Aim2 Couples with Ube2i for Sumoylation-mediated Repression of Interferon Signatures in Systemic Lupus Erythematosus

Abstract

Objective.

Systemic lupus erythematosus (SLE) involves kidney damage, and the inflammasome-caspase-1 axis had been demonstrated to promote renal pathogenesis. The current study was designed to explore the function of Aim2 in SLE.

Methods.

Female WT, Aim2^−/−, Aim2^−/−Ifnar1^−/−, Aim2^−/−Rag1^−/−, and Asc^−/− mice at 8–10 weeks of age received one intraperitoneal injection of 500μl pristane or saline, and survival was monitored twice a week for 6 months.

Results.

The absence of Aim2 but not Asc led to enhanced SLE in pristane-injected mice. Increased immune cell infiltration and type I interferon (IFN-I) signatures in the kidneys of Aim2^−/− mice coincided with lupus severity, which was alleviated by blockade of Ifnar1-mediated signal. Adaptive immune cells were also involved in the glomerular lesions of Aim2^−/− mice after pristane challenge. Importantly, even in the absence of pristane, plasmacytoid dendritic cells in the kidneys of Aim2^−/− mice were significantly increased compared with control animals. Accordingly, transcriptome analysis revealed that Aim2 deficiency led to enhanced expression of IFN-I-induced genes in the kidney even at an early developmental stage. Mechanistically, Aim2 bound Ube2i, which mediates sumoylation-based suppression of IFN-I expression; deficiency of Aim2 decreased cellular sumoylation, resulting in an augmented IFN-I signature and kidney pathogenesis.

Conclusion.

This study demonstrates a critical role for Aim2 in an optimal Ube2i-mediated sumoylation-based suppression of IFN-I generation and development of SLE, the Aim2-Ube2i axis can thus be a novel target for intervention.

INTRODUCTION

Systemic Lupus Erythematosus (SLE) is an autoimmune syndrome with severe clinical manifestations (1). SLE is associated with significant organ damage resulting from abnormal activation of immune signals (2). Innate immune factors including type I interferon (IFN-I) have been implicated in the development and pathogenesis of SLE (3, 4).

Recently, an essential role for caspase-1 in pristane-induced murine lupus has been described, indicating an involvement of inflammasomes in the pathogenesis of SLE (5), because assembly of an inflammasome leads to the activation of caspase-1, which triggers pyroptosis and the secretion of pro-inflammatory cytokines IL-1β and IL-18, as well as other functional proteins such as Galectin-3 (6, 7). Contrary to the function of caspase-1, Aim2 deficiency was implicated in SLE pathogenesis due to increased IFN-I signaling (8). However, direct in vivo evidence for this function is still lacking in lupus mice, even though several recent studies have demonstrated that Aim2, Asc and Caspase1 knock-out mice all produce higher amounts of IFN-I upon DNA stimulation (9–13).

Somewhat surprisingly, in the current study, we found that Aim2^−/− mice developed severe SLE upon pristane induction. In contrast, Asc^−/− mice did not develop any disease. This indicated that the enhanced lupus in Aim2^−/− mice was not due to a lack of inflammasome function. Instead, the severe lupus-like syndrome in Aim2-deficient mice was mainly attributed to the hastened glomerular IFN-I signal. Mechanistically, Aim2 was found to bind Ube2i, which mediates sumoylation-based inhibition of IFN-I transcription. When Aim2 was deleted, compromised Ube2i activity led to a stronger IFN signature and severe SLE. Therefore, our study uncovers the mechanism by which loss of Aim2 function drives lupus development, which is not associated with its inflammasome activity, but rather its function in facilitating Ube2i-mediated sumoylation-based inhibition of IFN-I signaling.

MATERIALS AND METHODS

Mice

C57BL/6 wild type (WT), Ifnar1^−/−, Rag1^−/− mice were from the Jackson Laboratory. Asc^−/− mice were from Dr. Vishva M. Dixit, Aim2^−/− mice were from Dr. Kate Fitzgerald. Animal care, use, and experimental procedures complied with national guidelines and were approved by the Animal Care and Use Committee at Institut Pasteur of Shanghai.

Quantification of anti-dsDNA and total IgG

Serum anti-dsDNA, -ssDNA, -MPO, -PR3, and total IgG levels were determined by enzyme-linked immunosorbent assay (ELISA) as described (14) (SIGMA, Jianglai Bio, eBioscience).

