Histone Deacetylase 9 Deficiency Protects against Effector T Cell-mediated Systemic Autoimmunity*

Co-repressor histone deacetylase 9 (HDAC9) plays a key role in the development and differentiation of many types of cells, including regulatory T cells. However, the biological function of HDAC9 in T effector cells is unknown. Systemic autoimmune diseases like lupus, diabetes, and rheumatoid arthritis have dysfunctional effector T cells. To determine the role of HDAC9 in systemic autoimmunity, we created MRL/lpr mice with HDAC9 deficiency that have aberrant effector T cell function. HDAC9 deficiency led to decreased lympho-proliferation, inflammation, autoantibody production, and increased survival in MRL/lpr mice. HDAC9-deficient mice manifested Th2 polarization, decreased T effector follicular cells positive for inducible co-stimulator, and activated T cells in vivo compared with HDAC9-intact MRL/lpr mice. HDAC9 deficiency also resulted in increased GATA3 and roquin and decreased BCL6 gene expression. HDAC9 deficiency was associated with increased site-specific lysine histone acetylation at H3 (H3K9, H3K14, and H3K18) globally that was localized to IL-4, roquin, and peroxisome proliferator-activated receptor-γ promoters with increased gene expression, respectively. In kidney and spleen, HDAC9 deficiency decreased inflammation and cytokine and chemokine production due to peroxisome proliferator-activated receptor γ overexpression. These findings suggest that HDAC9 acts as an epigenetic switch in effector T cell-mediated systemic autoimmunity.

In immune systems HDAC7 and HDAC9 play significant roles in development and differentiation. HDAC7 has a specific role in thymocyte development. HDAC7 and HDAC9 are key regulators in regulatory T cell functions (9,10). HDAC9 is expressed at higher levels in regulatory T (T-reg) cells than non-T-reg cells. HDAC9 expression significantly decreases during CD3/CD28 stimulation in human regulatory T cells compared with T effector cells (11). HDAC9 deficiency results in increased T-reg cells in lymphoid tissues in mice, with enhanced expression of Foxp3, CTLA4, and GITR and suppressive activity of T-reg cells (10). HDAC9-deficient mice are resistant to developing colitis (12).
The molecular mechanisms of pathogenesis of systemic autoimmunity include T cell-dependent B cell activation, autoantibody production, immune complex formation, direct infiltration of autoreactive T and B cells, and production of inflammatory mediators in the specific tissues (13,14). The effector T cells (including T helper Th1, Th2, Th17, and Tfh cells) are largely responsible for B cell activation, affinity maturation, class switching, and induction of long-lived plasma and memory cells. Conversely, T-reg cells are key mediators of peripheral tolerance by suppressing T effector cells and are known to be dysfunctional in systemic autoimmunity (15). Therefore, it has been hypothesized that aberrant differentiation of multipotential naïve CD4 ϩ T cells into distinct lineages (including Th1, Th2, Tfh, and Th17 cells) via epigenetic mechanisms may create specific T effector cell-mediated autoimmunity (3).
To delineate the role of HDACs in T cell-mediated autoimmune disease, we initially used a chemical inhibition approach in established models of systemic autoimmunity. Two HDAC inhibitors (trichostatin A and suberoylanilide hydroxamic acid) decreased systemic autoimmunity in MRL/lpr and NZB/NZW F1 mice (16 -18). We also demonstrated that trichostatin A reverses skewed expression of several important genes in human lupus T cells (19). Additionally, trichostatin A decreases systemic autoimmune disease in NZB/W F1 mice, in part by increasing T-reg cells (16). We have also observed site-specific hypoacetylation in different lysine residues of histone H3 and H4 that can be corrected by HDAC inhibitors in MRL/lpr mice in vivo (20). Collectively, these results suggest that HDACs play a key role in the pathogenesis of systemic autoimmunity, particularly lupus.
To identify which HDACs may play a key role in aberrant T cell differentiation in systemic autoimmunity, we initially screened class IIa HDACs in autoimmune MRL/lpr mice. HDAC9 was overexpressed in lupus-prone MRL/lpr mice in different T cell subsets compared with control MRL/MpJ mice. To understand the role of HDAC9 in T effector cell development and function beyond its well known function in T-reg cells, we studied MRL/lpr lupus-prone mice. MRL/lpr mice have aberrant Th1 cell and inducible co-stimulator (ICOS)positive Tfh function (21) and a normal percentage of CD4 ϩ / CD25 high T cells and Foxp3 mRNA and protein expression in T-reg cells (21). Effector CD4 ϩ T cells from MRL/lpr mice cannot be fully suppressed by T-reg cells due to diminished expres-sion of CTLA4, CD80, and CD86 (21). Therefore, these mice are valuable for elucidating the role of HDAC9 function in T effector cell-mediated autoimmunity independent of T-reg function.
