Context dependent effects of ascorbic acid treatment in TET2 mutant myeloid neoplasia

Loss-of-function TET2 mutations (TET2MT) are common in myeloid neoplasia. TET2, a DNA dioxygenase, requires 2-oxoglutarate and Fe(II) to oxidize 5-methylcytosine. TET2MT thus result in hypermethylation and transcriptional repression. Ascorbic acid (AA) increases dioxygenase activity by facilitating Fe(III)/Fe(II) redox reaction and may alleviate some biological consequences of TET2MT by restoring dioxygenase activity. Here, we report the utility of AA in the prevention of TET2MT myeloid neoplasia (MN), clarify the mechanistic underpinning of the TET2-AA interactions, and demonstrate that the ability of AA to restore TET2 activity in cells depends on N- and C-terminal lysine acetylation and nature of TET2MT. Consequently, pharmacologic modulation of acetyltransferases and histone deacetylases may regulate TET dioxygenase-dependent AA effects. Thus, our study highlights the contribution of factors that may enhance or attenuate AA effects on TET2 and provides a rationale for novel therapeutic approaches including combinations of AA with class I/II HDAC inhibitor or sirtuin activators in TET2MT leukemia. Using TET2- and ascorbic acid deficient model systems Guan et al show that long term treatment with ascorbic acid delays myeloid neoplasia in mice and reveal a complex interplay of post-translational modification of lysine residues that modulate TET2 activity in neoplastic evolution.

TET2 mutations (TET2 MT ) are common in myeloid neoplasia (MN), particularly chronic myelomonocytic leukemia and myelodysplastic syndrome (MDS) 1,8,9 , and in T/B -cell lymphomas 10 . TET2 MT are also found in the healthy elderly individuals with clonal hematopoiesis of indeterminate potential (CHIP) [11][12][13][14][15][16] . TET2 deficiency alone produces an initially benign phenotype characterized by a long disease latency and incomplete penetrance 1,9,15 . However, TET2 MT in CHIP suggest that they are often a first step toward progression to MN 17 . The incidence of both CHIP and TET2 MT neoplasia increases with age: up to 60% of MDS in octogenarian harbor TET2 MT1, 9,18 . Tet2 −/− and Tet2 +/− mice develop an initially mild myeloproliferative syndrome 19 . Acceleration of this process by radiation suggests that additional genetic hits may lead to a higher mutational burden enabled by a primary TET2 lesion 20 .
Ascorbic acid (AA) enhances the activity of TET2 likely by reducing catalytic site Fe(III) to Fe(II) 21,22 . Simultaneous 2oxoglutarate decarboxylation provides two additional electrons needed for dioxygenase activity. AA may thus improve the function of intact TET2 in cases with heterozygous TET2 MT , or increase compensatory enzymatic activity of TET1/3 in cases with biallelic TET2 inactivation. Such effects may restore/improve hydroxymethylation, and potentially reverse the epigenetic consequences caused by TET2 deficiency. In agreement with these theoretical predictions, recent reports have demonstrated the biologic activity of AA on TET2 MT hematopoietic cells in vitro, in Tet2 MT -mediated leukemia models [23][24][25][26] , and in melanoma 27 . Interestingly, patients with leukemia have lower serum AA concentrations compared to age-matched controls 28 , and their AA levels continue to decline with age 29,30 . However, this agedependence of serum AA levels have not been observed in healthy individuals 31 . The effect of AA in preventing leukemogenesis was recently reported 23,26 ; these studies relied heavily on bone marrow (BM) transplantation models, and not enough is known about the effect of AA on genetic models of the evolution of MN. A recent case report 32 did suggest that an acute supraphysiological dose of AA may benefit TET2 MT AML as a single agent. However, it remained unclear what is the contribution of TET dioxygenase activation and what are the TET2-dioxygenase independent effects of AA. In addition, other factors may affect TET2 activity and therefore either enhance or attenuate AA effects. For instance, post-translational context-dependent acetylation and deacetylation of TET2 lysine residues may enhance or reduce TET2 stability and activity 33,34 . Acetylation of TET2 Nterminal lysine residues prevents proteasomal degradation 34 . In contrast, TET2 catalytic domain lysine residues deacetylation by sirtuins may increase its activity.
In this study, we investigated the long term impact of AA treatment in the prevention of MN evolution in murine model and the mechanistic effects of AA on TET2 mutant human myeloid cells derived from the MN patients. In addition, we demonstrate that the catalytic domain lysine acetylation and missense mutations that closely mimics lysine acetylation have significant effect on TET dioxygenase activity. The loss of function in TET2 caused by catalytic domain lysine acetylation or missense mutation cannot be restored by AA treatment. However, addition of chemical probes regulating lysine acetylation amplifies AA mediated TET-dioxygenase activity in leukemia cells. Our study further rationalizes the targeted therapeutic application of AA alone or in combination with other agents in TET2 MT myeloid leukemia.

