Abstract

   IL-27 is an immunoregulatory cytokine whose essential function is to
   limit immune responses. We found that the gene encoding cholesterol
   25-hydroxylase (Ch25h) was induced in CD4^+ T cells by IL-27, enhanced
   by TGF-β, and antagonized by T-bet. Ch25h catalyzes cholesterol to
   generate 25-hydroxycholesterol (25OHC), which was subsequently released
   to the cellular milieu, functioning as a modulator of T cell response.
   Extracellular 25OHC suppressed cholesterol biosynthesis in T cells,
   inhibited cell growth, and induced nutrient-deprivation cell death
   without releasing high-mobility group box-1. This growth inhibitory
   effect was specific to actively proliferating cells with high
   cholesterol demand and was reversed when extracellular cholesterol was
   replenished. Interestingly, Ch25h expressing CD4^+ T cells that
   received IL-27 and TGF-β signals became refractory to 25OHC-mediated
   growth inhibition in vitro. Nonetheless, IL-27 treated T cells
   negatively affected viability of bystander cells in a paracrine manner,
   but only if the bystander cells were in the early phases of activation.
   In mouse models of skin inflammation due to autoreactive T cells or
   chemically induced hypersensitivity, genetic deletion of Ch25h or
   Il27ra led to worsened outcomes. Thus, Ch25h is an immunoregulatory
   metabolic switch induced by IL-27 and dampens excess bystander T
   effector expansion in tissues through its metabolite derivative, 25OHC.
   This study reveals regulation of cholesterol metabolism as a modality
   for controlling tissue inflammation and thus represents a mechanism
   underlying T cell immunoregulatory functions.

One Sentence Summary:

   Cholesterol 25-hydroxylase-expressing CD4^+ T cells limit pathogenic
   bystander T cell expansion by suppressing the cholesterol biosynthetic
   pathway with secreted 25-hydroxycholesterol.

INTRODUCTION

   Helper CD4^+ T cells are crucial for proper host defense, but also
   contribute to the pathology of a wide variety of diseases. The lineage
   commitment of helper T subsets is initiated by cytokines in the
   microenvironment that activate transcriptomic programs for a particular
   subset while repressing other fates. The actions of proinflammatory
   cytokines such as interleukin (IL)-12, IL-4, and IL-6 are linked to
   inflammatory T helper subsets (Th1, Th2, and Th17), respectively,
   whereas those of IL-27 are more nuanced ([62]1). IL-27 was originally
   reported to promote interferon (IFN)-γ production and T cell
   proliferation and considered as a proinflammatory cytokine ([63]2,
   [64]3). However, the genetic deletion of IL-27 signaling predisposed
   mice to death due to unconstrained inflammation after Toxoplasma gondii
   and Trypanosoma cruzi infection, suggesting IL-27 plays an
   immunoregulatory role ([65]4, [66]5). Subsequent studies have confirmed
   the regulatory roles of IL-27 in a series of infection, allergy, and
   autoimmune disease models ([67]3, [68]6–[69]8).

   Several mechanisms underlying the immunoregulatory actions of IL-27
   have been reported. The downregulation of master transcription factors
   such as GATA3 or RORγt ([70]9, [71]10) during Th differentiation has
   been reported. Moreover, the induction of T-bet^+CXCR3^+ regulatory T
   cells (Tregs), a suppressive population specialized for Th1
   inflammation, has been reported ([72]11). These examples are relevant
   to specific circumstances but do not explain the broad regulatory
   effects of IL-27. Studies focusing on identifying downstream
   immunoregulatory molecules directly induced by IL-27 have suggested key
   molecules such as IL-10 ([73]12, [74]13) and co-inhibitory receptors,
   including PD-1, activated protein C receptor, and podoplanin ([75]14,
   [76]15) in T cells. However, these molecules are not exclusively
   induced by IL-27 ([77]16–[78]18), suggesting the involvement of other
   factors selective to IL-27 that contribute to drive IL-27 mediated
   immunoregulatory effect.

   In this study, we sought to explore novel target molecules
   preferentially induced by IL-27 in helper T cells by referencing an
   IL-27-treated condition to other known lineages of effector or
   regulatory conditions (Th0, Th1, Th2, Th17, and iTreg). We identified
   cholesterol 25-hydroxylase (Ch25h) as an immunoregulatory molecule
   induced in cells treated with IL-27, and also by type I IFNs. Previous
   work has identified induction of Ch25h by type I IFNs in macrophages
   ([79]19, [80]20). Critically though, IFN-γ does not efficiently induce
   Ch25h in T cells because T-bet acts as a negative regulator of Ch25h.
   In contrast, IL-27 effectively induces Ch25h and its product 25OHC. We
   provide evidence that alterations of cholesterol metabolism in cells
   receiving an IL-27 signal are crucial for regulating the viability of
   bystander lymphocytes in the early phase of active proliferation. We
   also show Ch25h-expressing T helper cells limit skin inflammation in a
   T cell-dependent autoimmune disease model.

RESULTS

Ch25h is selectively induced by IL-27 in CD4^+ T cells via STAT1

   To identify potential target molecules relevant to the regulatory
   functions of IL-27, transcriptomic analysis was performed using in
   vitro differentiated CD4^+ helper subsets, including Th1, Th2, iTreg,
   Th17, and Th0, versus IL-27-treated CD4^+ T cells supplemented with
   transforming growth factor (TGF)-β, anti-IFN-γ, anti-IL-2, and
   anti-IL-4. The last condition was chosen to simulate IL-27-mediated
   regulatory conditions in vitro and maximize the likelihood of selecting
   for IL-27-induced events by antagonizing proinflammatory cytokine
   effects. Under these conditions, 146 genes were differentially
   expressed only in IL-27- and TGF-β-treated cells and not in the other
   subsets. Notably, Ch25h, which encodes cholesterol 25-hydroxylase, was
   the most differentially expressed gene ([81]Table S1 and [82]Fig. 1A).
   The specificity of Ch25h induction by IL-27 was further validated by
   individual pairwise comparisons with other subsets ([83]Fig. S1).
   Real-time PCR confirmed that Ch25h was expressed at the highest levels
   in IL-27- and TGF-β-treated cells among other representative in vitro
   Th subsets ([84]Fig. 1B).

Figure 1. Ch25h is selectively induced in CD4^+ T cells stimulated with IL-27
and TGF-β.

   Figure 1.
   [85]Open in a new tab

   (A) Heatmap of gene expression measured using mRNA-seq. Genes that are
   expressed in IL-27- and TGF-β-treated T cells, but not in other
   conditions, are selected and shown. Representative results from two
   independent experiments are shown. See also [86]Fig. S1.

   (B) Ch25h expression measured using RT-qPCR is shown relative to
   β-actin expression. Representative results from two independent
   experiments are shown. *p < 0.0001.

   (C) Ch25h expression measured using RT-qPCR in naïve CD4^+ T cells
   treated with TGF-β or IL-27 along with anti-CD3 and anti-CD28
   stimulation for 2 days. Representative results from two independent
   experiments are shown. **p < 0.0001.

   (D) Ch25h expression measured using RT-qPCR daily over the course of 3
   days in naïve CD4^+ T cells treated with or without IL-27 along with
   anti-CD3 and anti-CD28 stimulation. Representative results from two
   independent experiments are shown.

   (E) Ch25h expression measured using RT-qPCR in naïve CD4^+ T cells
   treated with IFN-γ, IFN-α, IFN-β, or IL-27 in conjunction with anti-CD3
   and anti-CD28 stimulation for 2 days. Representative results from two
   independent experiments are shown. *p < 0.0001, **p = 0.0002.

   (F) Ch25h protein expression detected using immunohistochemistry in
   wild-type and Ch25h^−/− naïve CD4^+ T cells treated with or without
   TGF-β and IL-27 along with anti-CD3 and anti-CD28 stimulation for 2
   days. Ch25h staining is in green, and DAPI is in blue. Scale indicates
   20 μm.

   (G) Ch25h expression measured using RT-qPCR in
   CD4^+CD25^−CD62L^hiCD44^lo naïve or CD4^+CD25^−CD62L^loCD44^hi memory T
   cells treated with or without IL-27 along with anti-CD3 and anti-CD28
   antibody stimulation for 2 days. Representative results from four
   independent experiments are shown. *p < 0.0001.

