6A in epidermal keratinocytes leads to an excessive influx of
neutrophils in skin inflammation

   To further understand the nature of the skin inflammation in Mettl3 cKO
   mice, we conducted flow cytometry analysis to profile immune cell
   compositions in the skin lesions of Mettl3 cKO mice and their
   sex-matched WT littermates (fig. S4). A drastic increase in the number
   of total immune (CD45^+) cells was observed in IMQ-treated Mettl3 cKO
   mice compared with WT mice, and neutrophils represented the major cell
   type (>50% of immune cells) that underwent population expansion
   ([139]Fig. 3K and fig. S5, A to C). Immunofluorescence staining with
   antibodies against the Ly6G and CD11b confirmed a more rapid
   accumulation of neutrophils in the skin lesions of Mettl3 cKO mice
   ([140]Fig. 3L). In contrast, the T cell numbers declined in Mettl3 cKO
   mice, while the numbers of other types of immune cells, including
   macrophages and dendritic cells, were not significantly altered (fig.
   S5, D to L). A higher degree of neutrophil recruitment was also
   observed in MC903-treated Mettl3 cKO mice than in WT mice ([141]Fig.
   3M). These results highlight neutrophil infiltration and accumulation
   as a major event that occurs during exacerbated psoriasis-like and
   AD-like skin inflammation in Mettl3 cKO mice. To investigate the
   mechanism underlying enhanced neutrophil accumulation, we performed
   RT-qPCR analysis of genes related to neutrophil chemotaxis and adhesion
   (such as Cxcl1, Cxcl2, Cxcl3, Cxcl5, Cxcl15, Csf3, and Icam1) in the
   skin lesions of both the IMQ-induced and MC903-induced inflammation
   mouse models. The expression of these genes was significantly elevated
   in the skin lesions of Mettl3 cKO skin compared to the skin lesions of
   WT mice skin in both models ([142]Fig. 3, N and O, and fig. S6, A and
   B).

   To assess the potential keratinocyte-intrinsic function of METTL3, we
   treated HaCaT cells, an immortalized human keratinocyte cell line, with
   a potent METTL3 inhibitor—STM2457 ([143]20). As expected, this led to a
   significant reduction in m^6A levels (fig. S6C). When HaCaT cells were
   treated with STM2457 along with a cocktail of inflammatory factors (the
   M5 cocktail, which includes IL-17A, TNF-α, IL-1α, IL-22, and oncostatin
   M) ([144]21) that is known to mimic a psoriatic inflammatory
   environment, it resulted in significantly elevated expression of
   neutrophil chemotaxis–related and adhesion-related genes, such as
   CXCL1, CXCL2, CXCL3, CXCL6, CSF3, and ICAM1, in comparison to the
   dimethyl sulfoxide (DMSO) control (fig. S6D). These results are
   indicative of a keratinocyte-specific role of METTL3, and by inference
   m^6A modification of mRNAs, in regulating the expression of genes
   involved in neutrophil recruitment.

   To examine the functional contribution of neutrophils to IMQ-induced
   skin inflammation in Mettl3 cKO mice, we used a neutralizing antibody
   against Ly6G to induce the depletion of neutrophils ([145]Fig. 3P and
   fig. S6E). Administration of the Ly6G antibody partially rescued the
   exacerbation of IMQ-induced skin inflammation in Mettl3 cKO mice, as
   evidenced by a significant decline in both the PASI score and TEWL
   value ([146]Fig. 3Q and fig. S6F). Moreover, histological analysis and
   immunofluorescence staining revealed a significant reduction in
   acanthosis, dermal cell infiltration, and epidermal Ki67^+ cell numbers
   in response to neutrophil depletion ([147]Fig. 3, R and S, and fig.
   S6G). Pustular psoriasis is a severe type of psoriasis characterized by
   the presence of pustules filled with neutrophils ([148]22). As
   anticipated, we found a significant decrease in m^6A modification and
   METTL3 expression in the keratinocytes of pustular psoriasis compared
   to those of the healthy skin ([149]Fig. 3, T and U).

   Accordingly, our mechanistic experiments indicate that METTL3, and by
   inference m^6A modification, had a regulatory effect on the expression
   of genes related to neutrophil recruitment. Thus, METTL3/m^6A may play
   a protective role in inflammatory skin conditions through inhibition of
   neutrophil infiltration.

Therapeutic potential of an m^6A activator in cutaneous inflammation

   Methyl piperidine-3-carboxylate (MP3C) has been recently reported to be
   an agonist that targets the m^6A methyltransferase METTL3-METTL14-WTAP
   complex ([150]23). We next investigated if MP3C could activate m^6A
   methyltransferase activity in skin lesions and alleviate the
   inflammatory skin conditions. The efficacy of MP3C in activating m^6A
   methyltransferase activity was confirmed by quantifying the levels of
   m^6A in RNA isolated from HaCaT cells after treatment with or without
   MP3C (fig. S7A). In the next set of experiments, we evaluated the
   therapeutic effectiveness of MP3C in models of skin inflammation
   induced by IMQ or MC903. Our results demonstrated that topical
   application of MP3C effectively reduced cutaneous inflammation induced
   by IMQ or MC903 ([151]Fig. 4, A to D and G to J, and fig. S7, B and C),
   with an increase in m^6A abundance in the epidermis of lesional skin
   compared to the control group (fig. S7, D and I). In addition, enhanced
   filaggrin and loricrin expressions, as well as decreased inflammatory
   factor expression, were observed in psoriasis-like or AD-like skin
   lesions after topical application of MP3C ([152]Fig. 4, E and K, and
   fig. S7, E to G and J to L). Notably, CD45^+ cells and neutrophil
   numbers were significantly diminished after topical application of MP3C
   to IMQ-induced psoriasis-like skin lesions ([153]Fig. 4F and fig. S7H).
   This is consistent with the flow cytometry results for skin lesions in
   Mettl3 cKO mice.

Fig. 4. Activation of the m^6A modification dampens the severity of skin
inflammation.

   [154]Fig. 4.
   [155]Open in a new tab

   (A) Schematic diagram of strategy for application of MP3C or EtOH and
   IMQ. (B) Representative images (left) and PASI score (right) of mice
   treated with MP3C or EtOH after IMQ treatment for 6 days or at the
   indicated time. (C and D) H&E staining images (C) and their
   quantification (D) of mice treated with MP3C or EtOH after 6-day IMQ
   application. (E) RT-qPCR analysis of the expression of indicated genes
   in the 6-day IMQ-treated skin from mice with MP3C or EtOH application.
   (F) Flow cytometry of CD45^+ cells and neutrophils in the 6-day
   IMQ-treated skin from mice with MP3C or EtOH application. (G) Schematic
   diagram of strategy for application of MP3C or EtOH and MC903. (H)
   Representative images and EASI score of mice treated with MP3C or EtOH
   after MC903 treatment for 6 days (left) or at the indicated time
   (right). (I and J) H&E staining images (I) and their quantifications
   (J) of mice treated with MP3C or EtOH after 6-day MC903 application.
   (K) RT-qPCR analysis of the expression of indicated genes in the 6-day
   MC903-treated mouse skin with MP3C or EtOH application. (L) Schematic
   depicting the treatment of skin biopsies obtained from healthy donors
   (left) and RT-qPCR analysis of specified genes in these biopsies after
   a 6-hour treatment (right). (M) Schematic depicting the treatment of
   lesional psoriatic skin biopsies (left) and RT-qPCR analysis of
   specified genes in these biopsies after a 24-hour treatment (right).
   The figures were created with [156]BioRender.com [(L) and (M)]. Scale
   bar, 100 μm [(C) and (I)]. *P < 0.05; **P < 0.01; ***P < 0.001. The P
   values were determined by two-way ANOVA [(B), (H), and (L)], unpaired
   two-tailed Student’s t test [(D), (F), and (J)], or paired Student’s t
   test (M). Data are means ± SEM.

