Introduction

While the infiltration of immune cells into skin plays a critical role in the development of psoriasis, as evidenced by interleukin (IL) -23/IL-17 axis (Fitch, Harper, Skorcheva, Kurtz, & Blauvelt, 2007; Hawkes, Yan, Chan, & Krueger, 2018; Kim & Krueger, 2017), recent studies have revealed that cells constructing skin structure, such as keratinocytes, fibroblasts, and endothelial cells, also play pivotal roles in the development (Heidenreich, Röcken, & Ghoreschi, 2009; Lowes, Russell, Martin, Towne, & Krueger, 2013; Zhang et al., 2023) and maintenance (Arasa et al., 2019; Francis et al., 2024; Li et al., 2023; Ma et al., 2023; Tan et al., 2015; Zhu et al., 2020) of psoriasis. Among these cells, keratinocytes function as a barrier and produce cytokines, chemokines, and antimicrobial peptides against foreign stimuli, resulting in the activation of immune cells (Ni & Lai, 2020; Zhou, Chen, Cui, Shi, & Guo, 2022).

Semaphorins were initially identified as guidance cues in neural development (Kolodkin, Matthes, & Goodman, 1993; Pasterkamp & Kolodkin, 2003) but are now regarded to play crucial roles in other physiological processes including angiogenesis (Iragavarapu-Charyulu, Wojcikiewicz, & Urdaneta, 2020; Serini, Maione, Giraudo, & Bussolino, 2009), tumor microenvironment (Hui, Tam, Jiao, & Ong, 2019; Jiang et al., 2022; Nakayama, Kusumoto, Nakahara, Fujiwara, & Higashiyama, 2018; Rajabinejad et al., 2020), and immune systems (Garcia, 2019; Kanth, Gairhe, & Torabi-Parizi, 2021; M. Naito & Kumanogoh, 2023). Semaphorin 4A (Sema4A), one of the immune semaphorins, plays both pathogenic and therapeutic roles in autoimmune diseases (Cavalcanti et al., 2020; He et al., 2023), allergic diseases (Maeda et al., 2019), and cancer (Iyer & Chapoval, 2018; Liu et al., 2018; Y. Naito et al., 2023; Pan, Wang, & Ye, 2016). While Sema4A expression on T cells is essential for T helper type 1 differentiation in the murine Propionibacterium acnes-induced inflammation model and delayed-type hypersensitivity model (Kumanogoh et al., 2005), Sema4A amplifies only T helper type 17 (Th17)-mediated inflammation in the effector phase of murine experimental autoimmune encephalomyelitis (Koda et al., 2020). Accordingly, in multiple sclerosis, high serum Sema4A levels correlate with the elevated serum IL-17A, earlier disease onset, and increased disease severity (Koda et al., 2020; Nakatsuji et al., 2012). In anti-tumor immunity, research involving human samples and murine models suggests that Sema4A expressed in cancer cells and regulatory T cells promotes tumor progression (Delgoffe et al., 2013; Liu et al., 2018; Pan et al., 2016), while other reports reveal that Sema4A in cancer cells and dendritic cells bolsters anti-tumor immunity by enhancing CD8 T cell activity (Y. Naito et al., 2023; Suga et al., 2021). In addition to these roles in immune reactions, mice with a point mutation in Sema4A develop retinal degeneration, suggesting that Sema4A is also crucial for peripheral tissue homeostasis (Nojima et al., 2013). These reports suggest that the role of Sema4A can differ based on the disease, phase, and involved cells.

Herein, we investigated the roles of Sema4A in the pathogenesis of psoriasis by analyzing skin and blood specimens from psoriatic subjects and using a murine model.

Results

Epidermal Sema4A expression is downregulated in psoriasis

The analysis of previously-published single-cell RNA-sequencing data from control (Ctl) and psoriatic lesional (L) skin specimens (Kim et al., 2023) revealed detectable expression of SEMA4A in keratinocytes, dendritic cells, and macrophages in both Ctl and L. SEMA4A expression was low in neural crest-like cells, fibroblasts, CD4 T cells, CD8 T cells, NK cells, and plasma cells, making the comparison of expression levels impractical (Figure 1A and B; Figure 1-figure supplement 1A-C). While dendritic cells and macrophages showed comparable SEMA4A expression levels between Ctl and L, expression in keratinocytes was significantly downregulated in L (Figure 1C).

Epidermal Sema4A expression is downregulated in psoriasis.

(A) UMAP plots, generated from single-cell RNA-sequencing data (GSE220116), illustrate cell distributions from control (Ctl) and psoriatic lesion (L) samples (n = 10 for Ctl, n = 11 for L). (B) Subclustering of immune cells. (C) SEMA4A expression in keratinocytes, dendritic cells, and macrophages. ****padj < 0.001. NS, not significant. Analyzed using Python and cellxgene VIP. (D) Representative immunohistochemistry and magnified views showing Sema4A expression in Ctl, psoriatic non-lesion (NL), and L. Scale bar = 50 μm. (E) Mean epidermal (Epi) Sema4A intensity in immunohistochemistry (n = 10 per group). Each dot represents the average intensity from 5 unit areas per sample. (F) Relative SEMA4A expression in Epi (n = 10 for Ctl, L, n = 7 for L and NL) and dermis (Derm, n = 6 per group). (G) Proportions of Sema4A-expressing cells in blood CD4 T cells (left), CD8 T cells (middle), and monocytes (right) from Ctl and psoriatic (Pso) patients. (n = 13 per group in CD4 and CD8, n = 11 for Ctl and n = 13 for Pso in Monocytes). (H) Serum Sema4A levels in Ctl (n = 20) and Pso (n = 60). E to H: *p < 0.05, **p < 0.01, ****p < 0.0001. NS, not significant.

