T cell cholesterol transport links intestinal immune responses to dietary lipid absorption

Abstract

The intrinsic pathways controlling membrane organization in immune cells and their impact on cellular functions are poorly defined. We found that the nonvesicular cholesterol transporter Aster-A linked plasma membrane (PM) cholesterol availability in CD4 T cells to systemic metabolism. Aster-A was recruited to the PM during T-cell receptor (TCR) activation, where it facilitated the removal of accessible cholesterol. Loss of Aster-A decreased cholesterol accumulation in the PM, which enhanced TCR nanoclustering and signaling. Aster-A associated with stromal interaction molecule-1 (STIM1) and negatively regulated Ca²⁺ flux. Aster-A deficiency promoted CD4 T cells to acquire a Th17 phenotype and stimulated Interleukin-22 production, which reduced intestinal fat absorption and conferred resistance to diet-induced obesity. These findings delineate how immune cell membrane homeostasis links to systemic physiology.

INTRODUCTION

Impaired lipid flux can lead to cellular and tissue dysfunction in human diseases, yet the pathways that maintain membrane lipid homeostasis remain incompletely defined. Immune cells, which must adapt to evolving threats and distinct microenvironments, may be especially reliant on their ability to re-organize membranes for optimal function. The initiation and resolution of immune responses also likely requires timely recalibration of the cellular lipid repertoire. Whether and how immune cell membrane homeostasis impacts organ function and systemic metabolism is incompletely understood.

RATIONALE

Cholesterol is an indispensable lipid component of the mammalian plasma membrane (PM). We previously characterized the Aster family of nonvesicular lipid transporters, which transfer cholesterol from the PM to the endoplasmic reticulum. Despite established links between cholesterol abundance and immune signaling, it is unknown how immune cells finetune membrane cholesterol.

RESULTS

We traced fatty acid uptake in the small intestine of mice, and found that specific deletion of Aster-A in T cells reduced fatty acid absorption and conferred resistance to diet-induced obesity. Loss of Aster-A increased a gene expression signature associated with T helper-17 (Th17) cells among small intestine resident T cells. We found that Aster-A was highly expressed in Th17 cells and was indispensable for maintaining appropriate PM cholesterol levels in this cell type. Using specific cholesterol probes we determined that T cell receptor (TCR) activation transiently increased the accessible PM cholesterol pool, which recruited Aster-A to restrain excess PM cholesterol accumulation. In the absence of Aster-A, excess PM cholesterol was funneled into a distinct cellular lipid pool, the sphingomyelin-sequestered pool, leading to increased TCR nanoclustering. Consequently, Aster-A deficient Th17 cells exhibited elevated TCR signaling and effector cytokine (Interleukin(IL)-17/22) production. Proximity labelling in T cells revealed an interaction between Aster-A and STIM1–a core component mediating store-operated Ca²⁺ entry following TCR activation. We found that accessible PM cholesterol was required for TCR-induced Ca²⁺ influx, while Aster-A dampened this process. Aster-A thus restrained both the early signals directly triggered by the TCR and the later signaling events that depend on Ca²⁺ influx.

Single-cell transcriptomics and immune profiling further pinpointed increased IL-22 levels in Th17 cells from small intestines of T cell-specific Aster-A knockout mice. Acute administration of IL-22 prior to feeding was sufficient to suppresses intestinal fatty acid uptake. Conversely, blocking IL-22 with neutralizing antibodies, genetic ablation of Il22 from T cells, or antibiotic depletion of microbial signals that maintain gut Th17 cells, each restored dietary fat absorption or diet-induced weight gain.

CONCLUSION

Our work identifies a rapid, on-demand regulator of immune membrane homeostasis. Aster-A responds to TCR activation-induced PM remodeling and subsequently removes excess membrane cholesterol to restrain TCR signaling cascades. We propose that membrane lipid remodeling may serve to promote microenvironmental adaptation by controlling tissue T cell reactivity. Dysregulation and accumulation of T cell PM cholesterol, therefore has the potential to lead to aberrant gut Th17 effector function, alters immune-epithelial communication, and modulates intestinal and systemic nutrient metabolism.

Graphical Abstract

Nonvesicular cholesterol transport links intestinal T cell immunity to lipid absorption. T cell activation induces PM accumulation of “accessible” cholesterol and recruits Aster-A, which extracts cholesterol and transfers it to the ER to dampen TCR nanoclustering. Aster-A/STIM1 association at the PM further restrains Ca²⁺ influx during TCR activation. Aster-A guards against excessive intestinal Th17 responses, thereby coordinating dietary lipid flux and systemic metabolism. [Figure created in BioRender] (63 words).

The plasma membrane (PM) constitutes a major reservoir of unesterified cholesterol. The PM is thought to contain three cholesterol pools with distinct biochemical and biophysical properties (1): an essential pool required for membrane integrity and cell viability; a relatively inactive pool recognized by Ostreolysin A (OlyA) in which cholesterol is sequestered by sphingomyelin (SM) (2); and an active, or accessible pool that can be readily mobilized and that can be detected by domain 4 of Anthrolysin O (ALOD4) (3, 4).

Cellular and external cues, such as uptake of lipoprotein-derived cholesterol, activation of sphingomyelin degradation, and infection, regulate the size of the accessible cholesterol pool (5–11). Excess PM accessible cholesterol can move to the endoplasmic reticulum (ER) for esterification or be effluxed to extracellular lipoprotein acceptors (6, 12–15). However, the functions of accessible cholesterol within the PM are poorly defined. The sole exception is hedgehog signaling in the primary cilium, where Patched-1 inactivation by Hedgehog ligands increases accessible cholesterol and potentiates the activation of Smoothened (8, 16).

Nonvesicular sterol transport is recognized as being important for maintaining appropriate lipid distribution between cellular membrane compartments (17, 18). The Aster proteins play an indispensable role in moving accessible cholesterol from the PM to the ER (5, 12, 19). Increased accessible PM cholesterol facilitates recruitment of the GRAM domain of Asters to the PM and the formation of PM–ER contact sites, through which the ASTER domain channels excess cholesterol to the ER (12, 19). Aster proteins are required for the efficient internalization, storage, and utilization of diet- and lipoprotein-derived cholesterol in various metabolic and steroidogenic organs (6, 12, 15, 20). Of particular interest, Aster-A is highly expressed in lymphoid organs (12), suggesting a yet-to-be-identified function for nonvesicular sterol transport in the immune system. How immune cell cholesterol homeostasis impacts tissue-specific functions and systemic metabolism remains unexplored.

Studies involving chemical or indirect genetic approaches have reported that increasing PM cholesterol abundance can enhance T cell receptor (TCR) signaling (21–25). At the same time, cryo-EM structural studies have revealed that the resting αβTCR core contains two free cholesterol molecules that may serve to maintain the inactivate state (26, 27). Such studies imply that PM cholesterol availability must be tightly controlled during immune activation. However, it remains unknown whether T cells engage specific machinery to monitor and redistribute PM cholesterol. How membrane cholesterol dynamics are integrated with TCR signaling cascades has also not been defined.

Results

Dietary fatty acid absorption is decreased in mice lacking Aster-A in T lymphocytes.

We generated Aster-A–deficient mice to analyze potential contributions of this cholesterol transporter to systemic metabolism. We observed that whole-body deletion of Aster-A impaired fatty acid uptake by the small intestine (SI). We gavaged Aster-A–deficient (A^−/−) or littermate control (A^+/+) mice with [³H]triolein (18:1), a major fatty acid species in human and mouse diets, and measured the appearance of radioactivity in SI segments, plasma, and tissues (28, 29). A^−/− mice showed lower ³H counts in the jejunum and lower total SI ³H levels after 2 h (Fig. 1A and fig. S1A), and a delay in the plasma ³H appearance compared to A^+/+ controls (fig. S1B). In mice where Aster-A was deleted specifically in the liver, using the Albumin^Cre (A^ΔAlb), or in the intestinal epithelium, using Villin^CreERT (A^ΔVil) (fig. S1C), the intestinal uptake of [³H]triolein was not altered (Fig. 1B-C), consistent with redundancy of Aster-A/C in the liver and low Aster-A expression in SI epithelium (fig. S1D) (6, 15). Thus, the decreased fatty acid absorption in A^−/− mice was not due to intrinsic effects on enterohepatic tissues.

Fig. 1. Aster-A deficiency impairs fatty acid absorption in the small intestine and regulates diet-induced obesity.

(A) Distribution of radioactivity in intestinal segments of Aster-A wild-type (A^+/+, n = 13) and global deficient (A^−/−, n = 14) littermate mice after an oral gavage of olive oil containing [³H]triolein (18:1) for 2 h. (B-C) Distribution of radioactivity in intestinal segments of liver/hepatocyte-specific (A^ΔAlb, panel B), or intestinal epithelium-specific (A^ΔVil, panel C) knockout mice and their respective cre-negative floxed (F/F) littermate controls after an oral challenge of olive oil containing [³H]triolein (18:1) for 2 h (n = 9 and 6 per group for liver-specific knockout mice; n = 4 per group for intestinal epithelium-specific mice). (D) Distribution of radioactivity in intestinal segments of male and female F/F (n = 29) and littermate A^ΔCD4 (n = 28) mice after an oral challenge of olive oil containing [³H]triolein (18:1) for 2 h. Data are pooled from at least three independent experiments. (E) Total radioactivity in small intestines from experiments in panel D, separated by sex (Female, n = 14 and 15 per group; Male, n = 15 and 13 per group). (F) Kinetics of radioactivity in plasma of F/F (n = 25) and littermate A^ΔCD4 mice (n = 23) after an oral challenge of olive oil containing [³H]triolein (18:1). (G-H) Radioactivity in heart (panel G, n = 13 per group) and liver (panel H, n = 10 per group) from F/F and littermates A^ΔCD4 after an oral challenge of olive oil containing [³H]triolein (18:1) for 2 h. Data are pooled from at least two independent experiments. (I) Visualization of neutral lipid accumulation (marked by LipidTOX-Red) in jejunum cross sections from F/F and littermate A^ΔCD4 mice after an oral challenge of olive oil for 2 h. Scale bar: 200μm. Inlay shows villi tip in higher magnification. Scale bar: 50μm. Right, mean lipid droplet (LD) size and total number of LD per image. n = 5 mice per group pooled from two independent experiments. (J) Body weight of male (left) and female (right) F/F and littermate A^ΔCD4 mice fed ad libitum with 60 kcal% high-fat diet (HFD) for 9 to 10 weeks. Male, n = 17 and 16 per group; Female, n = 15 per group. Data are pooled from at least three independent experiments. (K-L) Fecal non-esterified fatty acid (NEFA) (panel K) and total cholesterol (panel L) level from F/F and littermate A^ΔCD4 mice fed ad libitum with HFD for 10 d. n = 8 to 11 per group. Data are pooled from two independent experiments. Data are shown as mean ± SEM. For panels A-D, F, J, data points indicate mean value; for panels E, G, H, I-L, data points indicate individual mice.

Statistical analysis: for panels A-D, two-way ANOVA with Sidak’s multiple comparison test. For panel E, two-way ANOVA; For panels F and J, repeated measure two-way ANOVA with Sidak’s multiple comparison test. For panels G-I, K, L, two-tailed unpaired Welch’s t-test. ns, p>0.05, *p<0.05, **p<0.01, ***p<0.001, ****p<0.0001.

We sought to identify cell types with prominent Aster-A expression. In public mRNA expression datasets from humans and mice, secondary lymphoid organs (spleen and lymph nodes) exhibited abundant transcripts for Aster-A (GRAMD1A/Gramd1a) and low levels of transcripts for Aster-B and -C (Gramd1b and Gramd1c) (fig. S1E-F). Examination of immune cell types in the Immgen database (30) revealed high and specific expression of Aster-A in conventional T cells (fig. S1G). In mice, where Aster-A was depleted in T cells specifically using the CD4^Cre (referred to as Α^ΔCD4 mice), Gramd1a transcripts in the CD4⁺ and CD8⁺ T cells were reduced by more than 100-fold, and Aster-A protein levels were undetectable (fig. S2A-B) compared to control mice that carried only floxed Gramd1a alleles (F/F). A^ΔCD4 mice were born at Mendelian frequencies and exhibited normal development, suggesting no gross signs of early-onset immune disorders (fig. S2C–D).

To examine whether T cell–intrinsic Aster-A impacted fatty acid uptake by the intestine, we gavaged A^ΔCD4 or littermate F/F mice with [³H]triolein. A^ΔCD4 mice showed lower peak ³H uptake in the duodenum and jejunum and reduced total intestinal ³H absorption after 2 h (Fig. 1D), in both males and females (Fig. 1E and fig. S3A). Moreover, the magnitude of the reduction was comparable to that observed in A^−/− mice (Fig. 1A). A^ΔCD4 mice also exhibited a reduced rate of ³H appearance in plasma as well as lower ³H uptake in heart and liver (Fig. 1F-H). In line with the radiotracer data, an oral challenge of unlabeled olive oil (of which triolein is a major constituent) resulted in lower total plasma triglyceride levels in A^ΔCD4 mice (fig. S3B). Thus, T cell–specific Aster-A deletion reduced the rate of fatty acid uptake into intestinal epithelium and subsequent entry into the systemic circulation.

We did not observe a difference in the SI length or in gastrointestinal transit times between F/F and A^ΔCD4 mice fed a chow diet (fig. S3C-D). Histopathology and electron microscopy studies revealed that jejunal villus and crypt morphology, villus length, and brush border morphology were indistinguishable in F/F and A^ΔCD4 mice (fig. S3E-F). No evidence of inflammation was observed in the lamina propria (fig. S3E). Following olive oil gavage, villus tip enterocytes contained large unilocular lipid droplets in F/F mice, compared to markedly smaller sized droplets in A^ΔCD4 mice (fig. S3E-F). We quantified the size and distribution of enterocyte lipid droplets by image analysis of large cross-sections of jejunum stained with a neutral lipid probe. A^ΔCD4 mice exhibited a shift in distribution to smaller droplet size and fewer numbers of lipid droplets (Fig. 1I and fig. S3G), accompanied by reduced expression of lipid droplet-associated gene expression in jejunal epithelium (fig. S3H). Furthermore, mice that harbored the CD4^Cre allele but were heterozygous for the floxed Aster-A allele (CD4-Cre; A^F/+) showed no difference in [³H]triolein uptake compared to A^F/+ littermates (fig. S4A-C), thereby confirming that reduced fatty acid uptake in A^ΔCD4 mice was linked to loss of Aster-A function in T cells and not to the presence of CD4^Cre transgene. Furthermore, uptake of [¹⁴C]cholesterol was comparable between F/F and A^ΔCD4 mice (fig. S4D-E). Thus, the impact of Aster-A in T cells on nutrient absorption was specific to fatty acid uptake and unrelated to any apparent SI pathology.