   To address the translational relevance of MP3C in human inflammatory
   skin diseases, we conducted experiments using cultured skin biopsy
   tissues obtained from healthy donors. We found that M5 stimulation
   markedly increased the expression of inflammatory factors, including
   chemoattractants for neutrophils, T[H]17-related cytokines, and
   T[H]2-related cytokines such as TSLP and IL33. However, treatment with
   MP3C effectively suppressed these inflammatory responses ([157]Fig.
   4L). Incubation of lesional psoriasis skin cultures with MP3C resulted
   in a decreased expression of several psoriasis-related genes,
   chemoattractants for neutrophils, and adhesion molecule–encoding genes,
   including CXCL1, CXCL2, CXCL3, CXCL5, CXCL6, CXCL8, IL17A, ITNF, and
   IL6 ([158]Fig. 4M). These data suggest that MP3C has anti-inflammatory
   properties in human skin biopsies and may reduce inflammatory responses
   in the lesional skin from patients with psoriasis under ex vivo
   conditions.

   In summary, our study provides evidence that activation of m^6A
   modification could ameliorate skin inflammation and might serve as a
   potential therapeutic strategy for targeting m^6A in psoriasis and AD.

Mettl3 deficiency in keratinocytes leads to fatty acid metabolic
reprogramming

   To gain molecular insight into the effects of m^6A deficiency on skin
   inflammation, particularly its effects on epidermal keratinocytes, we
   performed methylated RNA immunoprecipitation sequencing (MeRIP-seq) and
   RNA-seq analysis on the lesional skin of patients with psoriasis and
   the healthy skin as well as the epidermis of Mettl3 cKO and WT mice
   treated with IMQ for 3 or 6 days ([159]Fig. 5A). Our analysis revealed
   a significant positive correlation between mRNA expression and m^6A
   modification levels in the lesional skin of psoriatic patients versus
   the healthy skin tissue ([160]Fig. 5B). Furthermore, Metascape analysis
   ([161]24) showed that genes down-regulated or hypomethylated in the
   psoriatic lesional skin were enriched in long-chain fatty acid
   metabolic process or the lipid biosynthesis process ([162]Fig. 5C).

Fig. 5. Lipid metabolism undergoes rewiring in psoriasis or in skin
inflammation by Mettl3 depletion.

   [163]Fig. 5.
   [164]Open in a new tab

   (A) Schematic illustration of sample collection for MeRIP-seq in
   psoriasis. The figure was created with [165]BioRender.com. (B) The
   volcano plot displays changes of RNA expression and m^6A modification
   levels between the lesional skin of patients with psoriasis and the
   healthy skin. Dashed lines define log[2] FC > 1 and log[2] FC < −1.
   Purple and green dots were peaks with |FC| > 2 and adjusted P value <
   0.05. (C) Top 10 enrichment clusters of down-regulated genes and
   hypomethylated genes in the lesional skin of patients with psoriasis
   analyzed using the Metascape database. One row per cluster uses a
   discrete color scale to represent statistical significance. (D and E)
   Volcano plot depicts significantly DEGs and DMPs (adjusted P < 0.05 and
   |FC| > 2) from MeRIP-seq between the epidermis of Mettl3 cKO and WT
   mice after IMQ treatment for 3 or 6 days (D) and the heatmap showing
   the top 20 enrichment clusters analyzed using the Metascape database
   (E). (F) GSEA performed by CEMiTool showing module activity for each
   sample group (n = 3). (G) Depiction of the gene interaction network of
   module 3 from G in (F), which related to lipid metabolism. (H) Top 10
   of GO terms enrichment of down-regulated DEGs in spontaneous skin
   lesions of Mettl3 cKO mice relative to WT skin. (I) Absolute
   quantification of fatty acids in the epidermis of Mettl3 cKO or WT mice
   after IMQ treatment using GC-MS. Lipid metabolism–related pathways are
   marked in red [(C), (E), and (H)]. PCC, Pearson’s correlation
   coefficient; D, day; IMQ, imiquimod; NES, normalized enrichment score;
   DEG, differentially expressed gene; DMP, differentially methylated
   peak; Hyper, hypermethylated gene; Hypo, hypomethylated gene; Down,
   down-regulated gene; Up, up-regulated gene. GC-MS, gas
   chromatography–mass spectrometry; CoA, coenzyme A. *P < 0.05. The P
   values were determined by two-tailed Student’s t test (I). Data are
   presented as means ± SEM.

   As observed in the human skin samples, a strong positive correlation
   was also observed between mRNA expression and m^6A modification levels,
   with most genes showing reduced mRNA levels and m^6A hypomethylation in
   the epidermis of Mettl3 cKO mice ([166]Fig. 5D). Metascape analysis of
   the genes that were down-regulated or hypomethylated in the epidermis
   of IMQ-treated Mettl3 cKO mice revealed lipid or fatty acid
   metabolism–associated clusters as being among the top 10 enriched
   clusters across multiple gene lists ([167]Fig. 5E). Furthermore, we
   performed a modular gene coexpression network analysis using CEMiTool
   ([168]25), through which seven individual modules were identified in
   our dataset ([169]Fig. 5F). Of note, the expression of module 3 was
   decreased in the epidermis of Mettl3 cKO mice after IMQ treatment
   compared to the epidermis of WT mice, and numerous pathways relevant to
   fatty acid metabolism or biosynthesis were enriched in this module
   ([170]Fig. 5G). In addition, GO enrichment analysis of DEGs
   down-regulated in the spontaneous skin lesions of Mettl3 cKO mice
   relative to the normal skin showed that lipid and fatty acid metabolic
   processes were the top terms ([171]Fig. 5H).

   On the basis of the above results, we aimed to investigate the impact
   of METTL3-mediated m^6A plays a role in skin inflammation via induction
   of fatty acid metabolism reprogramming and analyzed the fatty acid
   composition in the epidermis of Mettl3 cKO mice and WT mice after IMQ
   treatment. Our results revealed that the levels of palmitic acid (PA),
   stearic acid (SA), elaidic acid (EA), and oleic acid (OA) were
   significantly higher in the epidermis of Mettl3 cKO mice than in the
   epidermis of WT mice ([172]Fig. 5I). Conversely, a decreasing tendency
   was observed in the levels of very-long-chain fatty acids, such as
   erucic and lignoceric acids ([173]Fig. 5I). This suggests a potential
   disruption in fatty acid carbon chain elongation due to the loss of
   METTL3.

   The above results provide compelling evidence for the involvement of
   METTL3 in regulating the mRNA m^6A modification in genes associated
   with fatty acid metabolism in epidermal keratinocytes. Consequently,
   the loss of METTL3/m^6A leads to the reprogramming of fatty acid in the
   epidermis.

ELOVL6 is a critical downstream target of METTL3 in cutaneous inflammation

   To identify the functionally important target of METTL3 in skin
   inflammation, we performed a Venn diagram analysis using our RNA-seq
   and MeRIP-seq data to identify the potential target genes involved in
   lipid metabolism that exhibit down-regulation and hypomethylation in
   Mettl3 cKO compared to WT mice. Nine potential target genes were
   identified, namely, Acot1, Acox3, Agpat4, Cers4, Elovl3, Elov6, Hmgcr,
   Mgll, and Lpcat3 ([174]Fig. 6A). Among these genes, Elovl6 and Elovl3
   mRNA were previously reported to play critical regulatory roles in
   fatty acid metabolism ([175]26). Specifically, ELOVL6 converts C16
   saturated and monounsaturated fatty acids to C18 species ([176]27),
   while ELOVL3 catalyzes the synthesis of C20 to C24 fatty acids
   ([177]28).