Figure1-source data1

Excel file containing quantitative data for Figure 1.

Immunohistochemistry of Ctl and psoriasis demonstrated Sema4A expression in keratinocytes (Figure 1D). The staining intensity of Sema4A in epidermis was significantly lower in both non-lesions (NL) and L than in Ctl (Figure 1E). Relative mRNA expression of SEMA4A was also decreased in the epidermis of NL and L compared to Ctl, while it remained comparable in the dermis (Figure 1F). In contrast, the proportions of Sema4A-positive cells were significantly higher in blood CD4 and CD8 T cells, and monocytes in psoriasis compared to Ctl (Figure 1G; Figure 1-figure supplement 2). Serum Sema4A levels, measured by enzyme-linked immunosorbent assay (ELISA), were comparable between Ctl and psoriasis (Figure 1H). These findings demonstrate that the expression profile of Sema4A in psoriasis varies across cell types.

Psoriasis-like dermatitis is augmented in Sema4AKO mice

When psoriasis-like dermatitis was induced in wild-type (WT) mice and Sema4A knockout (KO) mice by imiquimod application on ears (Figure 2A), ear swelling on day 4 was more pronounced in Sema4AKO mice (Figure 2B) with upregulated Il17a gene expression (Figure 2C). Flow cytometry analysis of cells isolated from the ears revealed increased proportions of Vγ2+ T cells, Vγ2-Vγ3- double-negative (DN) γδ T cells, and IL-17A-producing cells of those fractions in Sema4AKO epidermis (Figure 2D and E; Figure 2-figure supplement 1). In Sema4AKO dermis, there was also an increase in the proportions of Vγ2+ T cells and IL-17A-producing Vγ2+ T cells (Figure 2D and E). These results suggest that Sema4A deficiency in mice accelerates psoriatic profile. IL-17A-producing T cells in skin-draining lymph nodes (dLN) remained comparable between WT mice and Sema4AKO mice (Figure 2F).

Psoriasis-like dermatitis is augmented in Sema4AKO mice.

(A) Experimental scheme. Wild-type (WT, green) mice and Sema4A knockout (KO, black) mice were treated with 10 mg/ear of 5% Imiquimod (IMQ) for 4 consecutive days. Samples for flow cytometry analysis were collected on Day 4. (B) Ear thickness of WT mice and KO mice on Day 4 (n = 15 per group). (C) Relative expression of Il17a in epidermis (n = 5 per group). (D and E) The percentages of Vγ3, Vγ2, Vγ2-Vγ3-γδ (DNγδ), CD4, and CD8 T cells (D) and those with IL-17A production (E) in CD3 fraction in the Epi (top) and Derm (bottom) of WT and KO ears (n = 6 per group, each dot represents the average of 4 ear specimens). (F) The percentages of IL-17A-producing γδ, CD4, and CD8 T cells in CD3 fraction in skin-draining lymph nodes (dLN) (n = 9 per group). B-F: *p < 0.05, **p < 0.01, ****p < 0.0001. NS, not significant.

Figure2-source data1

Excel file containing quantitative data for Figure 2.

Sema4A in keratinocytes may play a role in preventing murine psoriasis-like dermatitis

To investigate the cells responsible for the augmented ear swelling in Sema4AKO mice, bone marrow chimeric mice were next analyzed (Figure 3A). Mice deficient in non-hematopoietic Sema4A (WT→KO) displayed more pronounced ear swelling than mice with intact Sema4A expression (WT→WT) following imiquimod application (Figure 3B). Similarly, mice with a systemic deficiency of Sema4A (KO→KO) showed severe ear swelling compared to mice deficient in hematopoietic Sema4A (KO→WT) (Figure 3B). Ear swelling was comparable between WT→WT mice and KO→WT mice (Figure 3B). Flow cytometry analysis revealed increased infiltration of IL-17A-producing DN γδ T cells in the epidermis, as well as Vγ2+ T cells and IL-17A-producing Vγ2+ T cells in the dermis, in WT→KO mice compared to WT→WT mice (Figure 3C and D). These findings suggest that non-hematopoietic cells, possibly keratinocytes, are primarily responsible for the increased imiquimod-induced Sema4AKO mice ear swelling.

Sema4A in keratinocytes may play a role in preventing murine psoriasis-like dermatitis.

(A) Experimental scheme for establishing BM chimeric mice. (B) Day 4 ear thickness in the mice with the indicated genotypes (n = 14 for WT→WT, n = 13 for WT→KO, n =9 for KO→WT, n = 9 for KO→KO). (C and D) The percentages of Vγ3, Vγ2, DNγδ, CD4, and CD8 T cells (C) and IL-17A production (D) in CD3 fraction from Day 4 Epi (top) and Derm (bottom) of the ears from WT→ WT mice and WT→ KO mice (n = 6 per group, each dot represents the average of 4 ear specimens). B-D: *p < 0.05, **p < 0.01, ****p < 0.0001. NS, not significant.