Results
Long-term oral AA treatment slows myeloproliferation in Tet2 +/− mice. In vivo effects of AA in preventing leukemogenesis were established in transplant models of Tet2 knockout leukemia 23,26 , but remained unexplored in a direct genetic model of Tet2-deficiency. We therefore, performed several key experiments and probed the effects of AA treatment in Tet2 mt murine model. We cultured mononuclear cells purified from BMs and spleens of Tet2 +/+ , Tet2 +/− and Tet2 −/− mice in the presence or absence of AA and measured TDOP levels. AA increased TET-activity in mononuclear cells derived from mice with the 3 genotypic configurations as reflected in increased TDOP levels (Supplementary Figs. 1a-e and Supplementary Data 1) and decreased colony numbers (Supplementary Fig. 1f). To further explore the in vivo therapeutic efficacy of long-term AA supplementation in preventing disease evolution we treated Tet2 +/− mice (3.3 g/L, AA drinking water) for one year and analyzed the disease progression. Control Tet2 +/− mice receiving water, spontaneously developed splenomegaly and increased proportion of Gr-1 + /Mac-1 + myeloid lineage cells in the peripheral blood (PB), spleen and BM. This development was partially reversed with AA supplement (Fig. 1a-e and Supplementary Fig. 1g) as observed in lower spleen weight (Fig. 1c), monocyte counts (Fig. 1b) and myeloid lineage proportions (Fig. 1d-e). There were no significant effects of AA treatment on total counts of WBC, RBC, and neutrophils in PB or LSK/LK and Gr-1+/Mac-1+ myeloid lineage cell proportions in BM ( Supplementary Fig. 1g). We also exposed both cohorts of WT and Tet2 +/mice to sublethal ionizing radiation (IR), and maintained them with or without 3.3 g/L oral AA supplementation in drinking water. IR exposure had no effect on 1-year overall survival (OS) to WT mice on the other hand Tet2 +/− mice had only a 40% 1-year OS (Fig. 2a, b). Oral AA treatment improved the 1-year OS of IR exposed Tet2 +/− mice to 70%, and reduced BM myeloid-lineage cells fraction by~20% in average, compared to the untreated IR-Tet2 +/− mice (Fig. 2c).
To mitigate the impact of mice producing endogenous AA via L-gulonolactone oxidase (Gulo), we crossed homozygous Gulo −/− mice with Tet2 knockout mice to generate hybrid Tet2 +/− ;Gulo −/− mice (Fig. 2d). Gulo −/− mice are unable to synthesize AA, which is essential for the survival of mice 35 . Therefore, in our experiment, we supplied the mice with minimal sustenance dose of AA (0.033 g/L) in drinking water. AA treatment significantly increased 5hmC/5mC, 5fC/5mC, and 5caC/5mC ( Fig. 2e-g). We observed a dose dependent increase in the survival of Tet2 +/− ;Gulo −/− mice upon AA treatment. The median survival increased 36 days for 0.33 g/L and 116 days for 3.33 g/L AA supplementation compared to the minimal dose of AA supplementation (0.033 g/L) required for the sustenance of these mice, however, this effect does not achieve statistical significance in our experimental settings (Fig. 2h, i).
Impact of AA treatment on TET2 activation and proliferation. Partial loss of TET activity due to hypomorphic TET2 MT drives myeloproliferation in patients, and restoring TET2 function by overexpression of TET2 or by AA treatment should inhibit this proliferation. We next studied the ability of AA to increase TET2 activity in experimentally-generated TET2-overexpressing (TET2 OE ) and knockdown (TET2 KD ) cell lines. For this purpose, we ectopically expressed TET2 into both TET2 WT (MEG-01 and CMK) and TET2 MT (HEL) leukemia cell lines (Supplementary Data 2). TET2 OE cells have significantly higher TET-activity (up to eightfold) compared to the vector control cells ( Fig. 3a and Supplementary Fig. 2a). TET2 OE cells grew 2-4-fold slower than control cells correlating with increased TDOPs (Fig. 3a-d and Supplementary Fig. 2a). Similar growth perturbations were observed with AA treatment for control cells (Fig. 3b-d). In contrast, transduction of TET2 shRNA consistently reduced TET2 mRNA levels by 80-90% in K562 and MOLM13 cells and were associated with an approximately twofold decrease in 5hmC/5mC and 5fC/5mC (Fig. 3e, f and Supplementary Fig. 2b-c). Treatment of these cells with pharmacologic doses of AA increased 5hmC/ 5mC and 5fC/5mC ratios in both TET2 WT (3-5-fold) and TET2 KD (3-4-fold) (Fig. 3e, f and Supplementary Figs. 2b-c). AA treatment significantly reduced the growth of TET2 KD cells as well as the parental cells. However, the AA impact on the growth of TET2 KD cells were more pronounced (Fig. 3e-h and Supplementary Figs. 2d-e). Colony forming cultures were performed with purified human CD34 + cells transduced with TET2 shRNA or scrambled control. Following transduction, CD34 + cells were grown in methylcellulose media with or without AA supplementation. While TET2 shRNA increased colony formation, AA treatment reduced the growth of both control and TET2 KD CD34 + cells (Fig. 3i, j). We also studied AA-mediated up modulation of TET2 activity in primary BM derived mononuclear cells isolated from MN patients with TET2 WT or TET2 MT (Supplementary Data 3). In contrast to the genetically engineered mouse cells and human leukemic cell lines, AA only modestly increased TDOP levels in TET2 WT while TET2 MT showed mixed results. Even 250 µM AA treatment did not affect TDOP levels in a significant manner. Overall TET activity as reflected in 5hmC/ 5mC among TET2 MT cells were lower than in TET2 WT control (Fig. 3k). AA treatment had statistically insignificant effect on 5hmC/5mC on TET2 MT patients but the modest effect observed in TET2 WT patients were statistically significant as assessed using quantitative mass spectrometry (Fig. 3k, l). To better understand the mechanism of AA mediated TET2 activation and its utility in TET2 MT MN patients we further analyzed AA's effect on the dioxygenase activity of TET2.
