Tet1 is not required for myeloid leukemogenesis by MLL-ENL in novel mouse models

The Ten Eleven Translocation 1 (TET1) gene encodes an epigenetic modifying molecule that is involved in demethylation of 5-methylcytosine. In hematological malignancies, loss-of-function mutations of TET2, which is one of the TET family genes including TET1, are frequently found, while the mutations of TET1 are not. However, clinical studies have revealed that TET1 is highly expressed in some cases of the hematological malignancies including acute myeloid leukemia. Indeed, studies by mouse models using conventional Tet1 knockout mice demonstrated that Tet1 is involved in myeloid leukemogenesis by Mixed Lineage Leukemia (MLL) fusion gene or TET2 mutant. Meanwhile, the other study showed that Tet1 is highly expressed in hematopoietic stem cells (HSCs), and that deletion of Tet1 in HSCs enhances potential self-renewal capacity, which is potentially associated with myeloid leukemogenesis. To examine the role of Tet1 in myeloid leukemogenesis more precisely, we generated novel conditional Tet1-knockout mice, which were used to generate the compound mutant mice by crossing with the inducible MLL-ENL transgenic mice that we developed previously. The leukemic immortalization in vitro was not critically affected by conditional ablation of Tet1 in HSCs with the induced expression of MLL-ENL or in hematopoietic progenitor cells retrovirally transduced with MLL-ENL. In addition, the leukemic phenotypes caused by the induced expression of MLL-ENL in vivo was not also critically affected in the compound mutant mouse model by conditional ablation of Tet1, although we found that the expression of Evi1, which is one of critical target genes of MLL fusion gene, in tumor cells was remarkably low under Tet1-ablated condition. These results revealed that Tet1 was dispensable for the myeloid leukemogenesis by MLL-ENL, suggesting that the therapeutic application of Tet1 inhibition may need careful assessment.