Here we demonstrate that HDAC9 deficiency leads to decreased systemic autoimmunity with increased survival and decreased immunoproliferation, decreased activated T cells, decreased ICOS-positive T effector cells and double negative T cells (DNT), increased Th2 polarization, autoantibody production, and kidney disease in MRL/lpr mice. Our findings in this study further elucidate the role of HDAC9 in CD4 ϩ T cell effector-mediated systemic autoimmunity and may be useful in the future development of HDAC9 isoform-specific inhibitors for these conditions.

EXPERIMENTAL PROCEDURES
Human Samples-Adult subjects with an established diagnosis of systemic lupus erythematosus and healthy normal controls (age-, sex-, and race-matched) were studied. All subjects signed an informed consent form approved by the Wake Forest University Institutional Review Board before participation in this study.
Animals-MRL/MpJ and MRL/lpr mice were purchased from The Jackson Laboratory. HDAC9 Ϫ/Ϫ mice were a kind gift from Dr. Eric Olson (UT Southwestern Medical Center, Dallas, TX) (22). MRL/lprHDAC9 Ϫ/Ϫ (KO) are generated by backcrossing HDAC9 Ϫ/Ϫ to the MRL/lpr mice for 12 generations. Female-and age-matched control MRL/MpJ, MRL/lpr mice were used for the experiments. All procedures were approved by the Animal Care and Use Committee at Wake Forest University School of Medicine.
Cell Preparations-CD4 ϩ T cells were isolated from blood samples from healthy controls and patients with systemic lupus erythematosus by negative selection using magnetic beads as described previously (19). Similarly, murine CD4 ϩ T cells were isolated using a mouse CD4 isolation kit according to the manufacturer's instructions (Miltenyi Biotec).
Histopathology-Twenty-week-old female mice were sacrificed. Renal histopathological analyses were performed by a pathologist blinded to group assignments using a semi-quantified grading scheme as previously described (23). IgG and C3 deposits in the kidney were measured using a 1:100 dilution of FITC-conjugated goat anti-mouse IgG and C3 (Pierce). The degree of IgG and C3 deposition was determined in a blinded manner (17).
Autoantibody, Immunoglobulin, and Cytokine Measurements-Anti-nuclear antibodies and anti-dsDNA antibodies were measured using commercially available slides (Immuno Concepts) by indirect immunofluorescence using anti-goat mouse secondary antibodies. Anti-Sm and cardiolipin levels were measured by ELISA (Alpha Diagnostic International). Total IgA, IgM, and IgG1, IgG2a, IgG2b, IgG3, and IgE isotypes were measured by ELISA (Southern Biotechnology Associates). Serum immunoglobulins specific to anti-ds-DNA and NP were measured by ELISA (25). The inflammatory cytokines IL-2, IL-4, IFN-␥, and IL12 p70 in serum and supernatants were quantified by ELISA (BD Biosciences OptEIA TM ELISA kit).
Quantitative Real-time PCR-Total RNA was extracted with TRIzol reagent according to the manufacturer's protocol (Invitrogen). Reverse transcription reactions and TaqMan Universal PCRs or SYBR Green PCRs were performed in accordance with the manufacturer's instructions (Applied Biosystems) and analyzed on an Applied Biosystems 7000 Sequence Detection system.
RP-HPLC and Mass Spectrometry-Mass spectrometry analysis of mouse core histones was done as described previously (25). Briefly, acid-extracted histones were treated with propionylation reagent, digested with trypsin, and re-propionylated (25).
Chromatin Immunoprecipitation (ChIP) Assay-ChIP analysis was performed according to the Upstate (Millipore) protocol. Immunoprecipitated chromatins were de-cross linked, and DNA was isolated and analyzed by PCR with primers flanking either promoter or enhancer. For IL-4 and roquin, PCR products were resolved on a 2.0% agarose gel and visualized with ethidium bromide; band intensities were quantified using an Alpha Imager 2000. For the histone acetylation occupancy at the PPAR-␥ promoter, quantitative real-time PCR analysis was performed with Power SYBR Green PCR Master Mix (Applied Biosystems) using primers as previously described (26,27). Analysis was performed by the standard curve methods and the threshold cycle (C T ) for each product.