AA binds to TET2 CD and increases its activity. To better understand the mechanism of AA-mediated TET2 activation and its utility in TET2 MT MN patients we analyzed the nature of the interaction between AA and TET2 catalytic domain. DNA substrate and cofactors α-KG and Fe 2+ binding to TET2 are essential for its activity, but conditions needed to enable AA-mediated amplification of TET2 activity are poorly understood. Previously, using indirect internal fluorescence quenching assay, it was reported that AA binds to TET2 catalytic domain with an affinity of 8.7 × 10 3 M −1 , however, the binding condition lacked the substrate 22 . Nevertheless, DNA binding is co-operative and may induce structural fitness and enhance the dioxygenase activity 36 . Therefore, to explore the conditions of AA interactions with TET2 CD (TET2 catalytic domain) we performed a direct binding assay using surface SPR in the presence of cofactors and methylated cytosine containing DNA substrate. We expressed and purified TET2 CD with or without GST tag to homogeneity ( Supplementary Fig. 3a) and performed SPR binding analysis with AA. For this purpose, we either immobilized GST-TET2 CD on a CM-5 biosensor chip using anti-GST antibody and let AA flow over it, or we immobilized streptavidin and used it to capture biotin labeled 5 mC DNA, followed by binding to TET2 CD . Different concentration of AA in the presence of 5mC-DNA, 100 µM Fe 2+ and 10 µM of pseudo substrate NOG were used as analytes, without (1st scenario) or with (2nd scenario) TET2 CD . GST was used as a reference channel control. Analyses using Biaevalution™ software revealed that AA binds to immobilized GST-TET2 (1st scenario) with dissociation constant K d = 9.1 ± 2.3 µM in the presence of 5mC-DNA, Fe 2+ and pseudo substrate NOG (Fig. 4a). In the 2nd scenario, TET2 CD without GST tag was captured on 5mC-DNA-biotin immobilized on streptavidin sensor chip and different concentration of AA in the presence of 100 µM Fe 2+ and 10 µM of pseudo substrate NOG were used as analytes (Fig. 4b). Under these conditions, the dissociation constant was K d = 11.5 ± 2.2 µM. This value is nearly tenfold greater (stronger binding) than the previously reported K d for AA binding to TET2, which was estimated via UV-induced protein internal fluorescence quenching by AA in the absence of 5mC DNA 22 . In the present study, SPR measures interaction in the biologically relevant fully activated complex. In contrast, fluorescence quenching assays are prone to limitations of AA binding to proteins due to its strong absorbance in uv regions (Supplementary Fig. 3b). In addition, AA can quench the fluorescence intensity of disparate proteins such as bovine serum albumin (BSA), Glutathione-S-transferase (GST) and TET2 CD with quenching constants of 8.2 × 10 3 M −1 , 18.2 × 10 3 M −1 and 8.1 × 10 3 M −1 respectively (Supplementary Fig. 3c-f). These results are consistent with previously reported values for uv-induced fluorescence quenching constant for TET2 22 . This effect in part, may be due to a "sphere of action" quenching mechanism of AA with protein fluorescence 37 .
The binding constant of AA (9.1 ± 2.3 µM) with TET2 CD calculated from direct binding interaction in SPR was consistent with a half maximal effective concentration of AA (EC50 = 13.8 ± 1.3 µM) for enhancing TET2 CD activity as measured by the formation of 5hmC in an ELISA assay (Fig. 4c). Turning to the mechanism of AA enhanced TET2 catalysis, using isolated TET2 CD in the presence of α-KG, no effects of AA were observed on TET2 activity in the presence of Fe 2+ (Fig. 4d, e). The effect of AA was observed only when Fe 3+ were used. Thus, maintaining the redox state of iron Fe 2+ in the catalytic site to TET2 is most likely the mechanism of action of AA mediated increase in dioxygenase activity 21 . In such scenario, it is plausible to hypothesize that there can be several other direct or indirect biological factors that may circumvent the effect of AA in compensating loss of TET2 activity that may include the post translational modifications in the catalytic domain.