Introduction
Emerging evidence has demonstrated critical roles of epigenetic modifying molecules in oncogenesis [1]. Several findings over these decades suggest that cancer stemness is associated with aberrant epigenetic pathways caused by oncogenic driver mutations, including mutation of epigenetic modifier genes [2,3], similarly to the finding that stemness of embryonic stem cells (ESCs) is closely associated with epigenetic pathways [4].
Previous studies revealed that loss-of-function mutations of TET2 are frequently found in hematological malignancies, while the mutations of TET1 are not [10]. Tet2 deficiency in mice recapitulated myeloproliferative disease (MPD) and lymphoma in humans [11]. In contrast, Tet1 deficiency did not lead to MPD, but developed lymphoma with a long latency [12,13]. Meanwhile, the conventional knockout/knockdown of Tet1 suppressed leukemic transformation by MLL-AF9 through the downregulation of critical downstream molecules [14]. In addition, Tet1 deficiency suppressed a Tet2-loss-driven myeloid malignancy, but promoted the lymphoid malignancy [15]. In line with the results, high TET1 expression is found to be associated with a poor prognosis in cytogenetically normal AML [16]. These results suggested that TET1 is a tumor-suppressor gene in lymphoid malignancies, but an oncogene in myeloid malignancies.
Several studies have demonstrated that, at least in part, normal hematopoietic stem cells (HSCs) share molecular mechanism with leukemic stem/initiating cells and ESCs [17]. Interestingly, Tet1 is relatively highly expressed in HSCs and ESCs in comparison with their progenitor cells [7,13], suggesting the possible essential role of Tet1 in these stem cells. Also, a report suggested an important role of Tet1 in reprogramming [18]. However, Tet1 deficiency in ESCs did not abrogate pluripotency [12], and on the contrary, the deletion in HSCs enhances potential self-renewal capacity [13], reminiscent of the deletion of Tet2. HSCs are considered to be one of major target cells of driver mutation(s) that initiate AML [19]. Thus, a concept of Tet1 as an oncogene in myeloid malignancies might be contradictory in the leukemogenesis arising in HSCs, but this hypothesis remains to be investigated.
To clarify the role of Tet1 in myeloid leukemogenesis more closely, we have established a new mouse model with conditional ablation of Tet1 and induction of MLL-ENL in bone marrow (BM) cells. Using this model, we demonstrate that leukemic transformation by MLL-ENL in vivo and in vitro was not critically affected by the Tet1-ablation, although the expression of Evi1, which is one of critical target genes of MLL fusion gene, in tumor cells was remarkably low under Tet1-ablated condition. Also, leukemic immortalization by retroviral transduction of MLL-ENL in vitro was not inhibited by the conditional Tet1-ablation. Our results suggest that the therapeutic application of Tet1 inhibition in AML may need careful assessment. 4-hydroxytamoxifen (4-OHT) was dissolved in ethanol, and used at a final concentration of 0.1 μM. Tamoxifen (Sigma-Aldrich, St. Louis, MO) was dissolved in ethanol (100 mg/ml), and diluted in corn oil to 10 mg/ml. Starting at 8-12 weeks of age, or 5 weeks after bone marrow transplantation (BMT), tamoxifen was injected intraperitoneally at a dose of 100 mg/kg for 3 consecutive days. The time course started with the date of the third treatment as day 0. For drug selection in retroviral transduction, G418 (Thermo Fisher Scientific Inc., Waltham, MA) was used at a final concentration of 1 mg/ml.
The targeting vector was linearized by Not I digestion just upstream of the insertion site of the 1.7-kb fragment, and electroporated into B6 EGR-101 male ES cells, followed by positive and negative selection with G418 and ganciclovir, respectively, as previously described [20]. To detect the correct homologous recombination, PCR screening of genomic DNA from the engineered ES cells was performed using LA Taq (Takara Bio Inc., Otsu, Japan) according to the manufacturer's protocol. The primer sequences were as follows: 5'-GGCCTATCATGCT GTGCTGACCCTTTTGGGCT-3' and 5'-GCTCCCGATTCGCAGCGCATCGCCTTCTAT-3' for the first screening, and 5'-CTGCTAAAGCGCATGCTCCAGACTGCCTTGGGAAAAG-3' (first PCR) or 5'-GAAAAGCGCCTCCCCTACCCGGTAGAATTGACCT-3' (second PCR), and 5'-CCCACTA CACCACATTAGCAAAGGCATTACATGAGAGAAGCCT-3' for the second screening. To validate the screening, the karyotypes of PCR-positive clones were analyzed, and Southern blot analysis of genomic DNA from the clones with the correct karyotype was performed.
Chimeric male mice (F0), which were generated by the blastocyst injection of properly targeted ES cell clones, were bred with B6 wild-type female mice. F1 mice that were heterozygous for the targeted allele on a pure B6 background were bred with Flpe deleter mice (kindly provided by Dr. A. F. Stewart) to remove the FRT-flanked neo cassette. F2 mice that were heterozygous for the modified allele harboring the proximal loxP-FRT and the distal loxP sequences, as well as Flpe, were bred with B6 wild-type mice, to produce Tet1 fl/+ mice that were heterozygous for the floxed allele of Tet1 alone, and Tet1 fl/fl mice were generated by intercrossing of the Tet1 fl/+ mice. C57BL/6-Gt(ROSA)26Sor tm9(Cre/ESR1)Arte (CreER) mice were purchased from Taconic Biosciences (Rensselaer, NY), and inducible MLL-ENL transgenic, Tg-ME (ME), mice harboring wild-type (wt) and genetically engineered Tet1 alleles. The targeting vector was constructed by inserting 3 fragments (indicated by horizontal bold lines) corresponding to the wt regions (as shown), before the proximal loxP, between both loxP, and after distal loxP sequences, into the backbone vector. To generate the floxed (fl) allele, the neomycin resistance gene (neo) cassette flanked by the flippase (Flp) recognition target (FRT) sequences in the targeted allele was removed by mating with Flpe mice. The floxed allele can be recombinated between loxP sequences by Cre activity, resulting in the deleted (Δ) allele lacking exons 8 and 9. The restriction sites of BamH I (B), and the 5' and 3' probes, used for Southern blot analysis, are indicated with vertical dashed lines and short horizontal lines, respectively. Coding and non-coding exons are indicated by closed and open boxes, respectively. The exons coding a cysteine-rich (C) and a double-stranded β-helix (D) regions are shown. tk, thymidine kinase gene. (B) Southern blot analysis of ES clones. Each genomic DNA extracted from 6 candidate ES clones and wt ES cells was digested with BamH I and hybridized with the 5' probe, resulting in the wt 15-kb and targeted 6.4-kb bands. The blot image was cropped from S1 Raw images. https://doi.org/10.1371/journal.pone.0248425.g001
All animal studies were approved by the Animal Care Committees of Mie University (No. [24][25][26][27][28][29][30][31][32][33][34]. Mice were kept in the specific pathogen-free facility under standard conditions (temperature 22-24˚C; 12-h light/12-h dark cycle, and free access to food and water). In BMT experiments, recipient mice were housed in sterilized cages with filter-cap, and were administered neomycin (1 mg/ml (Merck Millipore, Burlington, MA)) in drinking water, for 4 weeks after BMT. Both male and female mice were subjected to experiments, except for that female mice were used for recipients in BMT. Animal health was monitored carefully and regularly in all experiments by trained staff. Hematological analyses of conditionally Tet1-ablated mice were performed, compared with the corresponding littermate controls, as previously described [22]. The mice were euthanized by carbon dioxide (CO 2 ) inhalation, to analyze or collect BM cells. The total number of mice used in this study was 108.

Retroviral transduction
The pMYs retroviral vector harboring MLL-ENL-internal ribosomal entry site (IRES)-neo was previously described [21]. Retroviral transduction was performed using RetroNectin (Takara Bio Inc.) with retroviral supernatants, which were harvested 48 hours after transfection from Plat E cells transfected with a retroviral construct, as previously described [23,24].