Figure3-source data1

Excel file containing quantitative data for Figure 3.

Sema4AKO epidermis is thicker than WT epidermis with increased γδ T17 infiltration

Even without imiquimod application, Sema4AKO ears turned out to be slightly but significantly thicker than WT ears on week 8 while their appearance remained normal (Figure 4A). While epidermal thickness of back skin was comparable at birth (Figure 4B), on week 8, ear epidermis of Sema4AKO mice was notably thicker than that of WT mice (Figure 4B), suggesting that acanthosis in Sema4AKO mice is accentuated post-birth. Dermal thickness remained comparable between WT mice and Sema4AKO mice at both times (Figure 4B). WT epidermis showed significantly higher Sema4a mRNA expression compared to the WT dermis (Figure 4C). Based on these observations, Sema4A appears to play a more pronounced role in epidermis than in dermis.

Naïve Sema4AKO epidermis is thicker than WT epidermis with increased γδ T17 infiltration.

(A) Ear thickness of WT mice and KO mice at week (wk) 8 (n = 15 per group) and representative images. (B) Left: representative hematoxylin and eosin staining of day 0 back and wk 8 ear. Scale bar = 50 μm. Right: Epi and Derm thickness in d 0 back (n = 5) and wk 8 ear (n = 8). (C) Relative Sema4a expression in WT Epi and Derm (n = 5 per group). (D) Relative expression of psoriasis-associated genes in Epi (top) and Derm (bottom) of WT mice and KO mice (n = 5 per group, #: not detected). (E) Representative dot plots showing CD69 and CD103 expression in the indicated T cell fractions from whole skin. The graphs show T cell counts per ear (top) and those with resident memory phenotype (bottom) (n = 7 per group). (F and G) The percentages of the indicated T cell fractions (F, n = 8 per group) and IL-17A production (G, n = 4 per group) in Epi (top) and Derm (bottom). E-G: Each dot represents the average of 4 ear specimens. A-G: *p < 0.05, **p < 0.01, ****p < 0.0001. NS, not significant.

Figure4-source data1

Excel file containing quantitative data for Figure 4.

Sema4AKO epidermis exhibited increased expression of Ccl20, Tnfα, and Il17a and a trend of upregulation of S100a8 compared to WT epidermis (Figure 4D). These differences were not observed in dermis (Figure 4D). Flow cytometry analysis revealed increased infiltration of γδ T cells in Sema4AKO ear (Figure 4E). These cells predominantly expressed resident memory T cell (TRM)-characteristic molecules, CD69 and CD103 (Figure 4E). Sema4AKO skin also had a higher number of TRM in both CD4 and CD8 T cells (Figure 4E). The percentages of Vγ2+ T cells, DNγδ T cells, and CD8 T cells in epidermis were higher in Sema4AKO mice than in WT mice, which was not the case in dermis (Figure 4F). The proportion of Vγ3+ dendritic epidermal T cells was comparable between WT mice and Sema4AKO mice (Figure 4F). Epidermal Vγ2+ T cells and DNγδ T cells in Sema4AKO mice showed higher IL-17A-producing capability (Figure 4G), while IFNγ and IL-4 production was comparable between WT mice and Sema4AKO mice (Figure 4-figure supplement 1A and B). Conversely, the frequency of IL-17A-producing T cells from dLN was comparable (Figure 4-figure supplement 1C). The production of IL-17A and IFNγ from splenic T cells under T17-polarizing conditions remained consistent between WT mice and Sema4AKO mice (Figure 4-figure supplement 2).

Taken together, it is suggested that T17 cells are specifically upregulated in epidermis, indicating that the epidermal microenvironment plays a pivotal role in facilitating the increased T cell infiltration observed in naïve Sema4AKO mice.

Sema4AKO skin shares features with human psoriatic NL

Previous literatures have identified certain features common to psoriatic L and NL, such as thickened epidermis (Figure 5-figure supplement 1) (Gallais Sérézal et al., 2019), CCL20 upregulation (Gallais Sérézal et al., 2019), and accumulation of IL-17A-producing T cells (Vo et al., 2019), which were detected in Sema4AKO mice.

Gene expression analysis using public RNA-sequencing data (Tsoi et al., 2019) with RaNA-seq (Prieto & Barrios, 2019) showed upregulation of keratinization and antimicrobial peptide genes in NL compared to Ctl (Figure 5A). Gene Ontology analysis highlighted an upregulation in biological processes, predominantly in peptide cross-linking involved in epidermis formation and keratinocyte differentiation, with a secondary increase in the defense response to viruses in NL (Figure 5B). While the expression of Keratin (KRT) 10 was comparable between NL and Ctl, upregulation in KRT5, KRT14, and KRT16 was observed in NL (Figure 5C).

Psoriatic NL shows aberrant differentiation of keratinocytes.

(A and B) The volcano plot (A) and Gene ontology (GO) analysis (B), generated from RNA-sequencing data (GSE121212) using RaNA-seq, display changes in gene expression in psoriatic NL compared to Ctl. (C) The difference in the expression of epidermal differentiation markers between Ctl and NL (n = 38 for Ctl, n = 27 for NL) was calculated with the transcripts per million values. **padj < 0.01. NS, not significant.