Statistics-Statistical analysis was performed using GraphPad Prism 3.03 software (GraphPad Software Inc.). Differences between parameters were analyzed using two-tailed unpaired Student's t test. Kaplan-Meier survival curves were determined by using log rank tests. p values of less than 0.05 were considered significant. Detailed descriptions of the methods are available in the supplemental material.

RESULTS
HDAC9 Is Overexpressed in Different Subsets of CD4 ϩ T Cells in MRL/lpr Mice and Human Lupus CD4 ϩ T Cells-To determine which HDACs play a role in systemic CD4 T cell-mediated autoimmunity, we performed real-time quantitative PCR in splenocytes from MRL/lpr mice and control MRL/MpJ mice. MRL/lpr splenocytes expressed higher levels of HDAC9 compared with MRL/MpJ mice (supplemental Fig. S1A). There was no significant difference in expression of HDAC4, -5, or -7 mRNA between the two groups of mice. We also measured HDAC9 expression during disease progression in splenocytes, kidney, and purified subsets of CD4 ϩ T cells from different age groups of MRL/lpr mice. HDAC9 expression increased in spleens and kidneys of MRL/lpr mice as a function of age and disease progression (supplemental Fig. S1, B and C). Similarly, HDAC9 was overexpressed in an age-dependent manner in CD4 ϩ T cells and DNT cells from MRL/lpr mice (supplemental Fig. S1, D and E). Similar to the murine model of lupus, HDAC9 was also significantly increased in CD4 ϩ T cells from lupus patients compared with healthy controls (supplemental Fig.  S1F).
To determine the role of HDAC9 in systemic autoimmunity, we generated MRL/lpr mice lacking HDAC9 genes by backcrossing HDAC9-deficient mice to MRL/lpr mice for 12 generations. To confirm that the mutation created a null allele on the MRL/lpr background, we performed real time PCR and Western blot analyses for HDAC9. The KO mice did not express HDAC9 mRNA or protein (supplemental Fig. S2A). To determine whether other HDACs might be up-regulated in mutant mice to compensate for the lack of HDAC9, we compared the expression of HDACs 1-8 by real time RT-PCR in splenocytes. We found no significant difference in the expression of these HDACs because of HDAC9 deficiency, consistent with previous reports (supplemental Fig. S2B) (22).
HDAC9 Deficiency Results in Histone Hyperacetylation at Lysine Residues H3K9, H3K14, and H3K18 of Histone H3-Although a mutational analysis of yeast revealed site-specific acetylation of different histones by different HDACs, the specific lysine residues on histones H3 and H4 that may be regulated by HDAC9 in mammalian systems are currently unknown (28). Therefore, we performed differential histone modification analysis using stable isotope labeling in combination with mass spectrometry in splenocytes from KO and MRL/lprHDAC9 ϩ/ϩ (WT) mice. Lysine residues H3K9 and H3K14 were hyperacetylated (ϳ2-fold), and H3K18 was hyperacetylated (ϳ5-fold) in splenocytes from HDAC9-deficient mice (supplemental Table  S1 and Fig. S3, A and B). To confirm our findings, we performed immunoblot analysis using site-specific anti-acetylated histone antibodies to determine whether these changes were specific to different organs. H3K9, H3K14, and H3K18 sites were hyperacetylated in splenocytes and kidney from KO compared with WT tissues, with no appreciable changes observed in liver or heart (supplemental Fig. S3C).
Loss of HDAC9 Improves Immunological, Serological, and Clinical Autoimmune Phenotypes and Prolongs Survival in MRL/lpr Mice-We monitored both WT mice and KO female mice for disease progression as they aged. We measured body weight, gross skin lesions, and proteinuria beginning at 10 weeks of age and continuing until the mice became moribund (for a survival study). We sacrificed mice at 20 weeks of age because 50% of MRL/lpr mice die spontaneously at this age (18). At sacrifice, we measured body weight and organ weights for spleen, lymph node, liver, kidney, and heart. Spleens and lymph nodes were significantly smaller in the KO mice (supplemental Fig. S4, A-C, and Table S2).
