To the editor,

TETs is diversely expressed in hematological malignancies and serves as potential therapeutic targets [1,1,2,3,4,5]. However, the roles of TET1 in T-ALL have not been fully unveiled, and potent TET1 inhibitors are needed to promote TET1 as a druggable target [6,1,2,3,4,5,6,7,8].

We analyzed gene expression from previous T-ALL cohorts and found TET1, but not TET2/3, is highly expressed in T-ALL and drug-resistant patients (Fig. 1a, b, Additional file 2: Fig. S1a–c). Validation in T-ALL cell lines confirmed TET1 upregulation in T-ALL cells compared to normal T cells and further upregulation in dexamethasone-resistant cells (Fig. 1c). TET1 expression negatively correlates (even though not significant) with poor prognosis (Fig. S1d). Silencing of TET1 impaired the proliferation of T-ALL cells (Fig. 1d, Additional file 2: Fig. S1e–i), indicating crucial roles of TET1 in T-ALL. However, treatment of T-ALL cells with two reported TET1 inhibitors only showed minimum proliferation inhibition (Additional file 2: Fig. S2), due to low potency of the inhibitors 6, suggesting more potent TET1 inhibitors are needed.

Fig. 1
figure 1

a Fragments Per Kilobase Million (FPKM) of TET1, TET2, TET3 in a cohort with 124T-ALL patients and 12 normal T cell samples (CNCB, HRA000122). Data are mean ± SD (Two-tailed unpaired Student’s t test, ** p < 0.01, *** p < 0.001). b Transcripts per million (TPM) of TET1 in glucocorticoid-resistant and glucocorticoid-sensitive T-ALL samples (GSE5820). c Protein expression levels of TET1, TET2, TET3 in 8 T-ALL cell lines and normal T cells. d Effect of TET1 knock down on cell growth of Jurkat cell line. e Schematic of structure guided virtual screening and cell line-based drug screening with 2059 FDA or EMA approved drugs. Jurkat and CCRF-CEM cell lines were treated with each drug (5 μM) for 48 h, and cell viability was detected using CCK-8 assay. f Joint analysis of virtual screening and cell line-based screening identified auranofin as a potential hit drug. g Affinity between TET1 protein and auranofin assayed by surface plasmon resonance (SPR). Equilibrium binding analysis indicates a KD of 1.804 μM. h Dot blot analysis of global 5hmC in Jurkat cell showing that auranofin dose-dependently decreased cellular 5hmC level. Left: dot blot image; right: quantification of dot blot results. i Quantification of genome-wide 5hmC distribution in T-ALL Jurkat cells treated with or without 0.1 μM auranofin for 24 h. Auranofin treatment decreased global 5hmC levels. n = 5 replicates. j Dose-dependent inhibition of TET1 catalytic activity by auranofin with an in vitro fluorometric quantification assay. IC50 of 0.076 μM was determined, indicating potent inhibition of TET1 by auranofin. k Molecular docking showing that auranofin are in close conformation with 2-OG and Fe (II) in the binding pocket of TET1-CD. l SPR-based assay showing that the binding of auranofin to TET1 was competed by increasing concentration of TET1 substrate analogue NOG. m Dot blot analysis of 5hmC revealed that auranofin induced TET1 catalytic activity inhibition was attenuated by increasing concentration of 2-OG. Left: dot blot image; right: quantification of dot blot results

Full size image

To identify potent TET1 inhibitors, we performed virtual- and cell-based screening (Fig. 1e). Auranofin was identified from both screening (Fig. 1f, Additional file 1: Table S1–S2). SPR revealed that auranofin binds to TET1 (Fig. 1g), with reduced affinity to TET2/3 (Additional file 2: Fig. S3).

Methylation-related pathways were enriched after auranofin treatment in T-ALL (Additional file 2: Fig. S4a, b, Additional file 1: Table S3), which induced dose-dependent decrease in cellular 5hmC and increase in 5mC, as indicated by dot blot and genome-wide 5hmC/5mC sequencing (Fig. 1h-i, Additional file 2: Figs. S4c, S5). LC–MS/MS confirmed inhibition of TET1 catalyzed 5mC to 5hmC conversion by auranofin (Additional file 2: Fig. S6). In vitro assay determined a IC50 of 76 nM (Fig. 1j).

We generated structure of TET1-CD-auranofin by molecular docking to explore the inhibition mechanism (Fig. 1k). Structure showed that auranofin was in very close conformation with NOG (analog of TET1 substrate 2-OG) and Fe (II), suggesting a potential inhibition mechanism of auranofin through competitively preventing substrates binding to TET1. Results indicated that both NOG and Fe (II) compete with auranofin for binding to TET1-CD and increasing concentration of 2-OG or Fe (II) attenuated auranofin inhibition on TET1 (Fig. 1l-m and Additional file 2: Fig. S7). Given that, 2-OG and Fe (II) are conserved in many demethylases, selectivity of auranofin to TET1 was verified by comparing to TET2 and KDM6B. Auranofin showed reduced affinity and very low inhibition at 1 μM to both proteins (Additional file 2: Fig. S8), indicating at least 13-fold selectivity for TET1 over TET2 and KDM6B. Structure and electrophoretic mobility shift assay also showed that auranofin did not affect DNA substrate binding to TET1 (Additional file 2: Fig. S9). Collectively, auranofin inhibits TET1 by competing with substrates for binding to TET1.