Figure5-source data1

Excel file containing quantitative data for Figure 5.

In the murine model, relative expression levels of Krt10, Krt14, Krt16, and filaggrin were elevated in Sema4AKO epidermis. (Figure 6A). Immunofluorescence analysis showed that Sema4AKO epidermis had a higher density of keratinocytes positive for Krt5, Krt10, Krt14, and Krt16 compared to WT epidermis (Figure 6B). This upregulation was not observed in back skin at birth (Figure 6-figure supplement 1). Comparable transepidermal water loss between WT mice and Sema4AKO mice indicated preserved skin barrier function in Sema4AKO mice (Figure 6C).

Sema4AKO skin shares the features of human psoriatic NL.

(A) Relative gene expression of epidermal differentiation markers between wk 8 Epi of WT mice and KO mice (n = 5 for Krt14 and Krt16, n = 8 for Krt5, Krt10, Filaggrin, and Loricrin). (B) Left: Representative immunofluorescence pictures of Krt5, Krt10, Krt14, and Krt16 (red) overlapped with DAPI. Scale bar = 50 μm. Right: Accumulated graphs showing the numbers of Krt5, Krt10, Krt14, and Krt16 positive cells per 100 μm width (n = 5 per group) of wk 8 ear (right). Each dot represents the average from 5 unit areas per sample. (C) Transepidermal water loss (TEWL) in back skin of WT mice and KO mice at wk 8 (n = 5 per group). A-C: *p < 0.05, **p < 0.01. NS, not significant.

Figure 6-source data1

Excel file containing quantitative data for Figure 6.

Based on these results, it is implied that the epidermis of human psoriatic NL and Sema4AKO mice exhibit shared pathways, potentially leading to an acanthotic state. Combined with the observed acanthosis and increased T17 infiltration in Sema4AKO mice, Sema4AKO skin is regarded to demonstrate the features characteristic of human psoriatic NL.

mTOR signaling is upregulated in the epidermis of psoriatic NL and Sema4AKO mice

Previous reports have shown that mTOR pathway plays a critical role in maintaining epidermal homeostasis, as evidenced by mice with keratinocyte-specific deficiencies in Mtor, Raptor, or Rictor, which exhibit a hypoplastic epidermis with impaired differentiation and barrier formation (Asrani et al., 2017; Ding et al., 2016; Ding et al., 2020). We thus investigated mTOR pathway in both human and murine epidermis.

Immunohistochemical analyses highlighted the increase in phospho-S6 (p-S6), indicating the upregulation of mTOR complex (C) 1 signaling, in the epidermal upper layers of both L and NL compared to Ctl (Figure 7A). The activation of mTORC2 in the epidermis, inferred from phospho-Akt (p-Akt), was scarcely detectable in L, NL, and Ctl (Figure 7A). In the epidermis of WT mice and Sema4AKO mice, the upregulation of both mTORC1 and mTORC2 signaling was observed in Sema4AKO epidermis by immunohistochemistry (Figure 7B). Western blot analysis showed upregulated mTORC1 signaling in Sema4AKO epidermis compared to WT epidermis (Figure 7C). The enhancement of mTORC1 and mTORC2 signaling became obvious in the Sema4AKO epidermis after developing psoriatic dermatitis by imiquimod application (Figure 7D).

mTOR signaling is upregulated in the epidermis of psoriatic NL and Sema4AKO mice.

(A) Representative results of immunohistochemistry displaying cells positive for p-S6 (Ser235/236), S6, p-Akt (Ser473), and Akt in Ctl, NL, and L. The graphs of accumulated data show the mean intensity of p-S6 and S6 in the upper and lower epidermal layers (n = 9 per group). Scale bar = 100 μm. Each dot represents the average mean intensity from 5 unit areas per sample. (B) The mean intensity of p-S6 (Ser235/236) and p-Akt (Ser473), detected by immunohistochemistry in the epidermis of WT mice and KO mice, were analyzed. Scale bar = 50 μm. Each dot represents the average intensity from 5 unit areas per sample (n = 8 per group). (C and D) Immunoblotting of p-S6 (Ser235/236), S6, p-Akt (Ser473), and Akt in tissue lysates from epidermis without treatment (C) and with IMQ treatment for consecutive 4 days (D) (n = 5 per group, except for p-Akt and Akt in C, for which n = 4). A-D: *p < 0.05, **p < 0.01. NS, not significant.

Figure 7-source data1

Excel file containing quantitative data for Figure 7.

