TET2-mediated mRNA demethylation regulates leukemia stem cell homing and self-renewal

Summary

TET2 is recurrently mutated in acute myeloid leukemia (AML) and its deficiency promotes leukemogenesis (driven by aggressive oncogenic mutations) and enhances leukemia stem cell (LSC) self-renewal. However, the underlying cellular/molecular mechanisms have yet to be fully understood. Here we show that Tet2 deficiency significantly facilitates leukemogenesis in various AML models (mediated by aggressive or less aggressive mutations) through promoting homing of LSCs into bone marrow (BM) niche to increase their self-renewal/proliferation. TET2 deficiency in AML blast cells increases expression of Tetraspanin 13 (TSPAN13) and thereby activates the CXCR4/CXCL12 signaling, leading to increased homing/migration of LSCs into BM niche. Mechanistically, TET2 deficiency results in the accumulation of methyl-5-cytosine (m⁵C) modification in TSPAN13 mRNA; YBX1 specifically recognizes the m⁵C modification and increases the stability and expression of TSPAN13 transcripts. Collectively, our studies reveal the functional importance of TET2 in leukemogenesis, leukemic blast cell migration/homing, and LSC self-renewal as an mRNA m⁵C demethylase.

In Brief

Li and colleagues report that TET2 deficiency facilitates LSC homing into bone marrow microenvironment, giving rise to increased LSC self-renewal and fast leukemogenesis in various animal models. Mechanistically, TET2 deficiency enhances the stability of TSPAN13 mRNA as an mRNA m⁵C demethylase and thereby triggers the CXCR4/CXCL12 signaling.

Graphical Abstract

Introduction

In the normal hematopoietic system, gene expression is strictly orchestrated not only by genetic enhancers and promoters, but also by epigenetic and epitranscriptomic modifications. Evidence is emerging that the dysregulation of covalent chemical modifications in DNA or RNA leads to hematopoietic malignancies, including acute myeloid leukemia (AML) ^1–4. For instance, somatic depletions and loss-of-function mutations of tet methylcytosine dioxygenase 2 (TET2) have been characterized in 15–20% of AML, 30% of myelodysplastic syndrome (MDS), and 50% of chronic myelomonocytic leukemia (CMML) cases ^5–7. TET2 catalyzes the conversion of 5-methylcytosine (5mC) to 5-hydroxy/formyl/carboxyl-cytosine (5hmC, 5fC, 5caC) to mediate DNA demethylation and plays a central role in regulating gene expression at the transcriptional level ^8–10. The vast majority of TET2 mutations lose demethylation activity and have been recognized as the initial step of myeloid malignant transformation in AML ^4,11. Moreover, TET2 deficiency confers poor prognosis and drug resistance in AML patients ^12–16. In addition to oxidization of DNA 5mC modification, recent studies support a novel function for TET2-mediated demethylation of methyl-5-cytosine (termed as ‘m⁵C’ to distinguish from DNA 5mC) in RNAs, including messenger RNAs (mRNAs) ^17–20. However, the systemic investigation of the role and biological relevance of mRNA m⁵C is still in its infancy ^21–23. It is completely unknown whether and (if so) how TET2-induced mRNA m⁵C demethylation contributes to leukemogenesis or tumorigenesis in general.

AML, one of the most common and aggressive types of blood malignancies, is featured by the clonal expansion of immature leukemia stem cells (LSCs). Despite currently available treatments, over 70% of AML patients do not survive over 5 years ^24,25. Relapse remains the principal cause of treatment failure and occurs in over 50% of AML patients ²⁶. LSCs are at the apex of the AML cellular hierarchy and are considered the root cause of drug resistance and AML relapse ^27–29. Thus, eradicating LSCs is critical in curing AML. LSCs reside in specialized microenvironments called “niches” in the bone marrow (BM) to support their survival and maintain their self-renewal capacity ³⁰. It was known that the direct interplay between the chemokine receptor 4 (CXCR4) and its ligand CXCL12 (also called stromal cell-derived factor-1, SDF-1) primes the crosstalk between LSCs and BM stromal cells in the microenvironment ³¹. Pharmacological inhibition of CXCR4 prevents the anchorage of AML cells and promotes the mobilization of LSCs out of the endosteal niche, thereby improving their vulnerability to chemotherapy agents ³².

The function of TET2 in hematopoietic differentiation and LSC maintenance has been well documented ^{6,11,14,33–35}, but the underlying cellular/molecular mechanisms remain largely elusive. Although previous studies suggested that TET2 deficiency facititated leukemogenesis driven by aggressive mutations such as MLL-AF9 and AML1-ETO9a ^36,37, the impact of TET2 deficiency on other AML models driven by less aggressive mutations has yet to be determined. Moreover, the roles of TET2 in BM microenvironment/niche programming have yet to be investigated. Here we show that TET2 functions as RNA demethylase, mediating the post-transcriptional regulation of tetraspanin 13 (Tspan13) mRNA stability. As a result, Tet2 deficiency causes the activation of the Tspan13/Cxcr4 axis, which in turn leads to increased LSC homing/self-renewal, thereby promoting leukemogenesis. Overall, our studies uncover previously unappreciated roles and underlying molecular mechanisms of TET2 in regulating BM niche programming and LSC homing/self-renewal in AML pathogenesis, as an RNA demethylase.

Results

Decreased expression or deficiency of TET2 is associated with poor prognosis in AML patients

Given that previous studies predominantly focus on loss-of-function mutations of TET2, we tried to analyze the overall expression level of TET2 and its clinical relevance in AML. In the analysis of The Cancer Genome Atlas (TCGA) AML cohort, we found that lower expression of TET2 transcript is significantly correlated with shorter survival in AML patients (Figure 1A). TET2 expression is significantly suppressed in AML patients, including those carrying t(15;17), t(8;21), MLL-rearrangement (MLL-r)/t(11q23), and normal karyotype, relative to healthy controls (Figure 1B) ³⁸. Moreover, the expression of TET2 as well as TET3, but not TET1, is also decreased in adverse-risk AML patients relative to the patients with favorable-risk AMLs ³⁹ (Figures 1C and S1A–S1B). Thus, a lower expression level of TET2 is significantly associated with a poor prognosis in AML patients. Similarly, AML patients with TET2 loss-of-function mutations are also associated with shorter overall survival compared to patients with wildtype TET2 (Figure 1D). Notably, TET2 mutations frequently co-occurred with the mutations in other genes such as FLT3, DNMT3A, NPM1, and RAS, as well as translocations such as t(11q23)/MLL-rearrangements and t(8;21)/RUNX1-RUNXT1 in AML patients (Figures S1C–S1D). Furthermore, single-cell RNA-seq data from AML patients ⁴⁰ indicates that expression level of TET2 is significantly lower in primitive (immature) AML cell population compared to the relatively mature (differentiated) AML cell population (Figure S1E). In addition, TET2, but not TET1 or TET3, is expressed at a significantly lower level in normal hematopoietic stem/progenitor cells (HSPCs) relative to normal mature myeloid cells (Figures S1F–S1G). Such data suggest that TET2 might be required for the differentiation of HSPCs and LSCs.

Figure 1. TET2 deficiency predicts poor prognosis in AML patients and promotes primary leukemogenesis.

(A) Kaplan-Meier survival curves of TCGA AML patients TET2 high (n=44; roughly the top 1/4) and low/medium (n=119; the remaining patients) expression. The X-tile software ⁶⁹ was used to determine the optimal cutoff values for predicting survival.

(B) Violin plots showing the expression of TET2 in healthy donors and AML patients. The expression values, derived from Affymetrix exon arrays, were log₂ transformed.

(C) Violin plots showing the relative expression of TET2 in favorable and adverse AML patients in the Beat AML cohort.

(D) Kaplan-Meier survival curves of AML patients with or without TET2 mutation. WT, wild-type (without TET2 mutation in the Beat AML cohort, http://vizome.org/aml/). n = 54 for TET2 mutation group; n = 415 for TET2 WT group.

(E) Schematic outline of primary BMT models driven by various gene mutations and translocations.

(F-H) Kaplan-Meier survival curves of NRAS^mut & AE9a (F), NRAS^mut & PML-RARA (G), and DNMT3A^mut & NRAS^mut (H)-driven primary murine AML models upon Tet2 deficiency. Tet2 (ΔE8–10) strain was used for F and G; Tet2 (ΔE3) strain was used for H.

(I-K) The white blood cell (WBC) count (I), red blood cell (RBC) count (J), and platelet (PLT) count (K) in the recipient mice transplanted with Tet2 ^+/+ versus Tet2 ^−/− (ΔE8–10) primary NRAS/AE9a (N&A) or NRAS/PML-RARA (N&P) cells (n = 4).

(L) The spleen weight and representative spleens of DNMT3A^mut & NRAS^mut -driven primary murine AML models (n = 3).

(M and N) The WBC count in the recipient mice transplanted with primary DNMT3A^mut & NRAS^mut (M)- or FLT3^ITD (N)-transformed Tet^+/+ or Tet2^−/− mouse BM progenitor cells (n = 3).

(O) Effect of Tet2 deficiency on the engraftment of FLT3^ITD leukemia cells into peripheral blood (PB), spleen (SP), and BM of primary recipient mice. The samples were collected on day 52 post transplantation (n = 3).

(P) The representative spleens of FLT3^ITD-driven primary murine AML models. The samples were collected on day 52 post transplantation.

Unpaired two-tailed Student’s t-test (B, C, and I-O); log-rank test (A, D, and F-H); *p < 0.05, **p < 0.01, ***p < 0.001.

See also Figures S1.

Tet2 deficiency induces rapid primary leukemogenesis

To further evaluate the role of Tet2 in primary leukemogenesis, we have generated various AML models driven by AML1-ETO9a (AE9a, resulting from t(8;21) translocation) ⁴¹, MLL-AF9 (MA9, MLL-r/t(11q23)), PML-RARA (t(15;17) translocation), NRAS mutation, FLT3^ITD, and DNMT3A mutation, coupled with Tet2^+/+ and Tet2^−/− mouse models (Figure 1E). Such selection is clinically relevant because TET2 mutation and/or transcriptional suppression have been characterized in those AML subtypes (see Figures 1B and S1C–S1D). Notably, Tet2 depletion dramatically promoted primary leukemogenesis in the BM transplantation (BMT) recipient mice (Figures 1F–1H and S1H). Furthermore, Tet2 depletion exacerbated leukocytosis, anemia, and thrombocytopenia (Figures 1I–1K) in peripheral blood (PB) of the recipient mice. Tet2 deficiency also led to splenomegaly, increased white blood cell (WBC) count, and a significant increase in the engraftment of primary AML cells into PB, BM, and spleen (Figures 1L–1P and S1I–S1L). Histopathology demonstrated the heavier infiltration of Tet2^−/−-leukemic blasts into spleen and liver, which disrupted their normal architecture (Figures S1M and S1N). Wright-Giemsa staining of PB and BM revealed the notably larger and less differentiated blasts in the Tet2-deficient cohorts compared to the Tet2^+/+ counterparts (Figures S1M). Via flow cytometry, we verified that Tet2 deficiency significantly attenuated myeloid differentiation in primary AML cells (Figures S1O–S1P).

Tet2 deficiency enhances LSC self-renewal

To decipher the mechanism(s) by which Tet2 deficiency dramatically promotes leukemogenesis in vivo, we collected primary Tet2^−/− and Tet2^+/+ AML cells for RNA-seq analysis. We found that Tet2 deficiency significantly increased expression of the genes enriched in the “Hematopoiesis Stem Cell” pathway (Figures 2A–2B). To validate whether Tet2 deletion impacts the self-renewal of LSCs, we conducted in vitro colony-forming/re-plating assay. More colonies with larger sizes and more cells were observed in Tet2 deficient pre-LSCs transduced with NRAS/AE9a in the serial replating assay (Figures 2C and S2A–S2B). Meanwhile, we analyzed the percentage of LSCs (Lin⁻c-Kit⁺Sca-1⁺, LSK) by flow cytometry (Figure S2C). As expected, Tet2 deficiency in NRAS/AE9a transformed pre-leukemic cells led to a significant increase in the Lin⁻ and LSC subpopulations, especially after 3 rounds of replating (Figure S2D and 2D). Consistently, we verified that Tet2 deficiency attenuated myeloid differentiation of the pre-leukemic cells via checking myeloid differentiation markers Gr-1 and Mac-1 (Figure S2E). In the MA9-driven pre-AML model, Tet2 deficiency also resulted in an increase in Lin⁻ and LSC subpopulations (Figure S2F–S2G). Consistently, Tet2 deficiency significantly potentiated the colony-forming/replating ability in additional pre-AML models driven by FLT3^ITD, DNMT3A-mutant/NRAS-mutant, DNMT3A-mutant/ FLT3^ITD, or NRAS-mutant/PML-RARA (Figures 2E–2F and S2H).

