HDAC10 deletion promotes Foxp3+ T-regulatory cell function

Foxp3+ T-regulatory (Treg) cells are capable of suppressing immune responses. Lysine acetylation is a key mechanism of post-translational control of various transcription factors, and when acetylated, Foxp3 is stabilized and transcriptionally active. Therefore, understanding the roles of various histone/protein deacetylases (HDAC) are key to promoting Treg-based immunotherapy. Several of the 11 classical HDAC enzymes are necessary for optimal Treg function while others are dispensable. We investigated the effect of HDAC10 in murine Tregs. HDAC10 deletion had no adverse effect on the health of mice, which retained normal CD4+ and CD8+ T cell function. However, HDAC10−/− Treg exhibited increased suppressive function in vitro and in vivo. C57BL/6 Rag1−/− mice adoptively transferred with HDAC10−/− but not wild Treg, were protected from developing colitis. HDAC10−/− but not wild-type mice receiving fully MHC-mismatched cardiac transplants became tolerant and showed long-term allograft survival (>100 d). We conclude that targeting of HDAC10 may be of therapeutic value for inflammatory disorders including colitis and also for transplantation.


Results
HDAC10 deletion does not affect conventional T cell function. Mice lacking HDAC10 (Fig. 1A) were born at the expected Mendelian ratios and had no signs of disease or abnormal development up to one year of observation. HDAC10 −/− lymphatic tissues had similar CD4 + and CD8 + , as well as CD44 hi CD62L lo effector/ memory cells (Fig. 1B-F), as well as CD25 + and Foxp3 + T cell populations (Fig. 1G-I) compared to wild type (WT) mice. Other immune cell populations, such as B-cells (B220, CD19), natural killer cells (CD49b), and macrophages were comparable (data not shown). Next, we compared HDAC10 −/− and WT CD4 + T cell subset polarization. We observed very similar Th1 and Th17 formation ( Fig. 2A,B). HDAC10 −/− CD4 + CD25 − T cells showed a trend towards reduced Foxp3 + iTreg polarization upon CD3ε/CD28 mAb co-stimulation in the presence of TGF-β and IL-2 (Fig. 2C,D). To evaluate Tconv and CD8 + T cell function in vivo, we adoptively transferred CFSE-labeled WT and HDAC10 −/− splenocytes of C57BL/6 origin (H-2 b ) into C57BL/6-DBA2 (H-2 bd ) mice (Parent-to-F1 assay, Fig. 2E). After three days, we obtained splenocytes of the recipients and identified the adoptively transferred cells through the absence of the H-2 d alloantigen (Fig. 2F). We observed, that both WT and HDAC10-deficient CD4 + and CD8 + T cells proliferated equally well and produced IL-2 and IFN-γ in vivo ( Fig. 2F-H). In summary, deletion of HDAC10 produced viable mice without apparent illness and with functional CD4 + and CD8 + T cells.

HDAC10 deletion improves Treg function.
In previous studies of the other HDAC class IIb isoform, HDAC6, we showed that HDAC6 deletion or its pharmacologic inhibition promoted Foxp3 acetylation and increased Treg suppressive function in vitro and in vivo 8,10 . These findings led us to question if HDAC10 −/− Treg might exhibit a similar phenotype? Indeed, compared to WT Tregs, HDAC10 −/− Treg showed stronger suppressive function against effector T cells proliferating in vitro (Fig. 3A,B), mirroring the phenotype of HDAC6-deficient Tregs. This observation led us to consider how targeting of HDAC10 might enhance the Treg suppression, including whether increased Foxp3 acetylation was involved, as with certain other HDAC phenotypes, including HDAC6, HDAC9, and Sirtuin-1 6,10 . Of note, in addition to the increased HDAC10 −/− Treg function, we also observed, that if the cells used to co-stimulate effector T cells in the Treg suppressive function assays (an irradiated mixed splenocyte fraction from which CD90.2 + T cell have been removed) originated from HDAC10 −/− rather than WT mice, that effector T cell proliferation was higher (Fig. 3C,D).