Auranofin was highly cytotoxic to T-ALL cells, but not to normal T cells (Fig. 2a). Overexpressing either full length or the catalytic domain of TET1, but not the catalytic dead mutants or TET2/TET3-CD, in T-ALL attenuated the cytotoxicity of auranofin (Fig. 2b, c and Additional file 2: Figs. S10–S12), revealing that the auranofin-mediated cytotoxicity depends on the catalytic activity of TET1. In vivo T-ALL xenograft model showed that treatment with auranofin significantly inhibited progression, as well as bone marrow invasion, of T-ALL and prolonged mice survival (Fig. 2d–i), indicating therapeutic potential of auranofin for T-ALL.

Fig. 2
figure 2

a Relative cell growth of T-ALL cell lines and normal T cell after treatment with various concentration of auranofin. Cell numbers were counted at 48 h post treatment with auranofin. b Auranofin-induced T-ALL cell death was attenuated by full-length TET1 overexpression. Data are mean ± SD (Two-tailed unpaired Student’s t test, **** p < 0.0001). c Auranofin-induced T-ALL cell death was attenuated by TET1-CD overexpression. Data are mean ± SD (Two-tailed unpaired Student’s t test, **** p < 0.0001). d Schematic of cell-derived xenograft (CDX) validating the in vivo anti-T-ALL activity. A total of 5 × 106 luciferase-expressing Jurkat cells were injected through the tail vein to NCG mice. Auranofin (20 mg/kg/2 days) was intraperitoneally administered from day 14 to day 28 with PBS as Vehicle. e Time-lapse bioluminescence imaging of the xenograft mice. f Quantification of the total photon flux of the xenograft mice at day 28. Data are mean ± SD (Two-tailed unpaired Student’s t test, ** p < 0.01). g Representative flow cytometry analysis of bone marrow resident Jurkat cancer cells in Vehicle and auranofin treated mice. Jurkat cancer cells were stained with anti-human CD7 antibody. h Quantification of the bone marrow resident Jurkat cancer cells shown in g. Two-tailed unpaired Student’s t test, *** p < 0.001. i Kaplan–Meier survival curves the xenograft mice treated with Vehicle or auranofin. j Venn diagram depicting vital genes which are intersected among 5hmC and RNA downregulated and 5mC upregulated genes. 31 intersected genes including c-Myc were identified. k Western blot analysis showing the time-dependent down-regulation of c-Myc expression in Jurkat and CCRF-CEM cells. l qRT-PCR analysis showing that c-Myc mRNA decreased after knockdown of TET1. m qRT-PCR analysis showing that c-Myc mRNA increased after overexpression of TET1. n Auranofin-induced T-ALL Jurkat cell death was attenuated by c-Myc overexpression at various auranofin concentrations. Data are mean ± SD (Two-tailed unpaired Student’s t test, **** p < 0.0001). o Schematic diagram showing the proposed mechanism of action of auranofin in T-ALL. Auranofin inhibits TET1 enzymatic activity through competitive occupation of the binding pocket to its cofactor substrates 2-oxoglutarate (2-OG) and Fe (II). Inhibition of TET1 subsequently down-regulated the transcription and translation of c-Myc, at least partially through c-Myc DNA epigenetic remodeling, leading to T-ALL cells death

Full size image

Auranofin has been reported to exert anti-tumor activity via Thioredoxin reductases and ROS [9, 10]. While, mechanism in T-ALL is different as neutralizing ROS or genetic manipulation of Thioredoxin reductases did not affect auranofin cytotoxicity to T-ALL (Additional file 2: Fig. S13).

Auranofin treatment altered genome-wide distribution of 5hmC/5mC in the promoter region (Fig. 1i, Additional file 2: Fig. S5, Additional file 1: Tables S4, S5), enlightened potential mechanism of action of auranofin via epigenetic control of transcription and translation of certain genes. Conjointly analysis of 5hmC-Seal, WGBS, and RNA-Seq data, identified 31 genes with 5hmC/RNA down-regulation and 5mC up-regulation, among which c-Myc, a central oncogene in T-ALL [11, 12], was discovered (Fig. 2j, Additional file 2: Fig. S14, Additional file 1: Table S6). Auranofin treatment down-regulated both c-Myc transcription and translation (Fig. 2k, Additional file 2: Fig. S15a, b). TET1 correlates with c-Myc expression (Additional file 2: Fig. S15c, d), as validated by genetic manipulation of TET1 (Fig. 2l–m). Overexpression of c-Myc in T-ALL attenuated auranofin-induced cytotoxicity (Fig. 2n, Additional file 2: Fig. S15e), suggesting c-Myc as downstream effector.

Collectively, we confirmed TET1 is a promising therapeutic target for T-ALL and discovered potent TET1 inhibitor, auranofin, with anti-T-ALL activity in vitro and in vivo. Mechanistically, auranofin-induced TET1 inhibition epigenetically alters transcription and translation of c-Myc to induce T-ALL cell death (Fig. 2o).