Figure 2. Tet2 depletion significantly enhances LSC self-renewal while its overexpression attenuates LSC self-renewal and AML progression in various models.

(A) Enrichment plot of Ivanova hematopoiesis stem cell pathway which was activated in Tet2^−/− (ΔE8–10) primary NRAS/AE9a AML cells.

(B) IGV tracks of Tet2 in Tet2 ^+/+ or Tet2 ^−/− (ΔE8–10) NRAS/AE9a AML cells.

(C) Colony numbers from serial replating assays with Tet2 ^+/+ and Tet2 ^−/− (ΔE3) primary leukemia cells. The representative data from NRAS/AE9a model was shown.

(D) FACS plots depicting the percentage of LK (Lin⁻c-Kit⁺Sca-1⁻) and LSK (Lin⁻c-Kit⁺Sca-1⁺) subpopulations in Tet2 ^+/+ and Tet2 ^−/− (ΔE3) NRAS/AE9a AML cells.

(E) The representative colony images of Tet2 ^+/+ or Tet2 ^−/− (ΔE3) primary AML cells driven by FLT3^ITD, DNMT3A^mut & NRAS^mut, and DNMT3A^mut & FLT3^ITD.

(F) Effect of Tet2 depletion on the colony numbers of primary AML cells driven by FLT3^ITD (left), DNMT3A^mut & NRAS^mut (middle), and DNMT3A^mut & FLT3^ITD (right).

(G) Kaplan-Meier survival curves of secondary BMT recipient mice. The leukemic BM cells isolated from the primary BMT recipient mice of Tet2^+/+ + MA9 or Tet2^−/− (ΔE8–10) + MA9 were used as donor cells.

(H) Representative spleen images of the secondary BMT models. All the spleens were collected on day 27 post transplantation of Tet2^+/+ + MA9 or Tet2^−/− (ΔE8–10) + MA9 primary AML cells.

(I) The in vivo limiting dilution assay with Tet2^+/+ + MA9 or Tet2^−/− (ΔE8–10) + MA9 primary BM leukemia cells. CD45.2 recipients (n = 5 for each group) were transplanted with different doses of leukemia cells and the estimated LSC frequencies were displayed.

(J) TET2 protein levels in the CD34⁺ and CD34⁻ populations of primary AML patient samples as determined by intracellular staining.

(K) Comparison of mean fluorescence intensity (MFI) of TET2 protein levels between CD34⁺ and CD34⁻ populations in primary AML patient samples.

(L-M) Effect of TET2 overexpression on the colony number (L) and cell number (M) of Mono Mac 6 AML cells.

(N) Effect of TET2 overexpression on the colony number of AML PDX (PDX-148, carrying DNMT3A mutation) cells.

(O) The bioluminescence images of AML PDX model (PDX-129, carrying a complex karyotype) with or without TET2 overexpression. The images were photted on day 25 post transplantation.

(P) The statistical bioluminescence signals of AML PDX model with or without TET2 overexpression.

(Q) Kaplan-Meier survival curves of AML PDX models with or without TET2 overexpression.

(R) Kaplan-Meier survival curves of Mono Mac 6-induced AML xenograft models with or without TET2 overexpression.

Unpaired two-tailed Student’s t-test (C, F, L-N, and P); log-rank test (G, Q, and R); paired two-tailed Student’s t-test (K); *p < 0.05, **p < 0.01, ***p < 0.001.

See also Figures S2 and S3.

To assess the impact of Tet2 depletion on LSC self-renewal in vivo, we performed secondary BMT. We found that Tet2-deficient MA9 leukemic cells showed a dramatically higher engraftment/expansion in BM (Figure S3A) and developed AML with much shorter latency (Figure 2G) than did the Tet2-wildtype counterparts, accompanied by a significant increase in WBC count (in PB) as well as spleen size and weight (Figures 2H and S3B–3C). We next conducted in vivo limiting dilution assay (LDA) to quantitatively assess the effect of Tet2 deficiency on LSC self-renewal capability. Remarkably, Tet2 deletion significantly increased LSCs frequency in MA9 AML model by over 30-fold (Figure 2I). We further confirmed the robust impact of Tet2 depletion on LSC frequency in NRAS/AE9a- and NRAS/PML-RARA-transformed pre-leukemic cells as detected by in vitro LDA assays (Figures S3D–S3F). In addition, we determined the percentages of the Lin⁻ c-Kit⁺Sca-1⁻CD34⁺FcγRII/III⁺ granulocyte-macrophage progenitor-like (L-GMP; representing LSC population in MA9-induced AML⁴²) cells as well as differentiated myeloid cells in the BM of primary recipient mice of the MA9 AML model (Figure S3G). We showed that Tet2 deficiency evidently increased the undifferentiated L-GMP population and substantially decreased the differentiated (Gr-1⁺Mac-1⁺) cell population (Figures S3H–S3J). Altogether, our data demonstrate that TET2 loss increases LSC self-renewal and suppresses myeloid differentiation in vitro and in vivo in various murine AML models.

TET2 overexpression suppresses human AML blast cell repopulation and inhibits AML progression in vivo

Consistent with the suppressive role of Tet2 in murine LSC self-renewal, the expression level of TET2 is also significantly lower in human CD34⁺ primitive AML blasts compared to the CD34⁻ bulk AML cells (Figures 2J–2K). Moreover, forced expression of TET2 dramatically decreased the colony-forming/replating ability of primary AML PDX cells and Mono Mac 6 AML cells (Figures 2L–2N), implying that TET2 may also play a suppressive role in human AML repopulation and progression. To evaluate the in vivo role of TET2 in AML in a setting relevant to human disease, we utilized patient-derived xenograft (PDX) and ‘human-in-mouse’ xenograft AML models. As expected, forced expression of TET2 in AML PDX cells (isolated from a relapsed patient) significantly suppressed AML blast cell repopulation, decreased leukemia burden and improved the overall survival in recipient mice (Figures 2O–2Q and S3K). Consistently, overexpression of TET2 in Mono Mac 6 AML cells (carrying MA9 translocation) significantly mitigated splenomegaly, decreased WBC number, suppressed engraftment into spleen and BM of NRGS mice, and substantially prolonged their survival (Figures 2R and S3L–S3P). Collectively, our studies have demonstrated that TET2 plays a potent tumor-suppressive role in various human and murine AML models, in which TET2 suppresses the self-renewal/repopulation of AML blast cells or LSCs.

Tet2 deficiency accelerates LSC homing/migration in vitro

To determine whether Tet2 deficiency offered a profound growth advantage to facilitate rapid and aggressive leukemogenesis, we analyzed the effects of Tet2 deficiency on the growth/proliferation and cell cycle of primary AML cells in vitro. Strikingly, Tet2 depletion had very weak effects on the growth/proliferation and cell cycle of primary AML cells in liquid culture (Figures S4A–S4E). In contrast, when co-cultured with stroma cells, Tet2^−/− AML cells substantially outperformed Tet2^+/+ AML cells in proliferation/growth (Figures 3A–3D), implying that the microenvironment plays a pivotal role in regulating Tet2 deficiency-mediated rapid leukemogenesis in vivo.

Figure 3. Tet2 deficiency promotes the migration of AML blast cells or LSCs into surrounding stroma to maintain their self-renewal.

(A-B) Effect of Tet2 depletion on the proliferation of primary MA9 cells in single culture condition (A) and stroma cell (OP-9) co-culture condition (B). The numbers of days after culture or co-culture are indicated under the x-axis.

(C-D) The effect of Tet2 deficiency on the proliferation of MA9-transformed leukemic cells after co-culturing with OP9 stromal cells for 8 days, as determined by 5-ethynyl-2′-deoxyuridine (EdU) staining. C, the statistical results of the percentage of EdU positive cells; D, representative flow cytometry data.

(E) Scheme showing the single culture and co-culture of primary AML cells with OP-9 stromal cells before colony-forming assay (left) as well as the representative colony images (right). The AML cells were co-cultured with OP-9 cells for 5 days before colony-forming assay. For each group, 150 AML cells were seeded into 96-well plate.

(F-G) The colony number (F) and LSC frequency (G, in vitro LDA) of Tet2^+/+ and Tet2^−/− primary MA9 AML cells with or without co-culture.

(H) Scheme showing the transwell assay to evaluate the migration ability of Tet2^+/+ and Tet2^−/− primary AML cells.

(I-K) The effect of Tet2 deficiency on the migration ability of primary AML cells carrying MA9 (I), FLT3^ITD (J), and DNMT3A & NRAS mutations (K).

(L) Effect of TET2 KO on the migration ability of THP1 AML cells.

Unpaired two-tailed Student’s t-test (A-C, F, and I-L); **p < 0.01, ***p < 0.001.

See also Figures S4.

Considering that LSCs reside in the BM niche, which plays an essential role in maintaining LSC self-renewal ability, we sought to investigate the effect of Tet2 deficiency on LSC homing to BM microenvironment and on LSC frequency. Through in vitro LDA assays, we showed that Tet2 loss significantly increased LSC frequency in the single culture system, while co-culture of LSCs, particularly the Tet2-deficient LSCs, with stroma cells substantially potentiated their self-renewal ability (Figures 3E–3G). Additionally, employing the transwell assay, we found that Tet2/TET2 loss significantly promoted the migration of various types of murine and human AML cells into stromal cells (Figures 3H–3L). Collectively, our results suggest that TET2 deficiency can promote LSC homing/migration to stromal cells and subsequently increase LSC self-renewal in vitro.

Tet2 deficiency facilitates homing of LSCs in vivo

During BMT, the homing and/or anchoring of LSC into the BM niche plays a predominant role to maintain/enhance LSC self-renewal in vivo. We employed the in situ confocal immunofluorescence (IF) imaging and flow cytometry (Figure 4A) to investigate whether Tet2 deficiency-induced LSC self-renewal was due to the increased migration into BM niche. Osteopontin (OPN), an extracellular molecule secreted by osteoblast, can anchor LSCs to support their survival and self-renewal in BM microenvironment ⁴³ and represent the stem cell niche in the BM. Our in situ staining data showed that there were significantly more Tet2 deficient-AML cells migrating into the OPN⁺ BM-derived stromal cells relative to the Tet2 wildtype group (Figures 4B–4C). By flow cytometry, we further validated that Tet2 deficiency indeed enhanced the homing of primary MA9 AML cells (Figures 4D and S5A). Furthermore, we collected the samples at various time points (1, 4, 12, and 16 hours) to carefully evaluate the effect of Tet2 deficiency on the homing of AML blasts (i.e., immature AML cells, including LSCs). We were able to detect significantly increased homing of Tet2-deficient AML blasts starting at around 12 hours post-transplantation (Figures 4E and S5B). In the NRAS/AE9a-driven primary AML models and MA9-driven secondary AML models, Tet2 deficiency also significantly enhanced homing of AML blasts (Figures 4F–4G and S5C). Conversely, overexpression of TET2 in AML PDX cells substantially attenuated their homing into the BM of NRGS mice (Figure 4H). In the BM niche, AML blasts, particularly LSCs, are surrounded by various stromal cells which support their stemness and self-renewal ability ³¹. Our data suggest that co-culture with stroma cells notably increases the migration/homing, proliferation, and self-renewal ability of Tet2-deficient AML cells (see Figures 3E–3L and 3A–3D). Therefore, the increased migration and homing of primary AML cells into BM niche might be a major contributor to Tet2 deficiency-induced rapid and aggressive leukemogenesis in vivo.

Figure 4. Te2 depletion facilitates LSC homing in vivo.

(A) Schematic outline of the method to evaluate LSC homing in vivo.

(B) The in situ immunofluorescence (IF) of MA9 primary leukemia cells (CFSE-labeled; Green) and BM osteoblasts (Osteopontin, OPN; Red) in the femurs from Tet2^+/+ and Tet2^−/− (ΔE8–10) group. While arrows indicated CFSE-labeled leukemia cells. The samples were collected at 16 hours post transplantation.

(C) Quantitative and statistical analysis of in situ IF images in Figure 4B.

(D) Effect of Tet2 depletion on the homing of primary MA9 AML cells into the BM of recipient mice. The data were determined by flow cytometry. The samples were collected at 16 hours post transplantation.

(E) Comparison of the percentage between Tet2^−/− and Tet2^+/+ primary MA9 AML cells homing into BM of recipient mice via flow cytometry. The samples were collected at 1, 4, 12, and 16 hours post transplantation.

(F) Effect of Tet2 depletion on the homing of primary NRAS/AE9a AML cells. The data were determined by flow cytometry. The samples were collected at 16 hours post transplantation.

(G) Evaluation of effect of Tet2 depletion on homing via secondary MA9 BMT models. The samples were collected at 16 hours post transplantation.