HDAC10 co-precipitates with Foxp3. To detect differences in gene expression between WT and HDAC10 −/− Treg, we isolated RNA from naïve CD4 + CD25 + Tregs and conducted whole-mouse-genome oligoarrays studies (GeneChip ™ Mouse Gene 2.0 ST, Thermo Fisher Scientific). The effect of HDAC10 deletion on Treg gene expression was limited. Under non-stringent statistical criteria (Student t-test FDR <0.1, 1.5-fold differential expression), <1% of genes were differentially expressed (Fig. 4A, Supplementary Excel File). We noted that Granzyme-b mRNA was increased in HDAC10 −/− Treg (Fig. 4B), which was confirmed by qPCR (Fig. 4C). Foxp3 mRNA showed a trend to higher expression but missed statistical significance, as did other Treg-associated genes such as Ctla4, Tgfb1 and Il10. (Fig. 4C). Next, we examined if HDAC10 and Foxp3 interacted. We transfected 293 T cells with Foxp3 and HDAC10, and observed binding, regardless of whether the complex was immunoprecipitated by HDAC10 and blotted by Foxp3, or vice versa (Fig. 4D,E). The immunoprecipitation between Foxp3 and HDAC10 led us to hypothesize that HDAC10 may deacetylate Foxp3. We observed that Foxp3-K31 acetylation, mediated by co-transfection of 293 T cells with Foxp3 and the histone/protein acetyl transferase p300 23 , was decreased with HDAC10 co-transfection (Fig. 4F). However, we could not discern if this was the result of HDAC10 deacetylating Foxp3 or destabilizing p300 expression. 293 T cell overexpression systems are helpful for WT or HDAC10 −/− splenocytes were CFSE-labeled, and adoptively transferred (i.v.) into C57BL/6-DBA/2 (H-2 bd ) recipients. After three days, the adoptively transferred cells were identified by their absence of H-2 d MHC. CD8 + and CD4 + T cells lacking HDAC10 proliferated equally well compared to WT in vivo. (F) Gating strategy. (G,H) Pooled data of CD4 + and CD8 + T cell proliferation (G) and cytokine production after PMA/ionomycin activation (H). Data pooled from three independent experiments (Mann-Whitney test). Abbreviations: n.s., not significant. Data shown as median ± IQR. examining biochemical interactions but have limitations. Increased Foxp3 acetylation has been observed to lead to better Foxp3 preservation (higher total Foxp3) and DNA binding by us and other groups 3,4,24 . To evaluate if loss of HDAC10 leads to more Foxp3, we conducted Western blotting. We did not observe a significant change in total Foxp3 protein expression (Fig. 4G,H). Taken together, Foxp3 and HDAC10 expressed in 293 T cells were found to co-precipitate, but we could not confirm an increase in Foxp3 protein acetylation, and it remains unclear if this step is mechanistically important to the increased HDAC10 −/− suppressive function.
HDAC10 deletion has no effects on Treg metabolism. HDAC10 has been reported to be highly expressed in lung cancer tissues where it is associated with a poor prognosis 25 . HDAC10 also promoted AKT phosphorylation in lung cancer cell lines 21 , which could affect metabolism by switching to a glycolytic phenotype 26 , and potentially have detrimental effects on Treg function 27 . We evaluated HDAC10 −/− Tconv and Treg bioenergetic function with and without CD3/CD28 co-stimulation. We did not find a difference in basal glycolytic activity (Fig. 5A) or in Tconv oxygen consumption (Fig. 5B,C). We noticed a trend in HDAC10 −/− towards an increased uncoupled oxygen consumption rate (Fig. 5D,E), not unlike that seen in HDAC6 deficient Tregs 28 . Overall, the differences are small and unlikely to be a major contributor to HDAC10 −/− Treg function.