Crithidia luciliae assay

dsDNA-specific Crithidia luciliae slides (Inova Diagnostics) were used for determination of anti-dsDNA in mouse sera: 25μl of serum was added to wells and incubated for 30 minutes. Afterwards, serum was washed off and each well covered with a drop of Alexa Fluor-488-conjugated anti-mouse IgG followed by additional incubation. Finally, coverslips were mounted and fluorescent images were captured.

Evaluation of renal histopathological features and immune complex deposition

Paraffin-embedded sections of renal tissues were stained with hematoxylin and eosin (HE) or periodic acid-Schiff reagents-hematoxylin (PAS-H). Histopathological features of glomerular lesions were graded for severity in a double-blinded manner. The score for each animal was calculated via dividing the total score by the number of glomeruli observed. For detection of immune complex deposition, frozen kidney sections were stained with anti-mouse IgG-FITC (Santa Cruz) and anti-mouse C3-FITC (Cedarlane), respectively (14). The intensity of fluorescence was confirmed by flow cytometry.

Biochemistry

Urine samples were collected prior to euthanasia of mice and assessed for total protein with BCA kit, for Blood urea nitrogen (BUN) and Creatinine levels with ELISA kits (BioAssay Systems).

Flow cytometry

Isolation of total kidney cells from mice were performed as described (14). Cells were incubated with anti-mouse CD16/32 to block nonspecific Fc binding, then incubated with F4/80, MHCII, Ly6C, Ly6G, PDCA (eBioscience), CD11b, CD3, CD4, CD8, CD11c, CD45.2, B220 (BD Pharmingen), CD19 (Biolegend). Data were acquired using a BD FACS Fortessa and analyzed using FlowJo10 software.

Real-time PCR

Total RNA was extracted from kidney using TRIzol reagent (Invitrogen). Reverse transcription of mRNA and synthesis of cDNA was performed using TaqMan reverse transcription reagents (Promega). Real-time PCR was performed using the SYBR green quantitative PCR (qPCR) master mix (Toyobo) and the 7900HT fast real-time PCR system (Applied Biosystems). Relative quantification of genes was achieved via normalization against GAPDH. The primers used are listed in Supplementary Table 1.

RNAseq

Two kidneys from each E18 embryo were pooled as one sample, three such samples from each genotyped mice were preserved in RNAlater solution (Ambion) and subjected to RNA extraction after homogenization in Trizol (Sigma) according to the manufacturer’s instruction. RNAseq was performed as described (15).

Yeast two hybrid

We used the CLONTECH MATCHMAKER Two-Hybrid system to identify Aim2-interacting molecules. Briefly, the PGB-Aim2 plasmid was constructed and co-transformed with the Mouse Kidney Matchmaker cDNA Library for screening. Clones were selected with SD medium lacking Trp, Leu, His, but containing 30mM 3-amino-1,2,4-triazole. Plasmids from all the identified clones were isolated and co-transformed with PGB-Aim2 into Y190 yeast cells individually, confirmed genes were identified through sequencing analysis.

Plasmids and antibodies

Plasmids of mouse Aim2 and Ube2i were cloned from cDNA of bone marrow derived macrophages (BMDMs), ligated to pcDNA4.0. Sumo1, sumo2 and sumo3 plasmids were kindly provided by Dr. Jinke Cheng (Shanghai Jiao Tong University). Blots were stained with anti-FLAG (Sigma), anti-V5 (InvivoGen), anti-HA (Santa Cruz), anti-sumo1, -sumo2/3, -Ube2i, -Aim2, -Stat1, -P-Stat1, and -GAPDH antibodies (Cell Signaling Technology).

Co-immunoprecipitation and immunoblot

Transfected HEK293 cells were lysed and centrifuged at 14,000 rpm for 20 min at 4°C, 40μl supernatants were used as input, and the remaining supernatants were diluted with lysis buffer and incubated with respective antibodies and protein A/G magnetic beads at 4°C overnight. Beads were then washed and immunoprecipitated proteins were resolved with 12% SDS-PAGE, followed by immunoblotting.

Sumoylation assay

Transfected HEK293 cells or primary mouse kidney cells, BMDMs, or THP-1 cells with or without AIM2 silencing (16) were lysed with SDS loading buffer for immunoblotting against sumo1/2/3 using respective antibodies.

Statistics

Statistical analysis was performed with Prism 7 software (GraphPad) using the paired t test unless noted otherwise. Values were presented as mean ± SEM for 3 to 10 animals in each group as indicated. *P < 0.05 was taken as significant difference.