(H) Effect of TET2 overexpression on the homing of AML PDX cells into the BM of NRGS recipient mice. The samples were collected at 16 hours post transplantation.

For C-H, data were presented as mean ± SEM (n = 3). ns, not significant; *p < 0.05, **p < 0.01, ***p < 0.001; unpaired two-tailed Student’s t test.

See also Figures S5.

Tspan13 is responsible for Tet2 deficiency-enhanced LSC homing and self-renewal

To dissect the mechanism(s) by which Tet2 depletion promotes LSC homing and self-renewal, we performed RNA-seq using two Tet2 knockout (KO) models. One model lacked exons 8 to 10 ⁴⁴ (see Figure 2B), and the other lacked exon 3 ¹¹ (Figure S6A). In line with the Tet2^−/− (ΔE8–10) NRAS/AE9a AML model, deletion of exon 3 (Tet2^−/− (Δ3)) also dramatically increased expression of the genes associated with hematopoiesis stem cell signature in the NRAS/AE9a AML model (Figure S6B). Amongst all the core-enriched genes of stem cell pathways in the two Tet2^−/− NRAS/AE9a AML models, we identified 41 overlapped potential targets (Figures 5A–5B and S6C). Furthermore, we analyzed the expression correlation between TET2 and the 41 co-enriched genes in 3 distinct AML cohorts. We identified that only TSPAN13 and synaptic vesicle glycoprotein 2A (SV2A) exhibited a significantly negative correlation with TET2 in all the 3 human AML cohort datasets (Figures 5C and S6D–S6E). TSPAN13, but not SV2A, is expressed at a significantly higher level in patients with adverse-risk AML than in patients with favorable-risk AML prognosis (Figures 5D–5E), in an expected pattern opposite to that of TET2 (see Figure 1C). Tet2 deficiency-mediated upregulation of Tspan13 was confirmed in primary murine AML cells by RNA-seq and qPCR (Figures 5F–5G). TSPAN13 belongs to the tetraspanin family, which is implicated in adhesion, trafficking, and cell signaling within the niche of hematopoietic stem cells (HSCs) and LSCs ^45–48. To evaluate the functional involvement of TSPAN13 in AML pathogenesis, we used a lentiviral vector-based short hairpin RNA (shRNA) system to knock down the expression of Tspan13 (sh1 and sh2) in Tet2-deficient primary AML cells (Figure 5G). Depletion of Tspan13 significantly decreased the colony-forming ability of Tet2-deficient AML blasts (Figures 5H–5J). LDAs demonstrated that Tspan13 KD significantly decreased LSC frequencies in AML blasts (Figures S6F–S6G). Furthermore, Tspan13 depletion not only attenuated the LSC homing into the BM niche (Figure S6H–S6I) but also almost completely reversed the enhanced LSC homing in the BM microenvironment caused by Tet2-deficiency (Figured 5K). Additionally, genetic depletion of Tspan13 substantially reversed Tet2 deficiency-enhanced LSC self-renewal (Figure 5L). We also conducted BMT to define whether Tspan13 was required for AML development in vivo and showed that Tspan13 depletion dramatically inhibited AML progression driven by Tet2-deficient primary AML cells and led to a significantly prolonged overall survival (Figure 5M) and dramatically suppressed engraftment of primary AML cells in the recipient mice (Figure 5N). Altogether, our data suggest that Tspan13 is a functionally essential target of Tet2, and the upregulation of Tspan13 is necessary for Tet2-deficiency enhanced AML pathogenesis and LSC homing and self-renewal.

Figure 5. Identification of Tspan13 as a functional essential target of Tet2.

(A) Overlap of core enriched genes in Ivanova hematopoiesis stem cell pathway from two Tet2 KO (Tet2^−/− (ΔE8–10) and Tet2^−/− (ΔE3)) NRAS/AE9a AML models. The two Tet2^−/− NRAS/AE9a (i.e., Tet2^−/− (ΔE8–10) NRAS/AE9a and Tet2^−/− (ΔE3) NRAS/AE9a) AML models as well as their corresponding control cells (Tet2^+/+) were subjected to RNA-seq. The left circle (green) indicates the core-enriched genes in the Ivanova hematopoiesis stem cell pathway when comparing Tet2^−/− (ΔE8–10) NRAS/AE9a AML cells with their control cells; while the right circle (red) indicates the core-enriched genes in the same pathway when comparing Tet2^−/− (ΔE3) NRAS/AE9a AML cells with their control cells.

(B) Heatmap of expression level changes of the 41 overlapped core enriched genes in Ivanova hematopoiesis stem cell pathway upon Tet2 depletion (ΔE8–10) in the NRAS/AE9a AML model.

(C) Bubble plot depicting the correlation between TET2 expression and the 41 overlapped genes in 3 human AML datasets.

(D and E) Expression levels of TSPAN13 (D) and SV2A (E) in the favorable and adverse AML patients. The raw data were derived from the Beat AML cohort (http://vizome.org/aml/). The three lines inside the violin plots are the first quartile, median and third quartile; unpaired two-tailed Student’s t-test.

(F) IGV track showing the abundance of Tspan13 in Tet2^+/+ or Tet2^−/− (ΔE3) NRAS/AE9a primary leukemia cells.

(G) The knockdown (KD) efficiency of Tspan13 in Tet2^−/− (ΔE8–10) MA9 primary leukemia cells as determined by qPCR.

(H) The effect of Tsapn13 KD on the colony-forming ability of Tet2^−/− (ΔE8–10) MA9 primary leukemia cells.

(I) The representative colony images of Tet2^−/− (ΔE8–10) MA9 primary leukemia cells upon Tspan13 KD.

(J) The effect of Tsapn13 KD on the colony-forming ability of Tet2^−/− (ΔE3) NRAS/AE9a primary leukemia cells.

(K) The rescued effect of Tspan13 KD on Tet2 deficiency-induced homing. The in vivo homing was determined by flow cytometry.

(L) The rescued effect of Tspan13 KD on Tet2 deficiency (ΔE3)-induced increase of LSC frequency as determined by LDA with NRAS/AE9a primary leukemia cells.

(M) Survival curves of recipient mice transplanted with Tet2^−/− (ΔE8–10) MA9 primary leukemia cells with or without Tspan13 KD. n = 5 mice per group. The p values were calculated with a log-rank test (n=5).

(N) The effect of Tspan13 KD on the engraftment of Tet2^−/− MA9 cells into the BM and Spleen of recipient mice.

For G, H, J, K, and N, data were presented as mean ± SEM (n=3). *p < 0.05; **p < 0.01; ***p < 0.001; unpaired two-tailed Student’s t-test.

See also Figure S6.

Tet2 depletion activates the Tspan13/Cxcr4/Cxcl12 signaling to facilitate LSC homing

Considering that Tet2 deficiency-induced upregulation of Tspan13 is responsible for LSC homing, our following question was how Tspan13 influenced AML homing. The interaction between CXCR4 and CXCL12 was reported to be fundamental for LSC homing ^49,50, and multiple members in the Tetraspanin family, such as TSPAN3, TSPAN29, and TSPAN30, have been reported to regulate CXCR4 expression to modulate cell migration, adhesion, and homing ^46,51,52. We therefore checked the expression levels of the activated (Ser339 phosphorylated) Cxcr4 (p-Cxcr4) in both murine and human primary AML cells upon Tet2/TET2 depletion and overexpression. Our results showed that Tet2 deficiency potentiated Cxcr4 activation (Figure 6A), while forced expression of TET2 suppressed the activation of CXCR4 (Figure 6B). Furthermore, we checked the surface expression of activated Cxcr4 via flow cytometry and showed that Tet2 deficiency significantly increased the expression of surface p-Cxcr4 (Figure 6C), while depletion of Tspan13 dramatically decreased its surface expression (Figure 6D). To further verify our discovery that Tet2 deficiency-mediated LSC homing is attributed to the activation of the Cxcr4/Cxcl12 axis, we pre-treated stromal cells with a neutralizing antibody against Cxcl12 to block Cxcl12-induced chemotaxis and then co-cultured them with AML cells (Figure 6E). Pretreatment with Cxcl12 antibody could largely reverse Tet2 deficiency-induced AML migration (Figure 6F). Moreover, Tspan13 depletion-mediated inhibition of Tet2-deficient LSC migration was comparable to the trend mediated by Cxcl12 antibody (Figure 6G). Furthermore, we pretreated AML blasts with Cxcr4 inhibitor (Cxcr4i, AMD3100) to evaluate whether the Cxcr4/Cxcl12 axis was responsible for Tet2 deficiency-mediated LSC homing in vivo (Figure 6H). As shown in Figures 6I and 6J, inhibition of Cxcr4 could substantially reverse Tet2 deficiency-enhanced LSC homing in vivo. Therefore, our results demonstrate that Tet2 deficiency facilitates LSC homing mainly through activating the Tspan13/Cxcr4/Cxcl12 axis.

Figure 6. Tet2 depletion promotes LSC homing via activating the Tspan13/Cxcr4/Cxcl12 signaling.

(A) Protein level changes of Tspan13, p-Cxcr4, and Cxcr4 in primary AML cells upon Tet2 depletion.

(B) Protein level changes of TET2-CD-Flag, p-CXCR4, and CXCR4 in AML PDX cells and cell line (Mono Mac 6, MMC6) upon TET2 overexpression.

(C) Effect of Tet2 depletion on the expression levels of surface p-Cxcr4 in primary AE9a/NRAS AML cells.

(D) Effect of Tspan13 KD on the expression levels of surface p-Cxcr4 in primary Tet2^−/− MA9 AML cells.

(E) Schematic outline of transwell assay with stroma cells which were pre-treated with Cxcl12 antibody. The OP9 cells were pretreated with 1μg/ml Cxcl12 antibody for 24–48 hours before co-culture with AML cells.

(F) Effect of Cxcl12 antibody on the migration ability of Tet2^−/− MA9 primary AML cells.

(G) Effect of Tspan13 KD and Cxcl12 antibody treatment on the migration ability of Tet2^−/− MA9 primary AML cells.

(H) Schematic outline of a strategy to evaluate the effect of Cxcr4 inhibitor (Cxcr4i, AMD3100) on the homing. Before transplantation, the MA9 primary AML cells were treated with 0.1mg/ml AMD3100 for 20 hours.

(I-J) Statistical results (I) and representative flow cytometry (J) showing the effect of Cxcr4 inhibition on the homing of Tet2^−/− primary AML cells.

For C, D, F, G, and I, data were presented as mean ± SEM (n = 3 independent experiments). *p < 0.05, **p < 0.01, ***p < 0.001; unpaired two-tailed Student’s t-test.

TET2 negatively regulates TSPAN13 expression via mRNA oxidation

Next, we sought to delineate the underlying mechanism by which TET2 depletion promotes the expression of TSPAN13 (Figure 7A). We first evaluated the transcription efficiency of Tspan13 in Tet2^+/+ and Tet2^−/− primary AML cells, because Tet2-mediated DNA demethylation usually regulates gene expression at the transcriptional level ⁵³. Surprisingly, the transcription efficiency of Tspan13 was not elevated by Tet2 depletion as detected by the nuclear Run-on assay (Figures 7B and S7A), and the DNA 5hmC abundance of Tspan13 was not significantly influenced by Tet2 depletion either (Figure S7B), suggesting that Tet2 deficiency-mediated upregulation of Tspan13 is not dependent on DNA demethylation or transcriptional regulation. Recently, evidence is emerging that TET2 also exerts a novel biological function to fine-tune gene expression at the post-transcriptional level by inducing RNA m⁵C oxidization as an mRNA demethylase ^17–20. We thus assessed the potential effect of Tet2 depletion on the stability of Tspan13 mRNA. We found that Tet2 deficiency indeed significantly elongated the half-life of Tspan13 transcript in AML cells (Figures 7C and S7C–S7D).

Figure 7. Tet2 induces the degradation of Tspan13 and thus negatively regulates its expression as an mRNA m⁵C demethylase.

(A) The possible mechanisms through which Tet2 regulates Tspan13 expression.

(B) The transcription efficiency of Tspan13 in Tet2^+/+ and Tet2^−/− (ΔE3) NRAS/AE9a leukemia cells as determined by nuclear run-on assay.

(C) Effect of Tet2 deficiency (ΔE3) on Tspan13 mRNA stability in primary NRAS/AE9a AML cells.

(D) The calibration curves of cytosine (C), m⁵C, and hm⁵C standards as determined by QQQ-MS.

(E) Scheme of the in vitro transcription assay as well as the in vitro (cell-free) demethylation assay with TET2 protein and m⁵C-modified TSPAN13 mRNA.

(F and G) Assessing the m⁵C (left) and hm⁵C (right) level changes in TSPAN13 mRNA in the in vitro demethylation assay with TET2 protein as detected by QQQ-MS (F) or dot blot assays (G).