Tregs lacking HDAC10 alleviates autoimmune colitis. We questioned if the increased HDAC10 −/− Treg suppressive function observed in our in vitro studies translated into in vivo models of autoimmune disease and transplantation. We evaluated two autoimmune colitis models, rescue and prevention. In the colitis rescue model, B6/Rag1 −/− mice were adoptively transferred i.p. with 10 6 WT Tconv, and then observed for weight loss and www.nature.com/scientificreports www.nature.com/scientificreports/ clinically signs of colitis. By day 33, the mice had developed colitis and weight loss, and were randomly assigned to receive 5 × 10 5 Treg i.p. from either WT or HDAC10 −/− donor mice for colitis rescue. B6/Rag1 −/− mice receiving HDAC10 −/− Treg showed a trend to improved weight outcomes (Fig. 6A), reduced splenocyte counts and preservation of colon length (Fig. 6B,C), however, the differences were not statistically significant. In the colitis prevention model, B6/Rag1 −/− mice were adoptively transferred i.p. with 1 × 10 6 WT CD4 + CD25 − Tconv together with 5 × 10 5 CD4 + CD25 + Treg from either WT or HDAC10 −/− donors simultaneously. Here, the differences were stronger. B6/ Rag1 −/− mice receiving HDAC10 −/− Treg showed less weight loss (Fig. 6D), smaller spleens and less splenocytes (Fig. 6E,F), as well as less colon shortening and thickening (Fig. 5G,H). This was matched by less inflammation upon histologic examination, whereby colon samples from B6/Rag1 −/− mice receiving WT, but not HDAC10 −/− Treg showed more prominent colitis including transmural inflammation, ulceration, reactive epithelial changes and crypt abscesses (Fig. 7A,B). These observations were quantified by blinded analysis (Fig. 7C-H). In conclusion, our data show that HDAC10 −/− Treg can alleviate autoimmune colitis.  , and monitored allograft survival daily by palpation of ventricular contractions. In the absence of further treatment, both WT and HDAC10 −/− recipients rejected their cardiac allografts rapidly (Fig. 8A). However, addition of subtherapeutic injections of rapamycin (0.1 mg kg −1 d −1 i.p. for 14 days post-transplant) elicited a strong difference between WT and HDAC10 −/− recipients, where the latter did not reject their cardiac allografts for >100 days post-transplant (Fig. 8A). The cardiac allografts recovered from HDAC10 −/− did not only maintain their function >100 days (p < 0.01), but, unlike WT mice at 2 weeks post-transplant, also showed no signs of rejection and maintained normal cardiac histologic architecture (Fig. 8B). In conclusion, in contrast to WT recipients, HDAC10 −/− mice receiving a brief course of low-dose rapamycin maintained their fully MHC-mismatched cardiac allografts indefinitely.

Discussion
Our studies of the HDAC10 immune phenotype reveal an overall immunosuppressive effect with HDAC10 deletion. We observed that Foxp3 + Treg cells lacking HDAC10 had increased suppressive function in vitro, and HDAC10 −/− mice accepted MHC-mismatched cardiac allografts long-term when treated with 14 days of low-dose rapamycin, while B6/Rag1 −/− receiving HDAC10 −/− Treg showed increased resistance to development of autoimmune colitis in vivo. However, it is important to also consider additional subsets of immune cells that may be altered through HDAC10 deletion. Our in vitro studies had indicated that irradiated CD90.2 − splenocytes, including antigen presenting cells that provide co-stimulatory signals in the Treg suppression assays, showed decreased stimulatory effects on effector T cells when lacking HDAC10. This finding is consistent with prior www.nature.com/scientificreports www.nature.com/scientificreports/ reports on HDAC10 deletion on antigen-presenting cells 22 . It is possible that decreased co-stimulatory signals in HDAC10 −/− transplant recipients could contribute to the observed prolonged allograft survival. Furthermore, differential effects of rapamycin on HDAC10-deficient versus wild type mice need to be considered. That said, Treg are likely to play a somewhat important role, considering that the origin of Tregs (HDAC10 −/− vs WT) was the only variable in the colitis prevention model.
Our data are congruent with the effects of genetic or pharmacologic HDAC6 targeting, as well global pharmacologic class IIb HDAC inhibition, which is reassuring for the further development of HDAC isoform-selective immunosuppressive therapeutics. Indeed, two recent reports indicate that development of HDAC10-selective inhibitors is feasible based upon targeting of the interaction of a hydrogen-bond between a cap group nitrogen and the gatekeeper residue Glu272 that is responsible for potent HDAC10 binding 29,30 . The effects of such inhibitors on immune functions will need to be carefully characterized once they become available, since the screening assays used for their development involved FRET-based displacement of HDAC10-bound ligands rather than enzymatic inhibition. Such inhibitors may potentially act on the lysine deacetylase activity of HDAC10 at least in some contexts, whereas in others, effects on the recently identified polyamine deacetylase activity of HDAC10 may be dominant 31 .