Fig. 6. Elovl6 is identified as a target gene of METTL3.

   [178]Fig. 6.
   [179]Open in a new tab

   (A) Venn diagram of genes associated with metabolism of lipids and
   hypomethylated genes in cKO mice versus WT mice after IMQ treatment.
   (B) Integrative Genomics Viewer (IGV) tracks showing m^6A modification
   on Elovl6 mRNAs in the IMQ-treated epidermis. (C and D) SN ratio or
   Elovl6 mRNA level of the epidermis of cKO mice relative to WT mice
   treated with IMQ for 3 days (C) or with MC903 for 6 days (D). (E)
   Immunofluorescence staining of ELOVL6 in the skin treated with IMQ for
   6 days. (F and G) Expression of ELOVL6 in the 6-day IMQ-treated
   epidermis from C57BL6/J mice with MP3C or EtOH application. (H)
   Signal-to-Noise (SN) ratio of Elovl6 RNA in the epidermis. (I and J)
   Immunofluorescence staining of ELOVL6 in the MC903-treated skin from WT
   or cKO mice (I) or from C57BL6/J mice with MP3C or EtOH application
   (J). (K and L) ELOVL6 expression in the tail (K) or palm (L) skin of WT
   and cKO mice. (M and N) IGV tracks showing m^6A modification on ELOVL6
   mRNAs in the human skin tissue (M) or in HEKa (N). (O) SN ratio of
   ELOVL6 RNA in HEKa relative to the DSMO group. (P) Immunoblot of ELOVL6
   in HEKa after STM2457 treatment. (Q) Actinomycin D chase experiments
   for ELOVL6 mRNA expression in HaCaT cells. (R and S) Expression of
   ELOVL6 after IGF2BP1-3 was silenced in HaCaT cells. (T) RIP-qPCR tested
   the binding of IGF2BP3 and ELOVL6 mRNA in HaCaT cells treated with
   STM2457 or DMSO. Scale bar, 50 μm. *P < 0.05; **P < 0.01; ***P < 0.001.
   The P values were determined by two-way ANOVA [(C), (D), (Q), and (T)],
   unpaired, two-tailed Student’s t test [(C), (D), (F), (H), (K), and
   (O)], or one-way ANOVA (R). Data are presented as means ± SEM.

   Next, we sought to confirm if ELOVL3 and ELOVL6 are direct targets of
   METTL3 in keratinocytes of skin inflammation. We found that significant
   reductions in m^6A modification and the expression levels of Elovl6
   were observed in the epidermis of Mettl3 cKO mice compared to WT mice
   in both IMQ-induced and MC903-induced skin inflammation ([180]Fig. 6, B
   to E and I, and fig. S8A), while MP3C treatment restored the m^6A
   modification and the expression levels of Elovl6 ([181]Fig. 6, F to H
   and J). Moreover, reduced mRNA and/or protein levels of Elovl6 were
   also observed in the epidermis of spontaneously developed skin lesions
   in Mettl3 cKO mice compared to WT mice ([182]Fig. 6, K and L).
   Furthermore, we found that the m^6A level of ELOVL6 was significantly
   decreased in the lesional skin of patients with psoriasis or AD
   according to the results of MeRIP-seq analysis (P value = 0.05 for
   psoriatic versus healthy control; P value = 0.03 for AD versus healthy
   control) ([183]Fig. 6M). To determine if ELOVL6 mRNA is a bona fide
   target of METTL3-mediated m^6A modification in human epidermal
   keratinocytes, we used both adult human keratinocytes. As we expected,
   treatment of human epidermal keratinocytes with STM2457 resulted in
   down-regulation of the m^6A level in ELOVL6 mRNA, accompanied by a
   decrease in ELOVL6 protein levels ([184]Fig. 6, N to P, and fig. S8, B
   to E). In contrast, although reduced m^6A modification and mRNA
   expression of Elovl3 were observed in the epidermis of inflammatory
   skin lesions from Mettl3 cKO mice (fig. S8, F to H), the m^6A signal
   was barely detectable in ELOVL3 mRNA in HEKa and HaCaT cells (fig. S8,
   I and J), and ELOVL3 mRNA expression was not suppressed by STM2457
   treatment (fig. S8K). These data demonstrated that ELOVL6, but not
   ELOVL3, is a direct target of METTL3-mediated m^6A modification in
   epidermal keratinocytes.

   Then, we examined the impact of m^6A on ELOVL6 mRNA stability. As
   expected, in STM2457 pretreated cells, ELOVL6 mRNA stability decreased
   after actinomycin D treatment, which led to the inhibition of gene
   transcription ([185]Fig. 6Q). The potential function of m^6A readers in
   ELOVL6 mRNA stability was further explored. Among various m^6A reader
   proteins, insulin-like growth factor 2 mRNA-binding proteins (IGF2BPs,
   including IGF2BP1/2/3) have been previously reported to enhance the
   stability and translation of target RNAs ([186]2, [187]29). To
   investigate if IGF2BPs might be required for ELOVL6 mRNA stability in
   keratinocytes, we individually knocked down IGF2BP1/2/3 respectively in
   HaCaT cells and found that silencing IGF2BP3 significantly inhibited
   ELOVL6 expression ([188]Fig. 6, R and S, and fig. S8L). In addition,
   RIP-qPCR revealed the direct binding of IGF2BP3 to ELOVL6 mRNA, and
   this binding was found to be inhibited by STM2457 ([189]Fig. 6T). This
   is indicative of a dependence on METTL3 activity/m^6A modification.
   Together, these results identify IGF2BP3 as a potential reader of
   m^6A-modified ELOVL6 mRNA that promotes its stability.

ELOVL6 inhibition aggravates the severity of inflammatory skin conditions

   To further investigate the role of ELOVL6 in the pathogenesis of
   psoriasis and AD, we first determined the expression level of ELOVL6 in
   psoriasis and AD. We found a significant decrease in the mRNA levels of
   ELOVL6 in the lesional skin compared to nonlesional skin in patients
   with both psoriasis and AD by analyzing a publicly available RNA-seq
   dataset [accession no. [190]GSE121212 ([191]18)] ([192]Fig. 7A). Our
   immunofluorescence staining results further confirmed this markedly
   reduced expression of the ELOVL6 protein, particularly in epidermal
   keratinocytes from patients with psoriasis (including psoriasis
   vulgaris and pustular psoriasis) and AD ([193]Fig. 7B). Consistent with
   the decreased expression of METTL3 observed in the inflamed murine skin
   ([194]Fig. 1, G, H, L, and M), the mRNA and protein expression levels
   of ELOVL6 were found to be significantly reduced in the epidermis of WT
   mice treated with both IMQ and MC903 compared to their vehicle control
   groups ([195]Fig. 7, C and D).

Fig. 7. Inhibition of ELOVL6 exacerbates skin inflammation.