   (H) Ch25h expression measured using RT-qPCR in CD4^+ or CD8^+ naïve T
   cells treated with or without IL-27 along with anti-CD3 and anti-CD28
   antibody stimulation for 2 days. Representative results from three
   independent experiments are shown. *p < 0.0001.

   (I) Ch25h expression measured using RT-qPCR in wild-type and Stat1^−/−
   naïve CD4^+ T cells treated with or without IL-27 along with anti-CD3
   and anti-CD28 antibody stimulation. Representative results from three
   independent experiments are shown. *p < 0.0001.

   Since a combination of IL-27 and TGF-β was used to identify Ch25h, the
   effect of each cytokine on Ch25h expression was evaluated separately.
   Ch25h expression was detected only in the IL-27-supplemented condition,
   but not with TGF-β, confirming the specificity of Ch25h induction by
   IL-27 ([87]Fig. 1C). Next, the time course of Ch25h expression was
   recorded following IL-27 stimulation in CD4^+ T cells. Ch25h mRNA
   expression was detected on day 1, peaked on day 2, and declined on day
   3 in CD4^+ T cells ([88]Fig. 1D). Since IFNs have been shown to induce
   Ch25h expression in bone marrow-derived macrophages (BMDMs) ([89]19,
   [90]20), the ability of IFNs and IL-27 to induce Ch25h in CD4^+ T cells
   was compared. Using quantitative PCR (qPCR), type I IFNs (IFN-α and
   IFN-β), but not IFN-γ, induced Ch25h with the range of doses tested.
   Nonetheless, IL-27 induced higher levels of Ch25h expression than type
   I IFNs ([91]Fig. 1E). Immunocytostaining confirmed that IL-27 in
   combination with TGF-β upregulated Ch25h protein more than that with no
   cytokine control, and Ch25h protein was undetectable without
   stimulation ([92]Fig. 1F). These results indicated that IL-27 and, to a
   lesser extent, type I IFNs are inducers of Ch25h production in CD4^+ T
   cells.

   Next, our analysis was extended to examine whether Ch25h expression was
   observed in cells other than IL-27-treated naïve CD4^+ T cells with
   T-cell receptor (TCR) stimulation. Not only naïve CD4^+ cells but also
   memory CD4^+ T cells upregulated Ch25h when stimulated via TCR
   crosslinking with IL-27 ([93]Fig. 1G). Conversely, Ch25h expression was
   not inducible in CD8^+ T cells, indicating that CD4^+ T cells
   preferentially induce Ch25h expression in vitro under IL-27-stimulating
   conditions ([94]Fig. 1H).

   Since IL-27 signaling is known to be mediated by the transcription
   factor Stat1, requirement of Stat1 for Ch25h induction in T cells was
   tested. Quantitative PCR revealed that IL-27-induced Ch25h expression
   was completely abolished in Stat1^−/− CD4^+ T cells ([95]Fig. 1I). This
   result is consistent with previous observations in BMDMs ([96]19).
   These results indicate that CD4^+ T cells are selectively capable of
   expressing Ch25h in a Stat1-dependent manner under IL-27 stimulation.

T-bet negatively regulates Ch25h expression in CD4^+ T cells

   STAT1 is also known to mediate IFN-γ signaling and both IL-27 and IFN-γ
   induce T-bet. Therefore, we speculated that IL-27-induced T-bet might
   also participate in Ch25h expression. In Tbx21^−/− CD4^+ T cells, Ch25h
   was significantly upregulated compared with wild-type cells following
   IL-27 stimulation ([97]Fig. 2A). Consistent with this result, Tbx21^−/−
   T cells produced more 25OHC upon IL-27 stimulation than wild-type T
   cells ([98]Fig. 2B). Interestingly, IFN-γ, which is the key cytokine to
   induce T-bet in Th1 cells ([99]Fig. S2A), failed to induce Ch25h
   expression in wild-type T cells ([100]Fig. 1E). However, in the absence
   of T-bet, this cytokine significantly upregulated Ch25h expression
   ([101]Fig. 2A). These results indicate that T-bet is a negative
   regulator of Ch25h expression and subsequent 25OHC production.

Figure 2. T-bet negatively regulates Ch25h expression in CD4^+ T cells.

   Figure 2.
   [102]Open in a new tab

   (A) Ch25h expression measured using RT-qPCR in wild-type and Tbx21^−/−
   naïve CD4^+ T cells treated with IFN-γ or IL-27 along with anti-CD3 and
   anti-CD28 stimulation for 2 days. Representative results from two
   independent experiments are shown. *p < 0.0001.

   (B) Amounts of 25OHC secreted in the culture supernatant measured using
   GC-MS following stimulation of wild-type and Tbx21^−/− naïve CD4^+ T
   cells with IL-27 for 3 days. Representative results from four
   independent experiments are shown. *p < 0.0001.

   (C) Genome track views of the extended Ch25h locus showing the
   deposition of H3K4me1 (yellow), H3K27ac (red), and T-bet (blue) in
   wild-type and Tbx21^−/− naïve CD4^+ T cells. ChIP-seq data
   ([103]GSE96724) reported previously ([104]21) were re-analyzed, and
   peaks with significance (default p-value threshold of 1E^−5) are
   underlined.

   (D) Ch25h expression measured using RT-qPCR in wild-type (gray) and
   Tbx21^−/− (white) naïve CD4^+ T cells treated with various combinations
   of IL-27 and TGF-β along with anti-CD3 and anti-CD28 stimulation for 2
   days. Representative results from two independent experiments are
   shown. *p < 0.001.

   (E) Tbx21 expression measured using RT-qPCR in naïve CD4^+ T cells
   treated with various combinations of IL-27 and TGF-β along with
   anti-CD3 and anti-CD28 stimulation for 2 days. Representative results
   from three independent experiments are shown.

   To determine whether T-bet directly represses the transcription of
   Ch25h, we examined our previously published ChIP-seq data for histone
   marks and T-bet binding ([105]21). Multiple enhancer marks (H3K4me1 and
   H3K27ac) were detected in wild-type cells downstream of the Ch25h locus
   and increased in intensity and number in Tbx21^−/− cells ([106]Fig.
   2C). Furthermore, a subset of these enhancers was bound by T-bet
   ([107]Fig. 2C), suggesting the direct involvement of T-bet in
   repressing Ch25h expression in wild-type cells.

   When we tested the effect of TGF-β (0–20 ng/ml) in the presence of
   IL-27 (20 ng/ml), it was found that TGF-β enhanced Ch25h expression in
   a dose dependent manner ([108]Fig. 2D: gray bars with WT cells) whereas
   the expression of T-bet was progressively suppressed ([109]Fig. 2E). As
   T-bet was presumed to directly repress Ch25h expression ([110]Fig. 2A
   and [111]C), it was hypothesized that TGF-β might counter the effect of
   T-bet mediated repression of Ch25h. Indeed, in the absence of T-bet,
   Ch25h expression was always higher than wild-type T cells across all
   doses of TGF-β tested, and apparent dependency of Ch25h expression on
   TGF-β dosage disappeared in Tbx21^−/− cells ([112]Fig. 2D: white bars).
   These data indicate that the TGF-β-mediated enhancement of
   IL-27-induced Ch25h expression is likely mediated by the ability of
   TGF-β to downregulate T-bet expression ([113]Fig. S2B).

Ch25h-expressing CD4^+ T cells produce 25OHC that down-regulates gene
expression of cholesterol biosynthetic enzymes

   Since Ch25h encodes the enzyme that converts cholesterol into
   25-hydroxylcholesterol (25OHC) ([114]22), IL-27-induced Ch25h
   expression was expected to induce production of intracellular 25OHC and
   possibly subsequent secretion of 25OHC into the culture medium. Gas
   chromatography-mass spectrometry (GC-MS) analysis showed that activated
   CD4^+ T cells produce little 25OHC, whereas IL-27 promoted a roughly
   seven-fold induction ([115]Fig. 3A). Furthermore, the combination of
   IL-27 and TGF-β caused a massive increase in 25OHC production
   ([116]Fig. 3B). In the absence of Ch25h, Ch25h^−/− CD4^+ T cells were
   unable to produce 25OHC at all even when treated with IL-27 or IL-27,
   IFN-β and TGF-β combined, confirming an essential role of Ch25h for
   production of 25OHC ([117]Fig. 3B).