To understand how HDAC9 deficiency reduced lymphoproliferation, we used flow cytometry to analyze spleen and lymph nodes from KO and WT animals. HDAC9 deficiency significantly decreased the percentage and cell number of CD3 ϩ CD4 Ϫ CD8 Ϫ T cells (DNT cells), activated CD4 ϩ CD69 ϩ T cells and B220ϩ B cells, and CD138ϩ plasma cells as compared with WT animals (supplemental Table S3 and Fig. 1). The decreased number of DNT cells in HDAC9-deficient MRL/lpr mice is consistent with our previous histone deacetylase inhibitor studies in MRL/lpr mice (18).
We next sought to determine whether HDAC gene deletion would affect autoantibody production. Anti-nuclear antibodies, anti-Sm, and anti-cardiolipin antibody levels were not statistically different between the two groups of mice (supplemental Fig. S5, A, C, and D). However, KO animals had a significant reduced titer of high affinity anti-dsDNA autoantibodies, as determined by Crithidia immunofluorescence assay compared with WT mice (supplemental Fig. S5B).
It is generally agreed that serum IgG2a, IgG2b, and IgG3 play significant roles in renal disease, although the role of IgM in lupus autoimmunity is controversial (29). To determine whether HDAC9 deficiency alters the titers of circulating antibody isotypes, we compared serum levels of IgM, IgG1, IgG2a, IgG2b, IgG3, IgA, and IgE between the two groups of mice. We observed a significant reduction of IgM and IgG3, a trend for decreased concentrations of IgG2a and IgG2b, no difference in IgA or IgE, and a modest increase in IgG1 in the serum of KO compared with the WT mice (supplemental Fig. S5E). Importantly, compared with WT mice, KO mice had significantly reduced anti dsDNA antibody levels of IgM, IgG2b, and IgG3 isotypes as measured by isotype-specific ELISA (supplemental Fig. S5F). HDAC9 deficiency resulted in a serological decrease in the high affinity dsDNA antibodies and nephritogenic antibodies, parallel with the decrease in activated T cells and plasma cells.
When renal function and pathology were assessed, KO mice showed significantly decreased proteinuria ( Fig. 2A), glomerulonephritis (Fig. 2B), glomerular activity index score (Fig. 2C), C3, and IgG deposition (Fig. 2, D-F) compared with WT mice. Histopathological analysis revealed decreased infiltration of inflammatory cells in the liver and lungs of KO mice (supplemental Fig. S4D). In spleen and lymph nodes, we observed decreased cellularity and partial restoration of normal splenic architecture in KO mice (supplemental Fig. S4D). Consistent with an improved disease phenotype, KO mice survived ϳ6 weeks longer (28.56 Ϯ 1.31 versus 22.6 Ϯ 1.07 weeks) compared with the WT mice (p Ͻ 0.05) (Fig. 2G). In sum, HDAC9 deficiency protected against systemic autoimmunity as judged by lymphadenopathy, splenomegaly, autoantibody production, proteinuria, renal disease, and mortality rate.
HDAC9 Deficiency Modifies Genes Involved in Follicular and Extrafollicular CD4 ϩ T Effector Cells-Autoantibody production in mouse models of spontaneous autoimmunity depends on both follicular and extrafollicular CD4 ϩ T cells (30 -32). ICOS plays a critical role in Tfh development, germinal center formation, and extrafollicular response in addition to providing the second "co-stimulatory" signal in T cell activation (30). The decreased titers of high affinity dsDNA antibodies in KO mice suggest that a decrease in number or function of Tfh cells assists in B lymphocyte differentiation. Moreover, MRL/lpr mice deficient in ICOS have decreased circulating IgM and IgG3 isotypes similar to our results in KO mice (33,34). These observations suggest a direct link between ICOS and HDAC9 expression and disease progression in MRL/lpr mice.
To determine whether HDAC9 deficiency modulates ICOS expression, we performed quantitative real-time PCR in purified CD4 ϩ T cells. KO mice expressed significantly lower levels of ICOS mRNA compared with WT mice (Fig. 3A). Using flow cytometry analysis, we confirmed that the percentage of CD4 ϩ B220 Ϫ CD44 high cells expressing ICOS in KO mice was significantly decreased compared with WT mice (Fig. 3, B  and C).  Table 2.