(H and I) The abundance of m⁵C in the poly(A) RNA isolated from NRAS/AE9a (H) and MA9 (I) primary leukemia cells with or without Tet2 deficiency.

(J) Relative levels of Tsapn13 mRNA m⁵C modification in Tet2^+/+ and Tet2^−/− (ΔE3) NRAS/AE9a primary leukemia cells as determined by gene-specific m⁵C qPCR.

(K) The direct interaction between Tet2 protein and TSPAN13 mRNA as detected by CLIP-qPCR in C1498 AML cells. Here, we overexpressed Flag-tagged Tet2 catalytic domain (CD) and performed CLIP with Flag antibody.

(L) The direct interaction between Ybx1 (Flag-tagged) and Tspan13 mRNA as detected by CLIP-qPCR in C1498 AML cells.

(M) The effect of Ybx1 KD on Tspan13 mRNA stability in C1498 AML cells.

(N) The cumulative distribution function (CDF) plot showing the effect of Tet2 depletion on global mRNA stabilities in NRAS/AE9a primary leukemia cells.

(O) The MA plot showing the mRNAs with significantly increased stability (Half-life UP; Red) and decreased stability (Half-life Down; Green) in NRAS/AE9a primary leukemia cells upon Tet2 depletion.

(P) The schematic diagram showing how Tet2 regulates LSC homing and self-renewal via targeting the m⁵C/TSPAN13/CXCR4 axis in the BM microenvironment.

For C, F, and H-N, data were presented as mean ± SEM (n=3 independent experiments). **p < 0.01, ***p < 0.001; unpaired two-tailed Student’s t-test.

See also Figure S7.

To determine whether the effect of TET2 depletion on the stability of TSPAN13 mRNA is caused by TET2-mediated oxidation of m⁵C (into hm⁵C) in TSPAN13 mRNA, we first established the QQQ-MS method to quantitively and precisely determine the abundance of both m⁵C and hm⁵C (Figure 7D). Next, we conducted an in vitro (cell-free) demethylation assay with m⁵C-modified single-stranded RNA (ssRNA) and purified TET2 protein (Figure S7E). The 5mC-modified DNA was also included as a positive control. Our results confirmed that TET2 could catalyze the formation of both hm⁵C in RNA and 5hmC in DNA (Figures S7F–S7G). Furthermore, we employed the in vitro demethylation assay with m⁵C-modified TSPAN13 mRNA and TET2 protein, and verified that TET2 could remove the methyl group from m⁵C modification in TSPAN13 mRNA as detected by both QQQ-MS and dot blot assays (Figures 7E–7G). Distinct from the abundant 5hmC in DNA, the endogenous hm⁵C level in mRNA in cells is much lower, even below the limit of QQQ-MS detection (Figure S7H), consistent with previous reports ^17,54. Therefore, we mainly focused on the m⁵C decoration in mRNA. In the primary AML cells, Tet2 deficiency significantly increased global m⁵C abundance in poly(A) RNA (Figures 7H–7I), the opposite was true when we ectopically expressed Tet2 (Figure S7I). More specifically, we conducted gene-specific m⁵C qPCR to determine the m⁵C level in Tspan13 transcript and demonstrated that Tet2 loss significantly increased the m⁵C abundance in Tspan13 mRNA (Figure 7J). Moreover, Tet2 protein, either the catalytic domain (CD) or the full length, directly interacted with Tspan13 mRNA in cellulo (Figures 7K and S7J–S7M).

A recent study revealed that Ybx1 functions as an m⁵C reader and stabilizes the m⁵C-modified mRNA ⁵⁵. To evaluate whether Ybx1 was required to regulate Tspan13 mRNA stability via recognizing m⁵C residue, we first conducted CLIP-qPCR to determine whether Ybx1 could directly interact with Tspan13 mRNA. As expected, Ybx1 directly bound with the Tspan13 transcript (Figures 7L and S7N–S7O). In addition, genetic depletion of Ybx1 significantly accelerated the degradation of Tspan13 mRNA (Figures 7M and S7P). Finally, we conducted mRNA bisulfite sequencing and stability profiling to assess the effect of Tet2 deficiency on mRNA m⁵C modification and stability, respectively, at the whole transcriptome level. The mRNA bisulfite sequencing showed that Tet2 deficiency in primary AML cells led to a gain of m⁵C sites in over 7,000 transcripts (Figures S7Q–S7R). Moreover, the mRNA half-time profiling showed that Tet2 deficiency resulted in a global increase of mRNA stabilities (Figure 7N). Specifically, Tet2 deficiency led to a significant increase of half-time of 1,347 transcripts (Figure 7O), and most of those transcripts (>67%) were modified by m⁵C modification in Tet2-deficient AML cells (Figure S7R). Those overlapped transcripts are mainly enriched in the signal pathways related to metabolism, intracellular transport, and purine nucleotide binding (Figure S7S). Taken together, our results suggest that Tet2 deficiency leads to the accumulation of m⁵C in Tspan13 mRNA, and Ybx1 recognizes m⁵C modification, stabilizes Tspan13 transcript, and increases its overall expression, which in turn leads to the activation of the CXCR4/CXCL12 axis in BM microenvironment (Figure 7P).

Discussion

Loss-of-function mutations of TET2 are common across a range of blood malignancies (including AML) and confer resistance to standard therapeutics. In the present study, we found that besides loss-of-function mutations in TET2, transcriptional suppression of TET2 is also common in AML patients, and its decreased expression is associated with shorter overall survival in AML patients. Interestingly, in addition to its canonical role in DNA demethylation, TET2 also functions as an RNA demethylase to post-transcriptionally regulate expression of its mRNA targets such as Tspan13. We revealed a previously unappreciated role of TET2-mediated mRNA m⁵C demethylation in AML and uncovered the essential biological function of the TET2/m⁵C/TSPAN13/CXCR4/CXCL12 axis in LSC homing into the BM niche, resulting in the rapid and aggressive leukemogenesis upon TET2 deficiency. Overall, our study provides novel insight into the cellular/molecular mechanisms underlying leukemogenesis, LSC homing, and self-renewal, and highlights the therapeutic potential of reactivation of TET2 signaling in treating leukemia patients with TET2 mutations or transcriptional suppression.

TET2 in RNA demethylation: beyond the canonical transcriptional gene regulation

Most recently, covalent chemical modifications in RNAs, especially mRNA, are progressively being appreciated as key regulators of RNA metabolism, including stability, translation efficiency, alternative splicing, and nuclear export ^3,56. Albeit over 170 decorations have been detected in RNAs ⁵⁷, the biological roles of most epitranscriptomic modifications have not been fully defined. The mammalian TET proteins, TET1, TET2, and TET3, are originally discovered as dioxygenases catalyzing DNA demethylation and mainly affect gene expression at the transcriptional level ^8–10. More recently, TET2 has been reported as an RNA-binding protein ⁵⁸ in vitro and in vivo, and can oxidize m⁵C in abundant non-coding RNAs, such as tRNA, in mouse embryonic stem cells (mESCs) as well as mRNA during myelopoiesis ^17,59,60. In tRNA, both m⁵C and its oxidation product hm⁵C have been characterized and account for 10% and 0.02% of cytidine, respectively ⁶⁰. However current studies reporting the abundance of m⁵C and especially hm⁵C in mRNA are somehow controversial. Specifically, Guallar et al. ¹⁸ and Lan et al. ²⁰ have clarified hundreds of hm⁵C-modified mRNA transcripts, including the transcripts related to pluripotency, in ESC cells via an antibody-dependent approach. Moreover, they have observed that hm⁵C decoration is enriched in chromatin-associated and nascent pre-mRNA, suggesting a co-transcriptional mechanism of mRNA m⁵C modification. In contrast, Legrand et al. ⁵⁴ failed to detect any evidence for the occurrence of hm⁵C (the potential oxidated product of m⁵C) in mRNA in mESCs through LC-MS/MS analysis, which agrees with another independent study ¹⁹.

In our study, we have accurately and unbiasedly assessed the abundance of both m⁵C and hm⁵C modifications in mRNA via in vitro (cell-free system) and in cellulo strategies. TET2 deficiency leads to the transcriptome-wide accumulation of m⁵C residues, and the opposite is true when TET2 is overexpressed in either primary AML cells or cell lines. According to our results, m⁵C accounts for 0.02–0.05% of cytosines in mRNA. Though m⁵C abundance is lower than the most prevalent m⁶A modification (which usually accounts for 0.2–0.5% of adenosines) in mRNA, it is still much more abundant than other sparse decorations, such as internal N⁷-methylguanosine (m⁷G), N¹-methyladenosine (m¹A), and N⁶,2′-O-dimethyladenosine (m⁶A_m) ^61–63. We believe the dynamic and reversible m⁵C modification offers the potential of regulating mRNA fate post-transcriptionally. More systemic and comprehensive studies are warranted to dissect the landscape, biological functions, and mechanisms of m⁵C in mRNA. In line with the reported studies ^19,54, we could not detect the existence of hm⁵C modification in mRNA in AML cells either. The possible reasons could be (1) the hm⁵C modification in mRNA represents a signal which immediately induces degradation; (2) the hm⁵C itself is an unstable oxidation product or occurs transiently during TET2-mediated demethylation of m⁵C modification; and/or (3) the abundance of hm⁵C modification in mRNA is below the minimum detection limit of QQQ-MS.

Although m⁵C is more abundant relative to other rare chemical decorations in mRNA, the m⁵C-modified mRNA targets of TET2 are still poorly understood. Here we identified TSPAN13 as a functionally essential and m⁵C-dependent mRNA target of TET2 and elucidated its pathological role in LSC homing into the BM niche. TET2 deficiency-induced accumulation of m⁵C modification in TSPAN13 mRNA can be recognized by YBX1, which in turn stabilizes the transcript and increases its overall expression. Genetic depletion of TSPAN13 can substantially reverse TET2 deficiency-enhanced LSC homing and self-renewal, which underscores the importance of TET2 activity for mRNA metabolism and post-transcriptional regulation in AML. The m⁵C residue in mRNA is mediated by the DNMT2 and NSUN methyltransferase family in the nuclei ²¹. However, it’s still elusive whether the deposition of m⁵C modification in mRNA is co-transcriptionally, and whether there is any link or crosstalk between TET2-mediated epigenetic and epitranscriptomic regulation.

Targeting leukemic microenvironment to treat TET2-deficient leukemia

Despite improvements in diagnosis and treatment-adapted regimens, the survival rate of AML is still dismal ⁶⁴. A combination of cytarabine and anthracycline (e.g., daunorubicin) is the backbone of AML therapy and such approach has not substantially changed over the decades ⁶⁵. AML patients with TET2-deficiency are poorly responsive to those standard chemotherapy agents. In addition, the antiproliferative therapy is not sufficient to effectively treat leukemias with TET2 deficiency. Consequently, AML patients with TET2 deficiency are associated with shorter survival compared to those without TET2 deficiency.

Using various primary AML models, here we showed that Tet2 deficiency-induced rapid leukemogenesis is not simply attributed to cell proliferation benefits, but largely due to the increased LSC homing and self-renewal. We also found that existence of stromal cells could substantially enhance the colony-forming ability of TET2-deficient AMLs. Such phenotype might be attributed to the soluble factors, such as SCF, CXCL12, and ANGPT, which are secreted at high levels by BM stromal cells ⁶⁶. The direct interplays between the leukemia cells and BM stromal cells have been postulated to be essential for chemoresistance and relapse of leukemia ⁶⁷. The existence of BM stromal cells protected AML cells from chemotherapy drugs-induced cell killing and leads to drug resistance. The CXCR4/CXCL12 axis plays a central role in trafficking LSCs into the BM niche. TET2 deficiency in AML promotes the TSPAN13/CXCR4 axis via an mRNA m⁵C-dependent manner, and thereby enhances LSCs homing and self-renewal. Several small molecule compounds targeting CXCR4 have been developed ^32,68. Pharmacological inhibition of CXCR4 induced the mobilization of AML blast cells into circulation, sensitized AML to chemotherapy, and extensively elongated the survival in the treated mice, demonstrating the pivotal role of the CXCR4/CXCL12 axis in the pathogenesis of AML and drug resistance. It would be interesting to test whether restoration of TET2 activity can block the homing of LSCs into BM niche via silencing the TSPAN13/CXCR4 signaling, whether “priming” leukemia cells with TET2 activator agent can render them more sensitive to standard chemotherapeutic drugs and targeted therapy agents such as CXCR4 inhibitors.

Collectively, our findings suggest that targeting the leukemic microenvironment to disrupt the interaction between leukemia cells and stromal cells could be a promising therapeutic strategy to treat TET2-deficient AMLs, and such a therapeutic strategy could be extended to other cancer types with TET2 mutation and/or suppression.