   [196]Fig. 7.
   [197]Open in a new tab

   (A) Expression of ELOVL6 mRNA reanalyzed from the RNA-seq analysis
   ([198]GSE121212) of the paired lesional and nonlesional skin from
   patients with psoriasis (n = 27) or AD (n = 21). (B) Immunofluorescence
   staining of ELOVL6 in the lesional skin of patients with PSV, GPP, and
   AD or the healthy skin. (C and D) RT-qPCR analysis (left) and
   immunofluorescence staining (right) of ELOVL6 in the epidermis from WT
   mice treated with IMQ (C) or MC903 (D). (E) Schematic diagram of
   strategy for application of ELOVL6-IN and IMQ. (F) Representative
   images (left) and PASI score (right) of mice treated with ELOVL6-IN
   after IMQ treatment for 5 days (left) or at the indicated time (right).
   (G and H) H&E staining images (G) and their quantifications (H) of the
   mouse skin. (I) Heatmap showing the expression of indicated genes in
   the mouse skin by performing RT-qPCR. (J) Flow cytometry of CD45^+
   cells and neutrophils in the mouse skin. (K) Schematic diagram of
   strategy for application of ELOVL6-IN and MC903. (L) Representative
   images and EASI score of the dorsal skin of mice treated with ELOVL6-IN
   after application of MC903 for 5 days (left) or at the indicated times
   (right). (M and N) H&E staining images (M) and their quantifications
   (N) of the mouse skin. (O) Heatmap showing the expression of indicated
   genes after the 5-day MC903-treated mouse skin by performing RT-qPCR.
   Scale bars, 100 μm [(F) and (M)] and 50 μm [(B) to (D)]. The dashed
   line indicates the border between the epidermis and dermis [(C) and
   (D)]. *P < 0.05; **P < 0.01; ***P < 0.001. The P values were determined
   by paired, two-tailed Student’s t test (A), unpaired, two-tailed
   Student’s t test [(C), (D), (H), (J), and (N)], or two-way ANOVA [(F)
   and (L)]. Data are presented as means ± SEM.

   To assess the potential role of ELOVL6 in regulating cutaneous
   inflammation, we topically applied a selective ELOVL6 inhibitor
   ([199]30) to the dorsal skin of C57BL/6J mice ([200]Fig. 7E). After the
   IMQ treatment, mice treated with the ELOVL6 inhibitor displayed
   significantly more pronounced psoriasis-like skin features compared to
   the vehicle control mice ([201]Fig. 7, F to H, and fig. S9, A to D).
   Furthermore, the skin of mice treated with the ELOVL6 inhibitor
   exhibited more marked up-regulation of psoriasis-related genes (such as
   Il17a, Il6, Mmp9, and S100a8) and neutrophil
   chemotaxis/adhesion–related genes (such as Cxcl1, Cxcl2, Cxcl5, Csf3,
   and Icam1) than the skin of mice in the control group ([202]Fig. 7I and
   fig. S9E). Consistent with these findings, the results of flow
   cytometry also revealed a remarkable increase in the number of CD45^+
   cells and neutrophils in the ELOVL6 inhibitor–treated skin after IMQ
   application ([203]Fig. 7J and fig. S9F). Similarly, ELOVL6 inhibition
   resulted in worsened skin inflammation in response to MC903 treatment,
   as indicated by higher EASI scores and TEWL values, more severe
   epidermal hyperplasia, dermal cell and neutrophil infiltration,
   decreased expression of terminal differentiation markers, and elevated
   expression of AD-related or neutrophil-related genes such as Tslp,
   Il33, Il13, Il6, Cxcl1, Cxcl3, Cxcl5, Cxcl15, Csf3, and Icam1
   ([204]Fig. 7, K to O, and fig. S9, G to K). These results imply that
   inhibition of ELOVL6 function aggravates both psoriasis-like and
   AD-like skin inflammation.

   To further understand the role of ELOVL6 in epidermal function, we
   induced silencing or overexpression of ELOVL6 in HaCaT cells and
   evaluated the changes in the inflammatory response (fig. S10, A and B).
   Silencing ELOVL6 in human keratinocytes resulted in increased mRNA
   expression of neutrophil chemoattractants such as CXCL1, CXCL2, CXCL3,
   CXCL5, CXCL6, CXCL8, CSF3, and ICAM1 in response to M5 stimulation
   (fig. S10C). In contrast, overexpression of ELOVL6 partially suppressed
   the elevated expression of these neutrophil chemoattractants in
   M5-treated keratinocytes, whereas METTL3 activity was inhibited by
   STM2457 treatment (fig. S10D). On the basis of these data combined with
   the observed in vivo effect of ELOVL6 inhibition, we deduced that
   ELOVL6, similar to its upstream regulator METTL3, plays a role in
   protecting the skin from both psoriasis-like and AD-like inflammatory
   manifestations, partly by regulating neutrophil chemoattraction in
   epidermal keratinocytes.

PA exacerbates IMQ-induced or MC903-induced skin inflammation

   To underline the role of ELOVL6 in fatty acid metabolism rewiring in
   keratinocytes during cutaneous inflammation, we conducted a lipidomics
   study to compare the fatty acid composition of the epidermis in mice
   treated with the ELOVL6 inhibitor or its vehicle after IMQ application
   as well as human keratinocytes after silencing ELOVL6 expression under
   inflammatory conditions ([205]Fig. 8A and fig. S10E). The predominant
   fatty acid species in the mouse epidermis were identified as PA, SA,
   OA, and linoleic acid (LA) ([206]Figs. 5I and [207]8A). Furthermore,
   the epidermis of the ELOVL6 inhibitor–treated mice showed notably
   elevated levels of PA, SA, and OA compared to the controls, which was
   consistent with the changes observed in the epidermis of Mettl3 cKO
   mice ([208]Figs. 5I and [209]8A). We also found that ELOVL6 knockdown
   in human keratinocytes resulted in a significant elevation of the
   levels of C16 fatty acid but not C18 fatty acid. This effect was
   associated with a marked decrease in the ELOVL6 activity index, which
   was calculated using the following formula: (SA [C18:0] + OA
   [C18:1])/(PA [C16:0]) + (palmitoleic acid [C16:1]) (fig. S10, E and F)
   ([210]31). This implies the importance of ELOVL6 in prolonging the C16
   fatty acid profile in human keratinocytes. Moreover, ELOVL6-silenced
   HaCaT cells had a greater PA composition than control cells, with no
   significant difference observed in the proportion of SA, EA, or OA
   (fig. S10E). On the basis of these findings, we hypothesized that PA
   might play an essential role in cutaneous inflammation induced by the
   METTL3-m^6A-ELOVL6 axis.

Fig. 8. Topical treatment of PA results in worsening of skin inflammation.

   [211]Fig. 8.
   [212]Open in a new tab

   (A) Quantitation of FAs in the epidermis of mice topically treated with
   ELOVL6-IN or its vehicle using GC-MS. (B) Schematic diagram of strategy
   for application of PA and IMQ. (C and D) Representative images (C) and
   PASI score (D) of the mouse skin treated with PA or its vehicle after
   IMQ treatment for 6 days (C) or at the indicated time (D). (E and F)
   H&E staining images and their quantifications (F) of mice treated with
   PA or its vehicle after 6-day IMQ application. (G) Flow cytometry of
   CD45^+ cells and neutrophils in the lesional skin of mice treated with
   PA or its vehicle after 6-day IMQ application. (H) RT-qPCR analysis of
   the expression of indicated genes in the 6-day IMQ-treated skin from
   mice with PA or its vehicle application. (I) Schematic diagram of
   strategy for application of PA and MC903. (J and K) Representative
   images (J) and EASI score (K) of the mouse skin treated with PA or
   vehicle after application of MC903 for 6 days (J) or at the indicated
   times (K). (L and M) H&E staining images (L) and their quantifications
   (M) of the mouse skin treated with PA or its vehicle in 6-day
   MC903-induced skin inflammation. (N) RT-qPCR analysis of the expression
   of indicated genes in the 6-day MC903-treated skin from mice with PA or
   its vehicle application. (O) Proposed model of the regulatory and
   functional role of the METTL3-m^6A-ELOVL6 axis during skin
   inflammation. The figure was created with [213]BioRender.com. *P <
   0.05; **P < 0.01; ***P < 0.001. Scale bar, 100 μm [(E) and (L)]. The P
   values were determined by unpaired, two-tailed Student’s t test [(A),
   (F), (G), (H), (M), and (N)] or two-way ANOVA [(D) and (K)]. Data are
   presented as means ± SEM.

