Cyclical palmitoylation regulates TLR9 signalling and systemic autoimmunity in mice

Abstract

Toll-like receptor 9 (TLR9) recognizes self-DNA and plays intricate roles in systemic lupus erythematosus (SLE). However, the molecular mechanism regulating the endosomal TLR9 response is incompletely understood. Here, we report that palmitoyl-protein thioesterase 1 (PPT1) regulates systemic autoimmunity by removing S-palmitoylation from TLR9 in lysosomes. PPT1 promotes the secretion of IFNα by plasmacytoid dendritic cells (pDCs) and TNF by macrophages. Genetic deficiency in or chemical inhibition of PPT1 reduces anti-nuclear antibody levels and attenuates nephritis in B6.Sle1yaa mice. In healthy volunteers and patients with SLE, the PPT1 inhibitor, HDSF, reduces IFNα production ex vivo. Mechanistically, biochemical and mass spectrometry analyses demonstrated that TLR9 is S-palmitoylated at C258 and C265. Moreover, the protein acyltransferase, DHHC3, palmitoylates TLR9 in the Golgi, and regulates TLR9 trafficking to endosomes. Subsequent depalmitoylation by PPT1 facilitates the release of TLR9 from UNC93B1. Our results reveal a posttranslational modification cycle that controls TLR9 response and autoimmunity.

Fig. 1. PPT1 deficiency protects SLE mice from autoantibodies and nephritis.

C57BL/6 J (B6), Ppt1^+/+ B6.Sle1yaa and Ppt1^-/- B6.Sle1yaa mice were sacrificed at 16 weeks (n = 5 mice for B6; n = 15 mice for Ppt1^+/+ B6.Sle1yaa and Ppt1^-/- B6.Sle1yaa for the entire figure). A Spleen image (left), weight (center) and cell numbers (right) are shown. B Serum anti-DNA antibodies were measured by ELISAs. C Serum anti-RNP/Sm antibodies were measured by ELISAs at 16 weeks. D Serum total IgG were measured by ELISAs. E Glomerulus size was calculated from H&E-stained kidney sections (left). Each dot represents an individual glomerulus (n = 40 in B6; n = 44 in Ppt1^+/+ B6.Sle1yaa and Ppt1^-/- B6.Sle1yaa). F Representative images from kidney immunofluorescence sections are shown (left). IgG (top) and C3 (bottom) staining in individual glomerulus was quantified (n = 40 glomeruli in B6; n = 42 glomeruli in Ppt1^+/+ B6.Sle1yaa and Ppt1^-/- B6.Sle1yaa). G Splenic CD11b⁺ cell distribution is indicated by FACS plots (left), percentages (middle), and cell numbers (right). H Splenic CD4⁺ T cell activation is indicated by FACS plots (left), percentages (middle), and cell numbers (right). I Splenic CD8⁺ T cell activation is indicated by FACS plots (left), percentages (middle), and cell numbers (right). J Splenic plasma cell distribution is indicated by FACS plots (left), percentages (middle), and cell numbers (right). All data are pooled from three independent experiments (mean ± SEM.; P values were calculated by two-way Student’s t test).

Fig. 2. PPT1 inhibitor HDSF suppresses IFNα in SLE patients and SLE pathogenesis in mice.

A PBMCs of SLE patients were treated with DMSO or HDSF overnight. After CpG A stimulation, IFNα levels was evaluated by ELISAs (n = 17 individuals for DMSO/HDSF). B 8-weeks-old B6.Sle1yaa mice (n = 7 mice per group for the rest of figure) were treated with DMSO or HDSF for 8 weeks before sacrifice at 16 weeks. Spleen image (left), weight (center) and cell numbers (right) are shown. C Serum anti-DNA antibodies were measured by ELISAs at indicated weeks of treatment. D Serum anti-RNP/Sm antibodies were measured by ELISAs at the end of treatment. E Serum total IgG were measured by ELISAs at indicated weeks of treatment. F Glomerulus size at 16 weeks is calculated from H&E-stained kidney sections (left). Each dot represents an individual glomerulus (n = 40 glomeruli per group). G Representative images from kidney immunofluorescence sections are shown (left). IgG (middle) and C3 (right) staining in individual glomerulus was quantified (n = 50 glomerulus per group). H Splenic CD11b⁺ cell distribution is indicated by FACS plots (left), percentages (middle), and cell numbers (right). I Splenic CD4⁺ T cell activation is indicated by FACS plots (left), percentages (middle), and cell numbers (right). J Splenic CD8⁺ T cell activation is indicated by FACS plots (left), percentages (middle), and cell numbers (right). K Splenic plasma cell distribution is indicated by FACS plots (left), percentages (middle), and cell numbers (right). All data are representative of three independent experiments (mean ± SEM.; P values were calculated by two-way Student’s t test), except for those presented in (A), which were pooled from five independent experiments (P values were calculated by two-tailed Wilcoxon matched-pair signed rank test).