Figure 3. Ch25h-expressing CD4^+ T cells produce 25OHC that down-regulates
gene expression of cholesterol biosynthetic enzymes.

   Figure 3.
   [118]Open in a new tab

   (A) Amount of 25OHC secreted in the culture supernatant measured using
   gas chromatography-mass spectrometry (GC-MS) following stimulation of
   naïve CD4^+ T cells by anti-CD3 and anti-CD28 antibodies and the
   indicated cytokines for 3 days. Representative results from three
   independent experiments are shown. *p = 0.002.

   (B) Amount of 25OHC secreted in the culture supernatant measured using
   GC-MS following stimulation of wild-type and Ch25h^−/− naïve CD4^+ T
   cells with various combinations of IL-27, IFN-β and TGF-β in the
   presence of anti-CD3 and anti-CD28 antibodies for 3 days.
   Representative results from three independent experiments are shown. *p
   = 0.0002, **p = 0.0001, ***p = 0.001.

   (C) Amount of 25OHC secreted in the culture supernatant measured using
   GC-MS daily during the course of naïve CD4^+ T cell stimulation with
   coated anti-CD3 and anti-CD28 antibodies and IL-27 and TGF-β.
   Representative results from three independent experiments are shown.
   Statistical significance was calculated in comparison with the value on
   day 1. *p < 0.0001, **p = 0.0002.

   (D) Differentially expressed genes (DEGs) after 25OHC treatment are
   depicted and were subjected to pathway analysis using DAVID and
   mRNA-seq data. A total of 215 DEGs were identified by comparing 10 nM
   and 100 nM 25OHC conditions, and 1831 DEGs were identified by comparing
   100 and 1000 nM 25OHC conditions. The top four biological pathways
   identified are shown for low (blue) and high (orange) dose effects. The
   number of genes involved in the corresponding pathways and names of
   representative genes is also included. See also [119]Tables S2 and
   [120]S3.

   (E) Gene expression heatmap showing genes assigned to the sterol
   biosynthetic process identified in [121]Figure 3D. Cells were harvested
   following 24 h of 25OHC treatment and subjected to mRNA-seq analysis.
   Genes showing more than a two-fold decrease in treated cells compared
   to untreated cells are marked with black dots. See also [122]Fig. S3
   and [123]Table S4.

   (F) An outline of the cholesterol biosynthetic pathway. Gene symbols of
   representative enzymes and names of metabolic intermediates are shown.
   Gene symbols that showed decreased expression following 25OHC treatment
   are colored in red.

   (G) Transcription factor enrichment plot. Transcription factor-binding
   DNA motifs enriched in 25OHC-mediated downregulated genes among genes
   annotated as cholesterol biosynthesis were plotted by CiiiDER. The
   −log[10] p-values for SREBP1 and SREBP2 are 3.02 and 2.14,
   respectively.

   Next, sequential measurement of 25OHC amounts in the culture
   supernatant was conducted over 5 days following IL-27 stimulation.
   Secreted 25OHC was first detected on day 2, peaked at approximately 100
   nM on day 3, and was maintained until a gradual decrease occurred on
   day 5 ([124]Fig. 3C). By contrast, Ch25h mRNA expression was notably
   diminished by day 3 ([125]Fig. 1D). These results indicated that 25OHC
   is secreted within the first 2–3 days of IL-27 stimulation and remains
   present in the culture medium for at least another few days in vitro.
   These results showed that induction of Ch25h expression in CD4^+ T
   cells results in the production of 25OHC, which is released from T
   cells into their microenvironment following IL-27 signaling.

   Next, the biological consequence of the presence of 25OHC in culture
   medium was evaluated by measuring the transcriptomic responses. CD4^+ T
   cells were treated with a range of concentrations of 25OHC (0, 10, 100,
   and 1000 nM) over time (days 1, 2, and 3). Depending on dosage,
   distinct biological pathways were impacted by 25OHC. With low doses of
   25OHC, the biosynthesis of cholesterol and other metabolic pathways
   were affected ([126]Fig. 3D: blue bars and [127]Table S2). By contrast,
   with high doses, pathways related to the cell cycle, DNA replication,
   DNA repair, and centromeres were principally affected ([128]Fig. 3D:
   orange bars and [129]Table S3). The results pointed to metabolic
   pathways as a primary biological process impacted by 25OHC but as
   concentrations of this lipid increase, more profound influences on cell
   cycle and DNA replication pathways emerge.

   Given these findings, we next focused our transcriptomic analysis on
   key enzymes and sterol-censoring molecules related to cholesterol
   biosynthesis ([130]Fig. 3E, [131]F and [132]Fig. S3). As expected from
   pathway enrichment analysis, 25OHC down-regulated multiple cholesterol
   biosynthetic enzymes compared with other accessory components of the
   cholesterol pathway ([133]Fig. 3E and [134]Table S4, p = 0.01).

   Previous studies have shown that 25OHC suppresses cholesterol
   biosynthesis by inhibiting the transport of sterol regulatory
   element-binding protein (SREBP) and SREBP cleavage-activating protein
   (SCAP) complex to the Golgi. Translocation of SCAP-SREBP from the
   endoplasmic reticulum to the Golgi is the critical initial step for the
   cleavage and activation of SREBP as a transcription factor to initiate
   cholesterol synthesis and uptake ([135]23). In line with previous
   findings ([136]24), our computational analysis revealed that
   significant enrichment of SREBP1 and SREBP2-binding DNA motifs was
   associated with genes depicted as 25OHC-down-regulated genes that
   belong to the cholesterol biosynthesis pathway ([137]Fig. 3G). Based on
   these results, it seems plausible that exposure to 25OHC limits
   activation of transcription factor SREBP. Insufficient availability of
   SREBP, in turn, leads to downregulation of key metabolic enzymes
   required for cholesterol biosynthesis, leading to suppression of de
   novo cholesterol biosynthesis.

25OHC impairs cell proliferation and induces cholesterol-deprived cell death

   Since IL-27-stimulated T cells were identified as a source of 25OHC, we
   next sort to determine how secreted 25OHC might influence the cellular
   phenotype of the surrounding T cells via suppressing cholesterol
   synthesis. According to previous reports, 25OHC mediated suppression of
   sterol synthesis, abolished DNA synthesis and lymphocyte proliferation
   ([138]25–[139]28). 25OHC was exogenously added to the in vitro culture
   of T cells stimulated with TCR, and cell division was measured by
   carboxyfluorescein diacetate succinimidyl ester (CFSE) labeling. Among
   the doses of 25OHC tested, cell proliferation was significantly
   suppressed at 100 nM, and almost completely abrogated at 1000 nM
   ([140]Fig. 4A), in line with previous reports ([141]26–[142]28). When
   cell viability was evaluated by a combination of 7-aminoactinomycin D
   (7-ADD) staining and cell size (FSC), impaired viability was observed
   in a dose-dependent manner ([143]Fig. 4B). The effect of 25OHC on cell
   size became evident at approximately 100 nM ([144]Fig. 4C) and resulted
   in a decreased frequency of 7-ADD negative live cells ([145]Fig. 4D).
   The effect of 25OHC on cell viability was present at 24 h of cell
   culture, and became more evident at later time points ([146]Fig. 4E and
   [147]Fig. S4A). In the absence of TCR- or IL-2-dependent activation, T
   cells were resistant to the effect of 25OHC even at the highest
   concentration tested ([148]Fig. S4A; [149]Fig. 4C, [150]D). Taken
   together, these results suggested that 25OHC impairs cell proliferation
   and viability depending on the metabolic demands of T cells, i.e.
   whether cells receive activation signals such as antigen stimulation or
   IL-2.

Figure 4. 25OHC impairs cell proliferation and induces cholesterol-deprived
cell death.

   Figure 4.
   [151]Open in a new tab

   (A) CFSE dilution assay of naïve CD4^+ T cells after anti-CD3 and
   anti-CD28 stimulation with various concentrations of 25OHC.
   Proliferation as quantified by %Divided calculated by FlowJo along with
   the proportion of non-divided cells are shown in the graphs.
   Representative results from two independent experiments are shown. *p <
   0.05. NA, not applicable. NS, not significant.