   Previously, PA, but not myristic acid (C14:0), palmitoleic acid
   (C16:1), or OA, has been shown to augment phorbol 12-myristate
   13-acetate–induced neutrophilic folliculitis in mice when topically
   applied ([214]32). Moreover, PA could promote the secretion of
   proinflammatory cytokines, such as IL-6, IL-1, and TNF-α, in human
   keratinocytes ([215]33). Our detection of increased levels of PA in the
   METTL3-deficient or ELOVL6-deficient epidermis ([216]Figs. 5I and
   [217]8A) prompted us to investigate if topical application of PA
   exacerbates IMQ-induced or MC903-induced skin inflammation. Topical
   application of PA remarkably aggravated the skin symptoms induced by
   either IMQ or MC903, which were manifested in the form of clinical
   sores, disruption of barrier function, epidermal proliferation, dermal
   cell infiltration, and keratinocyte differentiation ([218]Fig. 8, B to
   F and I to M, and fig. S11, A to D and G to I). Moreover, PA treatment
   resulted in a significant increase in the expression of both
   inflammation-related genes (such as Il6, Mmp9, Cxcl2, Cxcl5, Cxcl15,
   and Csf3) and the abundance of neutrophils in IMQ-induced skin lesions
   ([219]Fig. 8, G and H, and fig. S11, F and E). Similarly, PA
   application was found to enhance neutrophil accumulation as well as the
   expression of inflammatory factors in MC903-induced AD-like skin
   inflammation ([220]Fig. 8N and fig. S11, J and K). PA treatment of
   HaCaT cells led to elevated expression of neutrophil
   chemotaxis/adhesion–related genes (fig. S11L). Together, these data
   suggest that epidermal keratinocytes respond to elevated levels of PA
   by up-regulating the expression of genes involved in neutrophil
   recruitment and, consequently, cutaneous inflammation.

   Overall, our results provide strong evidence for the down-regulation of
   the epidermal METTL3-m^6A-ELOVL6-PA axis in cutaneous inflammation and
   underscore the normalization of this axis to be an important aspect of
   therapeutic corrections for inflammatory skin diseases such as
   psoriasis and AD.

DISCUSSION

   Our findings highlight the crucial involvement of an m^6A
   modification–fatty acid metabolism pathway detected in epidermal
   keratinocytes in preserving skin homeostasis and mitigating excessive
   inflammation in response to disease-inducing stimuli. More
   specifically, we suggest a working model in which METTL3, expressed by
   keratinocytes in the healthy skin, is responsible for maintaining
   appropriate levels of m^6A modification on critical target genes, most
   notably ELOVL6 mRNA, that regulate fatty lipid metabolism. In disease
   states such as psoriasis and AD, decreased epidermal METTL3 expression
   as a result of genetic deletion and/or aberrant inflammation results in
   decreased ELOVL6 mRNA levels, probably due to an m^6A/IGF2BP3-dependent
   decrease in stability. Consequently, PA extension is impeded and it
   accumulates, triggering the up-regulation of genes related to
   neutrophil recruitment or cell adhesion and culminating in enhanced
   neutrophil aggregation in the skin. These events collectively
   contribute to the escalation of the skin inflammatory cycle and
   aggravation of disease severity ([221]Fig. 8O).

   Recently, accumulating research has demonstrated the impact of m^6A
   modification in psoriasis and AD. It seems that the role of m^6A
   modification in cutaneous inflammation is dependent on cell types. For
   example, a proinflammatory effect of m^6A modification was demonstrated
   in γδT cells and CD4^+ T cells ([222]10, [223]11), whereas an
   anti-inflammatory function was identified in macrophages and
   keratinocytes that was a result of alteration of the fates of noncoding
   RNAs ([224]34–[225]36). Furthermore, a recent study reported the
   aggregated IMQ-induced psoriasis-like skin inflammation in Mettl3^+/−
   mice compared to WT controls ([226]9). Our study not only supports the
   protective role of Mettl3 and m^6A in murine epidermal keratinocytes
   but also shows that this protective role is so crucial that its loss in
   keratinocytes can be sufficient to initiate skin inflammation even in
   the absence of any external stimuli. This is evident from the formation
   of spontaneous psoriasis-like and AD-like lesions in Mettl3-deficient
   mice. Furthermore, our data reveal mRNAs as important targets of
   Mettl3-mediated m^6A modification in keratinocytes.

   Our findings propose a mechanism by which the absence of m^6A
   modification in epidermal keratinocytes promotes cutaneous
   inflammation, at least in part, via disruption of lipid metabolism,
   particularly fatty acid metabolism. However, it should be noted that
   the ability of m^6A modifications to modulate fatty acid metabolic
   reprogramming is not unprecedented ([227]37–[228]40), and disruptions
   in fatty acid metabolism are known to contribute to the development of
   various inflammatory diseases, including asthma, rheumatoid arthritis,
   and autoimmune skin diseases such as psoriasis and AD
   ([229]41–[230]44). However, our work pinpoints m^6A modification as an
   important means to regulate fatty acid metabolism in the context of
   skin inflammation. We have identified ELOVL6, a member of the
   elongation of very-long-chain fatty acids (ELOVL) family of enzymes, as
   a key target of METTL3-mediated m^6A modification and have provided in
   vivo evidence for its functional involvement in suppressing
   psoriasis-like and AD-like skin inflammation. The ELOVL family of
   enzymes plays a role in the formation and maintenance of healthy skin
   structure and improvement of skin barrier function by catalyzing the
   formation of long-chain fatty acids ([231]45). Accordingly, depletion
   of Elovl6 has previously been shown to aggravate the mechanical damage
   to the skin and subsequently cause dermatitis by promoting the cellular
   death of keratinocytes ([232]46). ELOVL6 is known to be responsible for
   the elongation of PA (C16:0) and palmitoleate (C16:1n-7) ([233]22).
   Consistent with this known biochemical activity, we found that
   inhibition of ELOVL6 promoted PA accumulation in the epidermis. While
   we cannot exclude barrier dysregulation and keratinocyte death as
   possible causes of the exacerbated inflammation and neutrophil
   aggregation based on the findings so far, our results do indicate that
   the effects of PA application closely resemble those of METTL3
   deficiency or ELOVL6 deficiency at the tissue, cellular, and molecular
   levels and suggest a more direct mechanism of PA downstream of the
   METTL3-m^6A-ELOVL6 axis that regulates epidermal-neutrophil cross-talk
   and neutrophil chemotaxis during skin inflammation. Our in vitro
   experiments with human keratinocytes support this notion and implicate
   a keratinocyte-intrinsic mechanism whereby PA induces the expression of
   neutrophil chemotaxis/adhesion–related genes in these cells. These data
   propose an innovative mechanism for regulating metabolism and immunity
   in the context of skin conditions.