Inhibition of mTOR signaling modulates cytokeratin expression in Sema4AKO mice

To investigate the contribution of mTORC1 and mTORC2 signaling in the development of psoriatic features in the Sema4AKO epidermis, mTORC1 inhibitor rapamycin and mTORC2 inhibitor JR-AB2-011 were intraperitoneally applied to Sema4AKO mice for 14 days. Although epidermal thickness remained unchanged by the inhibitors (Figure 8A and B), relative gene expression of Krt5 was significantly upregulated and that of Krt16 was significantly downregulated after rapamycin application (Figure 8C). While the upregulation of Il17a in Sema4AKO epidermis was not clearly modified by rapamycin (Figure 8C), immunofluorescence revealed the decrease in the number of CD3 T cells in Sema4AKO epidermis by rapamycin (Figure 8D). We additionally conducted topical application of rapamycin gel and vehicle gel on the left and right ears of Sema4AKO mice, respectively. Although there were no detectable changes in epidermal thickness and epidermal T cell counts, the upregulation of Krt5 and downregulation of Krt16 was observed again (Figure 8-figure supplement 1). Conversely, the application of JR-AB2-011 resulted in decreased expression of Krt5, Krt10, and Krt14 with a trend towards increased Krt16 expression (Figure 8E). JR-AB2-011 did not influence the number of infiltrating T cells in the epidermis (Figure 8F). mTORC1 primarily regulates keratinocyte proliferation, whereas mTORC2 mainly involved in the keratinocyte differentiation through Sema4A-related signaling pathways. The observed similarities in Il17a expression following treatment with rapamycin and JR-AB2-011 suggest that the production of these cytokines is not significantly dependent on Sema4A-related mTOR signaling.

Inhibitors of mTOR signaling modulate the expression of cytokeratins in Sema4AKO mice.

(A and B) Epidermal thickness of Sema4AKO mice treated intraperitoneally with vehicle (Ctl) or rapamycin (A), and Ctl or JR-AB2-011 (B) (n = 5 per group). (C and D) Relative expression of keratinocyte differentiation markers and Il17a in Sema4AKO Epi (C), and the number of T cells in Epi and Derm under Ctl or rapamycin (D) (n = 5 per group). (E and F) Relative expression of keratinocyte differentiation markers and Il17a in Sema4AKO Epi (E), and the number of T cells in Epi and Derm under Ctl or JR-AB2-011 (F) (n = 5 per group). D and F: Each dot represents the sum of numbers from 10 unit areas across 3 specimens. A-F: *p < 0.05, **p < 0.01. NS, not significant.

Figure 8-source data1

Excel file containing quantitative data for Figure 8.

Discussion

Recent studies have highlighted the significance of IL-17A-producing TRM in psoriatic NL (Cheuk et al., 2014; Gallais Sérézal et al., 2018; Vo et al., 2019; Vu, Koguchi-Yoshioka, & Watanabe, 2021). This condition is also characterized by epidermal thickening with elevated CCL20 expression (Gallais Sérézal et al., 2019) and a distinct cytokeratin expression pattern, marked by increased levels of Krt5, Krt14, and Krt16 (Tsoi et al., 2019) (Figure 5C). Here, we identified that murine Sema4A deficiency could induce these features of psoriatic NL.

Sema4A expression in epidermis, but not in dermis, was diminished in psoriatic L and NL compared to Ctl. While Sema4A expression on blood immune cells was upregulated in psoriasis, its serum levels were comparable between Ctl and psoriasis. Despite the reported involvement of increased Sema4A expression on immune cells in the pathogenesis of multiple sclerosis (Koda et al., 2020; Nakatsuji et al., 2012), our results suggest that the diminished expression of Sema4A in skin-constructing cells plays a more prominent role in the pathogenesis of psoriasis than its increased expression on immune cells.

In our murine model, regardless of imiquimod application, IL-17A-producing T cells were increased in Sema4AKO skin although their frequency was comparable in dLN of WT mice and Sema4AKO mice. Additionally, the potential for in vitro T17 differentiation did not differ between T cells from WT mice and Sema4AKO mice. These findings suggest that the absence of Sema4A in the skin microenvironment plays a crucial role in the localized upregulation of IL-17A-producing T cells in Sema4AKO mice.

It is well-documented that upregulated mTORC1 signaling promotes the pathogenesis of psoriasis (Buerger, 2018; Karagianni, Pavlidis, Malakou, Piperi, & Papadavid, 2022; Ruf, Andreoli, Itin, Pluschke, & Schmid, 2014), and that rapamycin can partially ameliorate the disease activity in psoriatic subjects and murine models (Bürger et al., 2017; Gao & Si, 2018; Reitamo et al., 2001). Our results using Sema4AKO mice revealed that the inhibition of mTORC1 leads to the downregulation of Krt16 and that of mTORC2 leads to the downregulation of undifferentiated keratinocytes, while the Sema4AKO epidermal thickening and the upregulated IL-17A signaling is not reversed by mTOR blockade. It is plausible that the downregulation of Sema4A can lead to the upregulation of mTORC1 and mTORC2 signaling in keratinocytes, and the augmented signaling leads to the psoriatic profile of proliferation and differentiation, which is a part of the psoriatic NL disposition.

This study has limitations. Sema4A expression in skin cells other than keratinocytes was not thoroughly investigated. However, since single-cell RNA-sequencing has shown that Sema4A is predominantly expressed in keratinocytes, dendritic cells, and macrophages, with a notable reduction in keratinocytes in psoriasis, we infer that keratinocytes are the primary cells responsible for the psoriatic features resulting from Sema4A downregulation. We were not able to determine whether Sema4A functions as a ligand or a receptor in epidermis (Ito & Kumanogoh, 2016; Kumanogoh et al., 2002; Lu et al., 2018; Sun et al., 2017) in this study, either. Although both mTORC1 and mTORC2 signals are upregulated in Sema4AKO epidermis, we were not able to confirm mTORC2 signaling from human skin due to technical limitation and sample limitation, while the results from augmented mTORC2 signal in Sema4AKO mice and the normalization of Krt5 and Krt14 by mTORC2 inhibitor imply the involvement of mTORC2 signal in psoriatic epidermis. The role of Sema4A other than mTOR signaling, which may be involved in the regulation of T17 induction in skin, is not discussed, either. These limitations should be overcome in the near future.