   (B) Flow cytometric analysis to detect live cells (red dots) following
   anti-CD3 and anti-CD28 stimulation of naïve CD4^+ T cells for 3 days
   with various concentrations of 25OHC in the absence of additional
   cytokines. Dead cells are quantitated by staining with
   7-aminoactinomycin D. Numbers indicate the proportions of live cells.
   Representative results from four independent experiments are shown.

   (C) Dose-dependent effects of 25OHC on forward scatter profile
   comparing control and anti-IL-2 antibody-supplemented conditions in
   conjunction with T-cell receptor stimulation, as shown in [152]Figure
   4B. Values of untreated cells were set as 100%.

   (D) 25OHC-mediated dose-dependent changes in live cells in the
   experiment similar to (C). Values for untreated cells were set as 100%.

   (E) 25OHC-mediated changes in live cells over time. Values for
   untreated cells at day 1 were set as 100%. Representative results from
   two independent experiments are shown.

   (F) Cholesterol metabolism pathway.

   (G) Fractions of live cells treated with cholesterol, 25OHC, and
   7α,25OHC are shown. Representative results from three independent
   experiments are shown. *p < 0.0001.

   (H) Flow cytometric analyses to detect live cells treated with various
   concentrations of 25OHC in combination with or without
   cyclodextrin–cholesterol. Representative results from three independent
   experiments are shown.

   (I) The proportion of live cells treated with 25OHC in combination with
   or without cyclodextrin–cholesterol is shown. Representative results
   from two independent experiments are shown.

   Given that cells with high metabolic demands are susceptible to 25OHC
   and prone to cell death, we next asked if the 25OHC-stressed cells
   could perpetuate the inflammatory response further by releasing danger
   signals such as alarmins. High mobility group box-1 (HMGB-1) is
   released to extracellular space from necrotic cells, but no HMGB-1 was
   produced by 25OHC-treated cells ([153]Fig. S4B) indicating that this
   mode of cell death by 25OHC was not associated with alarmin release.

   As 25OHC is further metabolized to 7α,25-dihydroxycholesterol
   (7α,25OHC) by Cyp7b1 ([154]Fig. 4F), the question arose whether a
   downstream metabolite of 25OHC could also confer similar inhibitory
   effects as 25OHC. It was found that 25OHC, but not 7α,25OHC, induced
   cell death of CD4^+ T cells ([155]Fig. 4G). Moreover, CD8^+ T cells
   were also susceptible to cell death by 25OHC in a dose-dependent manner
   ([156]Fig. S4C). Therefore, 25OHC, but not 7α,25OHC, could render CD4^+
   and CD8^+ T cell populations vulnerable to cell death.

   Given that insufficient cholesterol supply is likely a mechanism
   underlying cell death induced by 25OHC, a rescue experiment was
   conducted by providing an exogenous source of cholesterol in the
   culture media in addition to 25OHC. We found that addition of
   cyclodextrin–cholesterol efficiently rescued cell viability at low to
   medium doses of 25OHC; although, prevention of cell death was limited
   at higher doses of 25OHC ([157]Fig. 4H, [158]I). This finding was
   consistent with our transcriptomic analysis ([159]Table S3) indicating
   that pathways related to the cell cycle or DNA repair were critically
   impacted at high concentrations of 25OHC, such that the restoration of
   intracellular cholesterol levels alone did not fully rescue cells
   exposed to high doses of 25OHC ([160]Fig. 4H, [161]I). These results
   collectively indicated that 25OHC-induced cell death is in part
   attributable to the reduction in intracellular cholesterol owing to the
   impaired de novo biosynthesis in activated T cells with high metabolic
   demands.

Ch25h-expressing CD4^+ T cells impair viability of bystander T cells at early
stage of TCR activation in a paracrine manner

   Our findings so far collectively connect IL-27 (cytokine) to Ch25h
   (enzyme) to 25OHC (metabolite) as a series of molecular events leading
   to impaired supply of cholesterol for avidly proliferating T cells with
   high metabolic demands. As Ch25h-expressing cells produce and release
   25OHC into culture medium ([162]Fig. 3A–[163]C), we considered the
   possibility that Ch25h-expressing WT cells might be negatively impacted
   by autocrine 25OHC during the course of IL-27 exposure. By contrast, we
   expected that Ch25h^−/− CD4^+ T cells might have growth advantage due
   to lack of autocrine 25OHC production. Contrary to our expectations,
   there was no significant difference in cell viability and proliferative
   activity between WT and Ch25h^−/− CD4^+ T cells cultured separately but
   in the same IL-27 included conditions ([164]Fig. S5). These results
   indicated that Ch25h-expressing cells are refractory to growth
   inhibitory effect of autocrine 25OHC.

   This unexpected finding led us to consider that the timing of cell
   activation and 25OHC exposure was relevant in terms of inhibition of
   cell growth and viability. Normally, secretion of 25OHC required two
   days following activation of CD4^+ T cells (TCR crosslinking plus IL-27
   and TGF-β; [165]Fig. 3C). To clarify the sensitivity of activated T
   cells, we added exogenous 25OHC (200 nM) to in vitro T cell cultures at
   different time points. The inhibitory effect of 25OHC on cell viability
   was most evident if 25OHC was present concomitant with cell activation
   but was reduced if 25OHC addition was delayed. In fact, 25OHC had no
   effect if added 48 or 72 hours after activation ([166]Fig. 5A). These
   results indicated that because 25OHC production is delayed in IL-27/
   TGF-β activated cells, the cells are largely protected from inhibitory
   action of autocrine 25OHC. However, T cells in early phase of
   TCR-mediated activation (<48 hrs) are susceptible to 25OHC-mediated
   impairment on cell viability.

Figure 5. Ch25h-expressing CD4^+ T cells impair viability of bystander T
cells at early stage of TCR activation in a paracrine manner.

   Figure 5.
   [167]Open in a new tab

   (A) CD4^+ T cells were stimulated with anti-CD3 and CD28 antibodies at
   day 0 and cultured with 200 nM 25OHC at the time points indicated.
   Three days later, the percent of living cells was analyzed. 25OHC
   non-treated CD4^+ T cells were simultaneously analyzed. The ratios of
   living cells in 25OHC-treated cells as a percent of non-treated cells
   are depicted. Representative results from two independent experiments
   are shown. *p < 0.05.

   (B) Schematic overview for experiments in [168]Fig. 5C.

   (C) In FACS plots, proportion of 7-AAD^− living cells (lower panels)
   and dilution of CFSE-labeled responder cells (upper panels) were
   analyzed. In histograms, % divided responder cells were calculated by
   FlowJo. Representative results from three independent experiments are
   shown. *p < 0.05. NS, not significant.

   Given the timing of 25OHC production and the refractoriness of
   activated cells, the possibility of paracrine versus autocrine effects
   of 25OHC on bystander cells seemed more likely. To test this,
   sequential in vitro cultures were performed as described in [169]Fig.
   5B. First, WT or Ch25h^−/− naïve CD4^+ T cells were activated with TCR
   crosslinking in the presence of IL-27+TGF-β for 2 days to generate
   effector cells. After removing effector cells, CFSE-labelled Il27ra^−/−
   cells were introduced as responder cells to the pre-conditioned media
   and the cells were cultured for another 3 days. During the second
   culture, TGF-β neutralizing antibody was added to make certain that the
   responder cells did not receive IL-27 signaling (due to the lack of
   receptor) or TGF-β (due to neutralization) to induce autocrine Ch25h
   production. In this experimental setting, Ch25h-expressing WT effector
   cells impaired the viability and proliferation of responder cells,
   indicative of a paracrine effect ([170]Fig. 5C). As a control,
   IL-27+TGF-β-treated Ch25h^−/− T cells (1^st culture) had no effect on
   bystander cells (2^nd culture), thus linking IL-27/TGF-β mediated
   induction of Ch25h and 25OHC production. These results collectively
   indicate that Ch25h-expressing CD4^+ T cells impaired viability and
   proliferation of bystander cells at early stages of activation in a
   paracrine manner.