Effect of TET2 acetylation on AA activity. Several posttranslational modifications of TET2 have been reported to alter its stability and activity in cancer cell lines, including lysine acetylation 33,34,38 . To see the distribution of most abundant lysine acetylation we ectopically expressed TET2 protein in HEK293T cells purified by immunoprecipitation and analyzed for lysine modification by LC-MS/MS ( Supplementary Fig. 4). Consistent with the earlier reports 33,34 , we found K53 and K1478 are the most abundant acetylated lysine residues on N-and Cterminal of TET2 protein respectively ( Supplementary Fig. 4c, d).
To test the effects of lysine acetylation on the catalytic activity of TET2 and its impact on the cellular proliferation, we treated different leukemia cell lines with several chemical probe that impact N-and C-terminal lysine acetylation. For this purpose, we used the inhibitors of either class I/II deacetylase or acetyl transferase P300/CBP or the activator of sirtuins the class III deacetylase. Treatment of CMK cells with trichostatin A (TSA), a known inhibitor of class I/II deacetylase (HDAC I/II) 39 , resulted in a dose-dependent increase in TET2 protein level and activity (Fig. 5a, b and Supplementary Figs. 5a-b). This increase in TET2 activity is further enhanced by AA treatment up to threefold ( Fig. 5b and Supplementary Fig. 5b). To further expand the understanding of acetylation mediated increase in TET2 activity, we treated TET2 WT (CMK) and TET2 MT (HEL and SIG-M5) leukemia cell lines with specific P300/CBP inhibitors C646 40 or HATi 41 and also with the sirtuin activator SRT1720 42 in the presence and absence of AA. The HAT inhibitors or the sirtuin activators both, significantly increased 5hmC/5mC ( Fig. 5c and Supplementary Fig. 5c-e). However, treatment with either SRT1720 or HAT inhibitors did not change TET2 protein levels (Supplementary Fig. 5f-g). Treatment of CMK cells with sirtuin activator SRT1720 resulted in a dose-dependent increase in 5 hmC that was further amplified by the cotreatment of AA ( Fig. 5c and Supplementary Fig. 5h). While the sirtuin inhibitor, sirtinol 43 , demonstrated a dose dependent decrease in TETactivity ( Fig. 5d and Supplementary Fig. 5i). To test the growth inhibitory effects of the enhanced TET-activity, leukemia cell lines were treated with sirtuin activator SRT1720 in the present or absence of AA ( Fig. 5e-g). SRT1720 treatment on its own had antiproliferative effect with LD50 ranging from 4-8 µM in leukemia cells (Fig. 5e). There was a~50% decrease in the LD50 of SRT1720 by the addition of AA (Fig. 5f). AA at concentrations up to 100 µM showed a moderate effect, but increasing SRT1720 induced extensive cell death (Fig. 5g). The increased levels of TDOP along with antiproliferative effect of SRT1720 and AA in SIG-M5, a TET2 null cells suggest that the effects may be due to activation of TET1/3 (Supplementary Figs. 5b and j) and other SIRT1 targets with anticancer activity 44 .
To test whether this observation is true for cells derived from MN patients, we isolated BM mononuclear cells from three different cases of TET2 +/− (Supplementary Data 3) and treated them either with AA alone or in combination with TSA or SRT1720. AA as a single agent had no effect on TET2 activity or anti proliferative effect, however when combined with either TSA or SRT1720, there was a significant increase in 5 hmC/5mC with corresponding growth suppression ( Fig. 5h-i, Supplementary Fig. 5k), suggesting the combination of either sirtuin activators or class I/II HDAC inhibitor may potentiate AA effects on MN cells.
Effect of AA on TET2 lysine mutations in myeloid neoplasia. We hypothesized that TET2 missense mutation of catalytic domain lysine residues mimicking acetylation may be found in    (Fig. 6a, Supplementary Data 4). In order to probe the functional consequences of these mutations, we used site directed mutagenesis and generated these missense mutations and ectopically expressed in HEK293T cells. Our data showed that TET2 WT and K1299E, K1310Q, K1533R, and K1905E mutant proteins expressed to same levels, with exception of K1491N mutant being very lowly expressed, implicating K1491N mutation may affect the protein stability ( Fig. 6b and Supplementary Fig. 6a). The activity analysis using dot blot for 5hmC and 5mC showed that K1299E, K1310Q and K1905E mutants are completely inactive and, interestingly, addition of AA treatment could not restore the loss of TET2 function ( Fig. 6c and Supplementary Fig. 6b). Consistent with prediction, lysine to arginine (K1533R) mutation has no effect on TET2 activity ( Fig. 6c and Supplementary Fig. 6b), presumably due to the similarity in these two amino acids. Structural analysis of the TET2 CD in complex with 5mC DNA substrate and the cofactors revealed that lysine residues K1299, K1310 and K1905 are at the protein-DNA binding interface (Fig. 6d).Therefore, mutations in TET2 catalytic domain lysine that mimic acetylation, would significantly affect its ability to bind DNA substrate 36 and therefore any mutation or post-translational modification may cause loss of activity which cannot be reversed by amplifying Fe(III)/Fe(II) redox reaction by AA.