In summary, epidermal Sema4A downregulation can reflect the psoriatic non-lesional features. Thus, targeting the downregulated Sema4A and upregulated mTOR signaling in psoriatic epidermis can be a promising strategy for psoriasis therapy and modification of psoriatic diathesis in NL for the prevention of disease development.

Materials and methods

Human sample collection

Psoriatic L and NL skin specimens were acquired from 17 psoriasis patients. Ctl specimens were obtained from 19 subjects who underwent tumor resection or reconstructive surgery. For epidermal-dermal separation, specimens were incubated overnight at 4℃ with 2.5 mg/mL dispase (Wako) in IMDM (Wako). Blood samples were collected from 73 psoriasis and 33 Ctl. In addition to ELISA (Nakatsuji et al., 2012), peripheral blood mononuclear cells were isolated using Ficoll-Paque PLUS density gradient media (Cytiva). Patient details are provided in Supplemental Table 1.

Mice

C57BL/6J WT mice were procured from CLEA Japan. Semaphorin 4A KO mice with C57BL/6J background were generated as previously described (Kumanogoh et al., 2005). We examined female mice in order to reduce the result variation. Neonatal mice and female mice aged 8 to 12 weeks, maintained under specific pathogen-free conditions, were used.

To develop psoriasis-like dermatitis, 10 mg of 5% imiquimod cream (Mochida) was applied to both ears for 4 consecutive days. Ear thickness was measured using a thickness gauge (Peacock).

For bone marrow reconstitution, 2 × 105 bone marrow cells obtained from the indicated strains were intravenously transferred to recipient mice that had received a single 10 Gy irradiation. The reconstituted mice were subjected to the experiments in 8 weeks.

In the specified experiments, skin specimens were separated into epidermis and dermis by 5 mg/mL dispase (Wako) in IMDM for 30 minutes at 37℃.

Transepidermal water loss measurements were performed on the back skin of mice at week 8 using a Tewameter (Courage and Khazaka Electronic GmbH), according to the manufacturer’s instructions.

Immunofluorescence and immunohistochemistry

Specimens were fixed in 4% paraformaldehyde phosphate buffer solution (Wako), embedded in paraffin, and sliced into 3-μm thickness on glass slides. After deparaffinization and rehydration, antigen retrieval was performed using citrate buffer (pH 6.0, Nacalai tesque) or TE buffer (pH 9.0, Nacalai tesque).

For immunohistochemistry, samples were incubated with 3% H2O2 (Wako) for 5 minutes. After blocking (Agilent), the specimens were incubated with the indicated primary antibodies under specified conditions (Supplemental Table 2). The specimens were applied with the EnVision Detection Systems Peroxidase/DAB, Rabbit/Mouse (Agilent) and counterstained with hematoxylin (Wako). For immunofluorescence, the blocked specimens were incubated with the indicated primary antibodies followed by the secondary antibodies (Table E4). Mounting media with DAPI (VectaShield) was used. Slides were observed using a fluorescence microscope (BZ-700, Keyence). The staining intensity was measured using Fiji software (ImageJ, National Institutes of Health) over lengths of 200 μm in human samples and 100 μm in murine samples. This measurement was taken from five areas of each specimen, and the average score is presented. The average thickness of epidermis and dermis from 10 spots is presented. The number of epidermal cells positive for the indicated cytokeratins was counted per 100 μm width, with the average number from 5 areas being presented. Additionally, the numbers of CD3-positive cells in murine ears were counted across 10 fields of 400 μm width, with the total sum being presented.

Quantitative reverse transcription polymerase chain reaction (qRT-PCR)

Total RNA was extracted from homogenized skin tissue using Direct-zol RNA MiniPrep Kit (ZYMO Research). cDNA was synthesized by High-Capacity RNA-to-cDNATM Kit (Thermo Fisher Scientific). qPCR was performed using TB Green Premix Ex Taq II (Takara-bio) on a ViiA 7 Real-Time PCR System (Thermo Fisher Scientific). The primers are listed in Supplemental Table 3. Relative gene expression levels were normalized to the housekeeping gene GAPDH using the ΔΔCT technique.

Flow cytometry analysis

Single-cell suspensions from murine skin-dLN and spleen were prepared by grinding, filtering, and lysing red blood cells (BioLegend). Skin specimens were minced and digested with 3 mg/mL collagenase III (Worthington Biochemical Corporation) in RPMI 1640 medium (Wako) at 37℃ for 10 minutes for epidermis, and 30 minutes for dermis or whole skin. Cells were surface-stained with directly conjugated monoclonal antibodies (Supplemental Table 4). Dead cells were identified using LIVE/DEADTM Fixable Dead Cell Stain Kit (Thermo Fisher Scientific). To evaluate cytokine production, cells were stimulated with PMA (50 ng/mL, Wako) and ionomycin (1000 ng/mL, Wako), plus Golgiplug (BD Biosciences) for 4 hours before surface staining. Fixation, permeabilization, and intracellular cytokine staining were performed using BD Cytofix/CytopermTM Fixation/Permeabilization Kit (BD Biosciences) according to the manufacturer’s protocol. In specified experiments, the numbers of each cell subset per ear were estimated using CountBright Absolute Counting Beads (Thermo Fisher Scientific). Sample analysis was conducted using BD FACSCanto II (BD Biosciences), and data were analyzed with Kaluza software (Beckman Coulter).