Ch25h expressed in T cells is important to restrain tissue inflammation in
experimental autoimmune dermatitis

   Our in vitro results point to the generation of Ch25h-expressing T
   cells that result from IL-27 signal to influence bystander cells during
   heterogenous T effector responses. To determine whether this regulatory
   axis of IL-27-Ch25h-25OHC is relevant in vivo, a T cell-mediated
   autoimmune model of skin disease termed experimental autoimmune
   dermatitis (EAD) was used ([171]29). In this model, Rag2^−/− mice were
   adoptively transferred with desmoglein 3 (Dsg3)-specific TCR transgenic
   CD4^+ T cells (H1 T cells), which expand in lymphopenic host, then
   infiltrate to the skin and attack Dsg3-bearing epidermal keratinocytes
   leading to interface dermatitis (ID). Several Th effector subsets
   including Th1, Th17 and others are identified during skin inflammation,
   among which IFN-γ is indispensable for inducing pathological changes
   ([172]29). The relevance of IL-27-dependent 25OHC production on the
   extent of inflammation was evaluated by transferring WT versus
   Il27ra^−/−-H1 T cells. The Rag2^−/− recipient mice receiving
   Il27ra^−/−-H1 T cells showed significantly greater clinical
   inflammation scores than WT H1 T cells ([173]Fig. 6A), with severe
   epidermal damage associated with increased degenerative keratinocytes
   ([174]Fig. 6B: red arrows). As expected, the mice that received WT H1 T
   cells induced mRNA expression of IL-27p28 in dendritic cells isolated
   from the inflamed skin and from skin-draining LNs ([175]Fig. S6A).
   These results collectively demonstrated that IL-27 signaling plays a
   key role in constraining skin inflammation in EAD.

Figure 6. Ch25h expressed in T cells constrains tissue inflammation in
experimental autoimmune dermatitis.

   Figure 6.
   [176]Open in a new tab

   (A) Desmoglein 3 (Dsg3)-specific naïve CD4^+ T cells
   (TCRVβ6^+CD25^−CD62L^hiCD44^lo) were isolated from H1 mice and
   Il27ra^−/−-H1 mice and transferred to Rag2^−/− mice (n = 4 in each
   group). Clinical scores were measured following the course of induced
   dermatitis. Representative results from three independent experiments
   are shown. *p < 0.05 (two-way ANOVA).

   (B) Histopathology (hematoxylin and eosin staining) of the palate from
   two groups of treated mice on day 35. Red arrows indicate keratinocyte
   degeneration. Bar indicates 100 μm.

   (C) Dsg3-specific naïve CD4^+ T cells were isolated from H1 mice and
   Ch25h^−/−-H1 mice and transferred to Rag2^−/− mice to induce skin
   inflammation (n = 5 in each group). Clinical scores were measured
   following the course of induced dermatitis. Representative results from
   three independent experiments are shown. *p = 0.002 (two-way ANOVA).

   (D) Skin inflammation observed in Rag2^−/− recipient mice transferred
   with Dsg3-specific naïve CD4^+ T cells derived from H1 (right panels)
   and Ch25h^−/−-H1 (left panels) donors.

   (E) Histopathology of the palate and esophagus of Rag2^−/− recipient
   mice receiving wild-type (middle) or Ch25h^−/− (left) H1 CD4^+ T cells.
   The palate and esophagus of wild-type mice are also shown (right). The
   scale bar indicates 50 μm.

   (F) Immunofluorescence staining of the palate with TCRVβ (red) and DAPI
   (blue) of Rag2^−/− recipient mice receiving wild-type (right) or
   Ch25h^−/− (left) H1 CD4^+ T cells. The scale bar indicates 50 μm.

   (G) Numbers of CD4^+ T cells in the palate of Rag2^−/− recipient mice
   receiving wild-type or Ch25h^−/− H1 CD4^+ T cells. Cells gated by a
   square 100 μm on a side were counted using the images in (F). *p =
   0.001.

   (H) Flow cytometric analysis of Dsg3-specific CD4^+ T cells. Cells in
   the skin, skin-draining LNs (SLNs), mesenteric LNs (MLNs), and spleen
   (Sp) were compared between mice transferred with wild-type or Ch25h^−/−
   H1 CD4^+ T cells. *p = 0.045.

   Next, involvement of IL-27-Ch25h axis in regulating autoreactive T
   cells was evaluated in EAD. Expression of Ch25h was observed in H1 T
   cells isolated from the inflamed skin, but not in skin-draining LNs,
   indicating the localized expression of Ch25h, limited to the site of
   inflammation ([177]Fig. S6B). This upregulation of Ch25h was
   significantly reduced in T cells unable to respond to IL-27
   (Il27ra^−/−-H1 T cells) during inflammation ([178]Fig. S6C).

   We confirmed these results using an alternative in vivo immunization
   model that did not require adoptive transfer of transgenic T cells. WT,
   H1 transgenic or Il27ra^−/−-H1 mice were immunized by Dsg3 peptide and
   T helper effector cells (CD4^+CD62L^lowCD44^hi) were sorted after 1
   week of immunization and subjected to qPCR analysis for Ch25h mRNA
   expression. Compared with WT, H1 mouse expressed higher levels of Ch25h
   mRNA in activated effector T cells isolated from skin but not lymph
   nodes ([179]Fig. S6D), and the induction of Ch25h mRNA was abrogated in
   Il27ra^−/−-H1 mice ([180]Fig. S6E). These results indicated that
   IL-27-dependent induction of Ch25h in T helper cells in vivo during the
   course of skin autoimmunity either generated by peptide immunization or
   adoptive T cell transfer.

   For further confirmation of our model, Ch25h^−/− mice were crossed with
   H1 tg mice. Transfer of Ch25h^−/−-H1 T cells resulted in significantly
   greater skin inflammation compared to transfer of WT H1 T cells
   ([181]Fig. 6C, [182]D). Histological examination revealed prominent
   tissue inflammation ([183]Fig. 6E) and increased number of autoreactive
   TCRVβ^+ CD4^+ T cells in the recipients of Ch25h^−/−-H1 T cells
   ([184]Fig. 6F: red cells). Furthermore, the number of H1 T cells in the
   inflamed tissue was significantly higher in the absence of Ch25h
   ([185]Fig. 6G). When percentages of H1 T cell fractions were measured,
   the impact of Ch25h deficiency to expand H1 T cells was evident only in
   the skin but not in secondary lymphoid organs ([186]Fig. 6H),
   indicating its functional relevance limited at the site of
   inflammation. These results indicated that Ch25h, induced in CD4^+ T
   cells via IL-27, is a key enzyme that limits the extent of skin
   inflammation by constraining the expansion of pathogenic cells in the
   lesional skin, most likely through secretion of 25OHC in the inflamed
   tissue.

Loss of Ch25h exacerbates skin contact hypersensitivity

   Finally, to broaden the relevance of Ch25h in controlling skin
   inflammation, we employed a contact hypersensitivity model in which
   IL-27 is known to play a critical role ([187]30). After sensitization
   in the abdominal skin with 1-fluoro-2,4-dinitrobenzene (DNFB), contact
   hypersensitivity reaction was assessed by ear thickness in WT and
   Ch25h^−/− mice following topical DNFB application. The kinetics of skin
   inflammation differed between the two mice, with prolonged inflammation
   being observed in Ch25h^−/− mice ([188]Fig. 7A, [189]B). Histological
   examination of inflamed skin showed enhanced inflammation in Ch25h^−/−
   mice ([190]Fig. 7C). Importantly, exogenous administration of 25OHC via
   intraperitoneal injection reversed the severity of skin inflammation,
   confirming that Ch25h-25OHC axis is relevant to constrain the degree of
   skin inflammation in the model of DNFB-induced contact hypersensitivity
   ([191]Fig. 7D, [192]E).

Figure 7. Absence of Ch25h exacerbates skin contact hypersensitivity.

   Figure 7.
   [193]Open in a new tab

   (A - C) Contact hypersensitivity was elicited by DNFB sensitization and
   reapplication to ear skin in WT (n = 4) and Ch25h^−/− mice (n = 4). Ear
   skin thickness was measured (A) and calculated as area under the curve
   (AUC) (B). Histopathology at 72 hours after elicitation is shown (C).
   Representative results from two independent experiments are shown. Bar
   indicates 100 μm.

   (D – E) DNFB-induced contact hypersensitivity was elicited after 25OHC
   (n = 4) or vehicle (n = 3) i.p. injection once a day for five days
   after sensitization. Ear skin thickness was measured (D) and area under
   the curve (AUC) was shown (E). *p < 0.05. Similar results were obtained
   from two independent experiments.