   Neutrophilic inflammation is a characteristic feature of both psoriasis
   and AD pathologies ([234]47). The number of neutrophils in the lesional
   skin from patients with psoriasis and AD was higher than that in the
   healthy skin, and both were correlated with the expression levels of
   neutrophil chemoattractant genes ([235]48). Although the role of
   neutrophils in AD is still under debate, a growing number of recent
   studies indicate that neutrophils may play a crucial role in chronic
   recurrent AD ([236]49, [237]50). In the present study, down-regulation
   of the METTL3-m^6A-ELOVL6 axis was detected in the epidermal
   keratinocytes of patients with psoriasis vulgaris, psoriasis pustulosa,
   and AD. These findings are consistent with our observation of increased
   neutrophil infiltration in the inflamed skin of Mettl3 cKO mice, ELOVL6
   inhibitor–treated mice, and PA-treated mice. Thus, the
   METTL3-m^6A-ELOVL6 axis may be instrumental in driving inflammation in
   human patients with skin diseases associated with neutrophil
   involvement. The anti-inflammatory properties of ELOVL6 are not
   restricted to the skin as similar effects have also been observed in
   allergic airway inflammatory conditions such as asthma ([238]26). Thus,
   the regulatory mechanisms elucidated in this study may extend beyond
   the skin and have broad implications for autoimmune diseases affecting
   other cell types and tissues too.

   Consistent with our findings, previous studies have also indicated
   higher levels of PA in the serum of patients with psoriasis than in the
   serum of healthy controls ([239]51, [240]52). Similarly, an inclination
   toward elevated PA levels was observed in the epidermis of lesioned
   skin in patients with AD compared to nonlesional skin ([241]53).
   Application of fatty acids, including PA, has been associated with
   hyperkeratinization and exacerbation of neutrophilic folliculitis
   ([242]32). These studies further corroborate the role of PA in
   autoimmune skin disorders. In addition, a study reported that
   deficiency in METTL3-mediated m^6A modification in tumor cells enhances
   the recruitment of tumor-associated neutrophils and, consequently,
   affects tumor pathogenesis ([243]54). However, if abnormalities in
   lipid metabolism contribute to this process remains unexplored. Hence,
   our study may serve as a useful reference for further mechanistic
   studies in cancer and other inflammatory diseases that are
   characterized by neutrophil involvement.

   Our analysis of MeRIP-seq data reveals that, in addition to genes
   related to fatty acid metabolism, one gene involved in ceramide
   synthesis, such as Cers4, may also play a significant role in skin
   inflammation regulated by RNA m^6A methylation in keratinocytes
   ([244]Fig. 6A). Ceramides, which are essential components of skin
   lipids, have been linked to various inflammation conditions, including
   asthma, Alzheimer’s disease, and skin inflammation ([245]55–[246]58).
   In inflammatory skin diseases including psoriasis and AD, aberrant
   ceramide expression profiles are recognized for their disruptive
   effects on epidermal barrier maintenance, epidermal self-renewal, and
   cutaneous immune responses ([247]59). Future investigations will
   further extend the repertoire of m^6A-regulated genes that contribute
   to skin inflammation, enhancing a comprehensive understanding of m^6A
   modification in skin inflammation.

   Overall, our findings underscore the crucial role of disrupted m^6A
   modification in keratinocytes and consequent reprogramming of fatty
   acid metabolism and increased neutrophil accumulation in autoimmune
   skin disorders such as psoriasis and AD. These insights have important
   implications for potential therapies targeting m^6A modification in
   such diseases as well as other conditions related to epithelial
   inflammation.

MATERIALS AND METHODS

Patients

   The skin samples used in this study were obtained from patients with
   psoriasis vulgaris, patients with psoriasis pustulosa, patients with
   AD, and healthy donors. RNA extraction or paraffin sectioning was
   performed on these samples. The procedure for the collection of
   samples, including skin biopsies, was approved by the Ethical Review
   Committees of Shanghai Skin Hospital (no. 2020-04) and Shanghai Tenth
   People’s Hospital (no. 2021KN114). Informed consent was obtained for
   all the procedures. All the patients were volunteers and did not
   receive compensation for their participation in this study and
   provision of their skin samples.

Mice and tamoxifen-mediated deletion of Mettl3 in mice

   C57BL/6J mice were obtained from the Shanghai Laboratory Animal Center.
   Mettl3^fl/fl mice were generated by Shanghai Model Organisms Center
   Inc. The KRT14-CreER^T2 transgenic mice (catalog no. NM-KI-190024) were
   obtained from Shanghai Model Organisms Center Inc. Mettl3^fl/fl mice
   were crossed with the KRT14-CreER^T2 transgenic mice to generate
   KRT14-CreER^T2; Mettl3^fl/fl mice. To achieve conditional KO of Mettl3,
   5-week-old KRT14-CreER^T2 and KRT14-CreER^T2; Mettl3^fl/fl; mice were
   intraperitoneally injected with tamoxifen (catalog no. T5648,
   Sigma-Aldrich) (75 μg per g body weight), which is dissolved in corn
   oil, for five consecutive days. After 9 days of rest, mice were
   euthanized to test the KO efficiency. All animal studies were performed
   with sex-matched and age-matched mice. Mice were housed in groups with
   a 12-hour light/12-hour dark cycle. They had ad libitum access to food
   and water. All animal handling and surgical procedures were conducted
   according to national guidelines and approved by the Tongji University
   Animal Care and Use Committee (protocol no. 2020-029).

Histological analysis and immunofluorescence staining

   For histological analysis, skin tissue was fixed in 4%
   paraformaldehyde, embedded in paraffin, sectioned (4 μm), and stained
   with hematoxylin and eosin (H&E). Sites were randomly selected and
   evaluated using the ImageJ 1.52a software for analysis of epidermal
   hyperplasia (acanthosis) or dermal cell infiltration (number of dermal
   cells per 10,000 μm^2 of tissue). For immunohistochemical analyses,
   paraffin-embedded sections were deparaffinized in xylene and rehydrated
   before antigen retrieval using boiling in an EDTA buffer [1 mM EDTA
   containing 0.05% Tween 20 (pH 9.0)]. Next, the sections were incubated
   in H[2]O[2] for 10 min, permeabilized with 0.5% Triton X-100 for 15
   min, and washed with phosphate-buffered saline (PBS), and this was
   followed by blocking in 3% bovine serum albumin in PBS at room
   temperature for 1 hour and overnight staining at 4°C with primary
   antibodies. The next day, a two/three-color fluorescence kit (Recordbio
   Biological Technology, Shanghai, China) was used to stain the sections
   based on the tyramide signal amplification technology according to the
   manufacturer’s instructions. The fluorescence-stained sections were
   then mounted in ProLong Gold Antifade reagent with DAPI
   (4′,6-diamidino-2-phenylindole) (catalog no. 4083, Cell Signaling
   Technology) and visualized under a microscope (Leica). Detailed
   information about the antibodies and their usage in this study is
   presented in table S3.

RNA isolation and RT-qPCR

   The skin tissues were homogenized in TRIzol (Takara) after they were
   ground in the presence of liquid nitrogen, while the cells were
   resuspended directly in TRIzol. Total RNA was isolated using the
   RevertAid First Strand cDNA Synthesis Kit (Takara) according to the
   manufacturer’s instructions and reverse transcribed into cDNA. RT-qPCR
   was performed using SYBR Green master mix (Takara) on a StepOnePlus
   Real-Time PCR System (Applied Biosystem).