Beyond targeting class IIb HDACs for therapeutic immunosuppression, our data also indicate a note of caution when considering HDAC10 as a therapeutic target in cancer, which has been suggested by several groups. www.nature.com/scientificreports www.nature.com/scientificreports/ Increased HDAC10 expression has been associated with several malignancies, including lung and gastric cancer and neuroblastoma, and was been linked with poor outcomes 17,25,[32][33][34] . A number of mechanisms have been proposed. E.g. Yang et al. reported that HDAC10 promotes AKT phosphorylation in lung cancer cells 21 , and AKT phosphorylation is known to propagate cell growth, proliferation and survival 35 . However, if HDAC10 inhibition improves Treg immunosuppressive capacity, this could have counterproductive effects, given the importance of Tregs in cancer pathology 36 . Beyond immunosuppressive concerns for anti-tumor HDAC10 targeting, HDAC10 has also been reported to impair matrix metalloproteinase-2 and −9 expression 37 , which were important in metastasis formation 38 .
In conclusion, HDAC10 deletion leads to stronger Foxp3 + Treg cells with increased suppressive function in vivo. Pharmacologic targeting of HDAC10 and/or class IIb HDACs in general may be a promising strategy to augment immunosuppressive therapy. Cell culture media and cell culture conditions. For standard cell culture medium, we used RPMI 1640 medium supplemented with 10% fetal bovine serum (FBS), penicillin (100 U × mL −1 ), streptomycin (100 µg × mL −1 ), and 55 nM β 2 -mercaptoethanol. All bioenergetic measurements used Agilent Seahorse XF base medium (cat. #102353-100) from Aligent Technologies (Santa Clara, CA) plus 10 mM D-glucose, 1 mM Na pyruvate, and 2 mM L-glutamine.

Methods
Antibodies, plasmids and reagents. We purchased CD4 (BD Bioscience, Pacific blue, clone RM4- Bioenergetic measurements. We measured bioenergetic functions using XF24 and XF96 analyzers (Seahorse Biosciences, North Billerica, MA). Plates were coated using Cell-Tak (BD Biosciences) 11 . Isolated T cells were plated in unbuffered XF Assay Media, and then incubated for 30-60 min at 37 °C without CO 2 . For our T cell studies, we used 1 × 10 6 cells per well for XF24, and 2 × 10 5 cells per well for XF96 assays. To enhance cell adherence, plates were spun at room temperature for 5 min at 400 g. Three baseline measurements of OCR and ECAR were taken and cells were then exposed sequentially to oligomycin A, cyanide-4-[trifluoromethoxy]phenylhydrazone (FCCP) and rotenone and/or antimycin A. We used 1.25 µM oligomycin, 0.5 µM FCCP and either 1 µM rotenone followed by 1.8 µM antimycin A in XF24, or 1 µM rotenone & antimycin A combined in XF96 experiments. Three readings were taken after each sequential injection. Instrumental background was measured in separate control wells using the same conditions without biological material. Seahorse reagents were purchased from Agilent Technologies (Wilmington, DE). Data were analyzed using Wave (Agilent), Excel and Prism.
Cardiac allografting. We transplanted BALB/c hearts (H-2 d ) into C57BL/6 recipients (H-2 b ); donor and recipient mice were female and aged 8-12 weeks. We chose female mice to minimize fighting injuries and age 8-12 weeks to enable sufficiently large blood vessels to enable microsurgeries. Some recipients received low-dose rapamycin (0.1 mg kg −1 day −1 ) for two weeks, as indicated. Allograft survival was assessed by palpation, and confirmed by histology after 100 days.
Cell isolation and flow cytometry. Spleen and peripheral lymph nodes were harvested and processed to single cell suspensions of lymphocytes. Red blood cells were removed with hypotonic lysis. We used magnetic beads (Miltenyi Biotec, San Diego, CA) for isolation of Tconv (CD4 + CD25 − ), Treg (CD4 + CD25 + ), and antigen presenting cells (CD90.2 − ). Cells of interest were analyzed flow cytometry. All flow cytometry data was captured using Cyan (Dako) as well as Cytoflex (Beckman Coulter, Brea, CA) and analyzed using the FlowJo 10.2 software. We used phosphate buffered saline with 2% FBS as flow buffer in all experiments.

Histology.