Resistance to diet-induced obesity in mice with Aster-A-deficient T lymphocytes

To understand the long-term metabolic impact of Aster-A loss in T cells, we challenged mice with a 60 kcal% high-fat diet (HFD). Compared to F/F littermates, both female and male A^ΔCD4 mice were resistant to HFD-induced weight gain (Fig. 1J). After 10 weeks of HFD, A^ΔCD4 mice showed comparable body length, but were phenotypically leaner compared to F/F controls (fig. S5A). Tissue weights of liver, epididymal white adipose tissue (eWAT), and inguinal white adipose tissue were lower in A^ΔCD4 mice compared to controls (fig. S5B). Analysis of body composition by MRI showed less fat mass in A^ΔCD4 mice but no difference in lean mass (fig. S5C-D). There were no body composition differences when mice were fed normal chow diet (13 kcal% from fat) (fig. S5E). Thus, T cell-specific loss of Aster-A conferred resistance to diet-induced obesity.

Loss of Aster-A in T cells blunted the inflammatory adipose tissue changes characteristic of obesity. Histological examination showed comparable fat cell size but reduced macrophage infiltration in eWAT of HFD-fed A^ΔCD4 mice (fig. S5F). Compared to F/F controls, A^ΔCD4 mice were resistant to adipose resident macrophages (F4/80^hiCD11b^int) loss and showed less accumulation of inflammatory macrophages (F4/80^intCD11b^hi) (fig. S5G-H). Furthermore, A^ΔCD4 mice had reduced fat deposition in the liver (fig. S6A), and maintained better glucose tolerance and insulin sensitivity when fed with HFD but not with normal chow diet (fig. S6B-E).

Lower adiposity in A^ΔCD4 mice could not be attributed to differences in food intake, respiratory exchange ratio, or energy expenditure (fig. S7A-C). However, A^ΔCD4 mice showed increased fecal non-esterified fatty acids compared to controls with HFD feeding (Fig. 1K), despite similar total fecal output and abundance of unabsorbed fecal cholesterol (fig. S7D and Fig. 1L). These results indicated overall reduced dietary fat absorption, in addition to altered uptake kinetics. Fecal triglyceride content and pancreatic lipase expression was similar between F/F and A^ΔCD4 mice (fig. S7E-F). We concluded that loss of Aster-A in T cells decreased intestinal fatty acid uptake, leading to resistance to obesity.

Loss of Aster-A in T cells enhances the Th17 signature in the small intestine

As loss of Aster-A in T cells impacted SI fat absorption, we performed transcriptional profiling of CD4⁺ and CD8⁺ T cells from jejunal lamina propria (LP) of mice that were acutely fed with olive oil. While few transcriptional changes related to T cell effector function were observed in the CD8⁺ T cells (fig. S8A), genes that were upregulated and enriched in A^ΔCD4 CD4⁺ T included those encoding the cytokines Il17a, Il17f, Il22, and other genes previously associated with gut Th17 lineage and function (Fig. 2A-C). We analyzed T cell lineage and cytokine producer frequencies in secondary lymphoid organs and SI-LP of normal chow-fed mice and found an increased frequency of IL-17A⁺ and IL-22⁺ CD4 T cells in mesenteric lymph nodes and SI-LP, respectively (fig. S8B-E). By contrast, tissue residency markers and frequencies of other T cell lineages were largely unaltered (fig. S8F). Among T helper cell subtypes, Aster-A transcripts (Gramd1a) were prominently expressed in the Th17 and Treg lineages in mice (Fig. 2D) and in Th17 cells in humans (fig. S9A-B). Aster-A protein levels were most abundant in Th17 cells generated in vitro in the presence of IL-1β and IL-23 (Fig. S9C), and they increased over the course of differentiation (fig. S9D-E). The preferential expression of Aster-A in Th17 cells corresponded with the predicted binding of transcription factors that enforce the Th17 lineage, including JUN, STAT3, EGR2, and SMAD4, at the promotor region of Gramd1a (fig. S9F).

Fig. 2. Aster-A maintains the homeostasis of PM accessible cholesterol in Th17 cells.

(A) Volcano plot showing differentially expressed genes in CD3⁺CD4⁺ small intestine lamina propria T cells from F/F and littermate A^ΔCD4 mice gavaged with olive oil for 2 h. Blue dots indicate differentially expressed genes (FC > 1.5, p < 0.1). Red dots indicate genes of interest. n = 3 to 4 biological replicates per group. (B) Heatmap of representative differentially expressed genes from panel A. (C) Gene set enrichment analysis of all differentially expressed genes in lamina propria CD4⁺ T cells comparing A^ΔCD4 to F/F. (D) mRNA expression of Gramd1a (Aster-A) in in vitro differentiated Th cell subtypes. npTh17: IL-6 + TGFβ; pTh17: IL-6 + TGFβ + IL-1β + IL-23. Data are representative of at least three independent experiments. (E-F) Western blot (panel E) and flow cytometry histogram (panel F) of ALOD4 in pTh17 cells (gated on live cells) differentiated from F/F and A^ΔCD4 naive CD4⁺ T cells. Mean fluorescence intensity (MFI) is shown in panel F. n = 5 per genotype from three independent experiments. Secondary anti-His Tag-PE antibody alone, without His-ALOD4 staining was used as negative control. (G) ALOD4 staining (Red) of F/F and A^ΔCD4 pTh17 cells (IL-6+TGFβ+IL-1β+IL-23) or loaded with 25 μM Methyl-β-cyclodextrin (MβCD)-Cholesterol for 1 h. Scale bar: 1 μm (left panels) and 10 μm (right panels). DAPI (Blue) staining indicates nucleus. Data are representative of at least two independent experiments. (H) Schematic diagram of the competitive bone marrow reconstitution and ALOD4 staining assay. (I) (Left) histogram of accessible cholesterol levels (by ALOD4) in small intestine lamina propria RORγt⁺ T cells and (right) quantified difference of in total CD4⁺, RORγt⁺, or RORγt⁻ T cells, comparing CD45.2⁺ F/F or CD45.2⁺ A^ΔCD4 to the congenic competitor (WT CD45.1⁺) in the same mice. n = 5 per group, representative of two independent experiments. Gating strategy is shown in fig. S21B. Data are shown as mean ± SEM. For panel F, data points and connecting lines indicate cells generated from littermate pairs; for panel I, data points indicate individual recipient mice.

Statistical analysis: for panel F, paired Student’s t-test. For panel I, two-way ANOVA with Tukey’s multiple comparison test. ns, p>0.05, **p<0.01.

Aster-A modulates the accessible cholesterol pool in Th17 cells.

Aster proteins mobilize the ‘accessible’ pool of cholesterol from the PM to the ER in several cell types (6, 12, 15, 20). To determine if Aster-A modulated PM cholesterol levels in T cells, we differentiated primary Th17 cells from F/F or A^ΔCD4 mice in defined serum-free media to limit extracellular cholesterol availability (fig. S10A), and monitored accessible cholesterol levels on the PM with the ALOD4 probe (3). A^ΔCD4 Th17 cells, but not naïve T cells, exhibited elevated accessible PM cholesterol compared to F/F controls (Fig. 2E-F and fig. S10B). Accessible cholesterol was distributed at ‘hot spots’ on F/F Th17 cells but was widely distributed on the PM of A^ΔCD4 Th17 cells, which resembled T cells forcibly loaded with exogenous cholesterol (Fig. 2G).

To test whether Aster-A contributed to membrane cholesterol homeostasis in the Th17 lineage in vivo, we generated mixed bone marrow chimeric mice using F/F or A^ΔCD4 donor and the congenic, wildtype CD45.1⁺ competitor (Fig. 2H). This scheme allowed us to conduct intrinsic comparison of the PM accessible cholesterol pool between Aster-A-sufficient and -deficient SI-LP CD4 T cells using a combination of ALOD4 and surface marker staining (Fig. 2H). ALOD4 levels on A^ΔCD4 RORγt⁺ Th17 cells were increased compared to that of the CD45.1⁺ competitor (Fig. 2I). In contrast, no difference was observed between F/F Th17 cells and the competitor, or in RORγt⁻ non-Th17 cells between any donor genotypes (Fig. 2I and fig. S10C). Thus, Aster-A regulated accessible cholesterol level in SI Th17 cells in a cell-intrinsic manner. Aster-A deficiency had no effect on RORγt⁺ frequency in SI-LP T cells in the mixed bone marrow chimera setting (fig. S10D), consistent with our observations in steady-state mice (fig. S8E). Aster-A-mediated nonvesicular cholesterol transport was thus required to maintain PM cholesterol homeostasis in Th17 cells in vitro and in vivo.

T cell receptor activation generates accessible PM cholesterol and recruits Aster-A

To investigate the physiological signals that modulated accessible cholesterol, we examined PM dynamics in Th17 cells after T cell receptor (TCR) activation by plate-bound anti-CD3/CD28. Control Th17 cells maintained constant accessible cholesterol levels over 4 h (Fig. 3A). Aster-A-deficient Th17 cells exhibited elevated ALOD4 binding post activation, followed by a delayed rate of cholesterol removal (Fig. 3A). TCR-independent activation with PMA/ionomycin, in contrast, led to prolonged and unresolved accessible cholesterol accumulation in Th17 cells of both genotypes (Fig. 3A). This finding indicated that T cell activation increased accessible PM cholesterol, and that Aster-A was required for the normalization of PM cholesterol after TCR activation.

Fig. 3. T cell receptor activation generates accessible cholesterol that rapidly recruits Aster-A to the PM.

(A) Western blot analysis of PM ALOD4 staining (accessible cholesterol level) of in vitro-differentiated F/F and A^ΔCD4 pTh17 cells (IL-6 + TGFβ + IL-1β + IL-23) that were re-activated with plate-bound anti-CD3/CD28 antibody (1 μg/ml) for indicated times. (B) Localization of mCherry-Aster-A in EL-4 cells unstimulated, TCR crosslinked, or treated with 1 μM Ionomycin. Plasma membrane TCRβ was stained with anti-TCRβ-AlexaFluor 647 antibodies. Scale bar: 2 μm. (C) Schematic graph of PM recruitment-dependent B-GRAM-H ddFP dimerization. (D) Time course of ddFP fluorescence signal (normalized to first 120 seconds of baseline signal) in response to TCR crosslinking. Trace of four independent recordings, average of 200 to 1000 cells per second per recording. (E) Kinetics of ddFP signal in a 60-minute timescale. n = 4 independent recordings for TCR activation. (F) Time course of ddFP fluorescence signal (normalized to first 120 seconds of baseline signal) in response to 1 μM ionomycin. Trace of five independent recordings, average of 200 to 1000 cells per second per recording. (G) Western blot analysis of ALOD4 (accessible cholesterol level) and OlyA (sphingomyelin-sequestered cholesterol level) staining of in vitro differentiated F/F and A^ΔCD4 Th17 cells (IL-6 + TGFβ + IL-1β + IL-23) that were reactivated with plate-bound 1 μg/ml anti-CD3/CD28 antibody or loaded with 25 μM MβCD-Cholesterol for 1 h. Data are representative of or pooled from at least two independent experiments. Data are shown as mean ± SEM.

Aster proteins are recruited to the PM by accessible cholesterol (5, 6, 10, 12, 19). To test if TCR activation was sufficient to recruit Aster-A to the PM, we expressed mCherry fused to the N-terminal GRAM domain of murine Aster-A in the EL-4 mouse T cell line. Following synchronized TCR activation, the mCherry signal rapidly mobilized from perinuclear ER towards cell surface TCR (Fig. 3B and Movie 1), indicating Aster-A recruitment to the PM. Ionomycin further provoked synchronized localization of mCherry signal to the PM (Fig. 3B and Movie S1).

To further resolve the kinetics, we utilized a dimerization-dependent system (ddFP) composed of a high sensitivity GRAM domain (B-GRAM-H) and a PM anchor fused to split fluorescent protein (30, 31), allowing simultaneous detection of accessible cholesterol generation and GRAM recruitment to the PM (Fig. 3C). B-GRAM-H-ddFP was expressed in EL-4 T cell line and human Jurkat T cell line (fig. S11A), and ddFP dimerization was detected using flow cytometry. Jurkat cells, but not EL-4 cells, showed increased ddFP signal in response to PM cholesterol loading and thus were used for subsequent experiments (fig. S11B). When Jurkat cells were activated by TCR crosslinking, the ddFP signal increased by 3 min post activation (Fig. 3D), and remained elevated for up to 1 h (Fig. 3E). Ionomycin treatment led to a larger increase in ddFP signal (Fig. 3F), tracking with a greater accumulation of the PM accessible cholesterol compared to TCR activation (Fig. 3A). These observations prompted us to explore whether a common mechanism altered PM cholesterol distribution during TCR-dependent and -independent activation.

Redistribution of cholesterol from outer to inner leaflet of the PM, which enables sensing and transport by the Asters, can be triggered by altered transmembrane phospholipid asymmetry (32), such as phosphatidylserine (PS) scrambling. TCR-mediated activation triggers non-apoptotic, low-grade, and reversible PS flipping to the PM outer leaflet, in contrast to ionomycin which generates persistent widespread PS scrambling (33–35). We confirmed these prior findings in primary Th17 cells, Jurkat cells, and EL-4 cells using Annexin-V detection of outer leaflet PS (fig. S12A). We further tested whether TCR activation- or ionomycin-induced GRAM recruitment depends on TMEM16F, a major Ca²⁺-dependent PS scramblase (36). 1PBC, a well-characterized TMEM16F inhibitor (37), eliminated the ionomycin-induced Annexin-V-high population, but the Annexin-V-low population induced by both TCR activation and ionomycin appeared to be TMEM16F-independent (fig. S12B-C). Accordingly, 1PBC largely abolished ionomycin-, but not TCR-induced GRAM recruitment (fig. S12D and F, quantified in fig. S12E and G). Thus, both TCR-dependent and independent T cell activation generated accessible cholesterol and recruited Aster proteins to the PM, and TMEM16F-mediated PS scrambling contributed to this phenomenon upon ionomycin exposure.

Aster-A restrains TCR nanoclustering and dampens proximal TCR signaling

PM cholesterol composition has been proposed to modulate membrane TCR activity and avidity by impacting TCR nanoclustering (22, 38). We tested whether loss of Aster-A impacted the accessible and sphingomyelin-sequestered cholesterol pools using ALOD4 and OlyA probes, respectively (4). A^ΔCD4 Th17 cells showed enhanced ALOD4 and OlyA binding to the PM compared to F/F Th17 cells at steady state, and both pools expanded upon TCR activation (Fig. 3G). These data suggested that excess accessible PM cholesterol in Aster-A deficient cells funneled into sphingomyelin-sequestered domains.

We investigated if the accumulation of accessible and sphingomyelin-sequestered PM cholesterol in A^ΔCD4 Th17 cells altered TCR nanoclustering or signaling. Serial extraction of isolated Th17 membranes by saponin and Brij97 allowed separation of nanoclustered complexes from monomeric TCR (39). Quantification of CD3ζ chain abundance in these fractions indicated that the loss of Aster-A increased the TCR nanocluster to monomer ratio (Fig. 4A-B). Accordingly, following reactivation, A^ΔCD4 Th17 cells exhibited enhanced proximal TCR signaling, as well as increased phosphorylation of S6 and ERK compared to control Th17 cells (Fig. 4C, quantified in fig. S13A). Consequently, A^ΔCD4 Th17 cells showed increased frequency of IL-17A and IL-22 effector cytokine production and gene expression, reduced Foxp3 expression (Fig. 4D-E), and a moderate increase in proliferation (fig. S13B). We conclude that Aster-A functions to restrain TCR nanoclustering and signaling in Th17 cells by removing PM cholesterol.

Fig. 4. A regulates TCR nanoclustering and effector responses of Th17 cells through gating PM cholesterol level.

(A-B) Western blot analysis of TCR (CD3ζ) clustering in in vitro-differentiated F/F and A^ΔCD4 pTh17 cells (IL-6 + TGFβ + IL-1β + IL-23). Representative blot (panel A) showing two biological replicates, and quantification (panel B) of Saponin- (nanocluster) versus Brij97- (monomer) extracted CD3ζ intensity. n = 4 from two independent experiments. (C) Western blot analysis of phosphorylation of LCK/Src, Zap70, SLP76, S6, and ERK1/2 following TCR activation. In vitro-differentiated pTh17 cells were ‘rested’ overnight without differentiation cocktail or TCR stimulus and then reactivated with plate-bound anti-CD3/CD28 antibodies (1 μg/ml) for indicated times. Histone H3 serves as overall input control. Data are representative of three independent experiments. (D) Frequency of in vitro-differentiated pTh17 cells producing IL-17A or IL-22 (Gated on live single cells). Representative flow plot is shown on the left and quantifications on the right. n = 6 to 7 per genotype from at least two independent experiments. (E) Gene expression from in vitro differentiated pTh17 cells. n = 3 to 4 per genotype from three independent experiments. (F) Diagram of experimental set up to probe PM accessible cholesterol level during T cell activation in vivo. (G) (Left) Histogram of ALOD4 staining in mesenteric lymph node CD4 T cells (CD45⁺CD3⁺CD4⁺) from F/F and A^ΔCD4 mice that were intraperitoneally injected with anti-CD3ε (20 μg/mouse) for indicated times or 20 μg isotype control antibody (0 h time point). (Right) Normalized Mean Fluorescence Intensity (MFI) of ATTO647N-ALOD4. n = 3 to 4 mice per genotype per time point, representative of two independent experiments. (H) Frequency of RORγt⁺ population in mLN CD4⁺ T cells at indicated time points after anti-CD3ε injection. n = 4 to 7 mice per group per time point, pooled from two independent experiments. Data are shown as mean ± SEM. For panels A, D, data points indicate cells generated from individual mice; for panel E, data points and connecting lines cells generated from littermate pairs; for panel G, data points indicate mean values; for panel H, data points indicate individual mice.

Statistical analysis: for panels B, D, two-tailed Welsh’s t-test. For panel E, multiple ratio paired t-test. For panel G, H, two-way ANOVA with Sidak correction for multiple comparison. ns, p>0.05, *p<0.05, **p<0.01, ****p<0.0001.

To understand whether this pathway operated in vivo, we intraperitoneally administered CD3ε-specific antibody to mice (Fig. 4F) – a challenge that elicits activation and expansion of mesenteric lymph node (mLN)- and intestine-associated Th17 cells (fig. S13C) (40). Following anti-CD3ε injection, ALOD4 staining of F/F mLN T cells showed unchanged accessible cholesterol levels at 8 h and a moderate increase at 24 h (Fig. 4G). By contrast, accessible PM cholesterol accumulated substantially in A^ΔCD4 mLN T cells 8 h to 24 h after injection (Fig. 4G). Furthermore, anti-CD3ε treatment markedly expanded RORγt⁺ CD4 T cells in the mLN of A^ΔCD4 mice at 24 h, consistent with our finding that Aster-A dampens Th17 reactivation (Fig. 4H). These data demonstrated that TCR activation led to increased PM accessible cholesterol, and that Aster-A is required for its timely removal to limit Th17 activation and expansion.

Spatial association of Aster-A with STIM1 in T cells

To gain further insight into mechanisms by which Aster-A modulated T cell function, we probed the Aster-A proximity interactome by harnessing Turbo-ID fused to the N-terminus of Aster-A (fig. S14A) (41). This fusion allowed identification of proteins proximal to Aster following its recruitment to the PM by cholesterol loading or TCR activation (Fig. 5A). Excess PM cholesterol induced Aster-A self-biotinylation (Fig. 5B and fig. S14B), consistent with the reported oligomerization of Aster-A at ER-PM contact sites (12, 19). Our screen revealed Stromal Interaction Molecule 1 (STIM1), a core component of the store-operated Ca²⁺ entry (SOCE) machinery and known PM-ER contact site protein (42–46), to be in proximity to Aster-A. Aster-A association with STIM1 increased following both cholesterol loading (Fig. 5B and fig. S14B) and TCR-activation (Fig. 5C, lane 3). This association was enhanced by Ca²⁺ chelators that prolong STIM1 localization at PM-proximal ER (Fig. 5C, lane 4) (47, 48). Despite only pulling down a small fraction of endogenous or overexpressed STIM1 with Aster-A, their coimmunoprecipitation was nevertheless increased upon cholesterol loading (fig. S14C-D). These data implied that Aster-A and STIM1 interacted transiently rather than formed a stable complex. In contrast, the intracellular TCR signaling components LCK and CD3ε did not show prominent interaction with Aster-A (Fig. 5C), suggesting that TCR nanoclustering in the setting of Aster-A deficiency was largely a result of preexisting PM cholesterol accumulation (Fig. 4A).

Fig. 5. Aster-A associates with STIM1 and modulates TCR-mediated Ca²⁺ signaling.

(A) Schematic graph showing proximal labeling strategy using N-terminal Turbo-ID-fused Aster-A. (B) Results of semi-quantitative proteomics showing differentially biotinylated proteins. EL-4 cells expressing Turbo-ID-Aster-A were treated with 10% lipoprotein-deficient serum (LPDS) media overnight, 10% fetal bovine serum (FCS) media overnight, or LPDS media and loaded with 100 μΜ MβCD-Cholesterol (Chol). (C) Western blot analysis of biotinylated proteins in Turbo-ID-Aster-A EL-4 T cells stimulated with Ionomycin (1 μΜ) or plate-bound anti-CD3/CD28 (5 μg/ml) in the presence or absence of BAPTA-AM (50 μM) and EGTA (10 mM) for 1 h. SA, streptavidin. (D) Analysis of cellular localization of HA-Aster-A, endogenous STIM1, and plasma membrane TCRβ in primary mouse Th17 cells. A^ΔCD4 Th17 cells were reconstituted with HA-tagged WT or R186W Aster-A and were either unstimulated or stimulated with TCR crosslinking [soluble 5 μg/ml Biotin-anti-CD3/CD28 followed by 10 μg/ml streptavidin (SA)]. (Left) Merged images of all three or two of the three channels, as well as HA-Aster-A/STIM1 colocalization (White). (Right) The relative pixel intensity of TCRβ-AlexaFluor647, HA-AlexaFluor 555, and STIM1-AlexaFluor488 from cross-section of the cell (white box). Scale bar: 2 μm. (E) (Left) SOCE traces (median of the ratio of Ca²⁺ indicators Fluo-4 to FuraRed) from F/F and A^ΔCD4 Th17 cells that were restimulated with soluble biotin-anti-CD3/CD28 (5 μg/ml), followed by SA-mediated crosslinking (10 μg/ml) and Ionomycin (1 μM). Trace is the average from n = 3 independent experiments. (Right) Area under curve (AUC) of each treatment is quantified. (F) (Left) Representative SOCE trace from Aster-A deficient (A^ΔCD4) Th17 cells reconstituted with 5P (ASTER domain loss of function, control), WT, R186W (GRAM domain loss of function), or G184L (GRAM domain gain of function) Aster-A. Trace representative of three independent experiments. (Right) AUC of each treatment is quantified. * represents comparison to 5P mutant, $ represents comparison to WT. n = 5 independent experiments. Gating strategy is shown in fig. S22. (G) (Left) Representative SOCE trace from WT Th17 cells restimulated as in panel E. Cells were treated with 20 μg/ml purified ALOD4 or the cholesterol non-binding mutant (D4-Mut) 30 min prior to time 0. (Right) Quantifications of peak SOCE. n = 4 to 6 independent experiments. $ indicate comparison to untreated (UT) condition. (H) (Left) Representative SOCE trace from Ctrl gRNA or Aster-A KO (GRAMD1A gRNA) human Jurkat T cell clones re-expressed with empty vector (EV) or human Aster-A. (Right) AUC of each treatment is quantified. n = 3 independent experiments. Data are shown as mean ± SEM. For panel G, data points indicate independent experimental replicates.

Statistical analysis: for panels E, paired two-way ANOVA with Sidak’s correction for multiple comparison. For panel F, H, multiple paired one-way ANOVA with Tukey’s correction for multiple comparison. For panel G, ordinary one-way ANOVA. ns, p>0.05, *p or $p<0.05, **p or $ $p<0.01, $ $ $p<0.001.

Previously identified residues that are required for Aster-B sensing and transfer of PM cholesterol are highly conserved in Aster-A (Fig. S15A) (19, 49). Homologous mutations in Aster-A resulted in corresponding alterations in PM localization patterns in response to cholesterol loading (Fig. S15B). Aster-A R186W, a loss of function GRAM domain mutation that diminishes cholesterol sensing capacity, failed to localize to the PM in response to cholesterol (Fig. S15B). Co-immunoprecipitation of STIM1 with R186W was reduced compared to WT Aster-A (fig. S14D). The observation that the abundant ER-resident protein calnexin did not co-immunoprecipitate with Aster-A suggested that recruitment of Aster-A and STIM1 to PM was required for their association, despite the fact that they are both ER-anchored proteins (fig. S14C). To visualize the localization of Aster-A relative to STIM1, we reconstituted A^ΔCD4 Th17 cells with HA-tagged Aster-A (Fig. 5D and fig. S14E). In resting Th17 cells, both STIM1 and Aster-A localized primarily to perinuclear ER (Fig. 5D, row 1). In response to synchronized TCR activation, wild-type Aster-A migrated to the PM, and in proximity with the STIM1 ‘ring’ or ‘cap’ structure (Fig. 5D, row 2). Aster-A R186W failed to translocate to the PM, but this did not adversely impact STIM1 localization (Fig. 5D, row 3). Thus, when accessible cholesterol was available on the T cell PM, Aster-A increased its proximity to STIM1 due to their co-localization at PM-ER contact sites.

Aster-A and accessible cholesterol regulate TCR-induced Ca²⁺ cascades.

The coupling of STIM1 with the PM Ca²⁺ channel ORAI1 (Calcium release-activated Calcium Modulator 1) is indispensable for SOCE and Th17 effector function (50–52). The association of Aster-A and STIM1 suggested that Aster-A may regulate SOCE in Th17 cells. Using flow-cytometry and imaging-based assays, we found that A^ΔCD4 Th17 cells showed increased TCR-dependent and ionophore-induced Ca²⁺ influx, without impacting STIM1 expression (Fig. 5E and fig. S16A-B). By comparison, Aster-A deficiency did not impact SOCE in naïve T cells (fig. S16B-C).

We asked if cholesterol sensing or transport by Aster-A was required for the modulation of SOCE. In addition to Aster-A R186W, we generated and validated two additional homologous mutations: G184L, a gain-of-function GRAM domain mutation, which increased Aster-A localization to PM; and 5P (I431P, S432P, N433P, Q434P, L435P), a loss-of-function ASTER domain mutation that is unable to transport cholesterol. 5P was used as a reconstituted but non-functional control (Fig. S15A-B). Reconstitution of primary A^ΔCD4 Th17 cells with wild-type Aster-A or G184L decreased PM accessible cholesterol and Ca²⁺ flux (Fig. 5F and fig. S16D-E). Conversely, reconstitution with loss of function mutations in GRAM domain (R186W) or ASTER domain (5P) elevated PM accessible cholesterol and Ca²⁺ flux (Fig. 5F and fig. S16D-E). Thus, regulation of Ca²⁺ flux required both recruitment of Aster-A to the PM and cholesterol transfer capacity. These data indicate that Aster-A negatively regulated TCR-induced SOCE in Th17 cells.

Increasing PM cholesterol by treating Th17 cells with methyl-β-cyclodextrin-cholesterol or low-density lipoprotein (LDL) promoted TCR-dependent SOCE, while removing PM cholesterol by 2-hydroxypropyl-β-cyclodextrin abolished it (fig. S16F-G). To understand the specific role of the accessible cholesterol pool in STIM1-dependent Ca²⁺ flux, we incubated wild-type Th17 cells with ALOD4 to sequester and immobilize accessible cholesterol on the outer leaflet of the PM. ALOD4 markedly dampened TCR-induced Ca²⁺ flux compared to treating with the non-binding mutant (53), while maximum flux induced by ionomycin was intact (Fig. 5H). Therefore, we propose a model in which accessible cholesterol generated by TCR activation supports SOCE, while the recruitment of Aster-A to the proximity of STIM1 curtails Ca²⁺ influx by modulating PM cholesterol composition.

Aster proteins are highly conserved in mammals (12). We generated an Aster-A-deficient Jurkat T cell clone by CRISPR-Cas9, and then re-constituted it with human Aster-A or an empty vector (fig. S16H). Stable expression of human Aster-A efficiently reduced both TCR- and ionomycin-triggered Ca²⁺ flux compared to empty vector-transduced cells (Fig. 5H), mirroring our observations in mouse Th17 cells (Fig. 5E). These data revealed a functionally conserved accessible-cholesterol-Aster-A/STIM1 axis that regulated Ca²⁺ responses in T cells.

IL-22 drives malabsorption in T cell Aster-A deficiency

We investigated the basis for dietary lipid malabsorption in A^ΔCD4 mice. To explore factors in SI Th17 cells that may affect absorption, we performed single-cell RNA-sequencing of SI-LP mononuclear cells. CD90⁺ (based on Thy1 expression, the gene which encodes CD90) clusters were subsequently extracted and further sub-clustered to reveal 23 distinct lymphocyte subpopulations including conventional T cell subtypes, γδ T cells, innate lymphoid cells, and NKT cells (Fig. 6A and fig. S17A). Four prominent RORγt (Rorc)-expressing populations were identified, denoting Th17 cells (cluster-8), RORγt⁺ Tregs (cluster-7), γδT-17 cells (cluster-14), and type-3 innate lymphoid cells (ILC3, cluster-15) (Fig. 6B and fig. S17A). Examination of cluster-8 revealed that SI-LP Th17 cells from A^ΔCD4 mice expressed higher levels of Il22, with subtle changes in Rorc and Il17a expression (Fig. 6C and fig. S17B). Other canonical cytokines produced by SI Th17 cells, such as Il17f, Il21, and Ifng, were either below detection or comparable between genotypes (fig. S17B). Other markers associated with mucosal Th17 effector or regulatory functions were not notably changed (fig. S17C-D).

Fig. 6. Aster-A deficiency leads to IL-22-mediated inhibition of fatty acid absorption.

(A) Uniform Manifold Approximation and Projection (UMAP) analysis of single-cell RNA-sequencing (scRNA-seq) data from small intestine lamina propria (SI-LP) CD90⁺ (based on Thy1 expression) cells from F/F and littermate A^ΔCD4 mice. Cluster annotations are listed on the right panel. n = 1 pooled from 3 mice per group (total 4239 cells in UMAP). (B) UMAP analysis of Rorc expression. (C) Violin plots of Rorc, Il17a, and Il22 expression in cluster-8 (Th17) cells from scRNA-seq analysis in panel A-B. (D) Representative flow plots of IL-17A⁺ and IL-22⁺ SI-LP T cells (CD45⁺CD90⁺CD3⁺CD4⁺) from F/F and littermate A^ΔCD4 mice gavaged with olive oil for 2 h. Quantified percentages are shown on the right panel. n = 6 per group, pooled from two independent experiments. Gating strategy is shown in fig. S24A. (E) Percentage of IL-22-GFP⁺ in Vβ14⁺ or Vβ14⁻ SI-LP T cells (CD45⁺ CD90⁺ CD3⁺ CD4⁺) from Aster-A^F/F Il22^GFP(F/F × IL-22-GFP) and littermates Aster-A^F/F CD4^Cre Il22^GFP (A^ΔCD4 × IL-22-GFP) mice gavaged with olive oil for 2 h. n = 7 to 9 per group, pooled from two independent experiments. Gating strategy is shown in fig. S24B. (F) Schematic diagram for the IL-22 treatment experiment in panel G. (G) (Left) Distribution of radioactivity in SI segments of C57/BL6J wild-type mice after an oral challenge of olive oil containing [³H]triolein (18:1) for 2 h. Mice were intraperitoneally injected with PBS or 1.5 μg IL-22 at 1 h prior to triolein challenge. (Right) Total radioactivity in SI. n = 5 to 6 per group, representative of two independent experiments. (H) Western blot analysis of phospho-STAT3 (Tyrosine 705) and total STAT3 (t-STAT3) in jejunal epithelial scrapings from F/F and littermate A^ΔCD4 mice fasted for 16 h (Fast) or gavaged with olive oil (Fat intake) for 2 h. n = 3 to 4 per group, representative of two independent experiments. (I) Schematic diagram for the IL-22 neutralization experiment in panel J. (J) (Left) Distribution of radioactivity in SI segments of F/F and littermate A^ΔCD4 mice after an oral challenge of olive oil containing [³H]triolein (18:1) for 2 h. Mice were treated with 250 μg anti-IL-22 (αIL-22) or control IgG antibody for 16 h. (Right) Total radioactivity in the SI. n = 9 to 11 mice per group, pooled from two independent experiments. For panels D, E, G (right), J (right), data points indicate individual mice; for panels G (left), J, data points indicate mean values. Data are shown as mean ± SEM. For violin plots, lines indicate median and quartiles.

Statistical analysis: for panels D, G (right), two-tailed Welsh’s t-test. For panels E, ordinary two-way ANOVA. For G (left), J, two-way ANOVA with Sidak’s correction for multiple comparison. ns, p>0.05, *p<0.05, **p<0.01, ****p<0.0001.

IL-22 is a mucosal Th17-associated cytokine in mice that maintains gut homeostasis and mucosal defense by increasing the production of anti-microbial peptides and promoting epithelial regeneration (54, 55). Consistent with our RNA-seq results (Fig. 2A and 6C), flow cytometry revealed increased IL-22⁺ frequency among A^ΔCD4 SI-LP CD4 T cells, which also often co-expressed IL-17A (Fig. 6D). We generated F/F and A^ΔCD4 mice harboring an IL-22-GFP reporter allele. A^ΔCD4 SI-LP CD4⁺ T cells were more frequently GFP⁺ compared to control mice (Fig. 6E), while the GFP⁺ percentages of non-CD4 T cells and innate lymphocytes were comparable (fig. S18A). Induction of gut Th17 cells and IL-22⁺ populations in laboratory mice depend on the colonization of Segmented Filamentous Bacteria (SFB) (56). Accordingly, increased IL-22 in A^ΔCD4 mice was mainly attributed to TCR Vβ14⁺ T cells, a major clonal response to SFB antigen (Fig. 6E) (57, 58). Thus, loss of nonvesicular cholesterol transport in Th17 cells led to increased IL-22 producing cells in SI-LP.

Chronic overexpression or loss of IL-22 alters nutrient metabolism (59–63). We assessed whether exogenous IL-22 could suppress the absorption of a bolus of triolein. A single administration of IL-22 for 1 h prior to [³H]triolein challenge markedly reduced radiolabel uptake across the length of SI and its subsequent appearance in circulation (Fig. 6F-G and fig. S18B). To determine if increased IL-22 production from A^ΔCD4 Th17 cells impacted SI epithelial homeostasis, we assessed signaling and transcriptional components downstream of the IL-22 receptor. Jejunal epithelial scrapings from A^ΔCD4 mice showed elevated STAT3 phosphorylation compared to controls in both fasted and oil-fed states (Fig. 6H), confirming that IL-22 signals to jejunal epithelium, where most dietary fatty acids were absorbed. Furthermore, the distal ileum of A^ΔCD4 mice showed elevated expression of IL-22-responsive antimicrobial peptides, Reg3b and Reg3g (fig. S18C). Thus, the consequences of increased IL-22 production from A^ΔCD4 Th17 cells were evident throughout the length of the SI.

To directly test the hypothesis that elevated IL-22 contributed to lower fatty acid uptake in A^ΔCD4 mice, we transiently blocked the effect of IL-22 with a neutralizing antibody (Fig. 6I). IL-22 neutralization effectively lowered Reg3b and Reg3g expression in A^ΔCD4 mice compared to mice administered control isotype IgG (fig. S18C). Notably, blocking IL-22 also reversed the impairment in fatty acid absorption in A^ΔCD4 mice (Fig. 6J). To further determine the contribution of T cell IL-22 to diet-induced obesity in A^ΔCD4 mice, we generated Il22^F/FA^F/FCD4^Cre and Il22^F/FA^F/F mice (fig. S19A). Il22^F/FA^F/FCD4^Cre and Il22^F/FA^F/F mice showed no difference in weight gain when challenged with HFD (fig. S19B). SI [³H]triolein uptake was also normalized between the two genotypes (fig. S19C), thus implicating the epistatic relationship between IL-22 and Aster-A in T cells. These data indicated that loss of Aster-A in T cells decreased fatty acid absorption and restrained diet-induced obesity via IL-22.

Gut-resident immune cells and IL-22 can influence the microbiota composition (63, 64), which may independently affect responses to diet (65, 66). However, F/F and A^ΔCD4 mice fed HFD had similar fecal bacterial composition (fig. S20A). Depletion of gut bacteria and fungus with broad spectrum antibiotics effectively reduced Il22 transcripts in SI and mLN (fig. S20B-D) and accordingly restored fatty acid uptake in A^ΔCD4 mice (fig. S20E-F). Furthermore, littermate F/F and A^ΔCD4 mice that were individually housed or co-housed to either deviate or normalize microbiota showed similar differences in HFD-induced weigh gain and fat mass (fig. S20G-H). Thus, gut microbiota was required for IL-22 induction but itself did not directly contribute to impaired fatty acid uptake in A^ΔCD4 mice. We conclude that the effect of T cell-specific loss of Aster-A changes fatty acid absorption through direct T cell-enterocyte communication mediated by IL-22.

Discussion

Aster proteins are recruited to ER-PM contact sites by accessible cholesterol and mobilize this pool to the ER (12). Here we showed that the accessible PM cholesterol pool expands immediately following TCR activation, leading to Aster-A recruitment. Some of this additional accessible cholesterol, if not removed promptly, appeared to be funneled into the sphingomyelin-sequestered domains (21, 22), which supported TCR nanoclustering and reduced the threshold for activation. We posit that the heightened Th17 effector function observed in vitro and in vivo in the absence of Aster-A arises from both the inability to rapidly remove accessible cholesterol generated by TCR activation, and the increased abundance of TCR nanoclusters formed due to chronic accumulation of sphingomyelin-cholesterol domains.

Excess ER free cholesterol is rapidly converted to cholesteryl esters by acyl-coenzyme A: cholesterol acyltransferase (ACAT) (67). Inhibition of ACAT1 causes PM cholesterol accumulation and potentiates TCR activity in cytotoxic T cells (25, 68). Our results in Th17 cells suggest that Asters most likely act upstream of ACATs in this regulatory axis. Loss of Aster-A alone does not seem to impact peripheral T cell composition, consistent with previous studies in which homeostatic T cell cholesterol content was perturbed by loss of SCAP/SREBP2, ACAT1, or SULT2B1B (21, 25, 69, 70). Instead, PM cholesterol accumulates in Aster-A-deficient SI-LP Th17 cells in a cell-autonomous manner. Dynamic expansion of the accessible cholesterol pool also occurred following T cell activation and was associated with the expansion of Aster-A-deficient RORγt⁺ Th17 cells in vivo. Thus, we propose that Aster-A is an indispensable guard of Th17 membrane homeostasis.

It was unknown whether Asters cooperate with other classes of membrane contact site proteins at the PM. We identified the proximity of Aster-A with STIM1 in activated T cells. The precise molecular details by which Aster-A affects STIM1 function in SOCE remain to be determined. Notably, two unbiased protein interaction and proximity maps also identified association of Aster-A with STIM1 and TCR components (71, 72). Prior microscopy studies in human epithelial and fibroblast cell lines suggested stable colocalization of STIM1 with GRAMD2A, but not GRAMD1A/Aster-A at PM-ER contact sites (73). Given that we failed to co-immunoprecipitate a substantial quantity of STIM1 with Aster-A, we posit that their interaction is likely transient or indirect. Furthermore, murine T cells show only trace expression of Gramd2, suggesting that STIM1-containing PM-ER contact sites must be defined by a different set of proteins in T cells. Recruitment of the Aster GRAM domain to the PM requires dual recognition of accessible cholesterol and anionic phospholipids such as PS (5, 10, 12, 19, 49, 74). Paradoxically, PS scrambling from inner to outer leaflet appears to contribute to ionomycin-induced GRAM domain recruitment to the PM in T cells. The GRAM domain can bind anionic phospholipids other than PS, including phosphatidylinositol 4,5-bisphosphate [PI(4,5)P₂] and phosphatidic acid, which could serve as alternative PM localization signals in the absence of PS (12, 49). STIM1 is known to reside in PI(4,5)P₂-rich domains (46, 75, 76), leading us to speculate that Aster-A could be recruited to STIM1 domains by these lipids. Reciprocal PS-PI(4,5)P₂ transport at the PM-ER contact sites has recently been found to finetune STIM1 clustering (77, 78). Thus, additional mechanisms of PM phospholipid remodeling may act in conjunction or sequentially with Aster for regulation of TCR signaling and SOCE.

Excess LDL cholesterol can be incorporated into membranes of immune cells and may promote immunopathology (79–81). In particular, enhanced Th17-associated inflammation has been observed in various mouse models of hypercholesterolemia (82–85). Cholesterol internalized by the LDL receptor-mediated endocytosis is liberated in the lysosome and then moves back to the PM where it can be mobilized to the ER by Asters (6, 18, 74). Our results raise the intriguing question of whether Aster-A functions to curtail hypercholesterolemia-induced inflammatory T cell responses by resolving the PM accumulation of LDL-derived cholesterol. Further studies are required to fully understand potential anti-inflammatory roles of Asters in hypercholesterolemia and atherosclerosis.

Finally, our findings show that loss of membrane homeostasis in immune cells impacts systemic lipid metabolism. Aster-A expression in T cells is required for appropriate fatty acid absorption and helps determine chronic metabolic responses to HFD. Although we specifically traced the uptake of labeled triolein, which is composed solely of 18:1 fatty acid, diet challenge and fecal analysis suggest a broad impairment of dietary fat absorption in Α^ΔCD4 mice. Importantly, increased IL-22 production mediates this T-cell-intestinal crosstalk. In line with the well-established role of SFB in inducing IL-22⁺ Th17 cells (56, 58, 86), Aster-A deficiency led to major changes in IL-22 frequency in the SFB-associated clonal type. These observations imply that Aster-A tunes gut Th17 reactivity to commensal bacteria under homeostatic conditions. Our results, along with recent studies demonstrating the metabolic protective role of SFB-elicited Th17 cells (62), support the notion that mucosal lymphocytes exert important restraints on intestinal lipid handling.

Materials and Methods

Mice

Aster-A^F/F (exon 12–16 floxed) and Aster-A^−/− (global knockout) mice were generated by C57BL/6 ES cell-based gene targeting as described previously (5, 6). Aster-A^F/F offsprings were backcrossed into C57BL/6N background and crossed with CD4^Cre (RRID:IMSR_JAX:022071) line to generate T-cell-specific knockout animals, with Albumin^Cre (RRID:IMSR_JAX:003574) line to generate liver-specific knockout animals, with Villin^CreERT2 (RRID:IMSR_JAX:020282) line to generate tamoxifen-inducible, intestine epithelium-specific knockout animals. For tamoxifen-induced knockout, mice were injected intraperitoneally with 1 mg tamoxifen in 100 μl corn oil (both from Sigma Aldrich) per day for 5 d and used for experiments 5 d later. All studies involving cre-flox mice were performed in age- and sex-matched littermates. For studies involving wildtype (WT/CD45.2) mice and congenic (CD45.1) mice, C57BL/6J mice (RRID:IMSR_JAX:000664) or CD45.1 congenic mice (RRID:IMSR_JAX:033076) were obtained from The Jackson Laboratories. CD45.1/.2 mice were generated by crossing WT C57BL/6J females with CD45.1 congenic males. IL-22^GFP mice (RRID:IMSR_JAX:035005) were purchased from Jax and crossed to CD4^Cre; Aster-A^F/F (A^ΔCD4) mice. Il22^hCD4.fl (Il22^F/F) mice were originally generated by Dr. Carlene Zindl and Dr. Casey Weaver (University of Alabama at Birmingham) (87) and were crossed to A^ΔCD4 mice. Unless otherwise specified, 8–14 week old female and male mice were used for in vivo experiments and 6–10 week old female and male mice were used for in vitro experiments. All mice were housed in a temperature-controlled room (22 °C with humidity approximately 50–65%) with a 12-h light–dark cycle (06:00 to 18:00) under specific pathogen-free conditions. Unless otherwise specified in the experimental procedures below, mice were housed in ventilated cages containing nestlets as environmental enrichments, with a maximum of five adult mice per cage. Unless otherwise specified, mice were fed an irradiated chow diet (PicoLab Rodent Diet 20, 5053) and provided water ad libitum. Mice were euthanized by isoflurane overexposure, followed by secondary methods (cervical dislocation or organ harvest). All animal experiments were approved by the UCLA Institutional Animal Care and Research Advisory Committee (Animal protocol B-09–010, R-03–166, R-99–131).

Measurement of intestinal fatty acid absorption

For acute fatty acid uptake assay, mice were fasted for 16 h and then intragastrically gavaged with 5 μCi [³H]triolein (PerkinElmer) emulsified in 150uL highly refined, low acidity olive oil (Sigma Aldrich). Blood was sampled using retro-orbital method at 0, 30, 60, 120 minute time points. 2 h after gavage, mice were euthanized and small intestine was excised (between the base of the stomach and the cecal junction), flushed with at least 20 ml 0.5 mM sodium taurocholate in PBS, and cut it into 2 cm segments. Segments were incubated with 500 μl of 1N NaOH at 65°C overnight and then mixed with UltimaGold scintillation cocktail (PerkinElmer), and radioactive counts were collected using a Beckman LS6000 scintillation counter. Liver and heart were also harvested and rinsed extensively to remove blood and weighed. Liver medial lobe and heart were incubated with 1 mL and 500 μl of 1N NaOH, respectively, at 65°C overnight and ³H was counted as above and normalized to tissue weight. Plasma was obtained from blood in K2-EDTA tubes followed by centrifugation at 2,000 xg for 15 minutes. 20 μl plasma sample was mixed with 5 ml scintillation cocktail for counting.

Diet studies

Normal chow diet (PicoLab Rodent Diet 20, 5053) with 13.1 kcal% fat was purchased from LabDiet. The composition of major fatty acids (w/w) are as follows: 46.4% C18:2, 5.6% C18:3, 0.4% C20:4, 8.4% ω−3 fatty acids, 15.4% total saturated fatty acids, 20% total monounsaturated fatty acids. Rodent diet with 60 kcal% fat (high-fat diet) was purchased from Research Diet (D12492). The composition of major fatty acids (w/w) are as follows: 34.0% C18:1, 28.7% C18:2, 19.6% C16:0, 10.6% C18:0, 2.0% C18:3, 1.3% C16:1. Mice were co-housed with littermate of both genotypes, single-housed, or grouped by genotype as indicated in individual experimental conditions. Unless otherwise stated, mice were fed indicated diet and given water ad libitum when they reach 8–10 weeks of age or 23g mean body weight for males, 19g mean body weight for females. Body composition (fat and lean mass) was measured by magnetic resonance imaging (EchoMRI) at the beginning and end of each study. Body weight was measured weekly. Tissue weights were measured at euthanasia. Food intake measurement, metabolic chamber studies, and fecal 16S sequencing was performed in single-housed male mice.

Indirect calorimetry using metabolic cages

4 weeks after high fat diet feeding, respiratory exchange ratios and energy expenditure were measured by indirect calorimetry using the Oxymax Comprehensive Laboratory Animal Monitoring System (Columbus Instruments). Data were analyzed using CalR application (88).

Glucose and Insulin tolerance assay (GTT/ITT)

Lean mass of mice was determined by MRI on the day prior to GTT/ITT. HFD-fed mice were fasted for 4 h, followed by intraperitoneal injection of 1 mg per gram of lean mass for glucose, or 2.5 units per kg lean mass for insulin (Humulin) and blood glucose was measured by glucometer (Bayer) at the indicated times.

Mixed bone marrow transplant

CD45.1/CD45.2 recipient mice were irradiated using X-ray irradiator with two fractioned doses of 550 cGy with at least 4 h interval. To harvest bone marrow, femurs and tibias were harvested from F/F, Α^ΔCD4, or competitor CD45.1 mice and removed of muscle tissues. Bone marrow single cell suspension was obtained by homogenizing bones using mortar and pestle. Unfractionated bone marrow cells from donor (CD45.2: F/F or Α^ΔCD4) and competitor mice (CD45.1) were mixed at 1:1 ratio and 8–10 million mixed bone marrow cells were intravenously injected through the tail vein into the recipients. Recipient mice were given 0.5 mg/ml Baytril (Enrofloxacin) in drinking water for 1–2 weeks prior to transplant and for 2 weeks after transplant to prevent opportunistic infection. Small intestinal lamina propria cells were harvested 8–10 weeks after transplant using the methods described below. Cells were stained with antibodies against CD45.1, CD45.2, CD3, CD4, ALOD4, and RORγt for flow cytometry analysis as described below. All antibody information is reported in Table S1.

Acute IL-22 treatment and in vivo IL-22 neutralization experiments

In vivo IL-22 neutralization was performed by injecting 250 μg functional grade anti-human/mouse IL-22 (clone IL22JOP, ThermoFisher) intraperitoneally at the time of fasting (16 h prior to [³H]triolein gavage). Control mice were injected 250 μg of isotype Rat IgG2a, κ (clone eBR2a). Acute IL-22 treatment was performed by intraperitoneal injection of 1.5 μg recombinant mouse IL-22 (Genscript) in PBS. 1 h later, mice were used for fatty acid absorption assay.

Depletion of microbiota by broad-spectrum antibiotics

Caging and food were sterilized by autoclaving/irradiation. Mice were given broad spectrum antibiotics (1 g/L Ampicillin, 1 g/L Neomycin, 1 g/L Metronidazole, 0.5 g/L Vancomycin, 1 g/L Gentamycin, and 0.01 g/L amphotericin B) in sterile drinking water for 2 weeks. Feces before and after antibiotics treatment were collected. Bacterial DNA was isolated from same quantity of feces using Quick-DNA Fecal/Soil Microbe Miniprep Kit (Zymo Research). Quantitative PCR of 16s rRNA was performed using primers in Table S2.

Intraperitoneal anti-CD3ε antibody

20 μg anti-mouse CD3ε (clone 145–2C11) or control polyclonal Armenian hamster IgG (both from BioXCell) was diluted in DPBS and injected peritoneally to mouse at indicated times prior to sacrifice. For ALOD4 staining, mesenteric lymph nodes were resected and processed to single cell suspension on ice in DPBS (Ca²⁺, Mg²⁺). Cells were stained with ALOD4 and live/dead stain for 1 h at 4 °C, followed by fixation and permeabilization (eBioscience FOXP3 Transcription Factor staining set), and antibody staining as described below.

Plasmids

Table S3 lists plasmids used and generated by this study. Full length mouse Aster-A coding sequence was cloned into: 1) 3xHA-TurboID pRetrox with TurboID (Addgene Plasmid #155203) located at N-terminus of Aster-A GRAM domain with a GGGGS linker; 2) MSCV-IRES-hCD2 (Addgene Plasmid #133061) vector with an N-terminal HA tag and GGGGS linker; 3) an inframe N-terminal mCherry and 3x GGGGS linker, all by Gibson assembly (NEB). Aster-A functional mutations were identified from homologous amino acids between mouse Aster-A and human/mouse Aster-B (19, 49) and cloned into MSCV-IRES-hCD2 vector by site-directed mutagenesis. C-terminal Myc-GGGGS-tagged mouse STIM1 was cloned from pcDNA3.1 mSTIM1 myc (Addgene Plasmid #17732) and inserted into pMSCV-Blast (Addgene Plasmid #75085) vector by Gibson assembly. Full-length human Aster-A was cloned from Jurkat cDNA and mCherry was tagged at N-terminal using methods above into a pMSCV-hygro (Addgene Plasmid #34565) vector. For CRISPR plasmids, guide RNA sequences (described below) were cloned into pSCAR_sgRNA_puro-mKate-lox2272 (Addgene Plasmid #162076) or pSCAR_sgRNA_hygro-tagBFP-lox5171 (Addgene Plasmid #162070) vectors using BsmBI sites.

Viral production and transduction

For retrovirus production for murine cells, 2.5 μg transfer plasmid and 0.63 μg pCL-ECO were co-transfected to 1×10⁶ Platinum-E cell line in 6-well plate using OptiMEM and FugeneHD reagent for 6–8 h. Media was replaced with appropriate culture media for the target cell, and virus supernatant was harvested at 48 and 72 h post transfection. For retrovirus production for human cells, 2.5 μg transfer plasmid was transfected to Phoenix-AMPHO cell line and virus was harvested at 48 h. For lentivirus production, 2.5 μg transfer vector, 1.875 μg psPAX2, 1.25 μg pMD2.G was co-transfected to Lenti-X cell line in a similar procedure as above. For integration deficient Cre lentivirus, 2.5 μg pLX-Cre-ppl-del, 1.875 μg psPAX2 D64V, 1.25 μg pMD2.G was co-transfected to Lenti-X cell line. For viral transduction of suspension cell lines, viral supernatant was harvested and cleaned through a 0.45 μm filter. Cells were spin infected in 12-well plates with 1–2 ml viral supernatant containing 8 μg/ml polybrene (for cell lines) or 10 μg/ml Protamine Sulfate (for primary T cells). Viral supernatant was replaced with normal growth medium 4–6 h later and antibiotics selection was performed 48–72 h post transduction. For some experiments, transduction was repeated 24 h later to increase efficiency.

Cell lines, construction of stable overexpression cells and CRISPR knockout cells

Platinum-E (RRID:CVCL_B488) retroviral packaging cell line was obtained from Cell Biolabs. Lenti-X 293T cell line (RRID:CVCL_4401) was obtained from Takara Bio. Phoenix-AMPHO (RRID:CVCL_H716), EL-4 (RRID:CVCL_0255), Jurkat (RRID:CVCL_0065) cell lines were obtained from the American Type Culture Collection.

To construct stable TurboID-expressing EL-4 lines, EL-4 cells were transduced with retrovirus containing pRetroX-3xHA-TurboID or pRetroX-3xHA-TurboID-GGGGS-mAster-A. Cells were selected with 10 μg/ml puromycin for 1 week and maintained in 0.3 μg/ml puromycin. To construct stable mCherry-tagged mAster-A expressing EL-4 line, EL-4 cells were transduced with retrovirus containing pMSCV-mCherry-GGGGS-mAster-A, mCherry positive cells were FACS-sorted and used without further subcloning. To construct stable HA-tagged mAster-A expressing EL-4 line, EL-4 cells were transduced with retrovirus-containing pMSCV-HA-GGGGS-mAster-A (WT or R186W mutant)-IRES-hCD2, cells were FACS-sorted by surface hCD2 expression. STIM1 was further introduced by infecting HA-Aster-A cell lines with virus containing pMSCV-STIM1-Myc-His-Blast and cells were selected with 10 μg/ml Blasticidin for >1 week and maintained on 1 μg/ml Blasticidin. To construct PM-RA and B-GRAM-H (B-GRAM-H-ddFP) expressing cell lines, EL-4 or Jurkat cells were transduced with lentivirus containing pLJM1-PM-RA-BlastR and pLJM1-B-GRAM-H-NeoR. Cells were selected with 2 mg/ml G418 and 50 μg/ml Blasticidin for 1 week and maintained in 0.5 mg/ml G418 and 10 μg/ml Blasticidin. Protein expression in overexpression or reconstitution experiments was confirmed by western blot.

To construct Aster-A knockout Jurkat human T cell lines, pSCAR-Cas9-GFP was introduced by lentiviral transduction and stable polyclonal Cas9-GFP expressing Jurkat cells were selected by FACS sorting. Cas9-GFP Jurkat cells were co-transduced with pSCAR_sgRNA1_puro-mKate-lox2272 or pSCAR_sgRNA2_hygro-tagBFP-lox5171 virus with two distinct gRNAs: gRNA1 targeting hAster-A exon 10 (5’-GTCTGTCAATAGCTCCTCAC-3’) and gRNA2 targeting exon 14 (5’-TCTCAATGAGCGACTTCACC-3’). The resultant deletion of exons in between will eliminate the majority of ASTER domain in addition to a frame shift. As control, Cas9-GFP Jurkat cells were transduced with the above sgRNA vector containing viruses where sgRNA1 (5’-GCACTACCAGAGCTAACTCA-3’) and sgRNA2 (5’-GTGCGAATACGCCACGCGAT-3’) do not target any human genome sequence. Cells were selected with 50 μg/ml Blasticidin, 10 μg/ml Puromycin, and 500 μg/ml Hygromycin for 1 week to allow deletion. Jurkat cells were single-cloned by FACS, and GRAMD1A knockout clones were selected by PCR and subsequently confirmed by the absence of hAster-A protein by western blotting (anti-hAster-A, Bethyl labs). Control clone was selected to match growth rate, morphology, and cell size of the KO clone. Subsequently, Cas9-GFP, mKate, and tBFP was removed by transducing single clones with integrase deficient virus and isolated by GFP/mKate/tBFP-triple negative FACS sorting. hAster-A KO Jurkat clones were further reconstituted with HA-tagged hAster-A, or HA alone empty vector (EV) using amphotropic retrovirus and were selected by 500 μg/ml Hygromycin for 2 weeks.

Plasma and tissue lipid quantification

Plasma lipids were measured by colorimetric assay with L-Type TG M kit (Wako). Tissues were homogenized by dounce homogenizer in JP lysis buffer (10 mM Tris-HCl pH 7.4, 150 mM NaCl, 1 mM EDTA, Halt protease inhibitor cocktail (ThermoFisher)) and Folch method was used to extract lipids from materials containing 0.5–1 mg tissue protein (89). Protein content was measured by Bicinchoninic Acid Assay Protein Assay Kit (Pierce). Extracted lipids were dried under inert gas and resuspended in isopropanol and quantification was performed using the previously described kits. For cholesterol measurement in culture media, 1 ml of media was used for lipid isolation and quantification using Total Cholesterol E Kit (Wako).

Measurement of Fecal NEFA and lipid species

Littermate mice were fed with high fat diet for 10 d and then single-housed with fresh cellulose bedding. Feces accumulated for 48 h were collected, snap frozen, and pulverized using mortar and pestle. 100 mg fecal homogenate was used for lipid extraction with 19 volumes of 2:1 (v/v) chloroform:methanol using the Folch method. Solvent was removed by drying under nitrogen gas, resuspending in isopropanol, and measured using NEFA-Hr kit (Wako).

Thin layer chromatography was performed using lipid extraction from 25 mg fecal homogenates. Fecal lipids were extracted as described above, resuspended in chloroform, and dotted on a Silica Gel 60 TLC plate (Millipore Sigma). Samples are stacked using solvent 1 system (CHCl₃/MeOH = 2:1) and neutral lipid species were subsequently separated using solvent 2 system (Heptane/Isopropyl Ether/glacial Acetic Acid = 15:10:1). A mixed lipid standard (mouse liver lipid extract supplemented with 20 mM 18:1 fatty acid) was ran along the samples to identify triglycerides and free fatty acids. Separated lipids were visualized by exposing to iodine vapor for 30 minutes.

Protein isolation from the intestinal epithelium

Intestines were harvested immediately following euthanasia and extensively flushed with PBS containing 1 mM dithiothreitol (DTT), Halt phosphatase and protease inhibitor cocktail. The proximal 1/6 of the intestine was designated as duodenum, middle 3/6 as jejunum, and last 2/6 as ileum. Indicated intestinal section was excised and longitudinally opened to expose the lumen. Epithelial scrapings were obtained by gently scraping the luminal tissue with two glass slides. To prevent protein degradation, scrapings were immediately snap frozen in liquid nitrogen. Scrapings were cut and weighed on dry ice, and approximately 100 mg sample was resuspended in 1 ml DS buffer [150 mM NaCl, 1.5 mM DTT, 50 mM Tris HCl pH 7.4, 1.25 mM EDTA, Halt phosphatase and protease inhibitor cocktails, 1 mM phenylmethylsulfonyl fluoride (PMSF)] and immediately homogenized on ice using motorized Dounce (850 rpm, 15 strokes). Homogenates were further sonicated (3 sec × 3 at 50%) and lysates were cleared (2000 xg, 10 minutes, 4°C).

Total cell protein lysate

Cells were washed with PBS by centrifugation for at least 3 times, and pellet was lysed with RIPA buffer (containing protease and phosphatase inhibitor cocktails) on ice for 30 minutes. Lysate was cleared by centrifugation at 12,000 xg for 15 minutes. In some experiments, boiled 2X LDS sample buffer was directly added to cell pellets and DNA was sheared by sonication.

Western blot analysis

Protein content was determined by the Bicinchoninic Acid Assay Protein Assay Kit (Pierce) or Pierce 660 kit. An equal amount of protein was loaded into the NuPAGE 4–12% Bis-Tris gels (Invitrogen). Proteins were transferred to PVDF blotting membrane (Amersham Hybond). Membranes were blocked for 1 h at room temperature with 5% non-fat milk in TBST buffer (TBS + 0.1% Tween 20) and incubated with primary antibodies listed in Table S1 in 5% non-fat milk or 5% bovine serum albumin (BSA). IR800-, Cy3-, Cy5-, Horseradish peroxidase-conjugated anti-mouse, anti-rat and anti-rabbit IgG were used as secondary antibodies. The blot membrane was visualized using the Immobilon Forte Western HRP Substrate, SuperSignal West Femto Maximum Sensitivity Substrate, or directly imaged for fluorescent signal using ChemiDoc MP (BioRad).

RNA extraction and gene expression analyses

For tissue RNA isolation, approximately 50–100 mg frozen tissue was immediately homogenized in Trizol Reagent using preloaded 4 mm metal beads and Qiagen Tissue Lyzer. For RNA isolation from cells, cells pellets were lysed with 0.5–1 ml TriZol reagent for 5 minutes. Total RNA was extracted using the TRIzol method, quantified by NanoDrop (ThermoFisher) and reverse transcribed. cDNA was quantified by real-time PCR using iTaq Universal SYBR Green Supermix (Bio-Rad) on a QuantStudio 6 Flex 384-well qPCR system (Applied Biosystems). Gene expression levels were determined by using ΔΔCt method. Housekeeping gene 36b4 was used for normalization and every sample was analyzed in triplicate. Primers used for real-time PCR are listed in Table S2.

Purification and fluorophore conjugation of ALOD4 and OlyA

pALOD4, pALOD4 non-binding mutant (G501A/T502A/T503A/L504A/Y505A/P506A) and pOlyA plasmids were obtained from AddGene or gifted from the Radhakrishnan lab (Table S3) and proteins were purified according to a previously published method (4). Briefly, plasmids encoding specified proteins were expressed in BL21 (DE3) pLysS Escherichia coli (Invitrogen). Cell pellets were lysed by sonication in lysis buffer containing 50 mM NaH₂PO₄, pH 7.0, 300 mM NaCl, 1 mg/ml lysozyme, 1 mM DTT, 2 mM PMSF and Halt protease inhibitor cocktail. The protein was bound to HisPur Ni-NTA Agarose resin (Thermo Scientific), washed twice with 50 mM imidazole, eluted with 300 mM imidazole, and subjected to size exclusion chromatography on Superdex 200. Fractions containing specified protein were concentrated to 1 mg/ ml and stored at −80 °C with glycerol until use. For ALOD4 labeling, ALOD4 was conjugated to Alexa Fluor 555-maleimide, Alexa Fluor-488-maleimide (ThermoFisher), or ATTO647N-maleimide (SigmaAldrich) followed by affinity purification using HisPur Ni-NTA Agarose resin and dialyzation. Labeling efficiency was determined using on Nanodrop (Thermo Fisher).

ALOD4 and OlyA staining

For confocal imaging, cells were plated on Shi-Fix coverslips or poly-D-lysine coated coverslips. Cells were washed with DPBS (Ca²⁺, Mg²⁺) with 0.2% BSA three times. The ALOD4 working solution (20 μg/ml in DPBS + 0.2% BSA) was passed through a 0.45 μm filter before addition to cells and incubating while rotating at 4 °C in the dark for 1–2 h. Coverslips were washed three times with cold DPBS before being fixed with 3% paraformaldehyde (PFA) for 10–15 minutes at room temperature. After fixation, slides were washed with DPBS three times. For unconjugated ALOD4, cells were incubated with an anti-His antibody (27E8) overnight before incubation with fluorescently labeled secondary antibodies for 1 h at room temperature. Cells were mounted on slides with ProLong Diamond Antifade Mountant with DAPI (Invitrogen). Images were acquired using an Inverted Leica TCS-SP8-SMD Confocal Microscope.

For flow cytometry of ALOD4 staining, cells were washed with cold DPBS (Ca²⁺, Mg²⁺) for three times. Cells were incubated with ALOD4 working solution (20 μg/ml in DPBS) and Fixable Viability Dye eFluor450 (eBioscience) while rotating at 4 °C in the dark for 1–2 h. Cells were washed three times with DPBS containing 0.2% fatty acid-free BSA and fixed for 10–15 minutes with Fix/Perm buffer (BD) at 4 °C before proceeding to the remaining flow cytometry staining procedure. For flow cytometry analysis of plasma membrane ALOD4 following anti-CD3ε mediated activation, plasma membrane accessible cholesterol density was estimated by normalizing ALOD4-ATTO647N signal to forward scatter (FSC) parameter to regress out the increased cell size and surface area using the following conversion formula:

PM accessible cholesterol density = Intensity<ATTO647N>/(4π(Intensity<FSC>)²)

For western blot analysis, cells were washed extensively with DPBS (Ca²⁺, Mg²⁺). Equal numbers of cells were incubated with ALOD4 and OlyA working solution (both 20 μg/ml) as above. Surface ALOD4/OlyA-bound cells were lysed with RIPA buffer before being probed with anti-His (27E8).

Histology

Whole mouse tissue was fixed in neutral formalin solution for at least 48 h at room temperature, then rinsed extensively with water and preserved in 70% ethanol. Tissues were paraffin-embedded and cross-sectioned at 5 μm thickness. Hematoxylin and eosin (H&E) staining was performed by UCLA Translational Pathology Core Laboratory using routine procedures.

Immunofluorescence of the small intestine

Mice were fasted for 16 h to synchronize metabolic status, then were intragastrically gavaged with 150 μl refined low acidity olive oil. 2 h later, mice were euthanized and cardiac perfused with 10 mL cold DPBS. Small intestine was excised between stomach to cecum and flushed with 10–15 mL cold DPBS to remove luminal content, followed by a slow luminal flush of 10 mL 4% paraformaldehyde to flash-fix the luminal epithelium. Small intestine was then longitudinally opened and rolled into a Swiss roll with luminal facing out, immobilized in tissue cassette and fixed overnight in 4% PFA on a 4 °C rocker. Tissues were then dehydrated in 20% Sucrose in PBS for 24–48 h on a 4 °C rocker before embedding in OCT. Embedded tissue were sectioned at 18 μm thickness and attached to Superfrost Plus slides and further fixed with pre-chilled methanol in −20 °C for 20 minutes, air-dried and rehydrated with PBST (PBS containing 0.05% Tween-20). Slides were incubated in blocking/permeabilization buffer [PBS + 0.3% Triton X-100 + 10% normal goat serum block (BioLegend)] for 1–2 h at room temperature and stained overnight in blocking/permeabilization buffer containing anti-EPCAM-AlexaFluor 488 (G8.8). Slides were washed three to five times with PBST, stained with DAPI and lipidTOX-Red for 1 h at room temperature, and mounted with Prolonged Diamond anti-fade mounting reagent (all from ThermoFisher). For confocal imaging of the Swiss roll for lipid droplet quantification, tile scan images were acquired on a Zeiss LSM900 confocal microscope using Plan-Apochromat 10x/0.45 M27 and 30 μm Z-stack to capture the full thickness of the tissue sections. Following acquisition, the tile scan images were processed using stitch feature and then displayed as a maximum intensity projection.

Quantification of lipid droplets

To quantify lipid droplet size and number from enterocytes in the intestine confocal images, stitched maximum intensity projection images were deconvoluted in ImageJ with the DeconvolutionLab2 plugin using a theoretical PSF generated with the PSFGenerator plugin. Processed images were then loaded into CellProfiler (90) for lipid droplet quantification using a customized pipeline. Briefly, EPCAM signal was used to delineate cell membranes and mask lumenal artifacts. LipidTOX-stained lipid droplets were identified using the IdentifyPrimaryObjects module and measurements extracted with the MeasureSizeShape module. Droplet measurements were inspected and analyzed in R.

Live cell confocal imaging

0.25 × 10⁶ mCherry-Aster-A expressing EL-4 cells were resuspended in 500 μL warm Ringer’s solution (155 mM NaCl, 4.5 mM KCl, 2 mM CaCl₂, 1 mM MgCl₂, 10 mM D-glucose, 5 mM HEPES) and plated onto poly-D-lysine coated 35 mm No. 1.5 Glass Coverslip culture dish (MatTek). Cells were then stained with AlexaFluor 647 anti-TCRβ (non-blocking clone H57–597, 2.5 μg) for 30 minutes on ice. Confocal live cell imaging was performed using the Leica Confocal SP8-STED/FLIM/FCS microscope equipped with an environmental control stage (5% CO₂, 37 °C chamber and objective) and the 100x/1.40 objective. Live images (2048 × 2048 px per frame) were collected at 8000-Hz bidirectional scan speed with the resonance scanner using the following parameters: frame sequential mode, 15 second interval per acquisition, Pinhole = 1 Airy Unit (580 nm emission), pixel size = 102.8 px per micron with 5.84x digital zoom. mCherry channel was excited at 587 nm with 15% laser power and collected with a PMT gain of 1000V and a band pass of 595 to 640 nm. AlexaFluor647 channel was excited at 653 nm with 20% laser power and collected with a HyD gain of 200% and a band pass of 660 to 779 nm. Deconvolution was performed using the Leica LAS X Lightning Deconvolution module using the “adaptive” strategy. Bleach correction was performed using Histogram Matching Method (FIJI/ImageJ).

Dimerization-dependent fluorescent protein-fused GRAM domain recruitment assay

B-GRAM-H-ddFP expressing Jurkat cells were resuspended in Ringer’s solution and warmed up to 37 °C for 10–15 minutes. Live single cells were gated based on forward and side scatter parameters. ddFP signals were recorded using mCherry channel (Ex-561 nm/ Filter-585/15) on a BD LSR Fortessa flow cytometer. For TCR stimulation, 5 μg/ml anti-human CD3 was added for 15 minutes and then 2 minutes of baseline signal was acquired before addition of 10 μg/ml streptavidin (crosslinking agent, BioLegend) and recording for another 5–10 minutes. Following first segment of acquisition, cells were returned to 37 °C and were subsequently recorded at t = 15, 30, 60 minutes post activation (for 1 minute each) to obtain kinetics over a longer time frame. For ionomycin stimulation, 1 μM ionomycin was directly added following the 2-minute baseline recording. To pharmacologically inhibit phosphatidylserine scramblase TMEM16F, cells were pre-incubated with 10 μM 1PBC (Sigma Aldrich) for 30 minutes prior to recording. Kinetics platform in FlowJo v10 was used to plot time versus mCherry signal. Median with moving average was used to generate kinetics curve.

Electron microscopy

10-week-old F/F and A^ΔCD4 female mice were fasted for 16 h, then refed with refined low acidity olive oil for 2 h. Small intestine were harvested, contents were flushed with PBS and a 0.5 cm ring shaped segment from proximal jejunum (measured at 8–10 cm from the end of stomach) was harvested and directly placed into EM fixative (2.5% glutaraldehyde, 2% paraformaldehyde, 2.1% sucrose 0.1 M sodium cacodylate) at 4 °C overnight. The following day, the samples were washed 5 minutes, 5 times in cold 0.1 M sodium cacodylate and then postfixed with reduced osmium (2% OsO₄, 1.5% potassium ferricyanide, 0.1 M sodium cacodylate) for 1 h. Next, samples were washed 5 times, 5 minutes each in cold H₂O then treated with 1% thiocarbohydrazide for 20 minutes at room temperature. Samples were then rinsed with H₂O for 5 times, 5 minutes each, and stained again with osmium (2%OsO₄ in H₂O) for 30 minutes. After another round of H₂O washes, the tissue was stained with 1% uranyl acetate (SPI Chem) at 4 °C overnight. Tissues were dehydrated by incubating in a graded series of ethanol solutions (30%, 50%, 70%, 85%, 95%, 100%, 3 times each) for 8 minutes each. Then tissues were infiltrated with 33% (in acetone) EMbed 812 for 1 h, 66% overnight, then 100% for 4 h. Individual tissue pieces were then embedded in a flat mold and polymerized in a vacuum oven for 48 h. Resin blocks were then trimmed and 500 nm sections were collected onto small silicon wafers using a Leica UC6 ultramicrotome and a Diatome diamond knife. For backscattered electron microscopy (BSEM): Wafers were mounted onto SEM stubs with double-sided carbon tape and imaged using a Zeiss Supra 40VP SEM equipped with a backscattered electron detector. Accelerating voltage was set to 13KeV and working distance at approx. 5 mm. BSEM was performed at the Electron Imaging Center for Nanomachines which is a part of the California NanoSystems Institute at UCLA.

Tissue processing and immune cell isolation

For isolation of immune cells from lymphoid tissues (spleen, lymph nodes, thymus), tissues were homogenized between two frosted slides. Single cell suspension was obtained by filtering using a 70 μm strainer. In some experiments, red blood cell removal was performed using Hybri-Max Red Blood Cell Lysing Buffer (Sigma Aldrich).

For isolation of immune cells from small intestine lamina propria, tissues were carefully removed of mesenteric fat and Peyer’s patches. Then intestinal tissues were flushed of luminal contents, cut longitudinally open, and excised into 1-inch pieces. Cleaned tissues were extensively washed in cold HBSS buffer (HBSS + 25 mM HEPES) and then incubated in DTT buffer (HPSS + 25 mM HEPES + 3% fetal bovine serum + 1 mM DTT) for 10 minutes in an orbital shaker (200 rpm) at 37°C, extensively vortexed, and mucus in supernatant was removed through a 100 μm strainer. Epithelial layer was removed by two rounds of incubation with EDTA buffer (HBSS + 25 mM HEPES + 3% fetal bovine serum + 5 mM EDTA) for 10 minutes in an orbital shaker (200 rpm) at 37°C and followed by passing through 100 μm strainer. Remaining tissue was washed in HBSS, finely minced and digested (RPMI-1640, 0.5% fatty-acid free BSA, penicillin /streptomycin, glutamine, 1 mg/ml Collagenase D and 100 μg/ml DNase-I, 100 μg/ml Dispase II) for 35–55 minutes in an orbital shaker (150 rpm) at 37°C. Tissues were then vigorously vortexed and RPMI-1640 media containing 10% FCS was added. The digested tissue was filtered using 40 μm strainer. Cells were pelleted and loaded on 40%/80% Percoll gradient centrifugation (1,000 xg, 20 minutes at room temperature with no deceleration). Lamina propria mononuclear cells were collected from the interphase and washed with cold culture media or FACS buffer. For ALOD4 staining of ex vivo lamina propria cells, tissues were isolated and digested in 0.5% fatty-acid free BSA (Sigma) containing buffer in replacement of FCS.

For isolating adipose stromal-vascular fraction (SVF), mice were perfused with 10 mL cold PBS and adipose tissues were excised and weighed. Tissues were minced into 1 mm cubes and incubated in digestion buffer (HBSS + 10 mM HEPES + 10 mg/ml Collagenase D + 2% FCS + 10 mM CaCl₂, 4 ml/g tissue) for 40 minutes with orbital shaking (100 rpm) at 37 °C. Digested tissue was vortexed for 2 minutes, pushed through 70 μm strainer and pelleted (500 xg, 10 minutes). Supernatant containing mature fat cells were discarded and pellet containing SVF cells was washed again with cold culture media or FACS buffer.

Flow cytometry staining and analysis

Antibodies used for flow cytometry analysis are listed in Table S1. For the surface flow cytometry analysis, cell suspensions were resuspended in FACS buffer (DPBS, 2% FCS or 0.5% BSA, 2 mM EDTA), blocked with Fc block (anti-CD16/CD32), and subsequently stained with fluorophore conjugated antibodies. In some experiments, biotin-conjugated surface antibodies were used and subsequently stained with secondary streptavidin-PerCP-Cy5.5 (Biolegend). For non-conjugated ALOD4, a secondary PE anti-His Tag antibody (J095G46) was used and sample stained with secondary antibody but not ALOD4 was used as negative control. DAPI or LIVE/DEAD Fixable dyes were used to exclude dead cells. For intracellular cytokine staining, cells were cultured in 50 ng/ml PMA (Sigma-Aldrich), 1 μM ionomycin (Sigma-Aldrich) in the presence of 5 μg/ml brefeldin A (BioLegend) for 4–6 h and then were stained with Live/Dead Fixable dye according to manufacturer’s protocol. Cells were fixed and permeabilized using the Cytofix/Cytoperm kit (BD) and stained with antibodies diluted in permeabilization buffer. For intracellular transcription factor staining, cells were stained with Live/Dead Fixable dye and then fix, permeabilized according to FOXP3 fix/perm buffer set (eBioscience) protocol. Samples were acquired on a BD Verse or LSR-II flow cytometer. Data were analyzed using FlowJo v10. Unless specifically indicated, cells are identified based on Forward and Side Scatter (FSC/SSC) and single cells are gated using FSC-A/H/W and SSC-A/H/W parameters. Experiment-specific flow gating strategies are shown in supplemental figures.

T cell isolation and differentiation

Naive CD4 T cells were isolated from spleen and peripheral lymph nodes of 6–12 week old mice using naive CD4 T cell isolation kit (Miltenyi or BioLegend). For ‘pathogenic’ Th17 (pTh17) differentiation, naive CD4 T cells were activated using plate-bound anti-CD3/CD28 (5 μg/ml) in X-VIVO15 media (LONZA) containing 25 ng/ml mIL-6 (Peprotech), 5 ng/ml hTGFβ1 (Miltenyi), 20 ng/ml mIL-1β (Peprotech), 20 ng/ml IL-23 (R&D Systems), 10 μg/ml anti-IFNγ, and 10 μg anti-IL-4 (both from Biolegend) for 4 days. For restimulation experiments, day-4 pTh17 cells were removed from differentiation cocktail, washed and rested in fresh X-VIVO media overnight. Th17 cells were restimulated using 1–2 μg/ml plate-bound anti-CD3/CD28. Differentiation of other T-cell subtypes was performed as follows: Th0: 20 unit/ml IL-2 (BioLegend); Th1: 20 unit/ml IL-2, 10 ng/ml IL-12 (Peprotech), 10 μg/ml anti-IL-4; Th2: 20 unit/ml IL-2, 5 ng/ml IL-4 (Peprotech), 10 μg/ml anti-IFNγ; non-pathogenic Th17: 25 ng/ml mIL-6, 5 ng/ml hTGFβ1, 10 μg/ml anti-IFNγ, and 10 μg anti-IL-4; Th22: 25 ng/ml mIL-6, 20 ng/ml mIL-1β, 20 ng/ml IL-23, 400 nM FICZ (Invivogen), 10 μg/ml anti-IFNγ, and 10 μg/ml anti-IL-4; Treg: 50 unit/ml IL-2, 5 ng/ml hTGFβ1, 10 μg/ml anti-IFNγ, and 10 μg/ml anti-IL-4.

Retroviral transduction of primary Th17 cells

Naïve CD4 T cells were activated by plate-bound anti-CD3/CD28 (5 μg/ml) in differentiation X-VIVO15 media containing 25 ng/ml mIL-6, 5 ng/ml hTGFβ1, 20 ng/ml mIL-1β, 20 ng/ml IL-23, 10 μg/ml anti-IFNγ, and 10 μg/ml anti-IL-4 for 24 h. Activated cells were then spin infected (2,500 rpm, 90 minutes, 32°C) with retroviral supernatant containing 20 units/ml IL-2 and 10 μg/ml protamine sulfate (Sigma-Aldrich), and incubated with viral supernatant for additional 4–6 h. Then cells were returned to differentiation media. Spin-infection was repeated 24 h later. Successfully transduced T cells were identified by their surface expression of human CD2, which does not cross-activate mouse TCR.

TCR nanocluster and monomer purification

Membrane fraction was purified from primary Th17 cells using a modified serial lysis method (39). 5–10×10⁶ T cells were extensively washed in cold PBS, incubated in hypotonic lysis buffer (10 mM HEPES, 42 mM KCl, 5 mM MgCl₂, protease inhibitor cocktail, 1 mM PMSF), and mechanically disrupted using tight-fitting glass Dounce homogenizer. Whole cells and nuclei were removed by centrifugation for 10 minutes at 150 xg. Membrane patch was purified by ultracentrifugation at 100,000 xg for 30 minutes at 4 °C and pellet was washed once more with hypotonic lysis buffer. To extract TCR nanocluster, membrane pellet was resuspended in serial lysis buffer (20 mM Tris-Cl pH 8, 2 mM EDTA, 137 mM NaCl, 10%v/v glycerol, protease inhibitor cocktail, 1 mM PMSF) containing 1% saponin and rotated end-to-end at 4 °C for 15 minutes, followed by centrifugation at 14,000 xg for 15 minutes to obtain supernatant. The insoluble pellet was again resuspended in 1% saponin serial lysis buffer for a total of three rounds of extraction and supernatant was pooled. To extract TCR monomer, the insoluble pellet from previous step was further extracted for three rounds in serial lysis buffer containing 0.5% Brij97. Pooled extracts were reduced in LDS sample buffer and blotted with anti-CD3ζ.

Flow cytometry-based Ca²⁺ flux assay

Isolated naive CD4 T cells were cultured in X-VIVO15 media overnight before performing the Ca²⁺ flux assay. Differentiated Th17 cells were removed from differentiation cocktail, washed and ‘rested’ overnight in fresh X-VIVO15 media without further stimulation or IL-2. For flow cytometry-based assay, T cells were labeled with 1 μM Fluo-4-AM and 1 μM FuraRed-AM mixed 1:1 with 20% Pluronic F-127 in DMSO (all from ThermoFisher) in X-VIVO15 media for 30 minutes at room temperature with gentle rotation, and then washed and resuspended in Ringer’s solution. Labeled cells were warmed up to 37 °C for 10–15 minutes to allow cleavage of AM ester and were kept in 37 °C water bath during recording. Live single cells were gated based on forward and side scatter parameters. Fluo-4 (Ca²⁺ bound) and FuraRed (Ca²⁺ unbound) fluorescence were recorded on Ex-488nm/ Filter-530/30 and Ex-488nm/ Filter-695/40 channel on the following sequence: 1) 3 minutes of baseline acquisition; 2) 15 minutes following the addition of 5 μg/ml of biotin-anti-CD3ε and biotin-anti-CD28 for mouse T cells, or 5 μg/ml of biotin-anti-hCD3 for Jurkat cells; 3) 15 minutes following the addition of 10 μg/ml streptavidin, 4) 5 minutes following addition of 1 μM Ionomycin (Sigma-Aldrich). Kinetics platform in FlowJo v10 was used to plot time versus median Fluo-4/FuraRed ratio and to calculate peak value and area under curve.

Single-cell ratiometric Ca²⁺ imaging

Single-cell Ca²⁺ flux assay for T cells was modified from a previous publication (52). Overnight rested T cells were loaded at 1 × 10⁶ cells per ml with 1 μM Fura 2-AM for 30 minutes at 25 °C, resuspended in loading medium, and attached to poly-L-lysine-coated coverslips for 15 minutes. For [Ca²⁺]_i measurements, cells were mounted in a RC-20 closed bath flow chamber (Warner Instrument Corp., Hamden, CT) and analyzed on an Olympus IX51 epifluorescence microscope with Slidebook (Intelligent Imaging Innovations, Inc.) imaging software. Fura-2 emission was detected at 510 nm with excitation at 340 and 380 nm, and the Fura-2 emission ratio (340/380) was acquired at every 5-second interval after subtraction of background. For each experiment, at least 100 individual cells were analyzed using OriginPro (Originlab).

Biotin affinity purification and sample preparation for proximity labeling

EL-4 lines stably transduced with pRetroX-3xHA-TurboID-Aster-A or pRetroX-3HA-TurboID (empty vector) were constructed as described above. Tight TRE promoter has a leakage expression in EL-4 cell line, which resulted in moderate expression of 3x-HA-TurboID-Aster-A and stable cell lines were therefore used without doxycycline induction.

TurboID or TurboID-Aster-A expressing EL-4 cells were cultured in 10% FCS complete culture media or starved with a cholesterol-depletion media containing 1% Lipoprotein-deficient serum, 5 μM Simvastatin, 50 μM Mevalonate for 18–24 h. Depleted cells were either loaded with 100 μΜ MβCD-Cholesterol or maintained in fresh cholesterol-depletion media for 1 h and 15 minutes. Cholesterol-rich samples are maintained in 10% FCS complete media for the same time. 50 μM Biotin was added to the culture media for the last 30 minutes to facilitate labeling. Isolation of biotinylated proteins were performed based on a modified protocol (41). Cells were collected and washed extensively with cold PBS and incubated with lysis buffer (RIPA buffer containing 1X Halt protease and phosphatase inhibitor and 1 mM PMSF) with gentle agitation for 30 minutes on ice. Lysate were cleared by centrifugation at 12,000 xg for 15 minutes and protein concentration was determined by BCA assay. Equal quantities (900 μg) of cleared lysates in lysis buffer were incubated with 75 μl Streptavidin magnetic beads (Thermo Fisher Pierce) overnight at 4 °C with gentle rotation. The beads were then washed at room temperature twice in lysis buffer for 5 minutes each, once with 1 mL 1 M KCl for 2 minutes, once with 1 mL 0.1 M Na₂CO₃ for ~10 seconds, once with 1 mL 2 M urea in 10 mM Tris-HCl (pH 8.0) for ~10 seconds, and again twice with 1 mL lysis buffer for 2 minutes.