Western blotting

   Skin tissues and cells were lysed in a radioimmunoprecipitation buffer
   (catalog no. 89900, Invitrogen) with a protease inhibitor (catalog no.
   78442, Thermo Fisher Scientific). The lysates were heated at 99.9°C for
   10 min, and the protein samples were separated by SDS–polyacrylamide
   gel electrophoresis gel and transferred to polyvinylidene membranes
   (catalog no. IPVH00010, Millipore). The membranes were blocked with 5%
   nonfat milk prepared in Tris buffer saline plus 0.1% Tween 20 (TBST)
   for 1 hour at room temperature and then incubated with primary
   antibodies overnight at 4°C, and this was followed by incubation with
   goat anti-rabbit IgG (H+L) or goat anti-mouse IgG (H+L) as secondary
   antibodies for 1 hour the next day. Signals from blotted membranes were
   detected with the Odyssey Imaging System after the membranes were
   washed three times with TBST. Detailed information about the antibodies
   and their usage in this study is presented in table S3.

Epidermis separation

   Dorsal skin samples were excised from adult mice and incubated with
   Dispase II (3 mg/ml; catalog no. 17105041, Thermo Fisher Scientific)
   for 16 hours at 4°C, and then the epidermis were separated from the
   dermis under a microscope using a glass slide.

RNA sequencing

   For RNA-seq, Ribo-Zero Gold (Illumina) was used for cDNA library
   construction. To construct the library, the generated cDNAs were
   amplified by PCR. The libraries were validated on the Agilent 2100
   bioanalyzer and sequenced on NovaSeq 6000 (Illumina).

MeRIP sequencing

   MeRIP-seq was performed based on a previously described protocol
   ([248]60). In brief, total RNA from the skin tissue of patients or the
   epidermis of mice together with m^6A+ control RNA (GLuc) and m^6A−
   control RNA (CLuc) was randomly fragmented to 200 to 500 nt through
   sonication (Covaris S220). Subsequently, 1/20 volume of the fragmented
   RNA was retained as input RNA for total RNA-seq library construction.
   The rest was incubated with Protein A beads (catalog no. 10001D, Thermo
   Fisher Scientific) that were coupled with the polyclonal anti-m^6A
   antibody (catalog no. E1610S, NEB) in an immunoprecipitation (IP)
   buffer [10 mM tris-HCl, 150 mM NaCl, RNasin (4 U/μl) (pH 7.4), and 0.1%
   NP-40] for 4 hours. After IP, the RNA reaction mixture was washed twice
   in the IP buffer, twice in a low-salt IP buffer [50 mM NaCl, 10 mM
   tris-HCl (pH 7.4), and 0.1% IGEPAL CA-630 (catalog no. I8896,
   Sigma-Aldrich) in nuclease-free H[2]O], and twice in a high-salt IP
   buffer [500 mM NaCl, 10 mM tris-HCl (pH 7.4), and 0.1% IGEPAL CA-630 in
   nuclease-free H[2]O) for 5 min each time. RNA extraction followed by
   phenol-chloroform extraction and ethanol precipitation was performed to
   purify the m^6A-enriched RNA fragments. Last, the purified RNA from IP
   and input RNA was subjected to RT-qPCR or library construction through
   the SMARTer Stranded Total RNA-Seq Kit version 2 (Takara, 634411). The
   libraries were sequenced on an Illumina NovaSeq 6000 platform (Novogene
   and Nanjing Jiangbei New Area Biophamaceutical Public Service
   Platform).

SN ratio calculation for MeRIP-qPCR

   The following equation was used to calculate the Signal-to-Noise (SN)
   ratio: SN ratio = 2^(−Ct[m6A-positive or target gene in IP] +
   Ct[m6A-positive or target gene in input])/2^(−Ct[m6A-negative gene in
   IP] + Ct[m6A-negative gene in input]).

IMQ-induced psoriasis mouse model and MC903-induced AD mouse model

   For the IMQ-induced psoriasis mouse model, mice aged 7 weeks were
   treated topically with a daily dose of 62.5 mg of IMQ cream (5%)
   (Shichuan MedShine Pharmaceuticals Co.) that was applied to the shaved
   back for the indicated time. For the MC903-induced AD mouse model, 1
   nmol of MC903 (catalog no. HY-10001, MedChemExpress), dissolved in 10
   μl of ethanol, was applied daily to the ears of 7-week-old mice. For
   dorsal application, 4 nmol of MC903, dissolved in 40 μl of ethanol, was
   applied daily to the backs of the mice for the specified duration.

TEWL test

   TEWL was measured on the dorsal skin of mice using the GPSkin Barrier
   device at the indicated time points prior to IMQ or MC903 treatment.
   Measurements were performed in triplicate for each time point, and mean
   values were calculated for further analysis.

Skin cell preparation and flow cytometry

   To obtain whole skin cell suspension, the skin of mice was minced with
   fine scissors after removal of subcutaneous fat. Cells were digested
   with 10 ml of RPMI 1640 medium (catalog no. 11875093, Gibco) containing
   0.25% collagenase (catalog no. C9091, Sigma-Aldrich), 0.01 M Hepes
   (catalog no. BP310, Thermo Fisher Scientific), 0.001 M sodium pyruvate
   (catalog no. BP356, Thermo Fisher Scientific), and DNase
   (deoxyribonuclease) (0.1 mg/ml; catalog no. DN25, Sigma-Aldrich) for 1
   to 1.5 hours at 37°C with rotation, and 1 ml of fetal bovine serum
   (FBS) was added per sample to terminate the digestion. Cells were then
   filtered through a 40-μm filter and were spun down and resuspended in a
   fluorescence-activated cell sorting buffer (2% FBS in PBS). The skin
   cells were incubated with TruStain FcX antibody to block Fc receptors
   for 10 min at room temperature. Isolated cells were stained with
   different cell surface markers (CD45 for leukocytes; CD45^+ and CD3^+
   for T cells; CD45^+, CD11b^+, and Ly6G^+ for neutrophils; CD45^+,
   CD11b^+, and F4/80^+ for macrophages; CD45^+ and CD11c^+ for DCs; and
   CD45^+, CD90.2^+, and γδT cell receptor [TCR^+] for dermal γδT cells or
   DETCs) for 30 min at 4°C. The cells were then incubated with the SYTOX
   Blue Dead Cell Stain (catalog no. S34857, Invitrogen) for 5 min, and
   10,000 counting beads (catalog no. [249]C36950, Invitrogen) were added
   per sample before analysis. Skin cells were analyzed using the FACSAria
   Fusion Sorter (BD Biosciences). Data were analyzed using the FlowJo
   v.10 software (TreeStar). Detailed information about on the antibodies
   and their usage in this study is provided in table S3.

In vivo neutrophil depletion

   For the neutrophil depletion assay, 7-week-old Mettl3 cKO mice were
   treated with 400 μg per 20 g body weight of a rat anti-mouse Ly6G
   antibody (catalog no. BP0075-1, BioXCell) or rat IgG (catalog no.
   BP0089, BioXCell) a day before IMQ treatment, and this was followed by
   treatment with 100 μg per 20 g body weight of a rat anti-mouse Ly6G
   antibody or rat IgG every other day starting from 1 day after IMQ
   treatment until the end of the experiment. The effectiveness of
   neutrophil depletion was assessed by flow cytometry in the mouse skin
   tissue.