Fig. 3. PPT1 controls the cytokine response by pDCs, B cells and macrophages.

A pDCs enriched from PBMCs of healthy volunteers were treated with DMSO or HDSF overnight. After CpG A stimulation, IFNα levels was evaluated by ELISAs (n = 9 individuals). B Sorted Flt3L-pDCs were treated with CpG A at the indicated timepoints. Ppt1 (left) and Ifna (right) qPCR results are shown (n = 3 mice per group). C Sorted Ppt1^+/+ or Ppt1^-/- BM pDCs were treated with TLR9/TLR7 agonizts (left). CpG A and R848 ELISA results are shown (n = 4 mice per group). D Ppt1^+/+ or Ppt1^-/- mice were infected with HSV (left). Serum IFNα ELISAs for HSV (right) are shown (n = 5 mice for HSV). E Ppt1^+/+ or Ppt1^-/- indicated immune cells were treated with CpG B treatment. TNF ELISA results for CD19⁺ B cells, Per.MΦ and BMDMs are shown (n = 4 mice per group for CD19⁺ B cells and Per.MΦ, n = 3 mice per group for BMDMs). F WT indicated immune cells were treated with DMSO or HDSF, followed by a CpG A (pDCs) or CpG B (B cells and macrophages) pulse. TNF ELISA results for sorted BM pDCs, CD19⁺ B cells, Per.MΦ and BMDMs are shown (n = 4 mice per group for sorted BM pDCs, CD19⁺ B cells and Per.MΦ, n = 3 mice per group for BMDMs). All data are representative of three (mean ± SEM.; P values were calculated by two-way Student’s t test), except for those presented in (A), which were pooled from three independent experiments (P values were calculated by two-tailed Wilcoxon matched-pair signed rank test).

Fig. 4. TLR9 and TLR7 are S-palmitoylated.

A Click chemistry reactions with azido palmitic acid (PA) were performed on RAW264.7 cells transduced with mTLR9-FLAG. A representative blot (left) and mTLR9 relative palmitoylation levels (right, quantified as the output-to-input ratio of signals from all TLR9 bands, n = 3 replicates). B ABE assays were performed in RAW264.7 cells transduced with mTLR9-FLAG (n = 3 replicates). C ABE sample preparation for mass spectrometry. D Selected MS spectra for mTLR9 C258 and C265. E ABE assays were performed on Tlr9^-/- RAW264.7 cells transduced with TLR9 C258A + C265A (Mut2). mTLR9 S-palmitoylation was quantified as the mTLR9 output to calnexin output ratio (n = 4 replicates). F Click chemistry reactions with azido palmitic acid (PA) were performed on RAW264.7 cells transduced with mTLR7-HA. A representative blot (left) and mTLR7 relative palmitoylation levels (right, quantified as the output-to-input ratio of signals from all TLR7 bands, n = 3 replicates). G ABE assays were performed in RAW264.7 cells transduced with mTLR7-HA (n = 3 replicates). H ABE assays were performed in THP-1 cells (n = 3 replicates). I ABE assay of hTLR9-FLAG in 293 T cells (n = 3 replicates). All data are pooled from four (E) and three (A, B, F–I) independent experiments (mean ± SEM. P values were calculated by two-way Student’s t test), except for those described in (C, D), which were obtained from one mass spectrometry run.

Fig. 5. DHHC3 palmitoylates TLR9 in the Golgi.

A ABE assays were performed in 293 T cells transfected with indicated murine DHHCs and mTLR9-FLAG. A representative blot (left) and mTLR9 relative palmitoylation levels (right, quantified as a ratio of mTLR9-FLAG signal compared to empty vector) are shown (n = 3 replicates). B ABE assays were performed in 293 T cells transfected with murine DHHC3 wild-type (WT) or C157S mutant, DHHC17 WT or C467S mutant, DHHC18 WT or C214S mutant, DHHC20 WT or C214S mutant, along with mTLR9-FLAG. A representative blot (left) and mTLR9 relative palmitoylation levels (right, quantified as a ratio of WT DHHC signal compared to respective mutants) are shown (n = 3 replicates) C TLR9 palmitoylation in DHHC3-deficient cells. ABE assays were performed on DHHC3-deficient RAW264.7 cells retrovirally transduced with mTLR9-HA. A representative blot (left) and mTLR9 relative palmitoylation levels (right, quantified as a ratio of WT signals compared to KO) are shown (n = 3 replicates). D ABE assays were performed in 293 T cells transfected with murine DHHC3 wild-type (WT) along with mTLR7-FLAG. A representative blot (left) and mTLR7 relative palmitoylation levels (right, quantified as a ratio of mTLR7-FLAG signal compared to empty vector) are shown (n = 3 replicates). E ABE assays were performed in 293 T cells transfected with indicated human DHHCs and hTLR9-FLAG. A representative blot (left) and mTLR9 relative palmitoylation levels (right, quantified as a ratio of hTLR9-FLAG signal compared to empty vector) are shown (n = 3 replicates). All data are pooled from three independent experiments (mean ± SEM, P values were calculated by two-way Student’s t test).