Limitation of the study

In the present study, we utilized mRNA bisulfite sequencing to identify m⁵C sites. However, the harsh bisulfite treatment may lead to random degradation of mRNAs. Furthermore, mRNA bisulfite sequencing is unable to distinguish m⁵C, hm⁵C, and m⁴C. Therefore, the development of more accurate methods for m⁵C sequencing is necessary to gain a better understanding of the role of m⁵C modification in mRNA metabolism. Considering that the CXCR4-CXCL12 signaling plays a pivotal role in mediating AML cell homing in the BM microenvironment, it is important to systematically investigate whether and how TET2 deficiency in AML blast cells remodels BM microenvironment to facilitate their homing, survival and proliferation/self-renewal. Given that loss of TET2 substantially triggers the activation of the CXCR4 signaling, it is worth assessing the therapeutic potential of CXCR4 inhibitors alone or in combination with other therapeutics for the treatment of TET2-deficient leukemia.

STAR Methods

RESOURCE AVAILABILITY

Lead Contact

Further information and requests for resources and reagents may be directed to and will be fulfilled by the Lead Contact, Jianjun Chen (jianchen@coh.org).

Materials Availability

All cell lines, plasmids, and other stable reagents generated in this manuscript are available from the Lead Contact under a complete Materials Transfer Agreement.

Data and Code Availability

• The RNA-seq, mRNA half-life profiling, and mRNA Bis-seq data have been deposited at GEO and are publicly available as of the date of publication. Accession numbers are listed in the key resources table.

• This paper does not report any original code.

• Any additional information required to reanalyze the data reported in this work paper is available from the Lead Contact upon request.

KEY RESOURCES TABLE

REAGENT or RESOURCE	SOURCE	IDENTIFIER
Antibodies
Anti-5mC antibody	Active Motif	Cat# 39649, RRID:AB_2687950
Anti-5mC antibody	Abcam	Cat# ab10805, RRID:AB_442823
Anti-5hmC antibody	Active Motif	Cat# 39769, RRID:AB_10013602
Anti-GAPDH antibody	Proteintech	Cat# 10494-1-AP, RRID:AB_2263076
Anti-alpha-Tubulin Antibody	Cell Signaling Technology	Cat# 3873, RRID:AB_1904178
Anti-vinculin Antibody	Santa Cruz Biotechnology	Cat# sc-25336; RRID: AB_628438
Anti-FLAG M2 antibody	Sigma-Aldrich	Cat# F3165; RRID: AB_259529
Normal Mouse IgG control antibody	Millipore	Cat# NI03-100UG, RRID:AB_490557
Rabbit IgG normal antibody	Millipore	Cat# NI01-100UG, RRID:AB_490574
Anti-CXCR4 antibody	Proteintech	Cat# 60042-1-Ig, RRID:AB_2091809
Anti-CXCR4 (phospho Ser339) antibody	GeneTex	Cat# GTX32281, RRID:AB_2887647
Anti-CXCR4 (phospho Ser339) antibody	Thermo Fisher Scientific	Cat# BS-12256R
Anti-TSPAN13 antibody	Abbexa	Cat# abx214152
Anti-Lamin A/C	Cell Signaling Technology	Cat# 4777, RRID:AB_10545756
Anti-Mouse CXCR4 PE	BioLegend	Cat# 146505, RRID: AB_2562782
Anti-Human CD45 APC	Thermo Fisher Scientific	Cat# 17-9459-42, RRID: AB_10714837
Anti-Human anti-CD33 PE	Thermo Fisher Scientific	Cat# 12-0339-42, RRID: AB_10855031
Anti-CXCL12 antibody	R&D Systems	Cat# MAB310-100, RRID: AB_2276927
Goat anti-rabbit IgG H&L (HRP)	Abcam	Cat# ab6721; RRID: AB_955447
Goat anti-mouse IgG H&L (HRP)	Abcam	Cat# ab6789; RRID: AB_955439
Anti- mouse CD16/32	eBioscience	Cat# 14-0161-82, RRID: AB_467133
Anti-mouse CD45.2 FITC	BioLegend	Cat# 109806, RRID:AB_313443
Anti-Mouse Lineage marks eFlour 450	eBioscience	Cat# 88-7772-72, RRID: AB_10426799
Anti-Mouse c-Kit APC	eBioscience	Cat# 17-1171-83, RRID: AB_469431
Anti-Mouse c-Kit APC-Cy7	BioLegend	Cat# 135136, RRID:AB_2632809
Anti-Mouse Sca-1 PE	eBioscience	Cat# 12-5981-82, RRID: AB_466086
Anti-Mouse Mac-1 PerCP-Cyanine5.5	eBioscience	Cat# 45-0112-82, RRID:AB_953558
Anti-Mouse Gr-1 PE	eBioscience	Cat# 12-5931-81, RRID:AB_46604
Anti-Mouse CD34 APC	BioLegend	Cat# 128612, RRID:AB_10553896
Anti-Mouse CD16/32 PE-Cy7	Thermo Fisher Scientific	Cat# 25-0161-82, RRID:AB_469598
Anti-Mouse CD8 APC	eBioscience	Cat# 17-0083-81, RRID:AB_657760
Anti-Mouse osteopontin PE	R&D Systems	Cat# IC808P, RRID:AB_10643832
Phalloidin, CF647 conjugate	Biotium	Cat# 00041-T
7-AAD	BD Biosciences	Cat# 559925, RRID:AB_2869266
Bacterial and virus strains
Stbl3 E. coli	ThermoFisher	Cat# C7373-03
Chemicals, peptides, and recombinant proteins
Hexadimethrine bromide (Polybrene)	Sigma-Aldrich	Cat# H9268; CAS: 28728-55-4
Puromycin dihydrochloride	Sigma-Aldrich	Cat# P8833; CAS: 58-58-2
G418 Sulfate	Thermo Fisher Scientific	Cat# 10131027; CAS: 108321-42-2
L-Glutamine (200mM)	Thermo Fisher Scientific	Cat# 25030-081
Ammonium Chloride Solution	STEMCELL Technologies	Cat# 07850
Sodium Pyruvate (100mM)	Thermo Fisher Scientific	Cat# 11360-070
Insulin, human recombinant zinc solution	Thermo Fisher Scientific	Cat# 12585014
Penicillin Streptomycin	Thermo Fisher Scientific	Cat# 15-140-122
Plasmocin prophylactic	Invivogen	Cat# ant-mpp
X-tremeGENE™ HP DNA Transfection Reagent	Roche	Cat# 6366236001
Benzonase	Sigma-Aldrich	Cat# E1014
Phosphdiesterase I	Sigma-Aldrich	Cat# P3243,
Alkaline phosphatase	Thermo Fisher Scientific	Cat# EF0651
RIPA Buffer	Sigma-Aldrich	Cat# R0278
Halt Phosphatase Inhibitor Cocktail	Thermo Fisher Scientific	Cat# 78420
Halt Protease Inhibitor Cocktail	Thermo Fisher Scientific	Cat# 78429
4 × Laemmli Sample Buffer	Bio-Rad	Cat# 1610747
Bio-Rad Protein Assay	Bio-Rad	Cat# 5000006
PageRuler Plus Prestained Protein Ladder	Thermo Fisher Scientific	Cat# 26620
Amersham ECL Prime Western Blotting Detection Reagent	GE Healthcare	Cat# 45-010-090
Pierce™ ECL Western Blotting Substrate	Thermo Fisher Scientific	Cat# PI32106
Fluoroshield with DAPI	Sigma-Aldrich	Cat# F6057
Recombinant mouse IL-3	PeproTech	Cat# 213-13
Recombinant mouse IL-6	PeproTech	Cat# 216-16
Recombinant mouse SCF	PeproTech	Cat# 250-03
Recombinant murine GM-CSF	PeproTech	Cat# 315-03
Recombinant Murine IFN-γ	PeproTech	Cat# AF-315-05
Recombinant Mouse IFN-β	R&D SYSTEMS	Cat# 8234-MB-010
ColonyGEL-Mouse Base Medium	Reachbio	Cat# 1201
FastDigest NotI	Thermo Fisher Scientific	Cat# FD0595
FastDigest XbaI	Thermo Fisher Scientific	Cat# FD0684
FastDigest BglII	Thermo Fisher Scientific	Cat# FD0083
FastDigest XhoI	Thermo Fisher Scientific	Cat# FD0694
FastDigest KspAI	Thermo Fisher Scientific	Cat# FD1034
CFSE	eBioscience	Cat# 65-0850-84
Actinomycin D	Sigma-Aldrich	Cat#A9415; CAS: 50-76-0
L-Ascorbic acid	Sigma-Aldrich	Cat# A0278, CAS: 50-81-7
DL-Dithiothreitol solution	Sigma-Aldrich	Cat# 3483-12-3; CAS: 3483-12-3
α-Ketoglutaric acid	Sigma-Aldrich	Cat# 75890, CAS: 328-50-7
Adenosine-5 Triphosphate (ATP)	New England Biolabs	Cat# P0756S
PMA	Sigma-Aldrich	Cat# P1585, CAS: 16561-29-8
Ionomycin	Sigma-Aldrich	Cat# I9657, CAS: 56092-81-0
Propidium Iodide	Sigma-Aldrich	Cat# P4170; CAS: 25535-16-4
NEBNext Magnesium RNA Fragmentation Module	New England Biolabs	Cat# E6150S
0.5M EDTA	Thermo Fisher Scientific	Cat# 15575020
Formaldehyde solution	Sigma-Aldrich	Cat# F8775, CAS: 50-00-0
1M Tris-HCl, pH 8.0	Invitrogen	Cat# 15568025
1 M Tris-HCI Buffer, pH 7.5	Thermo Fisher Scientific	Cat# 15567027
Sodium Chloride Solution, 5M	Sigma-Aldrich	Cat# S6546
Paraformaldehyde Solution	Fisher scientific	Cat# AAJ19943K2
IC Fixation buffer	eBioscience	Cat# 00-8222-49
Permeabilization Buffer (10X)	Thermo Fisher Scientific	Cat# 00-8333-56
Wright-Giemsa Stain	Polysciences	Cat# 24985
Wright-Giemsa Stain/Buffer	Polysciences	Cat# 24984
Pierce Protein A/G Magnetic Beads	Thermo Fisher Scientific	Cat# 88803
D-Luciferin Firefly, potassium salt	Goldbio	Cat# LUCK; CAS: 115144-35-9
Platinum™ SuperFi II DNA Polymerase	Thermo Fisher Scientific	Cat# 12361010
Recombinant TET2 protein	Active Motif	Cat# 31418
RNaseOUT Recombinant Ribonuclease Inhibitor	Thermo Fisher Scientific	Cat# 10777019
T4 DNA ligase	Fisher Scientific	Cat# M0202S
Biotin-16-UTP	Sigma-Aldrich	Cat# 11388908910
5-methylcytidine-5’-triphosphate (5mCTP)	TriLink Biotechnologies	Cat# N-1014-1
5-Hydroxymethylcytidine	Carbosynth	Cat# NH35521, CAS: 19235-17-7
5-Methyl-2’-deoxycytidine	Cayman Chemical Company	Cat# 16166, CAS: 838-07-3
Cytidine	Cayman Chemical Company	Cat# 29602, CAS: 65-46-3
Guanosine	Cayman Chemical Company	Cat# 27702, CAS: 118-00-3
5-(Hydroxymethyl)-2’-deoxycytidine	Cayman Chemical Company	Cat# 18162, CAS: 7226-77-9
5-(Hydroxymethyl)-2’-deoxycytidine-d3	Toronto Research Chemicals	Cat# H946632, CAS: 7226-77-9
5-Hydroxymethylcytidine-13C,D2	Toronto Research Chemicals	Cat# H947092, CAS: 19235-17-7
5-Methyl-2’-deoxy Cytidine-d3	Toronto Research Chemicals	Cat# M295902; CAS: 1160707-78-7
Guanosine-5’,5″-d2 Monohydrate	Toronto Research Chemicals	Cat# G837915, CAS: 478511-34-1
DNase I, RNase-free (1 U/μL)	Thermo Fisher Scientific	Cat# EN0521
TURBO DNase	Thermo Fisher Scientific	Cat# AM2239
RNase A (DNase free)	Thermo Fisher Scientific	Cat# EN0531
Proteinase K	Thermo Fisher Scientific	Cat# EO0492
Dynabeads M-280 streptavidin	Thermo Fisher Scientific	Cat# 11205D
TRIzol™ Reagent	Thermo Fisher Scientific	Cat# 15596-018
AMD3100	Selleckchem	Cat# S3013
Critical commercial assays
Oligo Clean & Concentrator kit	Zymo Research	Cat# D4061
Cell Counting Kit 8	Dojindo	Cat# CK04
Click-iT EdU Alexa Fluor 647 Flow Cytometry Assay Kit	Thermo Fisher Scientific	Cat# C10424
EdU Proliferation Kit (iFluor 488)	Abcam	Cat# ab219801
In-Fusion HD Cloning Plus CE Cat#	Takara	Cat# 638916
PCR Mycoplasma Detection Kit	Applied Biological Materials	Cat# G238
DNeasy Blood & Tissue Kit	QIAGEN	Cat# 69506
CD117 MicroBead Kit, mouse	Miltenyi Biotec	Cat# 130-091-224
Lineage Cell Depletion Kit	Miltenyi Biotec	Cat#130-090-858
Pan T Cell Isolation Kit II	Miltenyi Biotec	Cat# 130-095-130
RNA Clean & Concentrator-5	Zymo Research	Cat# R1014
ChIP DNA Clean & Concentrator kit	Zymo Research	Cat# D5201
RNeasy mini kit	Qiagen	Cat# 74104
QIAGEN Plasmid Mini Kit	Qiagen	Cat# 27104
PolyATract mRNA isolation system IV	Promega	Cat# Z5310
QuantiTect Reverse Transcription Kit	Qiagen	Cat# 205314
Maxima SYBR Green qPCR Master Mix	Thermo Fisher Scientific	Cat# K0253
SuperScript™ III First-Strand Synthesis System	Invitrogen	Cat# 18080051
MEGAscript™ T7 kit	Invitrogen	Cat# AM1333
MEGAclear™ Transcription Clean-Up Kit	Invitrogen	Cat# AM1908
EZ RNA Methylation Kit	Zymo Research	Cat# R5002
Deposited data
RNA-seq (Raw and analyzed data)	This manuscript	GSE206488
mRNA half-life profiling & mRNA Bis-seq	This manuscript	GSE231330
Experimental models: Cell lines
C1498	A gift from Dr. Marcin Kortylewski	N/A
Mono Mac 6	DSMZ	Cat# ACC 124; RRID: CVCL_1426
OP9	ATCC	Cat# CRL-2749; RRID: CVCL_4398
HEK293T	ATCC	Cat# CRL-3216; RRID: CVCL_0063
Experimental models: Organisms/strains
NRGS mouse	The Jackson Laboratory	Cat# JAX:024099; RRID: IMSR_JAX:024099NCI
NCI B6-Ly5.1/Cr mouse	Charles river	Cat# CRL:564; RRID: IMSR_CRL:564
B6(Cg)-Tet2tm1.2Rao/J (ΔE8-10)	The Jackson Laboratory	Cat# 023359 RRID:IMSR_JAX:023359
Tet2−/− (ΔE3)	A gift from Dr. Mingjiang Xu	N/A
Oligonucleotides
Primers for Real-time PCR, see Table S1	Integrated DNA Technologies (IDT)	N/A
Recombinant DNA
pcDNA3-TET2-CD	A gift from Dr. Yue Xiong	N/A
pMSCV-PIG-NRASG12D	This paper	N/A
pMSCV-Neo-AF9a	This paper	N/A
pMSCV-Neo-PML-RARA	This paper	N/A
pMSCV-Neo-MLL AF9	A gift from Dr. Michael Thirman	N/A
pcDNA3-Tet2	Wang et al.70	RRID:Addgene_60939
pMSCV-Flt3-ITD-Y591F/Y919F	Kazi et al.71	RRID:Addgene_74499
pMSCV-PIG-3×FLAG-NotI	This paper	N/A
pMSCV-Neo-3×FLAG	This paper	N/A
pMSCV-PIG-3×FLAG-Tspan13	This paper	N/A
pMSCV-PIG-3×FLAG-Ybx1	This paper	N/A
pMSCV-PIG-3XFLAG-mTet2 CD	This paper	N/A
pCDH-3×FLAG-TET2 CD	This paper	N/A
pLenti CMV Puro LUC (w168-1)	Campeau et al.72	Cat# Addgene plasmid # 17477; RRID: Addgene_17477
pMD2.G	A gift from Dr. Didier Trono	Cat# Addgene plasmid # 12259; RRID: Addgene_12259
psPAX2	A gift from Didier Trono	Cat# Addgene plasmid # 12260; RRID: Addgene_12260
pLKO.1 puro	Stewart et al.73	Cat# Addgene plasmid # 8453; RRID: Addgene_8453
pLKO.1-shTspan13-1	This paper	N/A
pLKO.1-shTspan13-2	This paper	N/A
pLKO.1-shYbx1-1	This paper	N/A
pLKO.1-shYbx1-2	This paper	N/A
pMSCV-MLL-AF9-Neo	A gift from Dr. Michael Thirman	N/A
pCL-Eco	Naviaux et al.74	Cat# Addgene plasmid # 12371; RRID: Addgene_12371
Software and algorithms
GraphPad Prism	GraphPad	https://www.graphpad.com/scientific-software/prism/
GelAnalyzer	GelAnalyzer	http://www.gelanalyzer.com/
RSEM-1.2.31	Li and Dewey75	https://deweylab.github.io/RSEM/
STAR 2.7	Dobin et al.76	https://github.com/alexdobin/STAR
igv-2.3.72g	Thorvaldsdottir et al.77	http://software.broadinstitute.org/software/igv/
GSEA 4.2.0	Subramanian et al.78	https://www.gsea-msigdb.org/gsea/index.jsp