RESULTS

Deficiency of Aim2 exacerbates autoimmunity and leads to severe renal damage after pristane challenge

Lupus exhibits a strong female bias, and Aim2 expression is significantly decreased in female versus male mice of certain strains (17). Moreover, Aim2 silencing or deficiency in macrophages leads to elevated production of IFNβ upon DNA stimulation (18, 19). We thus tested whether Aim2^−/− mice were prone to pristane-induced lupus (14), and found that the survival of Aim2^−/− mice was significantly lower compared with that of wild type (WT) controls (Figure 1A). In addition, autoantibodies against DNA and total IgG in the serum of Aim2^−/− mice were significantly higher than in WT mice (Figure 1B). Of note, even saline injection led to a significant increase in autoantibodies (IgG and IgM) against DNA, complement proteins C3 and C5b-9, and total IgG from Aim2^−/− mice (Figure 1B, Supplementary Figure 1A). In addition, the dsDNA-specific Crithidia luciliae assay further confirmed the increase in serum dsDNA autoantibody levels from both saline- and pristane-injected Aim2^−/− mice (Figure 1C).

Figure 1. Aim2 deficiency leads to spontaneous autoimmunity and more severe renal damage upon pristane challenge.

(A) Female, 8–10 weeks old mice were injected with 500μl saline or pristine, cumulative survival was analyzed. Data were analyzed using the Gehan–Breslow–Wilcoxon test, data are mean ± 95% confidence interval. (B) Experimental mice from (A) were sacrificed 6 months after saline or pristane injection, serological levels of anti-dsDNA IgG, -ssDNA IgG and total IgG were determined via ELISA. (C) dsDNA-specific Crithidia luciliae slides were incubated with mouse sera followed by detection with fluorescent conjugated anti-mouse IgG. (D) Relative levels of urine protein and BUN over urine creatinine are shown. (E) Paraffin-embedded kidney sections were stained with PAS-Hematoxylin. (F) Glomeruli were counted on stained sections in (E), with the operator blinded to mice genotypes. (G) Myeloperoxidase (MPO) and proteinase 3 (PR3) in mouse sera were determined via ELISA. Data in (A) are pooled from two independent experiments. Values in (B, D, F, G) are mean ± SEM from five to eight mice for each genotype. *p < 0.05, **p < 0.01, ***p < 0.001, ****p<0.0001, ns, not significant.

As expected, the relative levels of urine protein and blood urea nitrogen (BUN) were elevated in Aim2^−/− compared with WT mice (Figure 1D). Histopathologic examination of the kidneys revealed increased glomerular diameter and accentuated glomerular lobularity in pristane-injected compared to saline-injected Aim2^−/− mice (Figure 1, E and F, Supplementary Figure 1B). The Aim2-deficient mice displayed mesangial widening, mesangial hypercellularity and endocapillary hypercellularity with mononuclear cells filling the glomerular capillaries. Glomerular fibrinoid necrosis was not identified and glomerular crescents were extremely rare (<0.5%).

Furthermore, there were clearly more glomerular immune complexes consisting of IgG and C3 in the kidney of Aim2^−/− mice after pristane injection (Supplementary Figure 1, C and D), which was confirmed by flow cytometry (Supplementary Figure 1E) (20). In addition, Myeloperoxidase (MPO) antibody was found significantly elevated in the serum of Aim2^−/− mice after either saline or pristane injection, while the level of proteinase 3 (PR3) was comparable between genotypes (Figure 1G). Anti-neutrophil cytoplasmic antibodies (ANCA) are a marker for ANCA-associated vasculitis (AAV) that is characterized by pauci-immune necrotizing crescentic glomerulonephritis. However, circulating MPO antibodies have also been demonstrated in sera of patients with immune-mediated glomerular disease, such as endocarditis-associated glomerulonephritis (21), anti-glomerular basement membrane (GBM) disease (22) and lupus nephritis (LN) (23). Aim2^−/− mice did not show glomerular fibrinoid necrosis or prominent cellular crescents, the characteristic glomerular lesions of ANCA disease, which have been demonstrated in murine models of MPO-ANCA disease (24, 25). The elevated anti-dsDNA antibody levels, glomerular hypercellularity, lack of glomerular fibrinoid necrosis and only rare crescents, and glomerular IgG and C3 deposition in the Aim2^−/− mice are findings more characteristic of LN.