To test if the catalytic domain acetylation leads to loss of TET activity that cannot be restored by AA treatment, we performed in vitro acetylation experiment using recombinant TET2 CD and P300 in the presence of co-factor Acetyl-CoA and measured enzymatic activity and lysine acetylation by high resolution MS ( Fig. 6e-g, Supplementary Fig. 6c). We observed K1299, K1310, and K1905 were acetylated along with previously reported lysines (K1472/1473/1478) 33 . Nearly complete series of b-ions and y-ions were observed and unambiguously assigned to specific indicated peptide sequences. Pairs b or y ions (marked * y3/y4 and b7/b8 for K1299, y283 + /y293 + for K1310, y6/y7 and b4/b5 for K1905) are indicative of acetylation as these fragmentation ions contain a 42 Da mass shift from the natural lysine (Fig. 6g). In addition, the corresponding unmodified peptides for K1310 were also detected. However, unmodified peptide for K1299 and K1905 were not observed probably due to the tryptic digestion at unmodified . e Lysine acetylation inhibits TET2 activity. In vitro acetylation reactions were performed for TET2 CD with or without P300/CBP and in the presence or absence of Acetyl-CoA. Anti-acetyl Lysine antibody (Cell Signaling, Cat# 9441, 1:1000) was used for the detection. SDS-PAGE was also stained to visualize the total proteins. Acetylated and nonacetylated TET2 CD were used for in vitro activity assay in the presence of 5mC-DNA substrate, 2-OG and Fe 2+ . 5hmC and 5mC levels were assessed by 2D-UPLC-MS/MS. f In vitro acetylation reactions were performed for TET2 CD with P300/CBP in the presence (+CoA) or absence of Acetyl-CoA (-CoA). Acetylated (+CoA) and nonacetylated (-CoA) TET2 CD were used for in vitro activity assay. 100 µM of Fe 3+ and increasing concentrations (0, 100, and 200 µM) of AA were added in the reaction buffer. 5hmC and 5mC levels were assessed by dot blot. g Mass spectrometric analysis was performed for the acetylated TET2 CD protein. K1299, K1310, and K1905 were identified as acetylated lysine. h Schematics of the regulation of TET2 activity by acetyltransferases and deacetylases. Compounds used to regulate acetyltransferases and deacetylases activity are indicated. Data are shown as mean ± SEM of the triplicate and are representative of two independent set of experiments; statistical significance (p values) from two tailed t test are indicated (e-f). lysine residue. The acetylation of catalytic domain lysine residues results in the loss of activity of TET2 CD that cannot be reversed by addition of AA (Fig. 6f). Thus, acetylation of the lysine residues in TET2 can have a context dependent opposing effect on its activity (Fig. 6h), for example, N-terminal lysine acetylation by class I and II HDACs prevent ubiquitination, decrease protein degradation, and hence increased activity (Fig. 5a, b). On the contrary, acetylation of catalytic domain lysine residues decrease TET2 activity without impacting the protein level (Supplementary Figs. 5f-g) suggesting a direct involvement of these residues in its activity.

Discussion
Over the years AA has been extensively investigated as an anticancer agent due to its ability to target most common vulnerabilities of cancer cells presented in redox imbalance, oxygensensing and epigenetic reprogramming that include enhancing TET-dioxygenase activity 32,[45][46][47] . Several recent studies have suggested that AA may enhance the activity of αKG-dependent dioxygenases and may thus have therapeutic utility in TET2 MT MN 23,26 . However, most of these studies used supra-natural cytotoxic pharmacologic doses of AA and, hence it remains unclear if the effects in TET2 MT cases are due to AA mediated partial restoration of TET2-activity or TET2 independent effects of AA on redox activities and oxygen sensing. While the question of the utility of AA in TET2 MT MN remains unclear, here by using Tet2 +/− mice in combination with radiation as MN disease model, we provide first time evidence of the effect of AA treatment in preventing MN evolution. We show that AA delay the onset of MN and extends survival of TET2 +/− and AA deficient strain of Tet2 +/− ;Gulo −/− mice. In these models, which one can view as analogous to human CHIP, AA increases TDOP levels and slow myeloproliferation.
We observed a highly variable effects of AA treatment on TET2 activity depending on the model and the context of TET2 mutations and post-translational modification of TET2 protein. AA treatment increased 2-10 fold of TET activity in different cell lines, as reflected in global levels of TDOP. However, the effect of AA was modest in amplifying TET activity in primary tissue derived from either human or mice. We demonstrate that cell lines with naturally occurring TET2 MT are indeed sensitive to AA mediated increase in TET-activity in both heterozygous and biallelic settings. However, the results in a wide variety of primary cells with TET2 MT showed a spectrum of responses indicating the presence of modifying factors that may impact the AA effect on TET2 activity. Optimal use of AA in TET2 MT MN remains to be determined. We observed that AA increases levels of TDOP in many cell lines, BM samples from TET2 MT patients and Tet2 mt mice and that this effect is indeed associated with decreased proliferation of TET2-deficient cells. Irrespective of the cell type, TDOP levels correlated with TET2 levels; TET2 overexpression led to slower growth. Similarly, proliferative effects resulting from TET2 deficiency were partially reversed by AA. Interestingly, the effects of AA were also observed in Tet2 −/− cells, possibly, in part, due to its effect on Tet1 or Tet3 activity that can also be amplified by AA 26 and well established TET family independent effects of AA on cell growth and survival 47 . In either case, AA likely increases the activity of the all TET enzymes and thus may be effective irrespective of whether TET2 MT is hemi-or homozygous or biallelic configuration. These in vitro data imply that AA may have utility as either a low-intensity preventive treatment in TET2 MT CHIP or as a high dose treatment of MN 48 , consistent with promising in vivo effects of AA in murine models of TET2 MT MN 23,26 , in particular, TET2/FLT3-ITD induced murine AML model 23 .