EXPERIMENT MODEL AND SUBJECT DETAILS

Primary AML Patient and Healthy Donor Specimens

Human AML patient samples were obtained at the time of diagnosis, after treatment, or relapsed with informed consent at City of Hope Hospital, Cincinnati Children’s Hospital, or Dana-Farber/Harvard Cancer Center, in congruence with the approved protocol by the institutional review board (IRB). The characteristics of AML patients are detailed in Table S1. The samples from healthy donors were collected from the Healthy Donor Center in Cincinnati Children’s Hospital under the institute’s approved IRB protocol. Erythrocytes lysis was performed on the samples, and the mononuclear cells (MNCs) were cryopreserved at −150°C for further study.

Cell Culture

All cell lines were obtained from the sources listed in the key resources table. HEK293T and C1498 were cultured in endotoxin-free DMEM medium (Life Technologies) supplemented with 10% fetal bovine serum (FBS) (100–106, Gemini Bio-Products), 1% penicillin/streptomycin (15–140-122, Thermo Fisher Scientific), and 2.5 μg/mL Plasmocin prophylactic (ant-mpp, InvivoGene); Mono Mac 6 was maintained in RPMI 1640 supplemented with 10% FBS plus 2 mM L-glutamine (25030–081, Thermo Fisher Scientific), 1× non-essential amino acids (11140050, Gibco), 1 mM sodium pyruvate (11360070, Thermo Fisher Scientific), 10 mg/ml human insulin (12585014, Thermo Fisher Scientific); OP9 cells were cultured in alpha-MEM (32571101, Thermo Fisher Scientific) supplemented with 20% FBS plus 2.2 g/L sodium bicarbonate. All the cell lines were routinely tested for mycoplasma contamination using a PCR Mycoplasma Detection Kit (G238, Applied Biological Materials).

Animals

All congenic recipient mice were NCI C57 B6.SJL (CD45.1⁺, RRID: IMSR_CRL:564) purchased from Charles River Laboratories (Wilmington, MA). The immune-incompetent NOD.Cg-Rag1^tm1Mom Il2rg^tm1Wjl Tg (CMV-IL3, CSF2, KITLG) 1Eav/J (NRGS, RRID: IMSR_JAX:024099) mice were originally purchased from the Jackson Laboratory and bred under pathogen-free conditions. All animal experiments were done in accordance with the guidelines of the institutional guidelines and Institutional Animal Care and Use Committee (IACUC) protocols approved by City of Hope. All these mice were maintained on a 12-hour:12-hour, light: dark cycle with food and water ad libitum. For each experiment, 6- to 8-week-old mice were randomly allocated to experimental groups.

METHOD DETAILS

Plasmid Construction

The catalytic domain of TET2 (pcDNA3-TET2-CD) kindly provided by Dr. Yue Xiong, the full-length Tet2 (pcDNA3–1xFlag-Tet2) (60939, Addgene), Flt3-ITD (pMSCV-Flt3-ITD-Y591F/Y919F) (74499, Addgene), and murine cDNA made from MA9 leukemic cells were served as templates to generate 3×FLAG-tagged TET2 CD, Tet2 CD, Flt3-ITD, Tspan13, and Ybx1 fragments by PCR amplification with CloneAmp^™ HiFi PCR Premix (639298, Takara Bio). The PCR fragment of TET2 CD was ligated into pCDH empty vector and the rest of fragments were ligated to the MSCV vector with In-Fusion HD Cloning Plus Kits according to the manufacturer’s instruction (638909, Takara Bio). shRNAs targeting Tspan13 and Ybx1 were generated as previously described with mild revision ⁷⁹.The palindromic DNA oligo was annealed by gradient cooling to form a double-strand oligo and then ligated to the pLKO.1-puro vector cleaved by AgeI and EcoRI restriction enzymes with T4 DNA ligase at 16°C for 2 hours or overnight. All the primers used for molecular cloning were listed in Table S2. The plasmids were extracted using QIAprep Spin Miniprep Kit (27104, Qiagen) and then verified by Sanger sequencing (Eton Bioscience).

Lentivirus and Retrovirus Preparation and Infection

These assays were conducted as previously described with minor modifications ^2,80. All lentivirus particles for overexpression and knockdown plasmids were packaged using the 2^nd-generation package system. Briefly, 2.25 μg psPAX2, 0.75 μg pMD2.G), and 3 μg target plasmids diluted in Opti-MEM medium (31985070, Thermo Fisher Scientific) were incubated with X-tremeGENE^™ HP DNA Transfection Reagent (6366546001, Sigma-Aldrich). For retrovirus packaging, 1.5 μg plasmids in MSCV backbone and 1.5 μg pCL-ECO vector ((IMGENEX, San Diego, CA) incubated with DNA transfection reagent. The DNA mixture were then co-transfected into HEK293T cells in 6-cm cell culture dishes. The following morning, medium was replaced with 4 mL DMEM complete medium. The virus-containing supernatant were harvested at 48 h and 72 h post-transfection and filtered through a 0.45μm syringe filter or centrifuged for 15 min at 3000g at 4°C. The viral supernatant was added to cells in the presence of 4–8 μg/mL polybrene (H9268, Sigma-Aldrich). To infect the suspension cells, cells were subjected to two rounds of spinoculation at 32°C for 2h. The infected cells were selected with 2 μg/mL puromycin (for pCDH, MSCV-PIG and pLKO.1 backbone) or 1 mg/mL G418 (MSCV-neo backbone) (10131–02, Life Technologies) after 48-hour infection.

Cell Viability and Proliferation Assay

Cell viability and proliferation were determined with Cell Counting Kit 8 (CK04, Dojindo). The cells were seeded into 96-well plates at a density of 1 × 10⁴ cells per well at a final volume of 100 μl. Per the manufacturer’s instruction, 10 μl dye solution was added into the well at indicated time point. After incubation at 37°C for 2–3 hours, the absorbance at 450 nm was determined using BioTek Gen5 system (BioTeck, USA).

Cell Cycle Assay

The percentage of cells located at G0/G1, S, and G2/M phases was assessed by Propidium Iodide (PI, P4170, Sigma-Aldrich) DNA staining. 1 × 10⁶ cells were harvested, washed with PBS once, resuspended in 1 ml Krishan’s buffer supplemented with 0.05 mg/ml PI, 0.02 mg/ml ribonuclease A, 0.1% trisodium citrate, and 0.3% NP-40, and incubated at 37°C for 30 minutes. The samples were then applied to flow cytometry analysis directly.

Flow Cytometry Analysis

Flow cytometry analysis of stem cell and differentiation markers was performed as described previously ⁸¹. The mouse samples collected from peripheral blood, bone marrow, and spleen were depleted red blood cells with ammonium chloride solution (07850, STEMCELL Technologies) before staining. Cells were washed with PBS, re-suspended in ice-cold FACS buffer supplemented with anti–mouse CD16/32 (14–0161-82, eBioscience) to block non-specific binding and desired antibodies at 4°C for 30min at dark and mix every 10 minutes. Then cells were washed with PBS and resuspended in IC Fixation Buffer (eBioscience). The samples were analyzed on BD FACS FortessaX-20 (BD Biosciences, USA) and the results were analyzed with FlowJo V10 Software.