Myeloid immortalization assay
Myeloid immortalization assays using serial replating were performed as described [21]. Briefly, the purified CD34(-) or CD34(+) KSL (150) and MP (1500) cells were directly sorted into 1.5 ml of methylcellulose medium prepared from M3234 (StemCell Technologies, Vancouver, Canada) according to the manufacture's protocol, supplemented with 25 ng/ml SCF, 10 ng/ml each of IL-6, mouse IL-3, mouse GM-CSF (Miltenyi Biotec), and 0.1 μM of 4-OHT. The sorted cells in 1 ml of the mixture were immediately plated in 35 mm dishes. Alternatively, retrovirally transduced KSL, CD34(+) KSL, and MP cells were plated in the same M3234based medium supplemented with G418 in addition to the cytokines. After culturing for 5 (fo MP cells), 6 (for CD34(+) KSL cells), and 10 days (for retrovirally transduced cells and CD34 (-) KSL cells), colonies were enumerated, and single-cell suspensions (0.3-1×10 4 cells) of colonies were subsequently replated in α-MEM-based medium containing 1% methylcellulose (Shin-Etsu Chemical Co., Ltd., Tokyo, Japan) supplemented with only the same cytokines. Every 5-7 days, replating of cells collected from colonies was repeated in the same way using the α-MEM-based methylcellulose medium. The immortalized cells were harvested from colonies in the third plating, and were cultured in α-MEM supplemented with 20% FBS and the same cytokines.

BMT and leukemogenesis assay
Hematopoietic reconstitution of BM and leukemogenesis assays using BMT using B6 mice aged 10-15 weeks were performed as previously described [21], with some modifications. Briefly, for BM chimera mice, 1×10 6 BM cells derived from the tested genotype of mice that were positive for CD45.2 were transplanted into lethally irradiated (7.5 Gy) wild-type recipient B6 mice that were positive for CD45.1. The recipient mice were periodically monitored, and analyzed morphologically and immunophenotpically at the indicated intervals as previously described [24].
For the mixed-BM chimera mice for use in leukemogenesis assays, a mixture of 1×10 6 BM cells each derived from wild-type CD45.1-positive mice and the tested genotype of CD45.2-positive mice were also transplanted into wild-type CD45.1-(BMT #1 (#1)), or CD45.1/CD45.2-(BMT #2 (#2)) positive recipient female B6 mice that were lethally irradiated in the same way. Mixed-BM chimera mice were treated with tamoxifen after engraftment. For secondary transplantation, 3×10 3 MPD cells harvested from BM were transplanted into sublethally irradiated (5.25 Gy) recipient B6 mice. To assess the mixed-BM chimera mice hematologically, PB samples collected from tail vein were measured 5 weeks after BMT to confirm the engraftment of transplanted cells, and also every 4 weeks after the treatment of tamoxifen. Mouse health was monitored for signs of illness such as less activity, scruffiness, weight loss, and hind-limb paralysis every day. The mice showing these signs were considered moribund, and were immediately euthanized by CO 2 inhalation with all efforts to minimize animal suffering (n = 11 (#1), 13 (#2), and 7 (secondary BMT)), however, some mice unexpectedly died without showing suspicious signs of illness (n = 10 (#1), 6 (#2), and 6 (secondary BMT)). The control mice in #1 were euthanized in the same way at the end of monitoring (n = 9). The monitoring was ended after all the mice transplanted with Tg-ME cells met the criteria of humane endpoints described above. The analyses of BM and spleens from the tested mice were performed as previously described [24], showing the development of lethal MPD in the mice transplanted with Tg-ME cells.

RNA extraction and reverse transcription (RT)
Total RNA was extracted from cells using TRI Reagent LS (Molecular Research Center Inc., Cincinnati, OH), followed by treatment with RQ1 RNase-free DNase I (Promega, Madison, WI). RT was performed using SuperScript II (Thermo Fisher Scientific Inc.) with random hexamers, as previously described [24].

Statistical analysis
Unpaired Student's t-tests, and a one-way analysis of variance (ANOVA) followed by Tukey-Kramer test as a post hoc test, were used to compare two groups, and more than two groups, respectively. A two-way repeated measures ANOVA followed by Shaffer's modified sequentially rejective Bonferroni procedure as post hoc analyses was used to compare several groups at different timepoints. Data from RT-qPCR were log2-transformed before the statistical analysis. To compare the gene expression from CD34(-) KSL-derived cells, after a one-way ANOVA was used to compare four genotypes of samples at the first round of plating and a (D) RT-PCR of the immortalized cells with Tet1 Δ/Δ and Tet1 +/+ using primers spanning exons 7-11. M, 100 bp ladder marker; NTC, negative control. (E) Dot blot assays of genomic DNA from the immortalized cells with Tet1 Δ/Δ and Tet1 +/+ . 5-hmC, 5-hydroxymethylcytosine; 5-mC, 5-methylcytosine. The blot and gel images were cropped from S1 and S2 Raw images.
https://doi.org/10.1371/journal.pone.0248425.g002 two-way ANOVA was used to compare two pairs of two genotypes samples at the first and third rounds of plating were performed, the obtained p values were adjusted using the Holm-Bonferroni correction. The probability of overall survival of the mice was estimated using the Kaplan-Meier method and compared by the log-rank test. All statistical tests were performed using the R software program (version 3.5.2, https://www.r-project.org/), except that two-way repeated measures ANOVA tests were performed using the ANOVAKUN software (version 4.8.4, http://riseki.php.xdomain.jp/index.php?ANOVA%E5%90%9B), which was executed on the R software program.