Earlier reports studying anthocyanidin synthase a 2-oxoglutarate Fe(II)-dependent oxygenase, catalyzes the penultimate step in the biosynthesis of the anthocyanin class of flavonoids can directly interact with AA 49 . The functional relationship of AA to human TET2 activity has been investigated using indirect methods in the absence of DNA as a substrate 22 . Using SPR-based binding analyses we report here that AA binds human TET2 catalytic domain in a dose-dependent manner leading to an increase in dioxygenase activity (EC50 of 13.8 ± 1.3 µM) consistent with its binding affinity (K d = 9.1 ± 2.3 µM) in cell free in vitro conditions. Earlier reports of a nearly tenfold lower affinity, may have been due to the indirect nature of the binding assays and the lack of the activated TET2 complex 22 . Interestingly, the mean AA levels in plasma of healthy controls approach 50 µM 31 , significantly below the earlier reported affinity 22 . Therefore, EC50 of AA for TET2 determined in present study is more consistent and meaningful with the physiologic AA levels in serum 31 . Furthermore, we show that AA may increase Fe 3+ to Fe 2+ redox reaction in the catalytic site, since Fe 3+ by itself fails to activate TET2. AA cannot further amplify the TET2 activity if Fe 2+ is present in excess. However, when costimulated with AA, Fe 3+ efficiently restores TET2 dioxygenase activity, strongly indicating that AA can enhance the activity of most Fe 2+ and αKG -dependent dioxygenases either by reducing the enzyme-bound Fe 3+ to Fe 2+ or by acting as an alternate oxygen acceptor during uncoupled decarboxylation cycles. These observations clarify the molecular underpinnings of AA/TET2 interactions and further our understanding of the in vivo activity of AA in TET2 MT MN.
Our results suggest AA effects on TET2 activity may be attenuated by several layers of regulations of TET2, explaining potential variability of AA treatment results in different models. For example, phosphorylation of tyrosine or serine/threonine residues of TET2 enhances its stability and activity 48 and acetylation of N-and C-terminal lysines exert opposite effects on TET2 activity 33,34 . N-terminal lysine acetylation increases TET2 protein level by preventing proteasome mediated degradation 30 . Consistent with these reports, cells treated with TSA, a class I and II HDAC inhibitors have increased TET2 protein levels and 5hmC. We show that AA can act synergistically with TSA and further amplifies TET activity. However, C-terminal lysine acetylation inhibits TET2 as reflected in global 5hmC/5mC 33 . Lysine acetylation can prevent the TET2 catalytic cycles due to inefficient recruitment of 5mC-DNA substrate, and therefore, AA fails to activate TET2 in such scenario. Effects of AA may thus depend on context of TET2 lysine acetylation.
TET2 lysine residues K1299, K1310 and K1905 are at the DNA binding interface of TET2 CD and are thus important for binding to the 5mC-DNA substrate. Consequently, these residues are susceptible to P300-mediated acetylation and thereby downmodulation of catalytic activity. Conversely, their acetylation can be reversed by sirtuins, a known nicotinamide adenine dinucleotide (NAD + )-dependent deacetylases (class III HDACs). Indeed, we have confirmed that activation of sirtuins by the small molecule SRT1720 increased 5hmC, an effect that is further amplified by AA. In contrast, dose-dependent sirtuin inhibition by sirtinol 43,50 leads to TET2 inhibition as apparent from the loss of 5hmC/5mC. Interestingly, a comprehensive analysis of the configuration of 1205 TET2 MT MN patients in a large cohort of MN patient data set 1 revealed that all lysine residues missense mutations (K1299E, K1310Q, K1491N, K1533R, and K1905E) are located on C-terminal of TET2. Functional analysis of these missense mutation suggests that K1299E, K1310Q, and K1905E mutations that may mimic lysine acetylation due to similarity in charge distribution have complete loss of TET2 activity and this loss cannot be reversed by AA treatment. On the other hand, K1533N TET2 mutant is poorly expressed and K1533R had no effect on its activity. This observation has mechanistic and clinical implications in cases of TET2 MT associated disorders including MN. For example, loss of sirtuin deacetylase activity may have significant effect on the acetylation status of catalytic domain lysine residues and thus rendering TET2 less permissive to AAmediated TET-activation. Moreover, our results suggest new context dependent combination approaches for the reactivating TET2, whereby either HAT inhibitors or sirtuin activators can be combined with AA. Other synergistic combinations, may also include hypomethylating agents azacytidine 51,52 and decitabine 53 , which has been reported to upregulate TET2 expression and activity.