The following antibodies were used for flow cytometry: CD45.2-FITC (109806, Biolend), anti-lineage markers-eFluor 450 (88–7772-72, eBioscience), c-Kit-APC (17–1171-83, eBioscience), c-Kit-APC-Cy7 (135136, BioLegend), Sca-1-PE (12–5981-82, eBioscience), Mac-1-PerCP-Cyanine5.5 (45–0112-82, eBioscience), Gr-1-PE (12–5931-81, eBioscience), 7-AAD (559925, BD Bioscience), CD34-APC (128612, Biolend), CD16/CD32-PE-Cy7 (25–0161-82, Thermo Fisher Scientific), CD8 (17–0083-81, eBioscience), CD45-APC (17–9459-42, Thermo Fisher Scientific), CD33-PE (12–0339-42, Thermo Fisher Scientific), CXCR4 (phospho Ser339) (BS-12256R, Thermo Fisher Scientific), and CXCR4-PE (146505, BioLegend).

EdU Proliferation Assay

The EdU proliferation staining was performed following the manufacturer’s recommendation using the EdU Staining Proliferation Kit (ab219801, Abcam). 2 × 10⁶ cells were seeded into a 6-well plate and incubated with 20 μM EdU at 37°C for 2 hours. The cells were then harvested, washed, and fixed with fixative solution for 15 min. Then the cells were washed and permeabilized for 15 min. Finally, washed cells were incubated with reaction mix to fluorescently label EdU for 30 min at room temperature protected from dark, washed, and re-suspended in FACS buffer for further analysis.

In Vitro Colony-Forming and Replating Assay

The serial colony-forming assay was employed as described previously with minor modifications ⁸². Mouse BM progenitor cells (Lin⁻ or c-Kit⁺) were enriched from 6- to 8-week-old C57BL/6J (CD45.2⁺) mice upon 150 mg/kg 5-fluorouracil (5-FU) treatment for 5 days with Lineage Cell Depletion Kit (130–090-858, Miltenyi Biotec) or CD117 MicroBeads (130–091-224, Miltenyi Biotec) and infected with MSCV-based retroviruses as before mentioned. The transduced cells were seed at a density of 1 × 10⁴ per 35 mm culture dishes (27150, STEMCELL Technologies) in ColonyGEL (1201, ReachBio Research Labs) in the presence of 10 ng/ml of murine recombinant IL-3, IL-6, GM-CSF, and 50 ng/ml of murine recombinant SCF, using G418 and/or puromycin as selectable markers. The colonies were incubated at 37°C in a humidified atmosphere with 5% CO₂ in air for 6 to 7 days. Serial replating was then conducted by collecting and replating colony cells every 7 days. Colony numbers were enumerated and compared for each passage.

Syngeneic Bone Marrow Transplantation (BMT)

For primary BMT, transformed cells (5 × 10⁵) were mixed with 1 × 10⁶ CD45.1⁺ BM ‘helper’ cells and transplanted into lethally irradiated 6- to 8-week-old NCI C57 B6.SJL (CD45.1⁺) recipient mice. For secondary BMT, leukemia cells sorted from bone marrow of primary leukemia mice were injected into sub-lethally irradiated recipient mice. 1–5 × 10⁵ CD45.2⁺ leukemic cells isolated from the BM of primary recipients were transplanted into CD45.1⁺ recipient mice with or without sub-lethally irradiated.

AML PDX Model and ‘Human-in-Mouse’ Xenograft Model

Both female and male NRGS mice were used as hosts for xenotransplantation experiments. For each experiment, both female and male mice were randomly assigned to each group. In the “Human-in-Mouse” model, Mono Mac 6 (1 × 10⁵) cells with or without human TET2 CD overexpression were re-suspended in PBS and intravenously transplanted into NRGS recipient mice. We also utilized AML PDX cell ⁸³, #2017–129, to further perform xenotransplantation. 2 × 10⁵ #2017–129 cells were implanted into NRGS mice, and AML progression was monitored using in vivo bioluminescence imaging. To assess AML burden, mice were euthanized at the same time and stained with anti-CD45 (17–9459-42, Fisher Scientific) and anti-CD33 (12–0339-42, Thermo Fisher Scientific) antibodies and subjected to flow cytometry.

In Vitro Co-Culture Assay and Migration Assay

These assays were performed as described previously with minor modifications ⁸⁴. Briefly, 1.5 × 10⁴ OP9 cells were seeded in the lower chamber in 600μl culture medium overnight. Then, a total number of 2 × 10⁵ murine AML cells (Tet2^+/+ or Tet2^−/−) were washed with serum-free PBS once, resuspended in the culture medium and seeded in the upper chambers of Transwell (3 μm, 3415, Corning). After 12–24 hours of incubation, migrated AML cells in the lower chambers were harvested and counted.

Limiting Dilution Assays (LDAs)

LDAs were conducted as previously described ⁸². For in vitro LDAs, the murine NRAS/AE9a, NRAS/PML-RARA, or MA9 AML cells were suspended in ColonyGEL medium supplied with murine 50 ng/ml SCF, 10 ng/ml IL-3 and IL-6, 1 mg/mL G418, and 2 μg/mL puromycin and plated in 48- or 96-well plates with 6–8 different doses of cell number for each group. The number of wells containing spherical colonies was counted after 5–7 days. For in vivo LDA, MA9 AML cells with or without Tet2 knockout were injected into non-irradiated 6- to 8-week-old CD45.1⁺ recipient mice with six doses of cells for each group. The number of recipient mice that developed full-blown leukemia within 80 days post-transplantation was counted in each group with each dose of cells. The frequency of LSCs was determined by the ELDA online tool: http://bioinf.wehi.edu.au/software/elda/ ⁸⁵.

Histopathology Analysis

Livers and spleens were dissected from mice and fixed with formalin overnight. Paraffin embedding and H&E staining were conducted by the Pathology Core at City of Hope. PB and BM smears were briefly fixed in methanol for 30 seconds, followed by two-minute incubation with Wright-Giemsa Stain (24985, Polysciences), 6-minute incubation with Wright-Giemsa Stain/Buffer (24984, Polysciences) mixture, and a 2-minute rinse in Wright-Giemsa Buffer. All the slides were examined with Widefield Zeiss Observer 7 microscope.

In vivo Bioluminescence Imaging

In vivo bioluminescence imaging was conducted to monitor in vivo engraftment of AML cells in recipient mice. The mice were weighed, injected intraperitoneally with 150 mg/kg BW D-luciferin (LUCK-2G, Goldbio) in PBS solution, and then anesthetized with isoflurane. Ten minutes after injection, the mice were imaged with Lago X (Spectral Instruments Imaging). The bioluminescent signals were quantified using Aura imaging software (Spectral Instruments Imaging). Total flux values were determined and presented in photons/second/cm²/steradian.

In vivo Homing Assay

Homing assays were performed as previously described ⁸⁶. Briefly, a total number of 5–10 × 10⁶ AML cells were labeled with 5(6)-carboxy fluorescein diacetate succinimidyl ester (CFSE) (65–0850-84, eBioscience) or stably expressed GFP and transplanted into the lethally irradiated CD45.1⁺ recipient mice or NRGS mice. Homing of the AML cells in the BM were measured at 16 h or at indicated timepoint after transplantation by flow cytometric analysis and confocal imaging. For the AMD 3100 inhibition group, MA9 Tet2^−/− cells were treated with 0.1mg/ml AMD 3100 for 20 hours before being labeled by CFSE.

Immunofluorescent Staining of Long Bones

Long bones (femurs or tibias) from the mice were processed, sectioned, and imaged as described previously ⁸⁷. Briefly, the fresh long bones were dissected from the leukemic mice and immediately fixed with 4% paraformaldehyde for 4 hours. The fixed bones were then decalcified in 0.5M EDTA (pH 7.4–7.6) for 24 to 48 hours and then incubated in ice-old cryoprotectant (CPT) solution at 4°C for 24 hours. The decalcified long bones were then embedded in tissue molds with embedding media (EBM). After drying at room temperature and stored at −80°C overnight, the bones were sectioned longitudinally using a microtome with a thickness of 40 μm. The sections were rehydrated, permeabilized, blocked, and incubated with PE-conjugated osteopontin (OPN) (IC808P, R&D Systems) overnight at 4°C. After washing, the slides were mounted with DAPI mounting medium to stain the nuclei. Imaging was performed using a confocal microscope (Zeiss, LSM880) and analyzed using the Zen program (Zeiss).

RNA Extraction and Real-Time Quantitative PCR (RT-qPCR) Analysis

Total RNA samples were extracted from cells using TRIzol reagent according to the manufacturer’s instructions. Reverse transcriptase reaction was performed with 200–1000 ng of total RNA or immunoprecipitated RNA samples in 10 μL reaction volume using the QuantiTect Reverse Transcription kit (205314, QIAGEN) following the manufacturer’s instructions. Quantitative PCR (qPCR) was then performed with Maxima SYBR Green qPCR Master Mix (2×) (FEPK0253, Thermo Fisher) on the QuantStudio 7 Flex PCR system (Thermo Fisher Scientific). GAPDH, ACTIN, and/or 18S rRNA were used as endogenous control and each reaction was run in triplicates. All the primers are listed in Table S2.

In vitro (cell-free) Demethylation Assay with TET2 Protein

To assess the enzymatic activity of TET2 on DNA and RNA, the synthesized 30-mer single-stranded RNA (ssRNA) with internal m⁵C modification (5’-CAG UAA CUG UGG UC(m⁵C) GGU AAC UGA CUU GCA-3’) (Dharmacon), or a single-stranded DNA (ssDNA) with a single 5mC situated in the same sequence context (Integrated DNA Technologies), or in vitro transcribed m⁵C-bearing TSPAN13 were served as substrates. Briefly, 2.5 μM RNA/DNA oligos or 250 ng TSPAN13 mRNA were incubated with 0.25 μg catalytic domain of TET2 protein (31418, Active Motif) in 50 μL reaction buffer containing 50 mM HEPES, pH 8.0, 100 μM NaCl, 75 μM Fe(NH₄)₂(SO₄)₂, 1 mM 1,4-Dithiothreitol (DTT), 2 mM L-ascorbate, 100 μM alpha-ketoglutarate, and 100 μM ATP at 37°C for 20–120 minutes. The RNA or DNA were purified using Oligo Clean & Concentrator kit (D4061, Zymo Research). The activity was detected by dot blot or QQQ-MS.

In vitro Transcription

The TSPAN13 mRNA was obtained from in vitro transcription by using MEGAscript T7 kit (AM1333, Invitrogen). Briefly, cDNAs were synthesized with total RNA extracted using SuperScript III First-Strand Synthesis System (18080051, Invitrogen). Then the TSPAN13 was amplified by PCR which contains a T7 promoter sequence at its 5′ terminus using Platinum SuperFi II DNA Polymerase (12361010, Thermo Scientific). Primers employed to generate DNA templates are listed in Table S2. Purified TSPAN13 DNA was subjected to an in vitro transcription reaction with MEGAscript T7 RNA polymerase (AM1354, Thermo Fisher Scientific) at 37 °C for 4 h in a 50 μL reaction mixture containing 5-methylcytidine-5’-triphosphate (5mCTP) (N-1014–1, TriLink Biotechnologies), according to the manufacturer’s instructions. DNA template was digested with TURBO DNase (AM2239, Thermo Fisher Scientific). Transcripts were then purified by MEGAclear Transcription Clean-Up Kit (AM1908, Thermo Fisher Scientific).

UHPLC-QQQ-MS/MS

The levels of m⁵C, hm⁵C, 5mC, and 5hmC measured by ultra-high-performance liquid chromatography coupled with triple-quadrupole tandem mass spectrometry (UHPLC-QQQ-MS/MS). The nucleic acids were digested in one-step procedure as previously described ⁸⁸. 200 ng-1000 ng DNA or RNA samples were digested with 50 μL digestion mix containing 2.5 U Benzonase (E1014, Sigma-Aldrich), 3 mU phosphdiesterase I (P3243, Sigma-Aldrich), 2 U alkaline phosphatase (EF0651, Thermo Fisher Scientific), 20 mM Tris-HCl, pH7.9, 100 mM NaCl, and 20 mM MgCl₂ at 37°C for 6 hours. The digestion mixture was diluted to 200 μl by LC-MS grade water and analyzed by Agilent 6410 Triple Quadrupole Tandem Mass Spectrometer with an Agilent 1290 HPLC. Nucleosides were separated by the Agilent RRHD SB-C18 column (2.1 × 50mm,1.8μm,1200 bar). 0.1% formic acid (FA) in ddH2O as Solvent A and 0.1% FA in acetonitrile as Solvent B were employed as the mobile phase. The amounts of nucleosides were calibrated by standard curves obtained from pure nucleoside standards (Cayman Chemical Company).