RIP assay

   RIP assay was performed using the Magna RIP Kit (Catalog no. 17-700,
   Merck) according to the manufacturer’s instructions. Briefly, 1.5 μg
   each of an anti-IGF2BP3 antibody and anti-rabbit IgG was incubated with
   50 μl of magnetic beads before adding cell lysates were added (~1 ×
   10^7 cells per sample). Then, the RNA-protein IP complexes were washed
   six times and incubated with proteinase K digestion buffer to remove
   proteins. Last, RNA was extracted by phenol-chloroform RNA extraction
   and purified for RT-qPCR analysis. Normalization of the relative
   enrichment was performed using the following formula: %Input = 1/10 ×
   2^Ct [IP] − Ct [input]. Detailed information about the IGF2BP3
   antibodies and their usage in this study is presented in table S3.

Human keratinocyte culture, M5 stimulation, and STM2457 treatment

   In this study, adult HEKa were isolated from human foreskin and
   cultured in EpiLife medium (Gibco) containing EpiLife Defined Growth
   Supplement (Gibco) and 0.06 mM Ca^2+ (Gibco). Cells at passages 3 to 5
   were used for subsequent experiments. The immortalized human
   keratinocyte cell line HaCaT cells (catalog no. CBP6033, Cobioer) were
   used for human keratinocyte-related research. Cells were cultured in
   RPMI 1640 medium (Gibco) containing 10% FBS (Gibco), penicillin (50 U
   ml^−1), and streptomycin (50 μg ml^−1) (Gibco) under standard culture
   conditions. Cells were cultured at 37°C and 5% CO[2]. The cytokine
   cocktail used to stimulate the cells in this study is M5, which is a
   cocktail of recombinant human IL-1α (catalog no. 200-01A, PeproTech),
   IL-17A (catalog no. 200-17, PeproTech), IL-22 (catalog no. 200-22,
   PeproTech), oncostatin M (catalog no. 300-10, PeproTech), and TNF-α
   (catalog no. 300-01A, PeproTech), each at 10 ng ml^−1. For STM2457
   treatment, 10 μM STM2457 (catalog no. HY-134836, MedChemExpress) was
   used to treat HaCaT cells for 48 hours before collection.

m^6A ELISA

   RNAs of HaCaT cells treated with STM2457 or DMSO for 48 hours was
   isolated as described above. m^6A modification on RNAs was measured
   with the EpiQuik m^6A RNA Methylation Quantification Kit (catalog no.
   P-9005, EpigenTek) according to the manufacturer’s instructions. ELISA
   on all samples was performed at the same time.

Topical application of MP3C or the ELOVL6 inhibitor

   For topical treatment with MP3C (catalog no. M193627, Aladdin Biochem),
   it was applied at 10 mg/kg daily for 5 days. C57BL/6J mice (7 weeks old
   at the start of the study) were treated topically with MP3C or its
   vehicle (ethanol, EtOH) daily for 5 days daily after treatment with a
   single application of IMQ or MC903 until the end of the study (6 to 8
   hours before each application of IMQ or MC903). For ELOVL6 inhibitor
   (ELOVL6-IN) treatment, 2 μg of ELOVL6-IN-2 (catalog no. HY-12146,
   MedChemExpress) (diluted with DMSO) was mixed with 2 mg of oil-in-water
   cream per gram of mouse body weight, as required for immediate
   application. C57BL/6J mice (7 weeks old at the start of the study) were
   treated topically with mixed ELOVL6-IN cream or vehicle for 3 days as a
   pretreatment procedure, and then it was applied daily until the end of
   the study (6 to 8 hours before each application of IMQ or MC903).

Human skin biopsy cultures

   Skin biopsy samples were processed according to a previously published
   study ([250]61). Briefly, human skin biopsy samples from healthy donors
   were incubated with or without 1 mM MP3C and M5/PBS for 6 hours after 2
   hours of starvation. For culture of psoriatic lesions, each skin biopsy
   specimen was equally divided into two equal portions, each containing
   epidermal and dermal structures, and then treated with or without 1 mM
   MP3C for 24 hours. The specimens were then cultured in RPMI 1640 medium
   (Gibco) containing 10% FBS (Gibco), penicillin (50 U ml^−1), and
   streptomycin (50 μg ml^−1) (Gibco). After the treatment, the specimens
   were snap frozen in liquid nitrogen and then subjected to RNA
   isolation.

Gene silencing experiments

   For ELOVL6 and IGF2BP1/2/3 silencing in HaCaT cells, small interfering
   RNAs (siRNAs) targeting human ELOVL6, IGF2BP1, IGF2BP2, or IGF2BP3 were
   synthesized by Hippo Biotechnology Co. Ltd. (Huzhou, China). Using the
   Lipofectamine RNAiMAX Reagent (Invitrogen), siRNAs targeting ELOVL6 and
   IGF2BP1-3 (table S2) were transfected into HaCaT cells according to the
   manufacturer’s instructions. Silencing efficiency was analyzed by
   RT-qPCR and Western blotting.

RNA stability assay

   Cells were seeded in 12-well plates overnight. After being pretreatment
   with STM2457 or DMSO for 48 hours, the cells were then treated with
   actinomycin D (5 μg/ml; S8964, Selleckchem) for 0, 3, or 6 hours. Total
   RNA was isolated using TRIzol and quantified by RT-qPCR.

Fatty acid composition

   Lipids from the mouse epidermis were extracted according to the method
   of Bligh and Dyer as previously described ([251]62). Briefly,
   chloroform/methanol (1:2, v/v) was used to extract the epidermis. To
   break the monophase, 1 M each of NaCl solution and chloroform was
   added. The samples were incubated on ice for a 10 min and centrifuged
   at 300g for 5 min. Then, the aqueous solution was discarded, and the
   chloroform phase was evaporated with nitrogen gas. Samples were
   incubated at 100°C for 45 min after addition of acetonitrile/6 N HCl
   (90:10, v/v). Liquid-liquid extraction ([252]63) with ethyl acetate was
   lastly performed, and the reconstituted samples were injected into an
   optimized gas chromatography–mass spectrometry (GC-MS) system.

RNA-seq and MeRIP-seq analysis

   RNA reads with adaptors and of low quality were removed using
   Trim_galore. Clean reads were aligned to the reference genome (UCSC
   hg38) using the STAR software. As described previously ([253]60),
   m^6A-modified spike-in (Gluc) was used to normalize mouse input and IP
   signals. The normalization of the genome coverage BigWig files of input
   and IP signals was performed as previously described ([254]60). Peak
   calling was first performed using MetPeak, and the peaks obtained were
   further filtered by calculating the IP/input signal ratio using scaled
   genome coverage BigWig files. Peaks with IP/input ratios > 1.5 were
   retained for further analysis. Differential gene expression analysis in
   this study was performed using the R package DESeq2 DiffBind v3 and was
   used to identify differential m^6A peaks. In our RNA-seq analysis, we
   determined DEGs using cutoffs: an absolute log[2] FC (|log[2]FC|) ≥ 1
   and a significance level of adjusted P value < 0.05. In our MeRIP-seq
   analysis, we determined differentially methylated peaks (DMPs) using
   cutoffs: an absolute log[2] FC (|log[2]FC|) ≥ 1 and a significance
   level of FDR (false discovery rate) < 0.05.

Statistics and reproducibility

   The in vivo experiments were repeated three times independently. The
   symbols in the figures indicate the exact n for each experimental
   group/condition. Each symbol represents an individual mouse. The in
   vitro experiments were performed in triplicate and were repeated two or
   three times independently of each other. To avoid technical bias, each
   sample was processed identically in each experiment, and internal
   controls and normalization methods were included. All data are
   presented as means ± SEM. The significance of differences between two
   groups was determined using unpaired, two-tailed Student’s t test, and
   one-way or two-way analysis of variance (ANOVA) was used for multiple
   group analyses using GraphPad Prism v.9.

ACKNOWLEDGMENTS