In summary, here we propose a complex interplay of various post translational modification of TET2 that have profound effects on its dioxygenase activity. Our results explain the effects of AA on TET2 function and likely other TET enzymes and provide a rationale for the long-term benefit of AA in preventing the evolution of MN.

Methods
Patient samples. Patient BM samples were obtained from healthy controls or patients with MN after informed consent in accordance with Cleveland Clinic IRBapproved protocol. Primary patient BM mononuclear cells were purified by Ficoll (Histopaque®-1077, SIGMA, 10771). Human cord blood was acquired from Cleveland Cord Blood Center, Cleveland, Ohio, and CD34 + cells were isolated by human CD34 MicroBead Kit (Miltenyi Biotec) according to the manufacturer's protocol.
Cell lines. Cell lines were purchased within 1 year of their use. Cells were used within 10 passages in fresh culture to prevent any drift. K562 and HEK293T were purchased from ATCC (Manassas, VA) and CMK, MEG-01, MOLM-13, HEL and SIG-M5 were purchased from DMSZ (Braunschweig, Germany). Cells were grown according to guidelines provided with them. Additional details are given in Supplementary Data 2.
Maintenance and analyses of mice. Animal care and procedures were conducted in accordance with institutional guidelines approved by the Institutional Animal Care and Use Committee. Tet2 −/− mice were generated as previously reported 19 . Gulo −/− mice 35 were used to study AA effects on the progression to MN in Tet2 +/− mice. Like humans, Gulo −/− mice need AA from food or drinking water for normal development, therefore, Gulo −/− mice were always maintained with 0.033 g/L AA in drinking water. Gulo −/− mice were crossed with Tet2 -/mice to generate Tet2 +/− ;Gulo −/− mice. After weaning (3 weeks of age), Tet2 +/− and Gulo −/− ;Tet2 +/− mice (C57BL/6 background) were divided into three groups each treated with 0.033 g/L (maintenance dose), 0.33 g/L (physiological dose) or 3.3 g/L (treatment dose) of AA in their drinking water. PB was collected by retro-orbital bleeding of mice and was smeared for May-Grunwald-Giemsa staining, and/or subjected to an automated blood count (Hemavet System 950FS). Total white blood cells were obtained after red cell lysis. For histopathology analyses, femurs were fixed in formaldehyde, decalcified, and paraffin embedded. Spleens and livers were treated similarly, omitting the decalcification step. Sections Fluorescence quenching. Different concentrations of AA were incubated with 0.4 µg protein in 100 ul HEPES buffer (50 mM HEPES, pH 6.5, 100 mM NaCl, 0.1 mM Fe 2+ , 1 mM DTT and 1 mM αKG). Fluorescence measurements were performed on an F-2500 Fluorescence Spectrophotometer (Hitachi). Proteins were excited at 280 nm wavelength and intrinsic fluorescence emission spectra was measured from 290 to 450 nm. Both excitation and emission slits were 5 nm. The response time was 0.08 s. The binding constant (Ka) was calculated according to the Modified Stern-Volmer equation 22 .
Surface plasmon resonance. Kinetic characterization of TET2 binding to AA was monitored by surface plasmon resonance (SPR) with a Biacore 3000 (GE Healthcare). Response units (RU), a measure of binding, were monitored as a function of time. To prepare a surface plasmon sensor chip, purified GST tagged TET2 CD (purity >90%) was captured by anti-GST antibodies (Biolegend, Cat# 640802, 10 µg/ml) 55,56 . Varying concentrations of AA (0-1000 µM) in the presence of substrate DNA, Fe 3+ and pseudo substrate NOG (25 µM; replacing the true substrate 2-oxoglutarate, 2-OG) was used as analyte. Alternatively, Biacore chips used immobilized biotin labeled 5mC DNA oligo substrate and GST tag free TET2 CD was captured, varying AA across 0-1000 µM in the presence of Fe 3+ and NOG (25 µM). In all SPR experiments, analyte solutions of different concentrations were passed over the sensor chip containing immobilized protein at a flow rate of 10 μl/min for 5 min, and dissociation was monitored while SPR buffer passed over the chips for an additional 5 min. Data were normalized against a reference channel containing immobilized GST. Surfaces were regenerated using two injections of 500 mM NaCl in HBS-P buffer at 20 µl/min for 30 s. Analysis, and fitting of data, was performed with BIA-Evaluation software, version 3.2 (Biacore Inc.), with the option for simultaneous Ka/Kd calculations. Sensorgram data were fitted using global fits to yield Ka and Kd simultaneously assuming a 1:1 Langmuir model. Goodness-of-fit was accepted based on the criterion of χ 2 ≤ 1% of the observed maximum response (R max ).