Generation of conditional Tet1-knockout mice
To conditionally ablate Tet1 in vivo, we designed a strategy to induce Cre/loxP-mediated excision of the exons 8 and 9 of Tet1, which led to a truncated product lacking the catalytic domain by a frameshift mutation (Fig 1A). The targeting vector was constructed, and the targeted ES clones (Fig 1B) were used to generate chimeric founder mice, from which two lines (57 and 71) of F1 mice were obtained. After removing the neomycin r cassette by Flpe deleter mice, Tet1 fl/+ mice were established, and were interbred to generate Tet1 fl/fl (T1 fl/fl ) mice. The CreER; Tet1 fl/fl (C/T1 fl/fl ) mice were generated by mating T1 fl/fl and CreER mice.

Validation of Tet1-ablation in hematopoiesis
First, to validate the genetic effect focused on hematopoiesis in C/T1 fl/fl mice, PB and c-Kit (+)-BM cells were analyzed in C/T1 fl/fl and T1 fl/fl mice treated with tamoxifen (Fig 2A). In these cells, efficient recombination in the floxed allele was detected at 4 weeks after treatment (Fig 2B). To examine the genetic effect more precisely, MLL-ENL-immortalized CD34(-)KSL cells highly expressing Tet1 (described in detail later) were analyzed, since the high expression of Tet1 is restricted to a small number of hematopoietic stem and immature progenitor cells. The Tet1 +/+ immortalized cells expressed the full-length transcript of Tet1 (an approximately 12.7-kb band, Fig 2C), whereas the Tet1 Δ/Δ immortalized cells did not. We also confirmed the expected alignment of exons in the Tet1 Δ/Δ immortalized cells by RT-PCR with primers spanning exons 7-11 ( Fig 2D). However, dot blot assays of DNA from the immortalized cells revealed little significant alteration in the levels of 5-mC or 5-hmC by Tet1-ablation (Fig 2E  and 2F), implying the limited involvement of Tet1 catalytic activity, and/or a redundant effect by Tet2 and/or Tet3. These results suggested that the Cre-mediated recombination was induced efficiently and correctly in hematopoietic cells, leading to genetic ablation of Tet1.

Hematological assessment of conditionally Tet1-ablated mice
We next assessed the biological effect of conditional Tet1-ablation on hematopoiesis. At the timepoints of 0 (just before the treatment), 4, 8, and 12 weeks after tamoxifen treatment, PB and BM (only 12 weeks) samples were collected from C/T1 fl/fl and T1 fl/fl mice (Fig 2A). However, no significant differences concerning white blood cells (WBCs) and hemoglobin (Hb) concentration in PB and the proportion of hematopoietic stem and progenitor cells (HSPCs) in BM were found (Figs 3A-3C and S1), while conditional Tet1-ablation did not affect the of 4 independent experiments (with 1 pair per each (total 8)) are shown in (B). Bar graphs show the mean and SD of data combined from two (A) and four independent experiments (C and D), respectively. �� p<0.005; n.s., not significant. (determined by two-way repeated measures ANOVA followed by Shaffer's modified sequentially rejective Bonferroni procedure (A), and two-tailed unpaired ttests (D), respectively). https://doi.org/10.1371/journal.pone.0248425.g003

PLOS ONE
Tet1 is not required for myeloid leukemogenesis by MLL-ENL Tet2 or Tet3 expression (Fig 3D). In addition, we generated BM chimera mice, in which the BM cells of recipient mice were replaced with C/T1 fl/fl or T1 fl/fl BM cells using BMT (S2A Fig), and treated the chimera mice with tamoxifen after confirmation of engraftment. In this experimental system, it was assumed that the donor-derived HSCs were subjected to proliferative stress, which was partly associated with aging, in the process of reconstitution of hematopoiesis before the conditional ablation of Tet1. However, no significant differences concerning WBCs in PB were found within an observation period of 34 weeks (S2B Fig). Furthermore, no lethal disease in the tamoxifen-treated mice receiving C/T1 fl/fl BM cells. These results suggested that Tet1 was dispensable in the maintenance of proper hematopoiesis, at least during the period of 34 weeks.

A role of Tet1 in leukemogenesis by MLL-ENL in HSPCs in vitro
Based on the finding that Tet1 is upregulated by MLL-AF9 [14], we analyzed the expression levels of Tet1 in retrovirally MLL-ENL-transduced KSL and MP cells (Fig 4A), and found similar increase of Tet1 expression, which was significant only in MP cells ( Fig 4B). Next, to examine whether Tet1 might be required for the maintenance of leukemic immortalization, C/T1 fl/fl CD34(+) KSL and MP cells were immortalized by MLL-ENL ( Fig 4A). CD34(-) KSL cells were not analyzed in these immortalization assays, due to difficulty in retroviral transduction of MLL-ENL into CD34(-) KSL cells, as previously described [21]. The 4-OHT treatment of the immortalized C/T1 fl/fl cells led to an approximately 80% reduction in Tet1 expression (Fig 4C), but clonogenicities in the treated cells were not significantly different from those in the control cells (Fig 4D). We also found no significant changes in the expression levels of the critical genes downstream of MLL-ENL or the epigenetic regulator genes associated with DNA methylation, Tet2, Tet3, and Dnmt genes, between the treated and the control cells (Figs 4C and S3). These results suggested that Tet1 was upregulated by MLL-ENL in HSPCs, but was not required for the maintenance of MLL-ENL-mediated immortalization in CD34(+) KSL and MP cells.

A role of Tet1 in leukemogenesis by MLL-ENL in LT-HSCs in vitro
In ME mice we established [21], induction of MLL-ENL led to leukemic transformation of CD34(-) KSL cells enriched for long-term HSCs (LT-HSCs), but not to the transformation of CD34(+) KSL and MP cells. To examine the role of Tet1 in MLL-fusion-mediated leukemogenesis in LT-HSCs, we generated CreER;Tet1 +/+ ;Tg-ME (C/ME) and CreER;Tet1 fl/fl ; Tg-ME (C/T1 fl/fl /ME) mice by mating the individual genetically engineered mice.
We first performed myeloid immortalization assays of CD34(-) KSL, CD34(+) KSL, and MP cells from T1 fl/fl , C/T1 fl/fl , C/T1 fl/fl /ME, and C/ME mice, respectively (Fig 5A). In the first plating, efficient recombination between loxP sites in Tet1 fl and/or Tg-ME alleles was confirmed (Fig 5B), accompanied by the similar reduction of Tet1 and/or induction of MLL-ENL among the cells harboring CreER, and Tet1 fl/fl and/or Tg-ME ( Fig 5C). Also, the four genotypes of CD34(-) KSL, CD34(+) KSL, and MP cells formed similar numbers of colonies, respectively. In the third plating, the induced MLL-ENL enabled C/T1 fl/fl /ME and C/ME CD34(-) KSL-derived cells to form numerous highly proliferative colonies, leading to immortalization, without significant differences in the number of the colonies between the two genotypes ( Fig 5D). Furthermore, both genotypes of the cells constituting the colonies exhibited similar morphological and immunophenotypic features of myeloid cells at various differentiation stages (Figs 5E and 6A). However, it should be noted that, in one of C/T1 fl/fl /ME CD34(+) KSL samples, a relatively higher MLL-ENL expression enhanced clonogenicity temporally, but failed to immortalize after additional replating (Fig 5C and 5D).
Next, to examine whether the conditional Tet1-ablation may affect the molecular mechanism of leukemic immortalization by the induced MLL-ENL, we analyzed the critical genes downstream of MLL-ENL and the epigenetic regulator genes associated with DNA methylation, in the first and third (only for CD34(-) KSL) plating (Figs 6B and S4). In the first plating, the expression levels of these genes tested were not significantly changed among T1 fl/fl -, C/T1 fl/fl -, C/T1 fl/fl /ME-, and C/ ME-derived cells in each subpopulation, except for a small reduction of Dnmt3a in C/T1 fl/fl -and C/T1 fl/fl /ME-derived cells in CD34(-) KSL, respectively. In comparison of C/T1 fl/fl /ME and C/ME CD34(-) KSL-derived cells in the third plating with those in the first plating, the Tet1 expression was non-significantly enhanced in the third plating of C/ME-derived cells, while the Tet1 expression did not increase with escape from the conditional ablation along serial plating in C/T1 fl/fl /MEderived cells. The MLL-ENL expression in the third plating was significantly enhanced in C/T1 fl/fl / ME-derived cells, but was still almost comparable to that in C/ME-derived cells (Fig 5C). The expression levels of Hoxa9, Meis1, and Evi1 in the third plating were significantly elevated in both genotypes, but were not significantly different between the two genotypes. Among the epigenetic genes tested, expression of Dnmt3b and the Dnmt1 was significantly more strongly enhanced in the C/T1 fl/fl /ME-and C/ME-derived cells, respectively, in comparison with each counterpart cells. Regarding Dnmt3b, our results were compatible with the finding in ES cells [26].
Taken together, these results indicated that the Tet1-ablation did not affect the leukemic immortalization by induced MLL-ENL in CD34(-) KSL, CD34(+) KSL, or MP cells.

The phenotype of MPD by the induction of MLL-ENL was not affected by conditional ablation of Tet1 in vivo
To examine whether the conditional Tet1-ablation may affect leukemogenesis by induced MLL-ENL in vivo, C/T1 fl/fl /ME and C/ME mice were preliminarily treated with tamoxifen,

Fig 6. Characterization of hematopoietic stem/progenitor cells after the induction of Tg-ME-derived MLL-ENL under simultaneous Tet1-ablation. (A)
The typical immunophenotype of the cells constituting the colonies derived from CreER; Tet1 fl/fl ; Tg-ME and CreER; Tet1 +/+ ; Tg-ME CD34(-) KSL cells at the end of the third round of plating in Fig 5. (B) The expression levels of Tet2, Tet3, Dnmt3b, Hoxa9, Meis1a, and Evi1 in the cells harvested at the end of the first (1R) and the third (3R, only in CD34(-) KSL-derived cells) rounds of plating in Fig 5, as assessed by RT-qPCR. Bar graphs show the mean and SD of three independent experiments in (B). ��� p<0.0005; �� p<0.005; � p<0.05; n.s., not significant; †, not significant in comparison of the other combinations. The p-values for the expression levels of the genes tested were determined by one-way ANOVA followed by Tukey-Kramer tests (CD34(+) and MP), or a combination of one-way ANOVA followed by Tukey-Kramer tests (1R) and two-way repeated measures ANOVA followed by Shaffer's modified sequentially rejective Bonferroni procedure (CreER; Tet1 fl/fl ; Tg-ME and CreER; Tet1 +/+ ; Tg-ME) adjusted using the Holm-Bonferroni correction (CD34(-) KSL). https://doi.org/10.1371/journal.pone.0248425.g006

Fig 7. The leukemogenesis by induction of Tg-ME-derived MLL-ENL under simultaneous Tet1 ablation in vivo. (A)
The experimental strategy for leukemogenesis assays enabling leukemic initiation in vivo. The two types (BMT #1 and #2) of the mixed-bone marrow (BM) chimera mice (from one mouse per each genotype per each experiment (total 8 (#1) and 4 (#2)) were prepared using BMT into CD45.1-, and CD45.1/45.2-positive recipient mice, respectively. Wt, wild-type; Tm, tamoxifen; ip, intraperitoneal injection; MPD, myeloproliferative disease. (B) Genomic PCR of PB cells from the mixed-BM chimera mice receiving CreER; Tet1 fl/fl ; Tg-ME (Tet1 fl/fl ) and CreER; Tet1 +/+ ; Tg-ME (Tet1 +/+ ) BM cells, just before (pre) and 4 weeks after (post) tamoxifen treatment. Bands derived from wild-type (wt), floxed (fl), and deleted (Δ) alleles at Tet1 locus (top panel), and those derived from germline (G) and recombinated (R) alleles regarding Tg-ME (bottom panel) are shown. (C) Chimerism analyses in peripheral blood cells just before (0) and 4 weeks after tamoxifen treatment in the cohort #2. P-values were determined by two-way repeated measures ANOVA followed by Shaffer's modified sequentially rejective Bonferroni procedure. (D) Survival curves of the mixed-BM chimera mice (cohort #2) receiving CreER; Tet1 fl/fl ; Tg-ME (n = 11) and CreER; Tet1 +/+ ; Tg-ME (n = 8) BM cells that were treated with tamoxifen. (E) Weight of the spleens from the moribund mice that developed lethal MPD. Data from two independent experiments were combined in (D). Bar graphs show the mean with SD of data combined from two independent experiments in (C) and (E). ��� p<0.0005; n.s., not significant. (determined by two-way repeated measures ANOVA followed by Shaffer's modified sequentially rejective leading to lethal MPD. To avoid ubiquitous induction of MLL-ENL, we beforehand prepared two types (#1 and #2) of mixed-BM chimera mice by BMT of mixture of equal ratios of BM cells from T1 fl/fl (#1), C/T1 fl/fl (#1), C/T1 fl/fl /ME (#1, #2), or C/ME (#1, #2) mice (CD45.2(+)), and wild-type mice (CD45.1(+)), into recipient mice (CD45.1(+) (#1) or CD45.1/CD45.2(+) (#2)) ( Fig 7A). In PB cells collected 4 weeks after the treatment of tamoxifen, efficient recombination between loxP sites in Tet1 fl and/or Tg-ME alleles was confirmed (Fig 7B), while no significant changes of chimerism in each genotype were found, except for a little increase in the chimera mice (#2) receiving C/T1 fl/fl /ME cells (Figs 7C and S5A). Over a period of about 200 days of observation after the treatment, the chimera mice receiving C/T1 fl/fl /ME or C/ME cells died with similar latency in each cohort (#1 and #2), while the mice (#1) receiving T1 fl/fl or C/ T1 fl/fl cells did not (Figs 7D and S5B). Almost comparable splenomegaly was found in the chimera mice receiving C/T1 fl/fl /ME or C/ME cells, in each cohort (Fig 7E). Immunophenotyping analyses of BM cells from the moribund mice revealed that the majority of the cells derived from CD45.2-positive donor cells, and expressed myeloid markers and c-Kit, and that these cells also expressed markers of immaturity, c-Kit/Sca-1, in some recipient mice (Fig 8A), similarly in each cohort. These features were also compatible with the morphological findings of BM cells showing MPD where myeloid cells at various stages of differentiation occupied BM (S5C Fig). In addition, in secondary transplantation of BM cells from moribund mice, both genotypes of MPD were aggressively reproduced with similar latencies (Fig 8B).