Cardiac allografts and colon samples were fixed in 10% neutral buffered formalin, routinely processed and embedded in paraffin. Histologic sections for light microscopy were cut to a thickness of 4 µm and stained with hematoxylin and eosin (H&E) and were reviewed by a pathologist (W.W.H.) blinded to treatment conditions. Histologic findings were characterized using a previously reported scoring system 40 : (1) degree of lamina propria inflammation graded 0-3; (2) degree of mucin depletion as evidenced by loss of goblet cells graded 0-2; (3) reactive epithelial changes (nuclear hyperchromatism, random nuclear atypia, increased mitotic activity) graded 0-3; (4) number of intraepithelial lymphocytes per high power field within crypts graded 0-3; (5) degree of crypt architectural distortion graded 0-3; (6) degree of inflammatory activity (infiltration of neutrophils within lamina propria and crypt epithelium, "cryptitis") graded 0-2; (7) degree of transmural inflammation graded 0-2; and (8) degree of mucosal surface erosion up to total surface ulceration graded 0-2. The total histopathologic score (0-21) was determined from the sum of the scores for each parameter to reflect the overall degree of inflammation within each specimen.
Immunoblotting. Immunoblotting was performed as previously reported 10 . Purified cells of interested were lysed in radioimmunoprecipitation assay (RIPA) buffer with Halt protease inhibitor (Thermo Fisher Scientific).
Protein concentration was determined by photometry (iMark ™ Microplate Absorbance Reader, Bio-Rad) using bovine serum albumin as standard (Cat. #500-0207, Bio-Rad Laboratories, Hercules, CA). Samples were mixed with Laemmli sample buffer containing 2-mercaptoethanol (Bio-Rad Laboratories, Hercules, CA), and loaded onto Mini-PROTEAN TGX TM 4 to 15% gradient gels (Bio-Rad) and underwent electrophoretic separation. Proteins were then transferred to PolyScreen PVDF Hybridization Transfer Membranes (PerkinElmer, Waltham, MA). Membranes were cut according to the molecular weights of the proteins of interest (as indicated with Precision Plus Protein Dual Color, Bio-Rad) and incubated with primary and horseradish peroxidase (HRP)conjugated secondary antibodies. We used Super Signal West Pico chemiluminescent substrate (Thermo Fisher Scientific) and X-OMAT Blue XB Film (Kodak, Rochester, NY) or the ChemiDoc TM imaging system (Bio-Rad), using Image Lab TM software (Bio-Rad) for image file export. Images were processed with Adobe Photoshop Creative Cloud (grayscale conversion and auto contrast function). Densitometric analysis was performed using ImageJ64 version 1.45 S (https://imagej.nih.gov/ij/). RNA extraction and quantitative polymerase chain reaction. RNA was extracted using RNeasy kits (Qiagen, Hilden, Germany), and RNA integrity and quantity were analyzed by photometry (Nanodrop 2000, Thermo Fischer). Reverse transcription and qPCR were performed as reported [9][10][11] . Isolated RNA was reverse transcribed to cDNA with random hexamers and amplified (PTC-200; MJ Research). Primer sequences for target genes were used for quantitative PCR amplification of total cDNA. All primers were purchased from Applied Biosystems. Differences in cDNA input were corrected by normalizing signals obtained with specific primers for 18 S rRNA. Relative quantitation of target cDNA was determined by the formula 2 −ΔCT , with ΔCT denoting fold increases above the set control value (resting Tconv). Data was analyzed using StepOnePlus TM (Applied Biosystems), Excel, and Prism.
Microarray. WT and HDAC10 −/− Treg were isolated and RNA extracted as noted above. Microarray experiments were performed using whole-mouse-genome oligoarrays (GeneChip ™ Mouse Gene 2.0 ST Array), and array data were analyzed using Transcriptome Analysis Console 4.0 software (ThermoFisher Scientific). Array data were subjected to robust multiarray average normalization. To assess differential gene expression, fold changes of up-and downregulated genes were calculated, and significance assessed via Student t-test. Data with a false discovery rate <0.1 and >1.5-fold differential expression were included in the analysis.
Quantification and Statistical analyses. Statistical analysis was conducted using GraphPad Prism 8 soft- ware. All data were tested for normal Gaussian distribution of variables using the D' Agostino-Pearson normality test. All normally distributed data were displayed as means ± standard error of the mean (SEM) unless otherwise