Fig. 6. PPT1 depalmitoylates TLR9 in lysosomes.

A ABE assays were performed on RAW264.7 cells transduced with mTLR9-HA after 4 h treatment of indicated TLR agonizts. A representative blot (left) and relative mTLR9 S-palmitoylation (right, quantified as the ratio of the mTLR9 output to the calnexin output, n = 3 replicates). B ABE assays were performed on RAW264.7 cells transduced with mTLR7-HA after 4 h treatment of R848. A representative blot (left) and relative mTLR7 S-palmitoylation (right, quantified as the ratio of the mTLR7 output to the calnexin output, n = 3 replicates). C ABE assays were performed on 293 T cells transfected with indicated depalmitoylating enzymes and mTLR9-FLAG (n = 3 replicates). D ABE assays were performed on 293 T cells transfected with PPT1 and mTLR7-FLAG (n = 3 replicates). E ABE assays were performed on Ppt1^+/+ or Ppt1^-/- Flt3L-pDCs after 4 h (top) and 24 h (bottom) of CpG A treatment (n = 3 replicates). F ABE assays were performed on Ppt1^+/+ or Ppt1^-/- RAW264.7 cells transduced with mTLR9-HA after 4 h of CpG A treatment (n = 3 replicates). G RAW264.7 cells transduced with mTLR9-HA were treated with DMSO or HDSF overnight. ABE assays were performed after the addition of CpG A (n = 3 replicates). All data are pooled from three independent experiments (mean ± SEM. P values were calculated by two-way Student’s t test).

Fig. 7. The palmitoylation cycle regulates TLR9 ligand binding.

A TNF production was measured by intracellular staining in CpG B-treated Tlr9^-/- RAW264.7 cells transduced with TLR9 Mut2 (n = 4 replicates). B Immunoprecipitation assay with CpG B-biotin were performed on Tlr9^-/- RAW264.7 cells transduced with TLR9 WT or Mut2. A representative blot (left) and relative ligand binding (right, calculated as the ratio of CpG B-bound on cleaved mTLR9 to the total mTLR9 in input) are shown (n = 3 replicates). C TNF production was measured by ELISAs in CpG B-treated Zdhhc3^+/+ or Zdhhc3^-/- RAW264.7 cells transduced with mTLR9-HA (n = 3 replicates). D Immunoprecipitation assays with CpG B-biotin were performed on Zdhhc3^+/+ or Zdhhc3^-/- RAW264.7 cells transduced with mTLR9-HA (n = 3 replicates). E TNF production was measured by ELISAs in CpG B-treated Ppt1^-/- or Ppt1^-/- RAW264.7 cells transduced with mTLR9-HA (n = 4 replicates). F Immunoprecipitation assays with CpG B-biotin were performed on Ppt1^+/+ or Ppt1^-/- RAW264.7 cells transduced with mTLR9-HA (n = 3 replicates). G RAW264.7 cells were treated with 2-BP or HDSF overnight. TNF production was measured by intracellular staining after addition of CpG B (n = 4 replicates). H Immunoprecipitation assays with CpG B-biotin were performed on RAW264.7 cells treated with 2-BP or HDSF (n = 3 replicates). I Cell fractionation of Zdhhc3^+/+ or Zdhhc3^-/- RAW264.7 cells transduced with mTLR9-HA showing the distributions of TLR9 in the indicated organelle markers. A representative blot (left) and the ratio of the full length mTLR9 in Golgi (right, quantified as the ratio of boxed fractions to all fractions, n = 3 replicates). J Immunoprecipitation of TLR9-HA from the DMSO or HDSF treated RAW macrophage lines in a followed by immunoblot of UNC93B1-FLAG. UNC93B1 levels in whole-cell lysates are also shown. A representative blot (left) and the ratio of the IP UNC93B1-FLAG in HDSF treated cells to the IP UNC93B1-FLAG in DMSO treated cells, n = 3 replicates. All data are representative of three independent experiment (mean ± SEM. P values were calculated by two-way Student’s t test).