P-selectin glycoprotein ligand 1 (PSGL-1 l0 ) CD4 ϩ T cells from MRL/lpr mice express CXCR4, localize to extrafollicular zones, and mediate production of IgG autoantibody through IL-21 (30). Similar to ICOS-deficient MRL/lpr mice, CD44 high CD62L low PSGL-1 l0 CD4 ϩ T cells were also reduced in KO compared with WT mice (Fig. 3, D and E). We also quantified the expression of other follicular and extrafollicular T cell genes including CXCR4, PD1, SLAM, CXCR5, and IL-21 mRNA expression by real time PCR in splenocytes. HDAC9 deficiency resulted in significant decreases in CXCR4 and IL-21, a significant increase in SLAM, and no difference in CXCR5 or PD1 mRNA expression (Fig. 3F). Roquin (Rc3h1) has been recently demonstrated to regulate stability of ICOS mRNA by promoting degradation through microRNA 101 (miR-101) (35,36). The sanroque M199R mutation in Rc3h1 causes T follicular effector cell-dependent, lupus-like autoimmunity via increased ICOS expression in T cells (35). The decreased ICOS expression of CD4 ϩ T cells from KO mice led us to hypothesize that HDAC9 may regulate the expression of roquin and miR-101.
We determined the expression level of roquin and miR-101 by real time PCR. KO HDAC9-deficient mice had significantly increased expression of both roquin (2-fold) and miR-101 (Fig.  3, G and J). Western blot analysis confirmed the overexpression of roquin in KO mice (Fig. 3H). To determine the molecular mechanisms behind the increased roquin expression associated FIGURE 2. HDAC9 deficiency decreases severity of proteinuria, kidney disease, and mortality rate in MRL/lpr mice. A, urinary proteins in MRL/lpr mice were reduced by HDAC9 deficiency. Urinary protein levels were assessed and graded semiquantitatively as described under "Experimental Procedures." B, hematoxylin and eosin-stained sections of kidneys from WT (top) and KO (bottom) mice are shown. Each panel is from a different animal representative of 10 -12 female mice examined from different groups at 20 weeks old. C, the glomerular index score of 20-week-old female WT and KO mice was determined by histological analysis of hematoxylin-and eosin-stained kidney slides according to previously known criteria by a pathologist blinded to groups; *, p Ͻ 0.05; **, p Ͻ 0.01. D, glomerular immune deposits were determined by direct immunofluorescence for complement C3 (top) and IgG (bottom) in frozen kidney sections from WT and KO mice and scored for intensity of staining in blinded fashion (E and F). G, the survival rate was determined between wild WT (n ϭ 22) and KO (n ϭ 20) mice by observing mice up to 40 weeks and sacrificing them when they were moribund.
with HDAC9 deficiency, we performed ChIP assays using antibodies against total acetylated H3 and site-specific Lys-9, Lys-14, and Lys-18-acetylated H3 antibodies followed by real-time PCR using primers for the roquin gene. We selected these sitespecific anti-acetyl antibodies for the ChIP assay to determine whether global increases in these acetyl histone modifications (supplemental Fig. 3, A-C) are localized to gene-specific pro-moter or enhancer regions. HDAC9 deficiency resulted in local accumulation of total acetylated H3 and site-specific Lys-18 acetylation but no change in Lys-9 and Lys-14 acetylation at the roquin promoter compared with WT mice (Fig. 3I).
The transcription factor BCL6 is involved in T follicular cell development and germinal center formation (3). Linterman et al. (37) recently reported that sanroque mice deficient in one

. HDAC9 regulates genes expressed in follicular and extrafollicular T effector cells in MRL/lpr mice.
A, shown is real time PCR analysis of ICOS mRNA in CD4 ϩ T cells from KO compared with WT mice. B, representative flow cytometric plots used for graph C for B220 Ϫ CD4 ϩ CD44 high ICOS ϩ T cells on splenocytes from KO are compared with WT mice. D, representative flow cytometric plots of CD4 ϩ CD44 high CD62L lo PSGL-1 lo T cells for graph E in splenocytes from KO are compared with WT mice. F, real time PCR analysis of mRNA expression of the indicated genes associated with follicular and extrafollicular T cells from KO are compared with WT mice. G, a real time PCR analysis of roquin mRNA in CD4 ϩ T cells from KO is compared with WT mice. H, shown is a Western blot analysis of roquin (from pooled samples of three mice from each group) that demonstrates increased expression in KO mice consistent with roquin mRNA expression; data are representative of two independent experiments. I, ChIP assays were performed in splenocytes from the KO and WT mice. Chromatin was immunoprecipitated with antibodies against total IgG, AcH3, Ac-H3K9, and AcK18. Primers flank the roquin gene, and precipitated DNA was analyzed by PCR. Non-immunoprecipitated samples served as an input control. A representative figure of two independent experiments from pooled samples of three mice in each group in each experiment is shown. J, real time PCR analysis demonstrates increased miRNA-101 in CD4 ϩ T cells from the KO compared with WT mice. miRNA-101 expression was normalized to control miRNA-U6. Real time PCR (K) and Western blot analysis (L) of BCL6 in splenocytes demonstrates decreased expression in KO compared with WT mice. Results represent the mean Ϯ S.E. of six animals per group. *, p Ͻ 0.05; **, p Ͻ 0.01(Student's t test).
