Modulation of TET2 Activity with Ascorbic Acid in Myeloid Neoplasia: Diagnostic and Therapeutic Implications

TET2 mutations (TET2^MT) are ubiquitous in myeloid neoplasia, but are most frequent in MDS and MDS/MPN. TET2 is an iron-catalyzed dioxygenase critical in active DNA demethylation that progressively oxidizes 5-methylcytosine (5mC) to 5-hydroxymethylcytosine (5hmC), 5-formylcytosine (5fC), and 5-carboxylcytosine (5CaC). As 5hmC residues are not recognized by DNMT3A, their methylation pattern is not copied onto the daughter DNA strand, leading to net DNA demethylation: TET2 can also therefore be considered a demethylase.

Ascorbic acid (AA) is required as a prosthetic group of TET-mediated cytosine demethylation. AA may increase the residual function/throughput of intact TET2 in cases with heterozygous TET2^MT and restore hydroxymethylation, thus reversing the expected epigenetic phenotype. We hypothesized that in heterozygous TET2^MT neoplasms, AA alone or in combination may constitute a rational targeted therapy to restore TET2 function.

We first assessed the effects of AA on TET2 in cell-free experiments. We combined purified wild-type TET2 (TET2^WT) with a synthetic poly mC substrate and 0-4 mM AA, and measured levels of 5mC, 5hmC, 5fC, and 5CaC via 2D-UPLC-MS/MS profiling. AA increased the ratio of 5hmC, 5CaC, and 5fC/5mC in a dose-dependent fashion (peak effect 4.5 to 9). To model heterozygous TET2^MT seen in myeloid neoplasia, we combined TET2^WT with a catalytically-dead TET2S1898F in a 1:1 ratio. AA again increased the ratio of 5hmC, 5fC, and 5CaC to 5mC (peak effect 3 to 6, p<.0001); AA had no effect on TET2S1898F alone.

To determine if AA TET2 upregulation could be reproduced in human cells, K562 and MOLM-13 AML lines were treated for 24 hours with 0-1000 uM of AA before subjecting them to 2D-UPLC-MS/MS analysis. AA increased 5hmC levels in K562 cells by 1.8-fold (500 uM, p=.04) and 1.7-fold (1000 uM, p=.15), and increased 5fC levels by 2.1-fold (500 uM, p <0.0001) and 3.1-fold (1000 uM, p=.035). Similarly, AA increased 5hmC levels in MOLM13 cells by 1.8-fold (500 uM, p=.01) and 2-fold (1000 uM, p<.0001), and increased 5fC levels by 3.6-fold (500 uM, p<.0001; 1000 uM, p=.01).

We then studied the effect of TET2 loss on leukemogenesis in dual Tet2+/- ; Gulo -/- knockout mice unable to produce endogenous AA, making them ideal to determine the effect of exogenous AA on in vivo TET2 function. Bone marrow samples isolated from Tet2+/- ; Gulo -/- mice had a 1.7-fold reduction in 5hmC (p<.0001), a 2-fold reduction in 5fC (p=.0009), and a 3.4-fold reduction in 5CaC (p<.0001). Treatment of Tet2+/- ; Gulo -/- mice with 3.3 g/L ascorbic acid (in the drinking water for 6 months) partially rescued these DNA oxidation products in their bone marrow cells, with significantly increased levels of 5fC (p = .02) and 5CaC (p =.04) compared to untreated Tet2+/- ; Gulo -/- mice. Compared to TET2^WT mice, Tet2+/- ; Gulo -/- mice had only a 1.2-fold reduction in 5hmC (p = 0.2), a 1.15-fold reduction in 5fC (p = .004), and a 1.7-fold reduction in 5CaC (p =.3).

Similar to the TET2 knockout mice results, 8/8 of TET2^MT patient samples profiled had reduced levels of at least one TET2-dependent DNA oxidation product. Intriguingly, 24-hour treatment with 250 uM AA treatment had no significant effect on any TET2 MT bone marrow samples; in TET2 WT samples, 250 uM AA increased 5fC levels by 2-fold (p =.014) and 5CaC levels by 3-fold (p =.0025), though it nonsignificantly reduced 5hmC levels by 80% (p =.069).

In our study, AA significantly increased the levels of TET2-dependent DNA oxidation products in purified TET2, Tet2+/- ; Gulo -/- knockout mice, TET2^WT human cell lines, and TET2^WT patient samples. Therapeutic AA may therefore play a role in treating myeloid neoplasia with IDH1 or IDH2 mutations, which have structurally-intact but functionally hypoactive TET2 from loss of the IDH-catalyzed TET2 substrate alpha-ketoglutarate. Given that AA was unable to restore TET2 activity in samples from TET2^MT patients, AA may not be beneficial in TET2^MT myeloid neoplasia.

Our study also tells a cautionary tale about using 5hmC levels alone to estimate TET2 activity. 5hmC levels were normal in 3/8 (37.5%) of patients with pathogenic TET2^MT; simultaneous measurement of multiple TET2-dependent products is therefore important to accurately assess in-vivo TET2 activity and determine who might benefit from therapeutic AA intervention.