Aim2 deficiency results in amplified immune cell infiltration and elevated levels of pro-inflammatory mediators in the kidney after pristane challenge

We further determined the profile of immune cell populations in the kidneys of Aim2^−/− and WT mice. First, a significant difference in the ratio of kidney/body weight between Aim2^−/− and WT mice after pristane injection was noted (Figure 2A). Moreover, the kidneys of Aim2^−/− mice contained more cells than that of WT mice after injection with either saline or pristane (Figure 2A). CD45.2⁺ cells were increased in the kidneys of Aim2^−/− mice in the saline group, and were further elevated after pristane injection (Figure 2A). When specific cell populations were analyzed, all were found to be significantly elevated by both percentage and absolute cell counts (Figure 2, B and C, Supplementary Figure 2, A and B). Strikingly, in the kidneys of Aim2^−/− mice, total dendritic cells, conventional dendritic cells and plasmacytoid dendritic cells (pDCs) were all astonishingly abundant compared to WT mice, independent of the challenge (Figure 2B, Supplementary Figure 2A).

Figure 2. Aim2 deficiency enhances renal infiltration of immune cells and expression of pro-inflammatory factors after pristane injection.

(A-C) The relative kidney to body weight from saline or pristane-injected Aim2^−/− and WT mice (Female, 8–10 weeks old) were monitored, and the total kidney cell number was counted. Isolated kidney cells were analyzed through flow cytometry after staining with the indicated antibodies: CD11c⁺ for total dendritic cells, CD11c^hiB220⁻ for conventional dendritic cells, CD11c^hiB220⁺PDCA⁺ for plasmacytoid dendritic cells, CD11b⁺F4/80⁺ for macrophages, CD11b⁺Ly6G⁺ for neutrophils, CD19⁺B220⁺ for B cells and CD3⁺ for T cells. (D-F) Quantitative real-time PCR was employed to monitor the expression of indicated pro-inflammatory factors in the kidney samples from saline or pristane-injected Aim2^−/− and WT mice. Data shown are mean ± SEM. In (A-C), a total of 3 to 4 mice were used in each group; in (D-F), 3 to 8 mice were used in each group. *p< 0.05, **p< 0.01, ***p< 0.001, ns, not significant.

As pDCs are powerful IFN-I producers (26, 27), and IFN-I is closely connected with the development and progression of lupus, these data implied an IFN-I-associated pathogenesis in Aim2^−/− mice. In addition, macrophages and neutrophils were more abundant in the kidneys of Aim2^−/− mice, especially after pristane injection (Figure 2B, Supplementary Figure 2A), indicating an enhanced inflammatory response in those mice. Of note, adaptive immune cells (B cells and T cells) were also significantly elevated in Aim2^−/− mice after pristane injection (Figure 2C, Supplementary Figure 2B). In the saline-injected group, all these cells showed an increase and CD4⁺ T helper cells were significantly more abundant in the kidney of Aim2^−/− mice (Figure 2C, Supplementary Figure 2B). Altogether, these data suggest that the severe renal damage observed in Aim2^−/− mice also involves recruitment of adaptive immune cells.

Since pDCs were so strikingly abundant in the kidney of Aim2^−/− mice, we reasoned that the IFN-I signature should be also high. Interestingly, however, we found that the expression of Ifnα and Ifnβ mRNA were only slightly elevated in the kidneys of Aim2^−/− mice (Figure 2D). Nonetheless, IFN-stimulated genes such as Isg15, Mx2, as well as Interferon regulator factor 7 (Irf7) and Ifi202b were all significantly elevated in Aim2^−/− kidney samples after pristane injection (Figure 2D, Supplementary Figure 2C). Of note, expression of Ifi202b and Irf7 was significantly elevated already in the kidneys of Aim2^−/− mice after saline injection, while Flt3 and Flt3L were only mildly increased (Supplementary Figure 2C). These data imply that type I IFN and ISG induction in vivo in chronic diseases such as SLE might be different from transient transfection of exogenous DNA into innate immune cells in vitro (9, 18).

Our further analysis demonstrated that pro-inflammatory cytokines Il6, Il18, Tnfa and Csf2 (GM-CSF) were all significantly elevated in the kidneys of Aim2^−/− mice compared with WT controls, while Il1β was only mildly higher (Figure 2E, Supplementary Figure 2D); these data are consistent with findings from SLE patients (28). We also found that transcription of Ccl2, Ccl3, Ccl5, Ccl20 and Cxcr3 were all markedly increased in the kidneys of Aim2^−/− mice (Figure 2F, Supplementary Figure 2D). Thus, deficiency of Aim2 in mice promoted the expression of pro-inflammatory cytokines and chemokines in the kidney, especially after pristane challenge.