allele of BCL6 exhibit a reduced lupus phenotype. To determine whether HDAC9 deficiency modulates BCL6 expression, we measured BCL6 mRNA and protein expression by real time PCR and immunoblot, respectively. BCL6 was significantly decreased at the mRNA and protein levels in splenocytes from KO compared with WT mice (Fig. 3, K and L). Taken together, our data indicate that HDAC9 plays a significant role in modulating genes that induce both follicular T cells and extrafollicular CD4 ϩ T effector cells, which play critical roles in autoantibody production in MRL/lpr mice.
ICOS is required for an effective T cell-dependent immune response (38). To determine whether HDAC9-deficient mice have impaired T cell-dependent immune responses because of ICOS deficiency, we immunized HDAC9-deficient mice (on a C57BL6 background) with a T-dependent antigen (NP-CGG) and two T-independent antigens (NP-LPS and NP-Ficoll). NPspecific IgM levels in HDAC9 Ϫ/Ϫ mice were significantly reduced at 0, 7, 14, and 21 days. IgG2a levels were reduced at day 7, and IgG3 levels were reduced at days 7 and 14 compared with controls, consistent with ICOS deficiency in these mice (supplemental Fig. S6). There were no differences in the levels of Ig subtypes between HDAC9 Ϫ/Ϫ mice and control mice when immunized with T-independent antigens.

HDAC9 Disruption Results in Decreased Th1 Cytokines and Increased Th2 Cytokine Production in Vivo and Increased Th2
Polarization and Decreased T Cell Activation in Vitro-Decreased serum levels of IgG3 with modest increases in IgG1 isotypes in the KO mice support the hypothesis that HDAC9 deficiency may decrease Th1 cytokines and favor Th2 polarization in MRL/lpr mice (29). To explore this hypothesis, we performed real-time PCR in splenocytes and isolated CD4 ϩ T cells. HDAC9 deficiency significantly decreased mRNA levels of IFN-␥ and IL-12 and increased expression of IL-4 and GATA3 but did not change T bet, TGF-␤1, IL-2, or IL-6 mRNA levels in splenocytes from MRL/lpr mice (Table 1). Consistent with mRNA data, KO mice had significantly decreased levels of IL-12 and IFN-␥ but increased IL-4 levels in serum compared with WT mice (Fig. 4A).
To determine whether HDAC9 deficiency promotes an altered cytokine-secreting capacity of Th cells in vitro, we stimulated naïve CD4 ϩ T cells with anti-CD3 and anti-CD28 mAbs for 24 h from both KO and WT mice. Mice with HDAC9-deficient T cells secreted significantly less IL-2 and IL-12 and had mild decreases in IFN-␥ but expressed increased IL-4 secretion ex vivo (Fig. 4, B-D). Next we cultured CD4 ϩ T cells from KO and WT mice in neutral and Th1-or Th2-polarizing media. HDAC9-deficient CD4 ϩ T cells produced more IL-4 mRNA and IL-4-producing Th2 cells under Th2-polarizing conditions, but there was no significant change in the Th1 cells in Th1polarizing conditions (Fig. 4, C-F). Collectively, these results demonstrate the role of HDAC9 on T cell polarization and IL-4 production.
To understand the molecular mechanisms of underlying IL-4 up-regulation, we performed ChIP assays to determine the level of histone acetylation at the IL-4 promoter. HDAC9 deficiency resulted in local accumulation of total acetylated H3 and sitespecific Lys-9, Lys-14, and Lys-18 acetylation at H3 at the IL-4 promoter (Fig. 4G).