DISCUSSION

   Cholesterol and its derivatives are essential for maintaining cellular
   functions in broad range of cells. In immune cells, cholesterol and its
   derivatives are reported to have anti-viral, inflammatory, and
   anti-inflammatory properties, but also have roles in immune cell
   guidance involving innate and adaptive cells ([194]31–[195]36). In this
   study, we showed that CD4^+ T cells have the potential to produce
   Ch25h, a cholesterol-metabolizing enzyme, and to secrete the
   metabolite, 25OHC, to the cellular milieu in response to IL-27.
   Extracellular 25OHC impaired the viability and proliferation of
   TCR/IL-2 stimulated T cells by impeding cellular cholesterol
   biosynthesis while metabolically quiescent naïve cells were spared from
   25OHC-mediated inhibition ([196]Fig.S4A). Importantly, 25OHC-mediated
   inhibition was effective only in a narrow window of early T cell
   activation (<48 hrs) and the T cells expressing Ch25h and secreting
   25OHC are themselves refractory to this metabolite. This is because
   activated T cells apparently have passed a vulnerable window of 48 hrs
   after which they then produce 25OHC ([197]Fig.3C). Instead,
   Ch25h-expressing T cells can mediate a paracrine regulatory role via
   25OHC production to suppress other cells, including bystander T cells.
   This effect can be demonstrable in terms of reduced bystander T cell
   proliferation in vitro, but only if the bystander cells are in the
   early phase of TCR-dependent activation. In contract, the cells that
   have proliferated for more than 2 days, most likely regardless of
   polarizing cytokine conditions, become refractory to 25OHC
   ([198]Fig.5A).

   Expression of Ch25h and production of 25OHC has been described as
   having diverse but critical effects in immune cells.([199]19, [200]37).
   For example, 25OHC activates the integrated stress response, which may
   contribute to its antiviral activity ([201]38). Ch25h and 25OHC inhibit
   viral entry and replication in several cell lines ([202]19, [203]20,
   [204]39), and render host cells resistant to cholesterol-dependent
   cytolysin-induced pore formation during bacterial infection ([205]40).
   Notably, the effect of 25OHC to block SARS-Cov-2 replication has been
   reported ([206]41). Also Ch25h/25OHC suppress immunoglobulin class
   switching to IgA in B cells ([207]42). On the other hand, the
   deficiency of Ch25h results in the overproduction of IL-1 from
   macrophages, indicating 25OHC inhibits IL-1 ([208]34, [209]43).
   Accordingly, Ch25h-deficient mice exhibit increased sensitivity to
   septic shock and exacerbated experimental autoimmune encephalomyelitis
   (EAE), because 25OHC limits AIM2-dependent inflammasome activation by
   regulating cholesterol synthesis and mitochondrial DNA release
   ([210]43, [211]44). Taken together, our findings reinforce the idea
   that the Ch25h–25OHC axis supports immunoregulatory function across
   broader immune cell populations including T cells. While our adoptive T
   cell transfer model point to the impact of T cell specific 25OHC
   production, the caveat of interpreting effector T cell function in
   immunodeficient host needs to be considered.

   On the other hand, there is no doubt that the role of Ch25h is rather
   complex and context dependent (i.e., dosage, cell type, disease model),
   and seemingly contradictory findings have been reported. For example,
   25OHC amplifies expression of genes encoding inflammatory cytokines
   including Il6, Il8 and Csf1 ([212]45). Also 25OHC plays a
   proinflammatory role by suppressing type 1 regulatory T (Tr1) cells
   ([213]46). 25OHC suppresses both proliferation of Tr1 cells as well as
   production of IL-10, leading to the proposal of a proinflammatory role
   of Ch25h/25OHC as a result of suppressing suppressor cells. What was
   not evaluated in the report is the effect of 25OHC on T effector cells,
   and in vivo outcome of 25OHC effect on inflammation, which is
   determined by balancing action of effector vs regulatory cell subsets.
   In this regard, we found that 25OHC impacted cell proliferation and
   viability of freshly activated effector cells with high metabolic
   demand. Using two distinct skin inflammation models in vivo, we found
   that the lack of Ch25h resulted in more severe and more prolonged
   inflammation, indicating a regulatory role of Ch25h-25OHC axis in vivo.
   Based on our in vitro result, we envision that Ch25h25-OHC axis has the
   potential of exerting a regulatory role in a paracrine manner in vivo
   as well. Although we are unable to pinpoint the precise cells that
   produce and respond to 25OHC in our vivo models, the emergence of
   Ch25h^+ T cells responding to IL-27 could be one of such 25OHC
   producers. This mechanism could shape tissue microenvironment during a
   resolution phase in which further expansion of newly activated T cells
   and other inflammatory immune cells are constrained.

   The molecular mechanisms underlying 25OHC-induced cell death and
   signals converging to regulate this process are intricate. An
   interesting observation in our study is the anti-IL-2 antibody-mediated
   rescue of cell death. Anti-IL-2 antibody restored cell death even at
   high concentrations of 25OHC, whereas exogenously added
   cyclodextrin–cholesterol could not save the cells. Removal of IL-2
   signaling resulted in slowed cell growth, and the reduction in the
   proliferative state effectively saved cells from cell death otherwise
   caused by 25OHC in vitro. Our pathway analysis of 25OHC-treated cells
   also revealed the cell cycle, DNA repair, and centromeres as the
   biological processes affected by high concentrations of 25OHC in line
   with previous reports describing suppressive effect of 25OHC on
   activated lymphocyte proliferation ([214]26–[215]28). Since notable
   concentrations of 25OHC (~ 300 nM) were detected in the serum of mice
   administered the TLR4 agonist ([216]42), it is reasonable to believe
   that states of immune cell activation and proliferation, rather than
   effector phenotypes such as Th1 and Th17 e.g., are key factors that
   determine whether a particular subset of immune cells is or is not
   affected by 25OHC. Indeed, local administration of 25OHC to lung
   modulated inflammation induced by asthma or injury ([217]47).

   The importance of cholesterol regulation is considerably evident in
   activated T cells because the demand for essential nutrients, including
   cholesterol, is very high. T cells reprogram their metabolism by
   simultaneously inducing SREBP2 and repressing liver X receptor
   signaling, with progression into the cell cycle sustained by SREBP2
   activation and cholesterol biosynthesis ([218]48–[219]50). Conversely,
   the down-modulation of cholesterol biosynthesis compromises cell cycle
   progression and metabolic reprogramming of activated T cells
   ([220]26–[221]28). The induction of Ch25h and production of 25OHC by
   IL-27 and TGF-β represent another aspect of cholesterol metabolism,
   which does not impede non-proliferating resting T cells that do not
   contribute to pathogenic immune responses. Instead, 25OHC-mediated
   suppression of cell proliferation and induction of cell death likely
   contribute to the effective removal of highly active pathogenic cells
   from inflamed tissues in the resolution phase without affecting other
   quiescent resident cells. In this regard, it is also interesting to
   note that in our T cell transfer dermatitis model, IL-27 appears to be
   expressed as much in draining LN DC as in skin DC, yet
   dermatitis-invoking H1 T cells only upregulated Ch25h in the skin. It
   is tempting to speculate that the combination of strong TCR engagement
   at the site of inflammation in conjunction with the availability of
   TGF-β in the skin might favor local induction of Ch25h by IL-27.

   The cell state selective action of 25OHC raises the intriguing
   possibility of pharmacologically modifying cholesterol synthesis to
   control autoimmune or inflammatory diseases. Recently, cholesterol and
   other cellular metabolites have emerged as promising new targets to
   modulate immune responses ([222]48). For example, pharmacological
   blocking of cholesterol esterification in tumor-infiltrating
   lymphocytes altered intracellular cholesterol distribution to boost the
   amount of cholesterol in the plasma membrane. This resulted in enhanced
   tumor immunity owing to enhanced TCR clustering and immunological
   synapse formation ([223]51). By contrast, our study suggested a way to
   constrain the immune response via Ch25h and 25OHC, which suppresses
   cell-intrinsic cholesterol synthesis. Unlike many available
   immunosuppressants that indiscriminately suppress all immune cells,
   designing a compound that mimics or induces 25OHC could selectively
   target proliferating pathogenic cells in autoimmune conditions. In this
   manner, the adverse effects encountered previously could be minimized
   while delivering targeted anti-inflammatory control.