The Evi1 expression was markedly lower in CreER;Tet1 Δ/Δ ;Tg-ME samples
To further examine whether the conditional Tet1-ablation may affect the molecular mechanism of leukemogenesis by induced MLL-ENL in vivo, we analyzed the critical downstream genes and epigenetic genes in MPD cells from the moribund chimera mice (Fig 9A). The Tet1 expression was significantly and remarkably lower in C/T1 Δ/Δ /ME MPD cells. However, the expression levels of other epigenetic genes did not differ between the two genotypes of MPD cells (Figs 9A and S6). Unexpectedly, the MLL-ENL expression was mildly but significantly higher in C/T1 Δ/Δ /ME MPD cells, accompanied with significantly higher expression levels of Meis1 and Hoxa9. Nevertheless, in C/T1 Δ/Δ /ME MPD cells, the Evi1 expression was significantly and markedly lower (median: approximately 3%) in comparison with C/T1 +/+ /ME MPD cells, while it was exceptionally higher in one C/T1 Δ/Δ /ME MPD sample. In comparison of the expression levels in the MPD samples with those of the third plating of CD34(-) KSLderived cells in myeloid immortalization assays, the lower expression levels of Evi1 in the C/ T1 Δ/Δ /ME MPD samples were highlighted, although those of MLL-ENL were higher in the MPD samples (Fig 9B).
Taken together, in the leukemogenesis by induced MLL-ENL in vivo, the conditional Tet1ablation led to marked reduction of Evi1 expression, but did not critically affect the MPD phenotype, although an unexpected increase of the MLL-ENL expression in C/T1 Δ/Δ /ME MPD cells might cancel out the phenotypic change. Furthermore, Tet1 was not involved in the maintenance of the leukemogenesis.

Discussion
The present study redefines the role of TET1/Tet1 in myeloid leukemogenesis by the MLLfusion gene more closely. The aberrantly increased and reduced expression of TET1 is found Bonferroni procedure (C and E), and by log-rank tests (D), respectively). The gel images were cropped from S2 Raw images. https://doi.org/10.1371/journal.pone.0248425.g007