Last, we observed that in the kidneys of Aim2^−/− mice, Baff and April were elevated significantly compared with WT controls; interestingly, T cell-associated cytokines Ifnγ and Il17a did not show a significant difference between Aim2^−/− and WT kidney samples (Supplementary Figure 2E). Therefore, lupus development in Aim2^−/− mice involves some but not all aspects of adaptive immunity.

Blocking type I interferon signal rescues pristane-induced lupus in Aim2^−/− mice

Since the number of pDCs and the expression of IFN-I-related factors were both increased in the kidneys of Aim2^−/− mice, we hypothesized that the observed increase in ISG expression was responsible for lupus development in Aim2^−/− mice. To test this possibility, we crossed Aim2^−/− mice with Ifnar1^−/− mice and induced lupus. Of note, Aim2^−/−Ifnar1^−/− mice were all protected from pristane-induced lethality, having lower titers of anti-dsDNA and total IgG in serum, and lower levels of proteinuria/creatinine and BUN/creatinine compared with Aim2^−/− mice (Figure 3, A–C). H&E and PAS-H staining showed that the size and morphology of glomeruli in Aim2^−/−Ifnar1^−/− mice were similar to those in WT mice, having no clear hypercellularity, mesangial expansion, or interstitial mononuclear cell infiltration as in Aim2^−/− mice (Figure 3D, Figure 1E). In addition, the percentage of IgG⁺ and C3⁺ cells were clearly decreased in Aim2^−/−Ifnar1^−/− mice (Figure 3E). Further, cell infiltration analysis in the kidneys from Aim2^−/− and Aim2^−/−Ifnar1^−/− mice showed that although the abundance of total immune cells was not significantly different (Figure 3F), infiltration of innate immune cells, including total dendritic cells, pDCs, macrophages and neutrophils were significantly diminished in Aim2^−/−Ifnar1^−/− mice (Figure 3G). Moreover, B cells and T cells in Aim2^−/−Ifnar1^−/− mice were significantly reduced compared with Aim2^−/− mice (Figure 3H). Furthermore, the expression levels of Irf7, Isg15 and Mx2 were all significantly decreased in mice deficient for both Aim2 and Ifnar1 compared with Aim2^−/− mice, although the expression of Ifnα and Ifnβ did not show a marked difference (Figure 3I). Accordingly, the expression levels of chemokines such as Ccl2, Ccl3, Ccl5, Ccl20, Cxcl10, and chemokine receptor Cxcr3 were all significantly decreased in Aim2^−/−Ifnar1^−/− mice (Figure 3J). Therefore, signaling mediated by Ifnar1 is central to the in vivo pathogenesis of nephritis in Aim2-deficient mice.

Figure 3. Kidney damage and systemic pro-inflammatory responses in pristane-injected Aim2 ^−/− mice depend on Ifnar1-mediated signaling.

(A) Female, 8–10 weeks old mice with indicated genotypes were injected and monitored as in Figure 1. (B-C) Disease parameters, including anti-dsDNA and total IgG levels in serum, urine protein excretion, BUN, and urine creatinine were assayed as in Figure 1 using samples from Aim2^−/− and Aim2^−/−Ifnar1^−/− mice. (D) H&E and PAS-H staining of paraffin-embedded kidney sections. (E-H) Flow cytometry analysis of cells expressing IgG, C3 and indicated markers in isolated kidney cells as in Figure 2. (I-J) Quantitative real-time PCR was performed to analyze the expression levels of the indicated factors in the kidney. All experimental samples were from pristane-injected Aim2^−/− and Aim2^−/−Ifnar1^−/− mice. Data in (A) are pooled from two independent experiments. Data in (B, C, D, J) are from four to seven mice; data in (E-H) are from three to four mice in each group. Data are mean ± SEM. *p < 0.05, **p < 0.01, ***p < 0.001, ns, not significant.