MATERIALS AND METHODS

Mice

   C57BL/6J mice were purchased from Sankyo Labo Service Corporation
   (Tokyo, Japan). Tbx21^−/− mice ([224]52) and Stat1^−/− mice ([225]53)
   were provided by Dr. Akihiko Yoshimura; Il27ra^−/− mice ([226]3) were
   provided by Dr. Hiroki Yoshida; Ch25h^−/− mice were purchased from the
   Jackson Laboratory (Bar Harbor, ME, USA). Rag2^−/− mice were purchased
   from the Central Institute for Experimental Animals (Kanagawa, Japan).
   Dsg3H1 mice were generated as previously reported ([227]29). The Keio
   University Institutional Animal Care and Use Committee approved all the
   experiments.

Reagents

   Cholesterol, 25OHC, and 7α,25OHC were purchased from Sigma-Aldrich (St.
   Louis, MI, USA) or Avanti Polar Lipids (Alabaster, AL, USA), resolved
   in DMSO, and stored at −80°C until use. Hexane, ethylacetate, and
   sodium sulfate anhydrous were purchased from Nacalai Tesque (Kyoto,
   Japan). 25-hydroxycholesterol and D6-25-hydroxycholesterol were
   obtained from Avanti Polar Lipids. Cholesterol-methyl-β-cyclodextrin
   was purchased from Sigma-Aldrich.

Cell culture

   Cell culture was performed in RPMI supplemented with 10% fetal bovine
   serum, 2 mM glutamine, 100 IU/ml penicillin, 0.1 mg/ml streptomycin, 1
   mM sodium pyruvate, and 2.5 μM β-mercaptoethanol. Naïve CD4^+ T cells
   (CD4^+CD62L^hiCD44^loCD25^− cells), memory CD4^+ T cells
   (CD4^+CD62L^loCD44^hiCD25^− cells), naïve CD8^+ T cells
   (CD8^+CD62L^hiCD44^loCD25^− cells), and memory CD8^+ T cells
   (CD8^+CD62L^loCD44^hiCD25^− cells) were isolated using a MoFlo XDP cell
   sorter (Beckman Coulter, Brea, CA, USA) or FACS Aria II (BD
   Biosciences, San Jose, CA, USA) and cultured in plate-bound anti-CD3
   and anti-CD28 (10 μg/ml each) in combination with 20 ng/ml IL-12 and 10
   μg/ml anti-IL-4 (11B.11; NCI Frederick, Frederick, MD, USA) for Th1; 20
   ng/ml IL-4 and 10 μg/ml anti-IFN-γ (XMG1.2; Bio X Cell, Lebanon, NH,
   USA) for Th2; 2 ng/ml TGF-β, 20 ng/ml IL-6, anti-IFN-γ, and 10 μg/ml
   anti-IL-2 (S4B6, BD Biosciences) for Th17 (TGF-β); 20 ng/ml IL-1β, 20
   ng/ml IL-6, 50 ng/ml IL-23, anti-IFN-γ, anti-IL-2, 5 μg/ml anti-TGF-β1
   (R&D Systems, Minneapolis, MN, USA), and 5 μg/ml anti-TGF-β2/1.2 (R&D
   Systems) for Th17 (IL-23); 20 ng/ml TGF-β, 50 U/ml IL-2, and anti-IFN-γ
   for iTreg; anti-IFN-γ and anti-IL-2 for Th0, for 3 days. IFN-β, IFN-α,
   and IL-27 were used in combination with anti-CD3 and anti-CD28 antibody
   stimulation for 2 days unless indicated for Ch25h induction. Cytokines
   were purchased from R&D Systems. For some experiments, CD8^+ cells were
   isolated using magnetic beads (Miltenyi Biotec, Bergisch Gladback,
   Germany) and stimulated with anti-CD3 and anti-CD28 as mentioned above.
   The effect of Ch25h on CD4^+ T cells was determined by culturing naïve
   CD4^+ T cells in plate-bound anti-CD3 and anti-CD28 (3 μg/ml each) in
   combination with various concentrations of 25OHC (Sigma-Aldrich). In
   some experiments, IL-2 was neutralized using anti-IL-2 antibody (S4B6).
   Cell viability and HMGB-1 release were assessed using a flow cytometer
   and HMGB-1 Detection Kit (Chondrex, Redmond, WA, USA), respectively.

Experiments with conditioned culture medium

   Conditioned culture medium was prepared by culturing naïve WT and
   Ch25h^−/− CD4^+ T cells with IL-27 (20 ng/ml) and TGF-β (20 ng/ml)
   along with anti-CD3 and CD28 antibodies for two days. The supernatant
   was collected after centrifugation and used for the 2^nd in vitro
   culture of 3 days using Il27ra^−/− CD4^+ responder T cells that were
   labelled with carboxyfluorescein diacetate succinimidyl ester (CFSE,
   ThermoFisher). To block TGF-β signals, neutralizing anti-TGF-β1 and
   TGF-β2 antibodies (2 μg/ml for each, R&D) were also added. %Divided was
   calculated by FlowJo (BD).

Transcriptome analysis in Th subsets

   Total RNA was prepared from 1 million cells using the mirVana miRNA
   Isolation Kit (Thermo Fisher Scientific). Subsequently, 1 μg of total
   RNA was used to prepare the RNA-seq library using the TruSeq SRRNA
   Sample Prep Kit (Illumina, San Diego, CA, USA), according to the
   manufacturer’s instructions. The libraries were sequenced for 50 cycles
   (single read) using the Hiseq 2000 (Illumina). Sequence reads from each
   cDNA library were mapped onto the mouse genome (build mm9) using tophat
   2.0.3. The obtained data were used to calculate gene expression values
   normalized based on fragment per kilobase exon model per million mapped
   reads using Cufflinks 1.3.0. Transcripts with normalized values of more
   than 1 were considered to be expressed in cells. RNA-seq data for naïve
   CD4 T cells previously deposited in GEO ([228]GSE45975) were used in
   [229]Figure 1A.

Transcriptome analysis: 25OHC-affected genes

   Naïve CD4^+ T cells were treated with different concentrations of 25OHC
   (0, 10, 100, and 1000 nM). After 1, 2, or 3 days of 25OHC treatment,
   cells were collected, and total RNA was isolated using the RNeasy Mini
   Kit (Qiagen), according to the manufacturer’s instructions. The quality
   of extracted RNA was evaluated using the RNA integrity number and
   Agilent 2100 Bioanalyzer (Agilent Technologies, Santa Clara, CA, USA).
   The DNA libraries for mRNA-seq were constructed using the SureSelect
   Strand-Specific RNA Library Prep Kit (Agilent Technologies), according
   to the manufacturer’s instructions, and sequenced using HiSeq 2500
   (Illumina). The DNA sequence was trimmed using the trim galore program
   (version 0.4.4;
   [230]http://www.bioinformatics.babraham.ac.uk/projects/trim_galore/)
   with the cutadapt program (version 1.14), and the trimmed sequence was
   mapped to the mouse genome (mm9) using STAR (version 020201). The
   number of reads of each gene was calculated using featureCounts
   (version 1.6.2) with the mm9 annotation [version July 17, 2015 (UCSC)].
   The mapped read data were normalized using rlog methods in the Deseq2 R
   package (version 1.16.1). Multivariate analysis for differential
   expression was performed using Deseq2 with variable 25OHC
   concentrations and culture days. Finally, genes that showed significant
   differences (p < 0.01; log2 fold change, ≥ 1 or ≤ −1) by multivariate
   analysis were analyzed using DAVID (version 6.8;
   [231]https://david.ncifcrf.gov) with default conditions to detect the
   candidate pathways associated with these genes. In the analyses,
   normalized read counts of genes associated with the sterol biosynthetic
   process (GO ID:0016126) in 25OHC-treated CD4^+ T cells on day 1 were
   expressed as a heatmap. Genes not expressed in CD4^+ T cells were
   excluded from the analysis. The calculated value from non-treated cells
   was defined as 1. Enrichment analyses of transcription factor-binding
   DNA motifs were performed with CiiiDER (L. J. Gearing et. al. PLoS One
   14, e0215495 (2019)).