In vitro Th17 differentiation

Murine splenic T cells were isolated using Pan T Cell Isolation Kit II (Miltenyi Biotec). Two hundred thousand T cells per well were cultured in 96-well plates in the presence of T Cell Activation/Expansion Kit (Miltenyi Biotec) for 2 weeks. The medium was supplemented twice per week with the following recombinant cytokines: mouse recombinant IL-6 (20 ng/mL), IL-1β (20 ng/mL), and IL-23 (40 ng/mL) for the IL-23 dependent Th17 cell condition; IL-6 (20 ng/mL) and TGFβ1 (3 ng/mL) for the IL-23 independent Th17 cell condition; or IL-23 (40 ng/mL) for the IL-23 only Th17 cell condition. The cytokines listed were purchased from BioLegend (Supplemental Table 5). Afterward, the cultured cells were processed for flow cytometry analysis.

Western blotting

Murine epidermis was lysed using RIPA buffer (Wako) containing proteinase- and protease-inhibitor cocktail (Nacalai tesque). Protein lysates were separated by 10% SuperSepTM Ace (Wako), transferred onto polyvinylidene difluoride membranes (0.45 μm, Merck) by Trans-Blot Turbo Transfer System (Bio-Rad). The membrane was blocked with 5% bovine serum albumin and subjected to immunoblotting targeting the indicated proteins overnight 4℃, followed by the application of HRP-conjugated secondary antibody for one hour at room temperature (Supplemental Table 2). WB stripping solution (Nacalai tesque) was used to remove the antibodies for further evaluation.

Inhibition of mTOR

Rapamycin (4 mg/kg, Sanxin Chempharma), JR-AB2-011 (400 μg/kg, MedChemExpress) and vehicle were applied intraperitoneally to mice once daily for 14 consecutive days. Rapamycin and JR-AB2-011 were dissolved in 100% DMSO and diluted with 40% Polyethylene Glycol 300 (Wako), 5% Tween-80 (Sigma-Aldrich) and 45% saline in sequence. For topical application, 0.2% rapamycin gel and vehicle gel were prepared by the pharmaceutical department as previously described (Wataya-Kaneda et al., 2017). Rapamycin gel was applied to the left ear, and vehicle gel to the right ear (10mg/ear) of Sema4AKO, once daily for 14 consecutive days.

Data processing of single-cell RNA-sequencing and bulk RNA-sequencing

The raw count matrix data from the previously reported single-cell RNA-sequencing data from GSE220116 (Kim et al., 2023) were imported into Scanpy (1.9.6) using Python for further analyses. For each sample, cells and genes meeting the following criteria were excluded: cells expressing over 200 genes (sc.pp.filter_cells), cells with a high proportion of mitochondrial genes (> 5%), and genes expressed in fewer than 3 cells (sc.pp.filter_genes). Counts were normalized using sc.pp.normalize_per_cell, logarithmized (sc.pp.log1p), and scaled (sc.pp.scale). Highly variable genes were selected using sc.pp.filter_genes_dispersion with the options min_mean = 0.0125, max_mean = 2.5, and min_disp = 0.7.

Principal Component analysis was conducted with sc.pp.pca, selecting the 1st to 50th Principal Components for embedding and clustering. Neighbors were calculated with batch effect correction by BBKNN (Polański et al., 2020). Embedding was performed by sc.tl.umap, and cells were clustered using sc.tl.leiden. Some cells were further subclassified in a similar manner. Differentially expressed genes were identified on the cellxgene VIP, with significance defined as padj < 0.05.

Bulk RNA-sequencing data from GSE121212 (Tsoi et al., 2019) were re-analyzed using RaNAseq (Prieto & Barrios, 2019) for differential expression in Ctl versus psoriatic NL. Our analysis included normalization, differential gene expression (defining significance as padj < 0.05), and focused on Gene Ontology biological process analysis. The gene expression was calculated with the transcripts per million values.

Statistical analysis

GraphPad Prism 10 software (GraphPad Software) was used for all statistical analyses except for RNA-sequencing. Mann-Whitney test was used for two-group comparisons, while Kruskal-Wallis test followed by Dunn’s multiple comparisons test was applied for comparison among three or more groups. Statistical significance was defined as p < 0.05 (*), p < 0.01 (**), and p < 0.001 (***). In the analysis of RNA-sequencing data, statistical significance was defined as padj < 0.01 (**), and padj < 0.001 (***).

Study approval

All experiments involving human specimens were in accordance with the Declaration of Helsinki and were approved by the Institutional Review Board in Osaka University Hospital (20158-6). Written informed consent was obtained from all participants. All murine experiments were approved by Osaka University Animal Experiment Committee (J007591-013) and all procedures were conducted in compliance with the Guidelines for Animal Experimentation established by Japanese Association for Laboratory Animal Science.