Depletion of adaptive immune cells in Aim2^−/− mice rescues pristane-induced lupus nephritis

When Aim2^−/− mice were crossed with Rag1^−/− mice, lupus-associated lethality diminished with substantially lower levels of serum anti-dsDNA, total IgG, as well as proteinuria/creatinine and BUN/creatinine (Figure 4, A–C). In addition, the size and morphology of the glomeruli in Aim2^−/−Rag1^−/− mice were similar to those in WT mice, with clearly less damage than that seen in Aim2^−/− mice (Figure 4D, Figure 1E). Moreover, cells expressing IgG were also reduced in Aim2^−/−Rag1^−/− mice compared with Aim2^−/− mice, although the drop in C3⁺ cells was mild (Figure 4E). The abundance of various immune cells was lower in Aim2^−/−Rag1^−/− mice compared with Aim2^−/− mice (Figure 4, F–H). Accordingly, real-time PCR analysis showed that in the kidneys of Aim2^−/−Rag1^−/− mice, the expression of Ifnα, Ifnβ, Irf7, Isg15, Mx2, Ccl2, Ccl3, Ccl5, Ccl20, Cxcl10 and Cxcr3 was clearly decreased compared with Aim2^−/− mice (Figure 4, I and J). Together, these data demonstrated that abolishing Rag1 in Aim2^−/− mice rescued renal damage and attenuated the level of pro-inflammatory mediators, as well as immune cell infiltration to the kidney upon pristane challenge, indicating that B and T cells contribute to the pathogenesis of lupus in Aim2-deficient mice.

Figure 4. Kidney damage and systemic pro-inflammatory responses are alleviated in Aim2^−/−Rag1^−/− mice compared with Aim2^−/− mice.

(A) Indicated mice (Female, 8–10 weeks old) were injected and monitored as in Figure 1. (B-D) Serological anti-dsDNA and total IgG levels (B), Urine protein excretion, BUN and urine creatinine (C), as well as kidney sections (D) were monitored as in Figure 3. (E-H) Flow cytometry analysis of cells expressing IgG, C3 and indicated markers in isolated kidney cells as in Figure 2. (I-J) Real-time PCR analysis of indicated pro-inflammatory factors as in Figure 3. All experimental samples were from pristane-injected Aim2^−/− and Aim2^−/−Rag1^−/− mice. Data in (B, C, D, J) are from four to seven mice; data in (E-H) are from three to four mice in each group. Data are mean ± SEM. *p< 0.05, **p< 0.01, ***p< 0.001, ns, not significant

Lupus development in Aim2^−/− mice is independent of the inflammasome

To determine whether the development of lupus in Aim2^−/− mice was due to a lack of inflammasome function, Asc^−/− and WT mice were injected with pristane and monitored for 6 months. None of the Asc^−/− mice died or developed SLE. Histological appearance, sera and urine parameters, as well as the expression of type I IFNs and ISGs in the kidneys of Asc^−/− and WT mice were all similar (Supplementary Figure 3, A–C). These data indicated that the phenotype of Aim2^−/− mice after pristane injection was independent of the Aim2-Asc-Caspase-1 inflammasome cascade.

Aim2 deficiency promotes inflammation in the kidney from embryonic stage

As saline-injected adult Aim2^−/− mice carry more renal damage, more inflammatory cytokine/chemokine expression and immune cell infiltration in the kidney compared with WT mice, we reasoned that Aim2 might play an important role in suppressing inflammation under resting state. Because IFN-I signature genes can be self-inducing and inter-amplifying, we examined when the inflammatory response begins and what are the early driving factors. To this end, we analyzed cell infiltration in the kidneys of embryonic stage (E18) mice and found that CD45.2⁺, CD11c⁺, CD11c^hiB220⁻, CD11c^hiB220⁺, CD11b⁺/F4/80⁺ cells were increased in Aim2^−/− mice, while CD11b⁺Ly6G⁺ cells were not different between these two genotypes; CD4⁺ T cells and CD19⁺B220⁺ B cells were slightly decreased in Aim2^−/− kidneys compared with WT mice (Figure 5, A and B). Accordingly, RNA-seq revealed that a large set of ISGs were upregulated in Aim2^−/− samples compared to WT mice (Figure 5C). Of note, Ifi202b was dramatically elevated, and Mx2, Ccl9, Ccr5, Stat1, Stat2 and Stat5a were all increased in Aim2^−/− mice (Figure 5C). Expression of selected ISGs such as Ifi202b, Isg15, Irf7 and Mx2 were further confirmed by quantitative PCR (Figure 5D). Thus, Aim2 negatively regulates IFN-I signatures even at an early stage of embryonic development.

Figure 5. Aim2^−/− mice spontaneously develop a type I IFN signature at embryonic stage E18.