HDAC9 Deficiency Up-regulates PPAR-␥ in Spleen and Kidney, with a Concomitant Decrease in Inflammatory Cytokines
and Chemokines-Several elegant studies have demonstrated the role of PPAR-␥ in the suppression of inflammatory genes through ligand-dependent transrepression and direct interference with transcriptional activation (39). PPAR-␥ agonists decrease lupus-related nephritis though decreased IFN-␥ and nitric oxide production in MRL/lpr mice in vivo, which parallels the findings in previous studies of HDAC inhibitors (17,18,40,41). Huang et al. (42) demonstrated that IL-4 induces PPAR-␥ expression in macrophages. HDAC-deficient MRL/lpr mice had increased systemic IL-4 production compared with MRL/ lpr mice. To determine whether increased IL-4 levels correlate with increased PPAR-␥ expression in the KO mice, we measured mRNA and protein levels of PPAR-␥ in splenocytes and kidneys. The mRNA and protein levels of PPAR-␥ was up-regulated in splenocytes and kidneys from KO mice (Fig. 5, A-D).
To determine whether up-regulation of PPAR-␥ was due to a localized increase of histone acetylation at the PPAR-␥ promoter resulting in enhanced transcription, we performed ChIP experiments using antibodies against total and site-specific H3 acetylated antibodies in splenocytes. HDAC9 deficiency resulted in increased levels of total H3 acetylation and sitespecific Lys-9 and Lys-18 acetylation of H3 at the PPAR-␥ promoter compared with controls (Fig. 5E).
We have previously demonstrated that trichostatin A and suberoylanilide hydroxamic acid down-regulate PPAR-␥ target inflammatory chemokine and cytokine genes, including inducible NOS and MCP1 in MRL/lpr and NZB/W F1 mice in vivo (16,18). We next assessed known PPAR-␥ target genes that play a pathogenic role in lupus using quantitative real time PCR assays. CXCR3, CXCR4, CCL3, CXCL9, and IFN-␥ were decreased in splenocytes from KO mice compared with WT mice, whereas the lipid homeostatic genes CD36 and Fab4 (ap2) were significantly increased in vivo (Table 1). Consistent with decreased kidney disease, IL-6, inducible NOS, MMP-9,

DISCUSSION
The results of this study reveal a unique role of HDAC9 in CD4 ϩ T cell plasticity, inflammation, and autoimmunity by selectively modulating gene expression in a tissue-specific context. The major findings were as follows. 1) HDAC9 is overex-pressed in murine and human lupus T cells. 2) HDAC9 regulates site-specific acetylation (Lys-9, Lys-14, and Lys-18) at histone H3 globally in a tissue-specific manner. 3) HDAC9 deficiency results in decreased autoantibody production, immunoproliferation, proteinuria, and kidney disease and increases survival in MRL/lpr mice. 4) HDAC9 deficiency improves the autoimmune phenotype by decreasing follicular and extrafollicular T cell genes and inducing Th2 polarization. 5) HDAC9 deficiency up-regulates roquin and IL-4 but down-regulates Bcl6, which are key genes in Tfh and Th2 differentiation. 6) HDAC9 down-regulates ICOS, which is essential for follicular and extrafollicular response by up-regulating microRNA 101 via roquin. 7) HDAC9 deficiency leads to decreased expression of cytokines and chemokines by up-regulating PPAR-␥ through accumulation of acetylated histone H3 at the PPAR-␥ promoter.
This study is the first to demonstrate that HDAC9 selectively regulates H3 acetylation at lysine sites Lys-9, Lys-14, and Lys-18 Ac in mammalian systems. Moreover, we report the novel findings that PPAR-␥, roquin, and IL-4 are regulated by site-specific acetylation of H3 at their promoter sites. Wakabayashi et al. (26) demonstrated that H4K20 monomethylation at the PPAR-␥ promoter increased HDAC9 gene expression. In this study, we demonstrate that the site-specific acetylation of Lys-9, Lys-14, and Lys-18 of histone H3 at the PPAR-␥ promoter are additional epigenetic mechanisms that can regulate PPAR-␥ transcription. Several gene-specific knock-out studies in mice have shown different ways to elicit autoimmunity, per-

. Decreased inflammatory cytokines and chemokine production in HDAC9-deficient MRL/lpr mice is due to increased PPAR-␥ expression.