RT-PCR

   Total RNA was isolated using the RNeasy Mini Kit (Qiagen, Hilden,
   Germany) in combination with RNase-free DNase. cDNA synthesis was
   performed using TaqMan Reverse Transcription Reagents (Applied
   Biosystems, Foster City, CA, USA). Quantitative PCR was performed using
   StepOnePlus Real-Time PCR System (Applied Biosystems) with Universal
   SYBR Select Master Mix using the following primers: Ch25h forward:
   5′-GCGACGCTACAA GATCCA-3′, Ch25h reverse: 5′-CACGAACACCAGGTGCTG-3′,
   β-actin forward: 5′-CGATGCCCTGAGGCTCTTT-3′, β-actin reverse:
   5′-TGGATGCCACAGGATTCCA-3′, Tbx-21 forward: 5′-CGTGGAGGTGAATGATGGA-3′,
   and Tbx-21 reverse: 5′-TGAGTGATCTCTGCGTTCTGGTA-3′, IL-27p28 forward:
   5′-TGTCCACAGCTTTGCTGAAT-3′, IL-27p28 reverse:
   5′-AAGGGCCGAAGTGTGGTAG-3′. In some experiments, Ch25h expression was
   quantified with TaqMan probe commercially available. All gene
   expressions were normalized by β-actin expression and shown as relative
   expression levels.

Flow cytometry

   A single-cell suspension of spleen, skin-draining LNs, mesenteric LNs,
   and skin was appropriately stained using 7-aminoactinomycin D, anti-Vβ6
   chain-BV421 (RR4-7), anti-CD8α-PE/Cy7 (53–6.7), anti-CD25-PE (PC61),
   anti-CD44-FITC (IM7), anti-CD62L-APC (MEL-14), and anti-CD4-APC/Cy7
   (RM4-5). The data were collected by FACSCanto II (BD) and analyzed by
   FlowJo (BD).

Sample preparation for GC-MS

   To 1 ml of culture medium, 4 ml of n-hexane containing
   deuterium-labeled internal standards (1 μg D6-25OHC) was added and
   mixed using a vortex mixer. Sterols were extracted with an additional 4
   ml of n-hexane. Following dehydration with sodium sulfate, organic
   fractions were applied on Sep-Pak Silica cartridges (WAT023595; Waters,
   Mississauga, Ontario, Canada) and eluted with 1.5 ml of ethyl acetate.
   The eluents were dried under a nitrogen stream. The residue was
   dissolved in 10 μl of pyridine and 50 μl of BSTFA + TMCS (99:1; #33148;
   Supelco Inc., St. Louis, MO, USA) for trimethylsilylation. The
   derivatization reaction was performed for 30 min at 40°C.

GC-MS analyses

   GC-MS analysis was performed on a Shimadzu GC-MS QP2010 Ultra equipped
   with an AOC20i autoinjector and Rtx-5MS column (30 m × 0.25 mm × 0.25
   μm df) in the 70 eV electron ionization mode. The oven temperature
   program was as follows: 60°C for 3 min and 15°C/min to 330°C, where the
   temperature was maintained for 10 min. The carrier gas was helium with
   a constant flow speed of 39.0 cm/s. One microliter was injected in the
   10:1 split ratio with an injector temperature of 250°C; the MS
   interface temperature was held at 280°C. Selected ion monitoring for
   quantification was performed by recording the ions at m/z 456.40 for
   the 25-hydroxycholesterol-trimethylsilyl derivative and at m/z 462.40
   for D6-25-hydroxycholesterol.

Experimental autoimmune dermatitis

   CD4^+Vβ6^+CD25^−CD62L^hiCD44^lo Dsg3-specific TCR transgenic naïve T
   cells were isolated from Dsg3H1 and Ch25h^−/−Dsg3H1 mice and adoptively
   transferred (0.5 to 3 × 10^5 cells/head) into Rag2^−/− mice with B220^+
   B cells (2 to 6 × 10^5 cells/head) derived from rDsg3-immunized
   K5-Dsg1tg-Dsg3^−/− mice ([232]54). In some experiments, TCR transgenic
   naïve T cells were obtained from Il27ra^−/−Dsg3H1 mice. The clinical
   severity of skin phenotype was evaluated at indicated time points
   according to the scoring system ([233]Table S5). For qPCR, CD11c^+ DCs,
   and Dsg3-specific CD4^+Vβ6^+ T cells were isolated by FACSAriaIII (BD).
   Dsg3H1 mice were immunized with 1 μg Dsg3 peptide, RNKAEFHQSVISQYR,
   with CFA (Sigma-Aldrich) as an adjuvant.

Contact hypersensitivity model

   To induce contact hypersensitivity (CHS) responses, 25 μL of 0.5%
   1-fluoro-2,4-dinitrobenzene (DNFB; Nacalai Tesque, Kyoto, Japan)
   dissolved in acetone/olive oil (4:1) was applied (for sensitization) to
   the shaved abdomens of mice (sensitization), which were challenged with
   20 μL of 0.15% DNFB (applied to the ear) 5 days later. Changes in ear
   thickness were measured at specific subsequent time points by using a
   thickness gauge (Teclock PG-20, Nagano, Japan). In some experiments, 1
   μmol 25OHC or vehicle (DMSO diluted 30 folds with PBS) were
   intraperitoneally injected in mice once a day for five days after
   sensitization. Graphs of ear thickness was plotted along the time
   points for vehicle control vs 25OHC treatment, and area under the curve
   was calculated for each and the difference was statistically evaluated
   by t-test using GraphPad Prism (GraphPad Software, San Diego,
   California, USA). The difference in ear thickness between experimental
   groups were evaluated by two-way ANOVA.

Pathological and immunofluorescence analysis

   The palate, esophagus, and skin of Rag2^−/− mice with the experimental
   autoimmune dermatitis phenotype or CHS responses were fixed in 10%
   formalin and stained with hematoxylin and eosin. Sections were observed
   under a DMD108 microscope (Leica, Wetzlar, Germany). For
   immunofluorescence staining, 10 μm cryosections of the palate were
   fixed with paraformaldehyde and stained with anti-CD4 and DAPI.
   Sections were observed under a fluorescence microscope (Axio
   Observer.Z1; Carl Zeiss, Oberkochen, Germany). Briefly, for Ch25h
   staining, differentiated T cells were spun onto microscope slides using
   a cytospin apparatus and fixed with acetone. Sections were stained with
   rat anti-Ch25h monoclonal antibody (185-1; BML) and stained with
   Alexa488-conjugated goat anti-rat IgG antibody, Alexa647-conjugated
   anti-CD4 antibody (RM4-5), and DAPI.

Statistics

   All data are shown as mean ± SEM of triplicate assays and analyzed by
   Student’s t-test using GraphPad Prism (GraphPad Software, San Diego,
   California, USA) unless otherwise specified.

Supplementary Material

   1

   Fig. S1. Ch25h is selectively expressed in IL-27- and TGF-β-treated
   CD4^+ T cells, but not in other Th subsets.

   Fig. S2. Tbx21 is induced by the combination of IL-27 and TGF-β.

   Fig. S3. Downregulation of enzymes in the cholesterol synthetic pathway
   in CD4^+ T cells treated with 25OHC.

   Fig. S4. The manner of 25OHC-induced cell death of T cells.

   Fig. S5. Equivalent viability and proliferation of wild-type and
   Ch25h^−/− CD4^+ T cells.

   Fig. S6. Ch25h expressed in T cells is important to restrain tissue
   inflammation in experimental autoimmune dermatitis.

   Table S1. List of genes expressed in CD4^+ T cells stimulated with
   IL-27 and TGF-β, but not in other indicated Th subsets.

   Table S2. Pathway enrichment analysis (25OHC; 10 nM vs. 100 nM)

   Table S3. Pathway enrichment analysis (25OHC; 100 nM vs. 1000 nM)

   Table S4. The expression of genes encoding enzymes was lower than that
   of genes encoding non-enzyme molecules related to cholesterol
   biosynthesis.

   Table S5. Severity scoring criteria for experimental autoimmune
   dermatitis
   [234]NIHMS1855627-supplement-1.pdf^ (1.4MB, pdf)

Acknowledgments: