Caveolin-1 deficiency exacerbates ESS pathology.
Although previous studies suggested ameliorated EAE pathology in Cav-1-/- mice7, unexpectedly, Cav-1 deficiency significantly exacerbated ESS development in mice upon disease induction, as revealed by accelerated salivary hypofunction and higher titer of autoantibodies compared with wildtype (WT) ESS mice (Fig. 1A-B). This was associated with enhanced IgG deposition on the salivary epithelium (Fig. 1C). In long-term observation, at acute-chronic stage, massive lymphocytic foci, together with much severe tissue damage were observed in Cav-1-/- ESS mice, whereas age-matched WT mice exhibited only mild lymphocytic infiltration along ESS progression. Confocal imaging analyses also indicated the presence of Bcl-6+CD4+ Tfh cells and GC-like B cells within the infiltrating aggregates, suggesting ectopic lymphoid structure formation in the inflamed SG tissues from Cav-1-/- ESS mice (Fig. 1D). Quantitatively, Cav-1-/- ESS mice exhibited higher histological scores, larger infiltrating area (Fig. 1E), and increased apoptotic epithelial cell counts when compared with WT mice (Fig. 1F). Thus, Cav-1 deficiency promoted ESS development in mice.
Given the varied abundance of Cav-1 proteins in lymphoid and non-lymphoid organs 23, it is not clear whether Cav-1 expression in hematopoietic origin would be responsible for the ESS pathogenesis. Thus, we generated chimera mice with Cav-1 deficiency in the immune cells by bone marrow transfer, while the genotype in the lymphoid organ was validated upon mouse sacrifice (Fig. 1G). Consistently, WT recipients reconstructed with Cav-1-/- bone marrows showed higher autoantibody levels (Fig. 1H) and histopathological changes (Fig. 1I-J), which was comparable with those mice of global Cav-1 deletion. Together, these results suggest the importance of Cav-1 in regulating immune responses during ESS development.
Caveolin-1 constrains ICOS expression and follicular migration of Tfh cells.
Previous studies, including our findings, identified a central role of Th17 cells in both EAE and ESS pathogenesis 7,13,24. However, phenotypic analyses showed that Cav-1 deficiency did not increase Th17 cell frequencies and counts upon ESS development (Fig. 2A). This was consistent with previous findings in Cav-1-/- EAE mice7,25. Instead, we detected significantly increased Tfh cells in Cav-1-/- ESS mice, which was associated with their follicular displacement in the germinal centers (Fig. 2A, Fig. S1A). This may explain the expanded GC areas and GC B cells (Fig. S1A-B). Thus, we sought to determine Cav-1 expression levels in CD4 + T cells during ESS development. Upon ESS induction, we observed a transient increase of Cav-1 in CD4 + T cells at disease onset, followed by persistent downregulation during ESS progression (Fig. 2B). This was negatively correlated with increased Tfh cell responses 19. Since Cav-1-/- B cells could respond to T cell-dependent antigens 20, we next investigated whether Cav-1 deficiency in CD4 + T cells would be sufficient to mount humoral dysregulation. WT or Cav-1-/- CD4 + T cells were co-transferred with WT B cells into SG-antigen immunized immunodeficient NOD-scid IL2Rgnull (NSG) mice. Notably, Cav-1 deficiency in CD4 + T cells markedly promoted autoreactive B cell response (Fig. 2C-D), resulting in elevated autoantibodies production and IgG deposition on the salivary epithelium (Fig. 2E, Fig. S1C). These data suggest that Cav-1 deficiency in CD4 + T cells might promote Tfh cell response in vivo. To validate this finding, WT or Cav-1-/- CD4 + T cells were purified and cultured for Tfh polarization. Indeed, Cav-1-/- CD4 + T cells exhibited higher capacity of Tfh cell differentiation (Fig. 2F). A previous structural biology study implicates a central role of scaffolding domain of Cav-1 (CSD) as a docking site in signaling transduction 26. Cavtratin, a cell-permeable peptide of CSD, binds to caveolin-binding motifs 27. Herein, it dose-dependently restrained Cav-1-/- CD4 + T cell response to Tfh polarization (Fig. 2F). Collectively, these data indicate the role of endogenous Cav-1 in regulating Tfh cell differentiation. We next determined the featured molecules in Cav-1-/- Tfh cells, in which PD-1 and ICOS, but not CXCR5 expressions were found markedly increased (Fig. S1D). However, we did not observe the elevation of CD40 ligand and representative cytokine productions, in particular IL-17 and IL-21 in Cav-1-/- CD4 + T cells under the polarization conditions (Fig. S1E). Moreover, Cav-1 deficiency-mediated Tfh cell response was also found stable in long term, as revealed by flow cytometric analysis of BrdU-incorporated WT or Cav-1-/- donor CD4 + T cells in the recipient ESS mice over 10 weeks post adoptive transfer (Fig. S1F).
ICOS-ICOSL plays an indispensable role of in Tfh cell motility 28. Notably, flow cytometric analysis showed comparable ICOSL levels between WT and Cav-1-/- B cells from ESS mice (Fig. S1G). Thus, we hypothesized that increased ICOS expression in Cav-1-/- CD4 + T cells might largely strengthen their follicular migration. Consistent with the previous finding 28, anti-ICOS activating antibody rapidly induced WT Tfh cell polarization, as revealed by pseudopod protrusion and persistent movement (Fig. 2G-H). Quantitatively, Cav-1-/- Tfh cells showed higher shape index with augmented pseudopod extension (Fig. 2G) upon ICOS ligation, and thus resulted in increased centroid speed (Fig. 2I) and cell displacement in random directions (Fig. 2H, Fig. S1H). This was validated in ICOSL-mediated transwell assay in response to CXCL13 (Fig. S1I). Consistently, ICOS expression was significantly increased in CD4 + T cells obtained from draining cervical lymph nodes of Cav-1-/- ESS mice, accompanied with enhanced ICOS + CD4 + T cells present in the lymphocytic foci of SG (Fig. S1J-K). We further testified Cav-1-/- CD4 + T cell motility in vivo, in particular those at T-B borders, by two-photon intravital imaging analysis. In this adoptive transfer model, CD4 + T cell zone could be clearly distinguished in the spleen at 72h post transfer (Fig. 2J). Consistent with the findings in vitro, Cav-1-/-CD4 + T cells exhibited more polarized status, as reflected by less sphericity in morphology (Fig. 2K). By analyzing the cell tracks, we observed markedly increased motility of Cav-1 deficient CD4 + T cells in situ (Fig. 2K), as reflected by enhanced cell displacement (Fig. S1L) and velocity (Fig. 2M) by quantification, in particular those at the T-B borders (Supplementary Video 1–2). Since the follicular homing of CD4 + T cells, as the consequence of Tfh cell motility, was defined as the fundamental feature to mount B-cell response in vivo, we next evaluated their follicular homing capacity in the draining cervical lymph nodes under the identical conditions. CFSE-labelled WT and Qtracker-labeled Cav-1-/- naïve CD4 + T cells were co-transferred into WT ESS mice with active disease, while the draining cervical lymph nodes were analyzed after 72h. As expected, Cav-1-/- CD4 + T cells exhibited overt follicular displacement when compared with those WT counterparts (Fig. 2N). Together, these results demonstrated that Cav-1 critically restrained Tfh cell migratory capacity toward B cell follicles. To further assess the B-cell help functions other than T cell motility, we sorting-purified WT or Cav-1-/- Tfh cells from ESS mice, and co-cultured with cognate WT B cells for 72h. Interestingly, Cav-1 deficiency did not obviously affect the effector molecules of T-B interactions, as revealed by comparable plasmacytic differentiation and autoantibody productions (Fig. 2O). This was in line with the findings of CD40 ligand and cytokines above mentioned. Together, our data suggest that increased ICOS expression in Cav-1-/- CD4 + T cells were responsible for enhanced humoral autoimmunity and ESS pathology. To validate this notion, we performed anti-ICOS blocking antibodies treatment in Cav-1-/- mice upon ESS induction. The efficacy was validated by reduced intrafollicular ICOS + CD4 + T cell counts, which was associated with similar Tfh cell numbers and follicular homing coefficient to those in WT ESS mice (Fig. S2A-C). Consequently, the reduced Tfh cell response upon ICOS blockage led to decreased plasma cell numbers and anti-SSA IgG levels in Cav-1-/- ESS mice (Fig. S2D-E). Therefore, these data indicate a critical role of Cav-1 in ICOS expression and Tfh cell response.
Impaired PPARα expression in CD4 + T cells contributes to caveolin-1-mediated Tfh cell response.
Cav-1 is mainly distributed at the plasma membrane and cytoplasm, but not in the nucleus in endothelial cells and B cells 21,26, Similarly, this was also observed in CD4 + T cells (Fig. S3A-B). Thus, we reasonably addressed the question that enhanced Icos transcription in CD4 + T cells could be indirectly regulated by Cav-1. We first performed RNA-seq analysis of WT and Cav-1-/- Tfh cells for transcriptome comparison. Notably, peroxisome proliferator-activated receptors (PPARs) signaling pathway was significantly affected by the absence of Cav-1 (Fig. 3A). PPARs are a family of transcription factors including PPARα, PPARβ/δ and PPARγ 29. Real time-PCR analysis revealed that Cav-1 deficiency mainly affected the transcription of PPARα, while PPARγ and PPARδ were comparable with WT Tfh cells (Fig. 3B). Indeed, significantly reduced protein levels of PPARα were also found in Cav-1-/- CD4 + T cells (Fig. 3C). This was similar to the recent findings in hepatocytes 30. To investigate whether impaired PPARα expression contributes to Tfh cell response, we first performed genetic ablation of PPARα in CD4 + T cells by CRISPR/Cas9. Interestingly, PPARα-/- CD4 + T cells phenocopied Cav-1 deficiency and showed significantly augmented Tfh cell differentiation, regardless of cavtratin treatment (Fig. 3D), suggesting that PPARα served as the downstream of Cav-1 in CD4 + T cells. This could be also achieved by the treatment of GW6471, a selective PPARα antagonist, which markedly promoted Tfh cell differentiation in dose dependent manner (Fig. 3E). Functional studies using transwell assay further validated enhanced ICOS-mediated T cell motility upon PPARα antagonism (Fig. 3F). Given an intense search for PPARs ligands in the past decades, selective agonists were reported to activate differential PPAR members 31. Thus, based on the binding affinity, 8-hydroxyeicosapentaenoic (8-HEPE), 15-deoxy-D12,14-prostaglandin J2 (15-Deoxy) and GW074232 were used to differentially activate PPARα, PPARβ/δ and PPARγ respectively. In contrast to PPARα deficiency, PPARα agonist effectively suppressed Tfh cell differentiation, while no difference was observed in the presence of 15-Deoxy and GW0742 (Fig. 3G). To avoid the possible off-target effect, we adopted fenofibrate, the pharmaceutical agonist of both human and murine PPARα 33, which dose dependently suppressed both WT and Cav-1-/- Tfh differentiation in culture (Fig. 3H). Thus, these data suggest that PPARα would be responsible for Cav-1 downstream signal and serves as negative regulator of Tfh cell response. In the context of ESS development, similar to Cav-1 expression in CD4 + T cells, we also observed a transient increase of PPARα, but progressively decreased during disease development (Fig. 3I-J), in particular at disease chronic stages (Fig. 3K). However, PPARγ was not correlated with Cav-1 and PPARα expressions in CD4 + T cells (Fig. 3I, Fig. S3C-D). Consequently, treatment with PPARα antagonist GW6471 accelerated ESS development and Tfh cell responses (Fig. 3L-M). Together, these results demonstrated the functional importance of Cav-1/PPARα axis in restraining Tfh cell response.
PPARα represses ICOS transcription in Tfh cells.
A previous study suggested downregulation of lipid metabolic processes as a major consequence of Cav-1 deficiency, among which PPARα was responsible 30. Indeed, we observed significantly lower levels of lipid droplets in Cav-1 deficient CD4 + T cells (Fig. 4A), in particular in the effector population (Fig. S4A). Thus, we sought to investigate whether lipid metabolism would be involved in Cav-1/PPARα axis-mediated ICOS expression in Tfh cells. We first monitored the transcriptional regulation of Icos during Tfh polarization. Notably, the mRNA levels of ICOS were significantly increased as early as 16h in Cav-1-/- CD4 + T cells, while protein levels at 48h upon Tfh differentiation (Fig. 4B). Since lipid metabolism-mediated energy generation consisted of several rate-limiting steps 34, in which fatty acid β-oxidation (FAO) produced acetyl-CoA and entered mitochondrial tricarboxylic acid (TCA) cycle, we next determined whether Icos transcription was altered during this process. We first measured the mitochondrial respiration from fatty acids by using palmitic acid (16:0, PA) as sole extracellular substrate 35. After 48h Tfh polarization, as expected, Cav-1 deficiency significantly decreased the basal and maximal respiratory capacity (Fig. 4C, Fig. S4B). Accordingly, ACADM protein levels, the representative enzyme of FAO initiation, were found significantly decreased in the absence of Cav-1. Moreover, succinyl-CoA synthetase (SDH), and aconitase (ACO2), the key regulatory enzymes of TCA cycle, were also significantly decreased at 48h in Cav-1-/- CD4 + T cells (Fig. 4D). This was associated with decreased cellular fatty acid content, as reflected by free fatty acid assay (Fig. 4E). However, this was not seen at the early stages upon Tfh polarization. FAO and cellular fatty acid content were minimal or undetectable in both WT and Cav-/- CD4 + T cells at 16h in culture, while there was no obvious difference of enzyme expression levels. Together, these results validated that Cav-1 deficiency indeed impaired FAO process. However, the rapid Icos transcription in Cav-1-/- CD4 + T cells, prior to energy status transition, suggests its transcriptional regulation in relatively direct manner.
To validate this notion, we next looked into the target genes of PPARα in CD4 + T cells, which were previously identified for fatty acid transportation and β-oxidation through carnitine palmitoyltransferase (CPT) system 34. Cpt1a, a highly conserved PPARα target gene and metabolic regulator, catalyzes the long chain fatty acids from acyl-CoA to carnitine for translocation across the mitochondrial membranes 36. Indeed, PPARα agonist significantly increased Cpt1a expression 37. Thus, we first determined the expression levels of Cpt1a during Tfh differentiation. Consistent with the findings of the lipid metabolism above, Cpt1a expression was elevated in WT CD4 + T cells at 48h under Tfh differentiation, which was found much lower in the Cav-1-/- counterparts. However, there was no obvious difference of Cpt1a level at early phase upon Tfh polarization (Fig. 4F). We next performed Cpt1a overexpression in CD4 + T cells to restore, at least in part, the PPARα deficiency-mediated lipid metabolism. Surprisingly, overexpression of Cpt1a did not restrain, but rather promoted Tfh cell differentiation in PPARα-/- CD4 + T cells (Fig. 4G). Similar findings were also observed following irreversible Cpt1a antagonism by etomoxir, as the etomoxir treatment did not phenocopy PPARα or Cav-1 deficiency, but rather suppressed Tfh cell development (Fig. 4H), which may attribute to globally constrained energy generation and requirement. This notion was further supported by fatty acid-free culture conditions, in which Cav-1-/- CD4 + T cell retained higher capacity of ICOS expression and Tfh differentiation (Fig. 4I, Fig. S4C). In this context, nutrient addition by BSA further gave rise to both WT and Cav-1-/- Tfh cell differentiation. Thus, these results demonstrate that impaired lipid metabolism in Cav-1-/- CD4 + T cells would not affect the increased Icos transcription under Tfh polarized conditions.
Additionally, we also determined the glucose oxidation of WT and Cav-1-/- CD4 + T cells under Tfh polarization conditions. Using glucose as a substrate, we detected minimal oxygen consumption rates (OCR) at 16h, but markedly increased at 48h in both WT and Cav-1-/- CD4 + T (Fig. S4D). Among the glucose oxidation cascade, the phospho-PFK2 to PFK2 ratio represents the glycolytic rate 38, in which a higher ratio leads to increased gluconeogenesis 39. Consistent with the OCR findings, using glucose as the sole extracellular source of carbon, the ratio of p-PFK2:PFK2 was found comparable in WT and Cav-1-/- CD4 + T cells during Tfh differentiation (Fig. S4E), as well as the expressions of SDH and ACO2. These data suggest that glucose oxidation might not be involved in Cav-1-mediated Tfh response.
PPARα is also recognized to limit inflammatory responses via transcriptional repression29. This is achieved by a conserved mechanism that PPARs could form heterodimer with 9-cis-retinoic acid receptor (RXR), which binds to peroxisome proliferator response element (PPRE) at the promoter region of target genes 40. Indeed, in the transcriptomic screening analysis, both PPAR and RXR binding activities were found significantly reduced in Cav-1-/- CD4 + T cells (Fig. 4J). Thus, we next assessed the binding capacity of PPARα to Icos promoter. ChIP-PCR analysis indicated that PPARα rapidly bound to Icos promoter region 12h upon Tfh polarization, an effect could be largely abrogated by selective antagonist (Fig. 4K). This was further supported by interfering the intranuclear translocation. Time series imaging showed that intranuclear translocation of PPARα initiated at 30 min and persisted for hours (Fig. 4L). Early studies have suggested a role of COX-1 in the nuclear translocation of PPARs 41,42. We also validated this finding in CD4 + T cells, as SC-560, a selective COX-1 inhibitor effectively retained PPARα in the cytoplasm (Fig. 4M). Importantly, PPARα agonist-mediated Icos trans-repression was abolished by SC-560 (Fig. 4N). Conversely, selective RXR antagonist HX531 augmented, while RXR agonist LG100754 43 inhibited ICOS and Tfh cell differentiation in vitro (Fig. 4O, Fig. S4F). Thus, these data suggest that PPARα could serve as a repressor of Icos transcription.
Cav-1/PPARα axis critically regulates human Tfh cells.
We next sought to investigate whether Cav-1/PPARα axis also operates in human subjects. Purified CD4 + T cells from healthy donors were transduced with GFP-incorporated plasmids for Cav-1 deletion. Similar to the murine system, Cav-1-/- hCD4 + T cells also exhibited strong capacity towards Tfh differentiation, as well as increased ICOS expression (Fig. 5A-B). In addition, PPARα expression levels were also significantly decreased upon Cav-1 deficiency (Fig. 5C). Thus, we evaluated the effects of fenofibrate in suppressing Tfh cell responses in dose dependent manner, which yielded an IC50 of 7.024 µM (Fig. 5D). Upon RNA-seq analysis, fenofibrate significantly reduced Icos transcript copies at 16h post stimulation (Fig. 5E), which strongly supported our findings in mice. Consistently, human PPARα also rapidly bound to Icos promoter region under Tfh polarization, revealed by ChIP-PCR analysis, which was largely abolished upon PPARα antagonism by GW6471 (Fig. 5F). Consequently, blockage of intranuclear entrance of PPARα by SC-560 prevented the transcriptional repression of Icos (Fig. 5G). Functional assay further verified the ICOS-mediated migration could be inhibited by fenofibrate (Fig. 5H). To validate this phenotype under disease conditions, we analyzed the circulating Tfh (cTfh) cells from pSS patients. By measuring the protein expression of each patient, we first detected strongly positive correlation between PPARα and Cav-1 in CD4 + T cells (Fig. 5I). Interestingly, patients with higher frequencies of cTfh cells exhibited relatively lower levels of Cav-1 and PPARα in CD4 + T cells, rendering a negative correlation between Cav-1 with ICOS expression (Fig. 5J-K). Thus, these results demonstrate Cav-1/PPARα axis as a negative regulator in human Tfh cells, while targeting PPARα may be a promising approach in treating Tfh cell dysregulation.
Pharmaceutical activation of PPARα ameliorates ESS development
Fenofibrate has a similar affinity for both murine and human PPARα with high efficacies 33. We found that fenofibrate had a half maximal inhibitory concentration of 8.39 µM in suppressing murine Tfh cell differentiation (Fig. S5). This prompted us to explore the therapeutic potential of this pharmaceutical PPARα agonist on mice with established ESS. Previous studies reported that oral administration of 100 mg/kg body weight fenofibrate was sufficient to induce global PPARα activation and downstream signaling pathways in vivo44,45. Thus, we first treated ESS mice with fenofibrate at disease onset (Fig. 6A), as diagnosed by reduced saliva secretion and elevated autoantibodies. Fenofibrate effectively ameliorated salivary hypofunction and decreased serum levels of anti-SSA IgG, although anti-SSA IgM levels were not altered (Fig. 6B-C). We next sought to determine the therapeutic potential of fenofibrate treatment at disease chronic stages. As revealed by mild lymphocytic infiltrations in the SG, ESS mice immunized for 20wk were treated with fenofibrate for 10wk. Notably, histopathological findings showed significantly reduced tissue damages and inflammation in SG of the fenofibrate-treated ESS mice, as evidenced by diminished lymphocytic infiltration and few apoptotic epithelial cells (Fig. 6D). Phenotypic analyses further showed that fenofibrate treatment significantly suppressed Tfh cell responses, in particular GC-Tfh cell counts in the lymphoid tissues (Fig. 6E-G). This was consequently associated with restrained GC area, GC B cell and plasma cell counts (Fig. 6G-H). A recent study reported a humanized mouse model by transplanting PBMCs in NSG mice 46. Thus, we adopted this method by dividing PBMCs from each pSS patient into two groups for paired analysis, followed by PBS or fenofibrate treatments. Although NSG mice xenografted with PBMCs from pSS patients exhibited higher frequencies of Tfh cells than those from healthy donors, fenofibrate administration effectively suppressed human Tfh cells in matched control mice (Fig. 6I). Together, these results provide strong evidence to support the notion that targeting PPARα could be a promising therapeutic approach in pSS and related autoimmune disorders with dysregulated Tfh cell responses.