(A-B) Isolated cells from pooled kidneys of 4 individual embryos (E18) of Aim2^−/− and WT mice were analyzed by flow cytometry after staining with the indicated antibodies. (C) Heat map of differential gene expression in Aim2 deficient mice compared with WT control at E18 in the kidney. Red indicated the mean of genes expression at higher levels, and blue represented genes expression at lower levels in Aim2^−/− mice than control. Data are from one representative of two experiments with three independent biological replicates in each. (D) Quantitative real-time PCR was employed to monitor the expression of indicated factors in the embryonic E18 kidney from Aim2^−/− and WT mice. Data are mean ± SEM from 4–6 embryos. *p< 0.05, ns, not significant.

Aim2 interacts with Ube2i, promotes Ube2i-mediated sumoylation and inhibits type I IFN expression

In order to identify factors that contribute to Aim2-mediated negative regulation of IFN-I and/or ISGs expression, a yeast two hybrid screen was performed, which revealed that Ube2i was the most frequent binding partner for Aim2 (Figure 6A). The interaction between Aim2 and Ube2i was confirmed in yeast (Figure 6B), and by co-immunoprecipitation (IP) in HEK293 cells (Figure 6C). Although weak, the endogenous Ube2i bound Aim2 in BMDMs (Supplementary Figure 4A). Mapping experiments revealed that only full-length Aim2 was able to bind Ube2i; the truncated forms lacking either pyrin or HIN200 domain lost interaction with Ube2i (Supplementary Figure 4, B and C). As one of the unique ubiquitin-conjugating enzyme E2, Ube2i is essential for sumoylation of substrate proteins (29). Interestingly, we found that Aim2 promoted Ube2i-mediated sumoylation of proteins in HEK293 cells (Figure 6D, compare lane 3 to 2, lane 6 to 5, lane 9 to 8). More importantly, primary kidney cells from Aim2^−/− mice exhibited decreased sumo1- and sumo2/3-modified protein compared with WT controls (Figure 6E). In addition, resting BMDMs from Aim2^−/− mice showed less sumoylated substrates (Supplementary Figure 5A). Sumoylation regulates protein stability, interaction and activity, and is involved in innate immune responses (30). Loss of sumoylation in innate immune cells leads to potent IFN-I responses without exogenous stimuli (31, 32). Along this line, Aim2-deficient BMDMs showed spontaneous increase of ISGs such as Isg15 and Mx2 (Supplementary Figure 5, B and C), which was similar to that seen in Ube2i deficient cells (31). Of note, treatment with sumoylation inhibitor Ginkgolic acid (GA) or Ube2i silencing with siRNA rescued Aim2 deficiency-mediated ISGs upregulation (Supplementary Figure 5, B and C), implying that Aim2 inhibits IFN-I signaling through Ube2i. Further, AIM2-silenced human monocytic THP-1 cells showed less sumo2/3-mediated modification compared with Scramble controls, while the sumo1-mediated modification was relatively unchanged (Supplementary Figure 5D). Notably, for the negative regulation of IFN response, sumo2/3-mediated modification was more dominant than sumo1 (32), supporting the observation of a stronger effect on sumo2/3-mediated modification than sumo1 when AIM2 was silenced. Thus, Aim2 functions as an Ube2i chaperone, promoting sumoylation and inhibition of IFN-I production (Supplementary Figure 5E).

Figure 6. Aim2 binds Ube2i, promotes Ube2i-mediated sumoylation and inhibits type I IFN production.

(A) Yeast two-hybrid screening results for Aim2 binding partners in Mouse Kidney Matchmaker cDNA Library, with Ube2i as the most frequent prey molecule. (B) PGB-Aim2 and pGADT7-Ube2i were co-transformed into yeast Y190, the yeast formed clones on the SD/-Trp/-Leu/-His + 30mM 3-AT selection media as the middle image show. PGBKT7-p53 and pGADT7-T were co-transformed as positive control as the top image show. PGBKT7-lam and pGADT7-T were co-transformed as negative control as the bottom image show. (C) HEK293 cells were transfected with indicated plasmids for 36 hours followed by co-immunoprecipitation and immunoblot analysis with indicated antibodies. (D) HEK293 cells were transfected with indicated plasmids for 36 hours followed by in vivo sumoylation assay and immunoblotting with indicated antibodies. HA blot on the top panel indicates total sumoylated proteins, HA blot in the bottom shows the expression of transfected sumo1, sumo2, sumo3 plasmids. (E) Primary kidney cells from WT and Aim2^−/− mice were lysed directly followed by immunoblot analysis of sumo1 and sumo2/3 as indicated. In (C-E), the result is one representative of two independent experiments.