Dot Blot Assay

Dot blot assays for quantification of 5mC and 5hmC in DNA and m⁵C and 5hm⁵C in RNA were conducted as previously reported ^17,63. Genomic DNA was isolated with a DNeasy Blood and Tissue Kit (69506, QIAGEN) according to the manufacturer’s instructions. DNA was denatured in 0.1 N NaOH at 99°C for 5 minutes and neutralized with 0.66 M ammonium acetate. RNA was denatured in RNA incubation buffer (65.7% formamide, 7.77% formaldehyde, and 1.33× MOPS) at 65 °C for 5 min, chilled on ice immediately, and mixed with equal volume of 20 × standard saline citrate (SSC) buffer. The denatured samples were spotted onto an Amersham Hybond-N⁺ membrane (45–000-927, GE Healthcare) using an assembled Bio-Dot apparatus (Bio-Rad) and followed by UV-crosslink (254 nm UV for 5 min). Then the membrane was stained with 0.02% methylene blue in 0.3 M sodium acetate (pH 5.2), blocked with 5% non-fat milk in 1×PBST for 1 h, and incubated with anti-5mC antibody (1:5000, 39649, Active Motif) or anti-5hmC antibody (1:5000, 39769, Active Motif) overnight at 4 °C. Then the membrane was washed for three times with 1×PBST, incubated with HRP-conjugated secondary antibody (1:5000) for 1h at room temperature. Dot blot was developed with Amersham ECL Prime Western Blotting Detection Reagent (45–010-090, GE Healthcare) and captured by a chemiluminescent imaging system.

Gene-specific m⁵C RNA Immunoprecipitation and qPCR

Poly(A)⁺ RNA was enriched from total RNA with polyATract mRNA isolation system IV kit (Z5310, Promega) following manufacturer’s instructions. m⁵C RNA immunoprecipitation (MeRIP) was performed as described previously with some modifications ⁵⁵. 11 μg mRNA was fragmented into 100–200 nt in length using NEBNext^® Magnesium RNA Fragmentation Module (E6150S, New England Biolabs). The fragmented RNA was purified and 5 μg mRNA was incubated with 5 μg anti-m⁵C antibody (ab10805, Abcam) or mouse IgG (NI03, Millipore) in 400 μL 1× IP buffer (150 mM NaCl, 0.1% NP-40, 10 mM Tris-HCl, pH 7.4, 0.4 U/ml RNase inhibitor) at 4°C overnight and one-tenth of fragmented RNA was saved as input control. Then the mRNA was incubated with washed conjugated Protein A/G Magnetic Beads (88803, Thermo Fisher Scientific) at 4°C for three hours. The magnetic beads were then washed four times with 1× IPP buffer and the bound RNAs were recovered by proteinase K digestion. The IP RNA fragments were finally eluted in using RNA Clean & Concentrator (R1015, Zymo Research). Input RNA and MeRIP-ed RNA were further analyzed by qPCR. Cycle threshold (Ct) values (ΔΔCt method) were used to determine the relative enrichment of m⁵C in each sample.

Nuclear Run-on Assay

Nuclear run-on assay was conducted as previously described ⁶³. Nuclei were isolated from WT and Tet2-null cells with Nonidet P-40 lysis buffer (10 mM Tris-Cl, pH 7.4, 10 mM NaCl, 3 mM MgCl₂, and 0.5% NP-40). The enriched nuclei were suspended in 396 μL nuclear freezing buffer (50 mM Tris-Cl, pH 8.3, 40% glycerol, 5 mM MgCl₂, and 0.1 mM EDTA). Then 100 μL of 5× run-on buffer (25 mM Tris-Cl, pH 8.0, 750 mM KCl, 12.5 mM MgCl₂, 1.25 mM each of ATP, GTP, and CTP) was added into the nuclei along with 4 mL biotin-16-UTP (11388908910, Sigma-Aldrich) to initiate the nuclear run-on reaction. After incubation at 29°C for 30 min, the reaction was quenched by adding 0.75 μL 1M CaCl₂, and 3 μL RNase-free DNase I and incubating for 10 min at 29°C. RNA was recovered with the RNeasy mini kit (74104, QIAGEN) and eluted in 50 μL of nuclease-free water. One-tenth of RNA was saved as ‘‘total nuclear RNA’’. Dynabeads M-280 streptavidin (11205D, Thermo Fisher Scientific) resuspended in binding buffer (10 mM Tris-HCl, pH 7.5, 1 mM EDTA and 2 M NaCl) were mixed with an equal volume of the RNA samples and incubated firstly at 42°C for 20 min and then room temperature for 2 hours. After washing with 2× SSC, the beads were resuspended in 25 μL of nuclease-free water and subjected to qPCR analysis. The qPCR products were then applied on a 2% agarose gel analysis and captured by Gel DocTM XR System (BIO-RAD).

Protein Extraction and Western Blot Assay

For total protein extraction, washed cell pellets were lysed in RIPA buffer (R0278, Sigma-Aldrich) supplemented with 5mM EDTA, 1 × Halt phosphatase inhibitor cocktail (78420, Thermo Fisher Scientific), and 1 × Halt protease inhibitor cocktail (78429, Thermo Fisher Scientific) on ice for 20 minutes, sonicated, and centrifuged at 20,000 g at 4°C for 15 minutes. Protein concentration was quantified with Bio-Rad Protein Assay Dye Reagent Concentrate (5000006, Bio-Rad) and bovine serum albumin (Protein Standard II #5000007, Bio-Rad). Protein lysates were diluted with 4× Laemmli Sample Buffer (5000007, Bio-Rad) containing β-mercaptoethanol and denatured at 95°C for 10 min.

Western blot was conducted as previously described ^63,82. Briefly, equal amounts of protein extracts (20–50 μg) were separated by 6%−10% SDS-PAGE gels and onto polyvinylidene difluoride (PVDF) membranes. The blots were blocked with 5% non-fat milk in PBST and incubated with diluted primary antibodies in 1% (w/v) BSA in PBST overnight at 4°C. Then membranes were washed with 1 × PBST, incubated with secondary antibody, and detected by Pierce ECL Western Blotting Substrate (PI32106, Thermo Fisher Scientific) or GE Healthcare Amersham ECL Prime Western Blotting Detection Reagent (45010090, GE Healthcare). The following primary antibodies were used for Western blotting: anti-GAPDH (1:5000, 10494–1-AP, Proteintech), anti-Vinculin (1:5000, sc-25336, Santa Cruz Biotechnology), anti-FLAG (1:3000, F3165, Sigma-Aldrich), anti-CXCR4 (1:2000, 60042–1-IG, Proteintech), anti-CXCR4 (phospho Ser339) (1:2000, GTX32281, GeneTex; 1:1000, BS-12256R, Thermo Fisher Scientific), anti-TSPAN13 (1:500, abx214152, Abbexa), anti-Lamin A/C (1:2000, 4777, Cell Signaling Technology), and anti-α-Tubulin (1:5000, 3873S, Cell Signaling Technology). Secondary antibodies used in this study include Horseradish peroxidase (HRP) conjugated Goat Anti-Mouse IgG H&L (HRP) (ab6789, abcam) and Goat Anti-Rabbit IgG H&L (HRP) (ab6721, abcam).

RNA Stability Assay

The cells were treated with 5 μg/ml actinomycin D (A9415, Sigma-Aldrich) for indicated time to assess RNA stability of TSPAN13 transcript. Total RNA was extracted from cells using TRIzol reagent for RT-qPCR analysis. The turnover rate and half-time of mRNA were estimated according to previously described ^63,82. Since the actinomycin D closes mRNA transcription, the half-life of mRNA concentration at a given time $(dC/dt)$ is proportional to both the rate constant of mRNA decay ($k_{decay}$) and mRNA concentration ($C$) as shown in the following equation:

$$dC/dt=-k_{\text{decay}}$$

Thus, the rate constant for mRNA degradation $k_{decay}$ could be estimated by the derivation of the equation:

$$\text{In}(C/C0)=-k_{\text{decay}}\cdot t$$

$C_0$ is the concentration of mRNA at time 0, $t$ is the transcription inhibition time, and $C$ is the mRNA concentration at the time t. To calculate the half-time $\left(t_{1/2}\right)$ of RNA, which means $C/C_0=50\text{\%}/100\text{\%}=1/2$, the equation can be rearranged into the following equation:

$$\text{In}\left(1/2\right)=-k_{\text{decay}}\cdot t_{1/2}$$

from where:

$$t_{1/2}=ln2/k_{\text{decay}}$$

mRNA Half-time Profiling and Data Analysis

WT and Tet2-null NRAS/AE9a AML cells were treated with 5 μg/ml actinomycin D (A9415, Sigma-Aldrich) for 0, 4, and 8 hours. At each time point, cells were rapidly harvested, and total RNA was extracted using TRIzol reagent. The RNA samples were subsequently subjected to sequencing on Illumina NovaSeq 6000 system with a paired-end 150-base pair (bp) read length. Reads were mapped to mm10 genome by STAR ⁷⁶. Gene expression (RPKM) was calculated by RSEM. The degradation rate and half-life of RNA were calculated using the formula mentioned earlier. The average values of the half-lives were calculated as described previously⁸⁹.

mRNA Bisulfite-seq and Data Analysis

Total RNA extraction from WT and Tet2-null NRAS/AE9a AML cells was performed using TRIzol reagent (15596–018, Thermo Fisher Scientific) and then was enriched for poly(A) RNA using polyATract mRNA isolation system IV kit (Z5310, Promega) following manufacturer’s instructions with the following modifications. The purified RNA was then subjected to bisulfite treatment using EZ RNA Methylation Kit (R5002, Zymo Research). Briefly, 1μg mRNA per sample in 20 μL nuclease-free water was mixed with 130 μl of RNA Conversion Reagent in a PCR tube. Then the mixture was incubated in a thermal cycler following 70°C for 10 minutes and 54°C for 50 minutes to convert unmethylated cytosine residues to uracil. The converted RNA was then subjected to column-based purification and next-generation sequencing using Illumina HiSeq 2500 platform with 70 million single-end reads per replicate sample. Adaptors were trimmed and low-quality bases were removed by Cutadapt ⁹⁰. Clean reads were mapped to the mouse reference genome (mm10) using meRanGh from meRanTK ⁹¹. The m⁵C sites were determined using meRanCall, the methylation caller of meRanTK with a false discovery rate < 0.01. Sites with a coverage depth ≥ 30, methylation level ≥ 0.1, and methylated cytosine depth ≥ 5 were considered high-confidence m⁵C sites ⁹².

RNA-seq and Data Analysis

For RNA-seq, total RNA extraction from WT and Tet2-null NRAS/AE9a cells was performed using TRIzol reagent (15596–018, Thermo Fisher Scientific) according to the manufacturer’s instructions. Total RNA was subjected to depletion of DNA and rRNA. RNA integrity was assessed by 2100 Bioanalyzer (Agilent Technologies, Santa Clara, USA). RNA library preparation was conducted using KAPA Stranded mRNA-Seq Kit (Illumina Platforms) (Kapa Biosystems, Wilmington, USA) with 10 cycles of PCR amplification and purified by AxyPrep Mag PCR Clean-up kit (Thermo Fisher Scientific). Libraries were run on an Illumina HiSeq 2500 (Illumina, San Diego, CA, USA) instrument in a 51 bp single-end run, generated by the TruSeq SR Cluster Kit V4-cBot-HS (Illumina). Sequencing reads were trimmed for adaptor sequence, masked for low-quality sequence by Cutadapt ⁹⁰, and aligned to mm10 genome by STAR ⁷⁶. Per million mapped reads (RPKM) of each gene were calculated by RSEM ⁷⁵, and p < 0.05 was set as the threshold of the differential expressions. The reads distributed in a specific transcript were displayed by IGV ⁷⁷. Gene Set Enrichment Analysis (GSEA) and hallmark gene sets in Molecular Signatures Database (MSigDB) ⁷⁸ were hired to calculate enriched pathways.

QUANTIFICATION AND STATISTICAL ANALYSIS

Data were analyzed with GrapPad Prism 9 and were presented as mean ± SEM or mean ± SD as indicated. Statistical Significance was calculated using two-tailed, unpaired Student’s t-tests, paired t-test, one-way ANOVA, or two-way ANOVA. Pearson tests were performed for correlation analysis. Kaplan-Meier survival curves were plotted with GraphPad Prism 8 and the $P$ values were calculated using the log rank test. p < 0.05 was considered a significant difference. For Western blot results, representative figures from three biological replicates were shown.

Supplementary Material

2

Table S2. Sequence of oligos used in this study (Related to STAR Methods)

Highlights.

• Microenvironment is crucial for TET2 deficiency-induced fast leukemogenesis in vivo

• TET2 loss promotes LSC homing into BM niche and thereby increases LSC self-renewal

• TET2 loss activates the TSPAN13/CXCR4 axis to facilitate LSC homing

• TET2 post-transcriptionally inhibits TSPAN13 expression as an mRNA m⁵C eraser