PLOS ONE
Tet1 is not required for myeloid leukemogenesis by MLL-ENL to be associated with myeloid and lymphoid leukemogenesis [13][14][15], respectively, although loss-of-function mutations of TET2/Tet2 develops both leukemogenesis [11]. MLL-fusion protein upregulates the TET1/Tet1 expression in the myeloid leukemogenesis [14], as confirmed in the present study. Interestingly, TET1 is found to be crucial for the proliferation of human T-cell acute lymphoblastic leukemia, suggesting context-dependent roles of TET1/Tet1 in leukemogenesis [27]. Indeed, our results suggested that Tet1 may not be necessarily essential for MLL-fusion-mediated myeloid leukemogenesis.
Our leukemogenesis assays in vivo showed that the conditional Tet1-ablation did not affect the myeloid leukemogenesis phenotype by induced expression of MLL-ENL. In comparison with BMT of retrovirally MLL-fusion-transduced HSPCs [14,17,22], our mixed-BM chimera enabled leukemic initiation to occur without temporary culture in vitro, thereby excluded the bias of engraftment and some inevitable influence by culture in vitro, as in conditional/inducible MLL-fusion-mediated mouse models [28,29]. Meanwhile, compared with conventional Tet1 knockout mice [13][14][15], the Tet1 protein expression remained at the early phase of the conditional ablation. Thus, partly due to differences in the approaches, our results may differ from the finding [14] that Tet1 plays an essential role for the myeloid leukemogenesis.
Tet1 was dispensable for the leukemic immortalization by MLL-ENL in CD34(-) KSL cells enriched for LT-HSCs, in which the role of Tet1 had not been well-characterized, while, in retrovirally MLL-AF9-transduced Lin -BM cells, Tet1 is critically involved in the myeloid leukemogenesis [14]. Since the retroviral transduction into LT-HSCs is inferred to be very difficult [21], this discrepancy concerning the role of Tet1 may reflect the difference of a subpopulation of cells targeted by MLL-fusion genes.
Unexpectedly, the expression level of induced MLL-ENL in MPD cells was higher under the Tet1-ablated condition. Since the induced MLL-ENL expression at the third plating of CD34 (-) KSL-derived cells under the Tet1-ablated condition was comparable to that under the Tet1 +/+ condition in vitro, the higher expression of induced MLL-ENL in vivo implied that the enrichment of cells that highly expressed MLL-ENL was more enhanced in association with the Tet1-ablation, in the process leading to MPD. The unexpected genetic effect, which may accelerate MPD under the Tet1-ablated condition, might reduce the effect of deceleration by loss of Tet1. However, in contrast to the finding that Hoxa9 and Meis1a were markedly downregulated under Tet1-deficent conditions [14], our results showed that those genes were upregulated by MLL-ENL under Tet1-ablated conditions in the same way as Tet1 +/+ conditions. In addition, irrespective of the genotypes, secondary BMT reproduced the MPD in similar latencies. In retrovirally MLL-ENL-immortalized CD34(+) KSL/MP cells, the conditional Tet1ablation did not cause any significant changes in the clonogenicity or expression of critical downstream genes, although it should be noted that the ablation was incomplete presumably due to escape from recombination in a part of 4-OHT-treated cells, and/or the remaining expression of Tet1 transcripts after the recombination. These results suggested that Tet1 might not be essential for the MLL-ENL-mediated leukemogenesis, even after taking into consideration of the high expression of induced MLL-ENL in Tet1-ablated MPD cells.
The conditional Tet1-ablation was closely associated with very low Evi1 expression in myeloid leukemogenesis by the induced MLL-ENL in vivo, but not in the myeloid immortalization assays in vitro. Considering the epigenetic function of Tet1 [8][9][10] and the recent finding that loss-of-function Tet leads to a localized increase in DNA methylation [30], further analyses, such as comprehensive profiling of the methylation status of regulatory genomic regions of Evi1 [31], may be needed in the future. Since MLL-ENL upregulates EVI1/Evi1 through promoter binding [32] and epigenetic regulation by histone modifications [33], epigenomic analyses regarding Evi1 may be also helpful. Another hypothesis is that Tet1-ablation might lead to leukemic initiation by MLL-ENL more efficiently in relatively mature progenitor cells with low expression of Evi-1, since MLL-fusion genes upregulate Evi1 in KSL cells more strongly than in myeloid progenitor cells [32,34,35]. Thus, further analyses including hematopoietic stem/ progenitor-specific ablation of Tet1 will be also needed to test this hypothesis.
In conclusion, to examine the role of Tet1 in the myeloid leukemogenesis more precisely, we generated conditional Tet1-knockout mice, and constructed a mouse model enabling leukemic initiation, in combination with inducible MLL-ENL transgenic mice. The phenotypes of leukemic transformation by MLL-ENL were not critically affected by conditional Tet1-ablation. However, the low Evi1 expression in MPD cells under the Tet1-ablated condition may be associated with a role of Tet1 in leukemic initiation by MLL-ENL. Furthermore, Tet1 was dispensable in leukemic maintenance in secondary BMT, suggesting that the role of Tet1 may not be essential for, at least, MLL-ENL-mediated myeloid leukemogenesis. Also, our results suggested that the therapeutic application of Tet1 inhibition in AML may need careful assessment.
Supporting information S1 Checklist. ARRIVE guidelines checklist. (PDF) S1 Fig. The proportion of myeloid, B-, and T lymphoid cells in peripheral blood from conditionally Tet1-ablated mice. The blood samples were measured just before (0), 4, 8, and 12 weeks after tamoxifen treatment as shown in Fig 2A (n = 5 per each (the same mice as used in Fig 2A)). Bar graphs show the mean and SD of data combined from two independent experiments. n.s., not significant. (determined by two-way repeated measures ANOVA followed by Shaffer's modified sequentially rejective Bonferroni procedure). show the mean and SD of three independent experiments. � p<0.05; n.s., not significant; †, not significant in comparison of the other combinations. The p-values for the expression levels of the genes tested were determined by one-way ANOVA followed by Tukey-Kramer tests (CD34(+) and MP), or a combination of one-way ANOVA followed by Tukey-Kramer tests (1R) and two-way repeated measures ANOVA followed by Shaffer's modified sequentially rejective Bonferroni procedure (CreER; Tet1 fl/fl ; Tg-ME and CreER; Tet1 +/+ ; Tg-ME) adjusted using the Holm-Bonferroni correction (CD34(-) KSL).  Fig 5A) just before (0) and 4 weeks after tamoxifen treatment in the cohort #1. P-values were determined by two-way repeated measures ANOVA followed by Shaffer's modified sequentially rejective Bonferroni procedure. (B) Survival curves of the mixed-bone marrow (BM) chimera mice (cohort #1) receiving CreER; Tet1 fl/fl ; Tg-ME (n = 13), CreER; Tet1 +/+ ; Tg-ME (n = 8), CreER; Tet1 fl/fl (n = 4), and Tet1 fl/fl (n = 5) BM cells that were treated with tamoxifen. (C) The typical morphology of BM cells from the moribund mice that developed lethal MPD. Cells were stained with Wright-Giemsa. Magnification, 200×; scale bars, 50 μm. Bar graphs show mean with SD of data combined from two independent experiments in (A). n.s., not significant. (determined by two-way repeated measures ANOVA followed by Shaffer's modified sequentially rejective Bonferroni procedure).