TET2 shRNA transduction. A 1 × 10 5 cells/condition in a total volume of 1 mL were placed in a 1.5 mL Eppendorf tube. 2 µL of Polybrene (4 mg/mL) was added to each tube, followed by 20 µL shTET2 (MISSION® shRNA Lentiviral Transduction Particles, TRCN0000421134, SIGMA) or scrambled shRNA virus. Samples were centrifuged at 1000 × g at 32°C for 90 min. Supernatants were discarded, samples were resuspended in appropriate media with FBS supplementation, and placed in 6-well plates for further experiments.
Stable TET2 overexpression. HEL, MEG-01, and CMK cell lines were stably infected via lentivirus with either an empty pOZ vector or a pOZ-TET2 (TET2 OE ) vector. TET2 expression was confirmed by RT-PCR.
Quantitative real-time PCR. Total RNA was extracted from cells using RNA purification kits (Macherey-Nagel). High capacity cDNA reverse transcription kits (Applied Biosystems) were used to generate cDNA. CFX96 real-time PCR detection (Bio-Rad) using Taqman gene expression primers TET2 (Hs0032599_m1) and GAPDH (Hs02786624g_m1) was also used. All experiments were duplicated. Gene expression was normalized to GAPDH and compared to controls. For TET1 and TET3 mRNA detection, qRT-PCR was performed by using SsoAdvanced Universal SYBR Green Supermix (Bio-Rad, Cat. 1725270). Primers are listed in Supplementary Data 4.
Site-directed mutagenesis. In-Fusion® HD Cloning kit (Takara, 638916) was used for TET2 missense mutation cloning. Primers are listed in Supplementary Data 4. All sequence was confirmed by Sanger sequencing.
Cellular viability Assay. Cells were treated with small molecules in the presence or absence of varying concentrations of AA for 0-72 h and assessed using Vi-CELL XR cell viability analyzer (Beckman Coulter).
In vitro colony-forming assays. Mononuclear cells derived from BM or purified CD34 + cells treated with shTET2 and empty vector were seeded at 3000 cells per methylcellulose plate (Methocult™, H4435; STEMCELL Technologies). Colonies were scored on day 14.
MS analysis of TET2 acetylation. The LCMS/MS analysis of TET2 post translational modification were performed as described previously 58 . Briefly, protein bands were excised from the SDS-PAGE gel, washed, destained (50% ethanol, 5% acetic acid) and dehydrated in acetonitrile followed by reduction in DTT and alkylation with iodoacetamide prior to the in-gel digestion by adding 5 μL 10 ng/μL trypsin or chymotrypsin in 50 mM ammonium bicarbonate at room temperature for 18 h. The peptides were extracted (2x) in 30 µL of 50% acetonitrile with 5% formic acid. These two extracts were combined and vacuum dried and resuspended in 30 µl of 1% formic acid for LC-MS analysis. For in solution digestion, acetylated and mock control samples were processed for trypsin digestion and desalted during solid phase extraction. The LC-MS system, ThermoScientific Fusion Lumos mass spectrometer interfaced with dionex Ultimate 3000 UHPLC with Dionex column (15 cm × 75 µm id Acclaim Pepmap C18, 2 μm, 100 Å reversed-phase capillary chromatography column) were used for data collection. The extracts from trypsin digest (5 µL) were injected and the peptides eluted using acetonitrile and 0.1% formic acid gradient at a flow rate of 0.3 μL/min. The microelectrospray ion source is operated at 1.9 kV.
Mass spectroscopic data analysis. Analysis of MS raw files were performed by Proteome Discoverer. MS/MS spectra were searched against TET2 (NCBI Reference Sequence: NP_001120680.1) or an in-house constructed Human TET2 protein sequence. Modifications considered included acetylation of Lysine (K). Precursor tolerance was set to 10 ppm and the fragment tolerance was set to 0.02 Da. MS-MS Spectra assigned to acetylation were manually inspected to assure correctness: yand bseries fragmentation ions were labeled for inspection. Labelfree Quantification (total ion intensity) of peptides were performed using Qual Browser in Thermo Xcalibur software. Briefly, peptide retention time and peptide masses were obtained from the associated MS-MS spectra. Isotopically resolved precursor ion m/z ratios were extracted from the LC-MS and used for the selection by column retention time of interest. The area under the peak was used to quantify particular series of ion abundance.
Statistics and reproducibility. All statistical analyses were performed in Graph-Pad Prism 8.0 (https://www.graphpad.com/) unless otherwise described. The statistical significance were performed using two tailed student t test unless described otherwise. Each experiment was performed in triplicate at least twice wherever possible.
Reporting summary. Further information on research design is available in the Nature Research Reporting Summary linked to this article.

Data availability
All source data reported in this manuscript is the part of supplementary data as Supplementary Data 1. Any specific material request can be made to the corresponding authors. All engineered cell lines used in this manuscript are available upon request. The mass spectrometry proteomics data for lysine acetylation of TET2 post translational modifications have been deposited to the ProteomeXchange consortium via the PRIDE 59 partner repository with the dataset identifier PXD020550.