In solution tryptic digestion for mass spectrometry

The samples were heated at 95 °C in elution buffer (12 mM sodium lauroyl sarcosine, 0.5% sodium deoxycholate, 50 mM triethylammonium bicarbonate (TEAB) containing tris(2-carboxyethyl)phosphine (10 mM) and chloroacetamide (40 mM) for 20 minutes. The samples were then diluted 5-fold with 50 mM TEAB and digested with trypsin (1 μg) for 16 h at 37 °C. A 1:1 (volume/volume) ratio of ethyl acetate plus 1% trifluoroacetic acid (TFA) was added to the samples and samples were vortexed for 5 minutes. Samples were centrifuged at 16000 xg for 5 minutes at room temperature and the supernatant was discarded. The samples were desalted using a modified version of Rappsilber’s protocol (91) in which the dried samples were reconstituted in acetonitrile/water/TFA (solvent A, 100 μL, 2/98/0.1, v/v/v) and then loaded onto a small portion of a C18-silica disk (3M, Maplewood, MN) placed in a 200 μL pipette tip. Prior to sample loading the C18 disk was prepared by sequential treatment with methanol (20 μL), acetonitrile/water/TFA (solvent B, 20 μL, 80/20/0.1, v/v/v) and finally with solvent A (20 μL). After loading the sample, the disc was washed with solvent A (20 μL, eluent discarded) and eluted with solvent B (40 μL). The collected eluent was dried in a centrifugal vacuum concentrator and reconstituted in 0.1% Formic Acid (10 μL) for LC-MS/MS analysis.

LC-MS/MS analysis

Aliquots of each sample (5 μL) were injected onto a reverse phase nanobore HPLC column (AcuTech Scientific, C18, 1.8μm particle size, 360 μm × 20 cm, 150 μm ID), equilibrated in solvent A (water/acetonitrile/FA, 98/2/0.1, v/v/v) and eluted (300 nL/minute) with an increasing concentration of solvent B (acetonitrile/water/FA, 98/2/0.1, v/v/v: min/% F; 0/0, 5/3, 18/7, 74/12, 144/24, 153/27, 162/40, 164/80, 174/80, 176/0, 180/0) using an EASY-nLC II (Thermo Fisher Scientific). The effluent from the column was directed to a nanospray ionization source connected to a hybrid quadrupole-Orbitrap mass spectrometer (Q Exactive Plus, Thermo Fisher Scientific) acquiring mass spectra in a data-dependent mode alternating between a full scan (m/z 350–1700, automated gain control (AGC) target 3 × 10⁶, 50 ms maximum injection time, FWHM resolution 70,000 at m/z 200) and up to 15 MS/MS scans (quadrupole isolation of charge states 2–7, isolation window 0.7 m/z) with previously optimized fragmentation conditions (normalized collision energy of 32, dynamic exclusion of 30 s, AGC target 1 × 10⁵, 100 ms maximum injection time, FWHM resolution 35,000 at m/z 200).

Protein identification

For experimental mass spectrometry data acquisition and analysis workflow, the raw proteomic data were searched against a Uniprot database containing the complete mouse proteome using SEQUEST-HT (Version 2.4, Thermo Scientific), which provided measurements of relative abundance of the identified peptides. Decoy database searching was used to generate high confidence tryptic peptides (FDR < 1%). Tryptic peptides containing amino acid sequences unique to individual proteins were used to identify and provide relative quantification between different proteins in each sample. We performed protein identification of all treatment conditions in both empty vector (TurboID only) and TurboID-Aster-A expressing cells. This approach enhanced confidence in identifying proximity associations. Potential non-specific associations were defined as proteins identified in substantial abundance in both control TurboID only and in experimental groups and were excluded from further analysis. The remaining protein abundance was compared between experimental conditions and ranked as shown in Fig. 5B and in Data S3.

Turbo-ID biotinylated protein IP and Protein co-immunoprecipitation (Co-IP)

Confirmation of MS results in cholesterol depletion/repletion conditions were performed as described above using 300–500 μg total protein. In TurboID labeling involving TCR activation, EL-4 cells are either cultured in 10% FCS complete media or pre-treated with DMSO or 50 μM BAPTA for 20 min at 37°C. Then cells were added to prewarmed culture plate precoated with 5 μg/ml anti-CD3/CD28 with or without 10 mM EGTA. The plate was centrifuged at 1,200 rpm for 1 minute to facilitate cell attachment and TCR activation for 1 h. For ionomycin treatment, 1 μM ionomycin was added to cells plated on non-coated culture plates for 1 h. For all experiments, 50 μM biotin was added to the culture media for the last 30 minute to facilitate labeling. Proteins were eluted by boiling beads in 30 μL 3X LDS loading buffer containing 2 mM free biotin and 20 mM DTT on a shaking thermo block (1,000 rpm) at 95 °C for 10 minutes. For input control, 30 μg cleared lysate prior to beads incubation was diluted in 1X LDS loading buffer with 20 mM DTT.

Stable polyclonal EL-4 lines expressing HA empty vector, HA-mAster-A (WT), HA-mAster-A (R186W) or co-expression with Myc-mSTIM1 were constructed as above and treated in cholesterol replete or deplete conditions as described in the TurboID biotinylated protein IP section. After 1 h treatment, cells were collected by centrifugation at 1,200 rpm, 5 minute at 4°C and washed extensively with cold PBS. Cell pellets were then incubated with IP lysis buffer (20 mM Tris-HCl, 2 mM EDTA, 135 mM NaCl, 10% (vol/vol) glycerol, 0.5% Igepal CA-630, 2X Halt protease inhibitor cocktail, 2 mM PMSF, pH 7.5) for 30 minutes at 4°C with gentle rotation. Lysates are further sonicated with 20% amplitude, 3 cycles of 5 seconds on, 10 seconds off and insoluble fraction was removed by centrifugation at 12,000 xg for 15 minutes. Protein concentration was determined by BCA assay and 1 mg total protein was precleared with protein A/G PLUS beads (Santa Cruz Biotechnology) for 1 h at 4°C. Cleared lysates were incubated with 10 μg mouse Anti-HA (16B2) overnight at 4°C with gentle rotation. Antibody bound proteins were precipitated with protein A/G PLUS beads (Santa Cruz Biotechnology) for 3–4 h at 4°C while rotating. A/G beads were washed five times, 10 minutes each with 1x co-IP lysis buffer and proteins were eluted by incubating A/G beads in 20–40 μl 3X SDS sample buffer at 95°C for 10 minutes with 1,000 rpm shaking. Antibodies against HA Tag, STIM1, mouse Aster-A, or Myc Tag were used for immunoblotting.

Sample preparation and library construction for bulk mRNA sequencing

Small intestine lamina propria mononuclear cells were isolated using the method described above. 1 μM flavopiridol (RNA polymerase inhibitor) was included in all digestion steps to halt de novo transcription of stress-related genes as a result of tissue dissociation. Single cell suspension was FACS-gated on Single/Live cell/CD45⁺/CD90⁺/CD3⁺ and sorted into CD4⁺ and CD8⁺ populations directly into 700 μl Trizol LS reagent. Sorting strategy is shown in fig. S21A. Total RNA was extracted using TRIzol and the miRNeasy Mini Kit with on-column DNase I digestion (Qiagen).

RNA library preparations and sequencing reactions were conducted at GENEWIZ, LLC. RNA samples were quantified using Qubit 2.0 Fluorometer (Life Technologies) and RNA integrity was checked using Agilent TapeStation 4200 (Agilent Technologies). RNA sequencing libraries were prepared using the NEBNext Ultra II RNA Library Prep Kit for Illumina using manufacturer’s instructions. Briefly, poly-A mRNAs were initially enriched with Oligo dT beads. Enriched mRNAs were fragmented for 15 minutes at 94 °C. First strand and second strand cDNA were subsequently synthesized. cDNA fragments were end repaired and adenylated at 3’ends, and universal adapters were ligated to cDNA fragments, followed by index addition and library enrichment by PCR with limited cycles. The sequencing library was validated on the Agilent TapeStation (Agilent Technologies) and quantified by using Qubit 4.0 Fluorometer (Invitrogen) as well as by quantitative PCR (KAPA Biosystems).

Bulk mRNA sequencing and analysis

The sequencing libraries were clustered on flowcells. After clustering, the flowcells were loaded on the Illumina HiSeq instrument (4000 or equivalent) according to manufacturer’s instructions. The samples were sequenced using a 2×150bp Paired End configuration. Image analysis and base calling were conducted by the HiSeq Control Software (HCS). Raw sequence data (.bcl files) generated from Illumina HiSeq was converted into fastq files and de-multiplexed using Illumina’s bcl2fastq 2.17 software. One mismatch was allowed for index sequence identification. Adapter and quality trimming of raw FASTQ files was performed using Trimmomatic. FastQC was used to perform quality control of the FASTQ files before and after trimming. Trimmed FASTQ files were aligned to GRCm38/mm10 using STAR package. HTSeq-count was used to extract gene counts and differential gene expression analysis was performed using DESeq2 package (doi:10.1186/s13059-014-0550-8) in R 4.1.1. Normalized gene expression of the individual samples is shown in Data S1 and differentially expressed gene list is shown in Data S2.

Single-cell RNA sequencing

Small intestine lamina propria mononuclear cells were isolated using the method described in the bulk mRNA sequencing section. Single Live EPCAM⁻ Ter119⁻ CD31⁻ CD45⁺ cells were sorted from 10- to 12-week-old female littermates on a BD Aria FACS sorter. Sorting strategy is shown in fig. S23. 100,000 cells were sorted from each mouse (n = 3 /genotype) and pooled together before loading to a Chromium Single Cell Controller (10X Genomics). Libraries were prepared using a 10X Genomics 3’GEX Kits following the manufacturer’s protocols. Libraries were then sequenced on an Illumina NovaSeq S2 using 2 × 50 bp paired end sequencing with a targeted depth of 50,000 read pairs per cell.

Raw reads were aligned to mouse mm10 genome using CellRanger (v7.2.0). Counts from ambient RNA was removed using CellBender [(v0.3.0); Github: broadinstitute/CellBender] and doublets were removed using scDblFinder [(v1.18.0); DOI: 10.18129/B9.bioc.scDblFinder]. Preprocessing was performed using Seurat [(v5.0.1); Github: satijalab/seurat] in R 4.3.1. Specifically, we filtered for cells with 500<nUMI <30000, 200<nGenes <10000 and percent.MT < 5%. Normalization was done using SCTransform with percent.MT regressed out. Initial data integration was done using the SelectIntegrationFeatures(), PrepSCTIntegration(), FindIntegrationAnchors () and IntegrateData() functions with RPCA method and 3000 variable features. Dimensional reduction was done using top 50 principal components (PCs) and subsequent Harmony integration. Visualization was performed using RunUMAP() function. Cell clustering was performed using the FindNeighbors() function (50 harmony embeddings, k=35) and FindClusters() function with igraph and Leiden method (resolution=0.6). To isolate lymphoid lineage for further analysis, we subset clusters expressing Thy1 (CD90) from the original Seurat object. The same normalization and integration procedure was applied to lymphoid lineage clusters. To identify stable clusters, we performed bootstrapping of 80% of the cells 20 times, and tested stability of clusters from a wide range of parameters, including number of PCs used, k parameters in FindNeighbors() and resolution in FindClusters(). We decided to use 20 top PCs, k = 30 and resolution = 0.8 to maximizing cluster diversity while maintaining cluster stability. Cluster markers were identified using FindMarkers() function and manual annotation was performed using established cell markers. Cluster were manually annotated based on matching cluster markers (fig. S17A, list of markers is shown in Data S4) with cell type ULI RNA-seq data in the Immgen database and previously published small intestine mononuclear cell single-cell RNA sequencing results (GSE124880).

Access and analysis of previously published gene expression and sequencing datasets

Tissue panel gene expression (fig. S1E) was analyzed in BioGPS repository from aggregated datasets GSE1133, GSE10246, GSE15998. Immune cell expression panel (fig. S1G) was obtained from ImmGen ULI RNA-seq data (GSE109125 and GSE127267). Human single cell meta-analysis (fig. S9A-B) was performed using Swiss Portal for Immune Cell Analysis and scIBD database (92, 93). ChIP-seq data (fig. S9F) was analyzed in Cistrome database using datasets GSM978770, GSM978769, GSM978772, GSM1126691, GSM2706520, GSM2387500. Sequencing tracks were visualized with Integrated Genome Viewer (2.16.2).

Statistical Analysis

Data in all figures are representative of or pooled from at least two independent experiments or as indicated in the figure legends. Data are shown as Mean ± SEM. For experiments involving wildtype mice or mice of the same genotype, groups were assigned so that control and treatment groups have similar average body weight. In Cre-flox system or WT-KO genotypes, balanced numbers of littermate mice were used where possible and randomly assigned to experimental groups. Samples are processed in an order that alternates between WT/KO or control/treatment subjects. Formal blinding was not applied when performing the experiments or analyzing the data. Power calculations were not applied to predetermine sample size. Sample size was based on prior laboratory practice and literature reports on similar assays. Groups were compared using Prism software (v10.3.0). For statistics tests requiring normal distribution, datasets were first tested for normality and log normality using Kolmogorov-Smirnov test. If data distribution was not normal, it was first log-transformed, tested again using Kolmogorov-Smirnov for normality, before proceeding with the corresponding statistical tests. The statistical methods and correction used for each experiment are reported in the corresponding figure captions.

Data and materials availability:

All data needed to evaluate the conclusions in the paper are available in the manuscript or in supplementary materials. Sequencing data generated in this study have been deposited to Gene Expression Omnibus (GEO) under accession numbers GSE277067 (bulk CD4 RNA-seq), GSE277138 (scRNA-seq). Proteomics mass spectrometry data generated in this study have been deposited to Dryad database (94). All unique biological materials used are readily available from the authors or from standard commercial sources. Aster-A^F/F mice are available from Dr. Tontonoz under a material agreement with University of California, Los Angeles.