Real time PCR analysis of PPAR-␥ mRNA expression in splenocytes (A) and kidney (C) from WT and KO mice is shown. Results represent the mean Ϯ S.E. of six animals per group. *, p Ͻ 0.05; **, p Ͻ 0.01 (Student's t test). Shown is a Western blot analysis of PPAR-␥ protein expression in splenocytes (B) and kidney (D) from WT and KO mice. E, ChIP-quantitative PCR analyses of H3Ac, H3K9Ac, and H3K18Ac modifications on PPAR-␥ promoter are shown. ChIP assays were performed in splenocytes from KO and WT mice. Chromatin was immunoprecipitated with antibodies against total IgG, AcH3, Ac-H3K9, and AcK18. Quantitative ChIP was performed using SYBR Green dye using primers flanking PPAR-␥ promoter. The data are the averages of three replicates, and error bars represent the S.E. Representative results of two independent experiments from pooled samples of three mice in each group are shown. Tables 1 and 2 summarize the real time PCR data of PPAR-␥ dependent genes in splenocytes and kidneys in KO and WT mice. Results represent the mean Ϯ S.E. of 6 animals per group. *, p Ͻ 0.05; **, p Ͻ 0.01 (Student's t test).  AUGUST 19, 2011 • VOLUME 286 • NUMBER 33

HDAC9 and Autoimmunity
haps by the creation of specific pathways and networks of aberrant gene expression. Here, using a genetic loss-of-function approach, we describe a molecular pathway that protects autoimmunity by involving HDAC9, roquin, ICOS, IL-4, and PPAR-␥, which generates an epigenetic switch of CD4 ϩ T cells toward decreased Th1, T follicular cell, and ultimately T celldependent autoantibody production by B cells. The change in the epigenetic landscape generated by decreased HDAC9 in different CD4 ϩ T effector cells creates an anti-inflammatory loop by decreasing production of cytokines and chemokines that limit infiltration of inflammatory cells to target organs and autoantibody deposition in kidney tissues. The changes in gene expression by HDAC9 occur through multiple mechanisms via protein-protein interaction, non-histone protein acetylation, and histone acetylation, although these pathways are not necessarily mutually exclusive (43). A number of cellular proteins, including MEF-2, Foxp3, and ankyrin, repeat the proteins ANKRA and RFXANK, which have been reported to be associated with HDAC9 in vitro and in vivo (22,44,45). Several elegant studies have demonstrated that HDAC9 deacetylases Foxp3, USF-1, and ATDC proteins (8,43,46). Although IL-4, roquin, and PPAR-␥ gene expression are regulated via increased accumulation of acetylated histones H3 at their promoter sites, we cannot exclude the other potential mechanisms of protein-protein interactions and acetylation of transcription factors by HDAC9 deletion. Inhibition of several inflammatory cytokines and chemokines, including IFN-␥, IL-12, inducible NOS, MCP1, CXCR3, CXCR4, CXCR5, CXCR6, and CXCR9, are most likely due to up-regulation of PPAR-␥. The mechanisms by which HDAC9 deficiency reduces BCL6 expression and up-regulates miR-101 and other key genes remains to be established.
Although our study demonstrates several potential mechanisms to explain how HDAC9 deficiency protects autoimmunity, there are several limitations in this study. Our interpretation depends primarily upon the genetic knockdown by homologous recombination approach. The role of HDAC9 in follicular T cell development and Th2 polarization needs to be further confirmed by transgenic overexpression studies and siRNA approaches. Moreover, the role of HDAC9 in other immune cells, particularly regulatory T cells, macrophage, and B cells in systemic autoimmunity, merits further investigation. The pioneering study by Tao et al. (10) reported that increased expression of HDAC9 in T-reg cells decreases their suppressive capacity. Moreover, HDAC9 Ϫ/Ϫ mice had increased numbers of T reg cells with enhanced suppressive activity compared with wild type mice, and T reg cells derived from HDAC9 Ϫ/Ϫ mice show enhanced expression of Foxp3, CTLA4, and GITR (10). HDAC9 Ϫ/Ϫ mice are protected from colitis due to up-regulation of HSP70 in regulatory T cells (12). The potential beneficial effect of HDAC9 deficiency in systemic autoimmunity may also involve induction of regulatory T cells and needs to be studied (47). Although our immunization studies support the hypothesis that the beneficial effect of HDAC9 deletion is independent of B cells (supplemental Fig. S6), the role of HDAC9 in B cell development and autoantibody production remains to be fully investigated. These studies are currently under way in our laboratory.
Histone deacetylase inhibitors have been considered for the treatment of several autoimmune diseases, including rheumatoid arthritis and systemic lupus erythematosus, graft versus host disease, and colitis, but clinical efficacy remains to be established (18, 48 -50). Our present study provides a mechanism-based understanding of the role of HDAC9 in T cell-mediated autoimmunity and provides a rationale for the development of HDAC9 isoform-specific inhibitors for these conditions.