Data availability

The single-cell RNA-sequencing datasets generated by Kim et al. (Kim et al., 2023) and Tsoi et al. (Tsoi et al., 2019) used in this study are available in the NCBI Gene Expression Omnibus under accession codes GSE220116 and GSE121212, respectively. Values for all data points in graphs are reported in the Source data.

Acknowledgements

We express our gratitude to all patients for their participation in this study. This study was supported by a Grant-in-Aid for JSPS Fellows 22KJ2071 (to MK), a Grant-in-Aid for Scientific Research (KAKENHI) 16K19705 (to RW), and a collaborative research grant from Maruho Co (Osaka, Japan).

Supplementary materials

Sema4A is downregulated in the keratinocytes of lesional psoriasis in the single-cell RNA-sequencing data.

(A) Sample information for specimens from Ctl and psoriatic L (GSE220116). (B and C) Clusters of cells were identified by their expression patterns of signature genes.

Gating strategy in flow cytometry.

Gating strategy for human Sema4A expression in blood cells. Large and small cells were distinguished using forward scatter (FCS) and side scatter (SSC) in a dot plot panel, with dead cells being excluded. Monocytes were defined within the live large cell population as CD11c positive. CD4 and CD8 T cells were identified within the live small cell population as CD3 positive CD4 positive and CD3 positive CD8 positive populations, respectively. The empty histogram represents the flow cytometry minus one control for Sema4A.

Gating strategy in flow cytometry.

Gating strategy for murine T cells infiltrating the epidermis and dermis. After excluding dead cells, TCRγδ positive T cells were evaluated for the expression of Vγ2. TCRγδ positive Vγ2 negative population was further assessed the expression of Vγ3. The CD3 positive TCRγδ negative population was evaluated for the expression of CD4 and CD8. Each population was analyzed for cytokine production. The empty histogram represents the isotype control for IL-17A.

Expression of IFNγ and IL-4 is comparable between WT and Sema4AKO skin.

(A and B) The graphs presenting the percentages of IFNγ (A) and IL-4 (B) -producing Vγ2, Vγ2-Vγ3-γδ (DNγδT), CD4, and CD8 T cells in CD3 fraction in the Epi (top) and Derm (bottom) of WT mice and Sema4AKO mice (n = 4 per group, each dot represents the average of 4 ear specimens). (C) The graphs showing the percentages IL-17A-producing γδ, CD4, and CD8 T cells in CD3 fraction from draining LN (dLN) of WT mice and Sema4AKO mice (n = 6 per group). A-C: NS, not significant.

Figure 4-figure supplement1-source data1

Excel file containing quantitative data for Figure 4-figure supplement1.

Comparable T17 differentiation potential under Th17-skewing conditions between WT mice and Sema4AKO mice.

Splenic T cells were cultured for 2 weeks, followed by flow cytometry analysis. The accumulated data display the percentages of IL-17A-producing (right) and IFNγ-producing (left) γδ, CD4, and CD8 T cells within CD3 fraction under various conditions: IL-23 dependent Th17-skewing condition (top), IL-23 only Th17-skewing condition (middle), and IL-23 independent Th17-skewing condition (bottom) . NS, not significant.

Figure 4-figure supplement2-source data1

Excel file containing quantitative data for Figure 4-figure supplement2.

The epidermis of psoriatic non-lesion is thicker than that of control skin.

Epidermal thickness of Ctl and psoriatic NL (n = 10 per group). ***p < 0.001.

Figure 5-figure supplement1-source data1

Excel file containing quantitative data for Figure 5-figure supplement1.

Upregulation of cytokeratin expression related to psoriasis is not detected at birth in Sema4AKO mice.

Representative immunofluorescence pictures of Krt5, Krt10, Krt14, and Krt16 (red) overlapped with DAPI, and the accumulated graphs showing the numbers of Krt5, Krt10, Krt14, and Krt16 positive cells per 100 μm width (n = 5 per group) in the epidermis of day 0 back. Scale bar = 50 μm. Each dot represents the average from 5 unit areas per sample. NS, not significant.

Figure 6-figure supplement1-source data1

Excel file containing quantitative data for Figure 6-figure supplement1.

Topical application of rapamycin gel yields partially similar results to intraperitoneal treatment.

(A) Comparison of epidermal thickness between vehicle (Ctl) gel-treated right ears and rapamycin gel-treated left ears of Sema4AKO mice (n = 10 per group). (B) Relative expression of keratinocyte differentiation markers and Il17a in Sema4AKO Epi under Ctl gel or rapamycin gel treatments (n = 5 per group). (C) The number of T cells in the epidermis (left) and dermis (right), under Ctl gel or rapamycin gel treatments (n = 5 per group). Each dot represents the sum of numbers from 10 unit areas across 3 specimens. A-C: *p < 0.05, **p < 0.01. NS, not significant.

Figure 8-figure supplement1-source data1

Excel file containing quantitative data for Figure 8-figure supplement1.

Patient information

Antibodies used for immunohistochemical, immunofluorescence, and western blot analyses

Primer sequences of real-time quantitative PCR used in human sample experiments

Antibodies used for flow cytometry analysis

Mouse recombinant cytokines