DISCUSSION

In the current study, we demonstrate that independent of its inflammasome function, Aim2 acts as a negative regulator of IFN-I and ISGs expression. We further found that Aim2 binds Ube2i, a protein that mediates sumoylation-based suppression of IFN-I production. Similar to Ube2i deficient mice (31), Aim2^−/− mice developed SLE like syndrome spontaneously, which was exacerbated by pristane challenge.

Because inflammasome-mediated kidney damage has been shown to contribute to the pathogenesis of SLE (5, 14), it is possible that although the IFN-I signal is high in Aim2^−/− mice, lack of inflammasome activation may decrease the likelihood of inflammatory damage. Our data indicated that the IFN-I signal is dominant, which is in agreement with a recent work showing that upon Francisella novicida infection, the detrimental IFN-I signaling dominates protective AIM2 inflammasome response (33). Nonetheless, the finding of abnormally high IFN-I-related signatures at early developmental stage in the kidneys of Aim2^−/− mice demonstrates that even in the absence of exogenous DNA challenge, Ube2i activity is compromised, resulting in spontaneous autoimmune phenotypes in Aim2^−/− mice.

As a critical cellular source of IFN-I, plasmacytoid dendritic cells are strongly involved in the pathogenesis of SLE (34). Intriguingly, our present work showed that after injection with either saline or pristane, adult Aim2^−/− mice had significantly more pDCs in the kidney. Remarkably, at the embryonic stage, more pDCs were found in the kidney of Aim2^−/− mice compared with WT control, which was further magnified at the adult stage, and the expression of ISGs showed the same bias (Supplementary Figure 6, Figure 5). Interestingly, the expression of Aim2 was dramatically decreased when the mice reach adult age, which fits very well with our findings that mice with lower Aim2 were prone to SLE. Moreover, since the expression of a panel of chemokines was significantly elevated in the kidneys of Aim2^−/− mice, the accumulation of pDCs was likely due to chemokine-mediated infiltration of such cells from immune organs. Indeed, the abundance of pDCs in the inguinal lymph node was increased, while slightly decreased in the spleen and bone marrow from Aim2^−/− mice compared with WT controls (Supplementary Figure 7). It should be noted that although an increase of Ifnα and Ifnβ expression in the kidneys of resting adult Aim2^−/− mice was observed, the elevation was only mild (Figure 2D). Therefore, the very early trigger of an IFN-I signature in the E18 kidney may be due to trace amounts of IFN-I in Aim2^−/− progenitor cells, which leads to further differentiation and maturation of pDCs. Since the level of Ifnar was also moderately increased in Aim2^−/− mice (Supplementary Figure 2D), and an increased IFN-I signaling via increased pSTAT1 levels in both kidney cells and BMDMs from Aim2^−/− mice was detected (Supplementary Figure 8), it is rather possible that the transcription of ISGs has been amplified through the IFN-I-IFNAR-STAT1-ISGs cascade.

As a typical ISG, the Ifi202b gene was strongly elevated in the kidney samples of Aim2^−/− mice (Supplementary Figure 2C, Figure 5, C and D). Since Ifi202b encodes the SLE susceptibility protein p202 (35), which can also drive the expression of IFN-I in the presence of STING (36), it is possible that the disease observed in Aim2^−/− mice is partially attributed to p202 elevation. In addition, the production of IFN-I is regulated by interferon regulatory factors (IRFs), and many of them are involved in SLE (37). In our current work, Irf7 was significantly elevated in the kidneys of Aim2^−/− mice in both saline- and pristane-injected mice compared with WT controls (Supplementary Figure 2). In E18 kidney samples, Irf7 also showed a significant elevation in Aim2^−/− mice. As almost all of the IRFs are ISGs, these data suggest that once a trace amount of IFN-I is produced, an activation loop with feed-forward effects may amplify the inflammatory responses locally and cause severe renal damage.

In summary, our work uncovers an important and novel mechanism for Aim2-mediated inhibition of IFN-I expression in the development and pathogenesis of lupus, pointing to the Aim2-Ube2i complex as a new therapeutic target for SLE.

Abbreviations used:

SLE

Systemic Lupus Erythematosus

Aim2

Absent in Melanoma 2

Asc

Apoptosis-associated speck-like protein containing a CARD

Ube2i (Ubc9)

Ubiquitin-conjugating enzyme E2I

pDC

plasmacytoid Dendritic Cell

IFN-I

Type I Interferon