PTP4A2 promotes leukemogenesis through inhibiting the p53 tumor suppressor signaling pathway in leukemia-initiating cells

Key Points

• •PTP4A2 promotes the pathogenesis of AML through inhibiting LIC apoptosis and senescence.
• •PTP4A2 interacts with p53 and dephosphorylates it at serine 392, thereby inhibiting p53 activity in LICs.

Visual Abstract

Abstract

Acute myeloid leukemia (AML) is an aggressive hematological malignancy that is sustained by leukemia-initiating cells (LICs). Although protein tyrosine phosphatase 4A2 (PTP4A2), also known as phosphatase of regenerating liver 2 (PRL2), is highly expressed in AML, the mechanisms by which PTP4A2 promotes leukemogenesis are largely unexplored. In this study, we demonstrate that PTP4A2 promotes AML by inhibiting the p53 tumor suppressor pathway in LICs. Using KMT2A-MLLT3–driven AML as a model, we found that PTP4A2 deficiency activates p53 and induces LIC apoptosis and senescence, thereby extending the survival of recipient mice repopulated with Ptp4a2^−/− LICs. Mechanistically, PTP4A2 directly interacts with p53 and dephosphorylates it at serine 392, decreasing p53 stability and activity to enhance LIC proliferation and survival. Collectively, our findings identify p53 as a potential PTP4A2 substrate in leukemia cells and uncover a novel mechanism by which PTP4A2 enhances LIC maintenance.

Introduction

Acute myeloid leukemia (AML) is an aggressive hematological cancer associated with poor prognosis.¹ The disease arises from the malignant transformation of hematopoietic stem cells (HSCs) or myeloid progenitors, resulting in the formation of leukemia-initiating cells (LICs), also known as leukemic stem cells (LSCs).² AML is divided into several subtypes based on mutations and/or chromosomal abnormalities. A wide range of chromosomal abnormalities, including inv(16) (CBFβ-MYH11), t(8;21) (RUNX1–RUNX1T1), and 11q23/MLL (KMT2A) rearrangements, as well as mutations in NPM1, FLT3, and TP53 genes, have been identified in AML.3, 4, 5 According to the European LeukemiaNet classification, AML is categorized into 3 prognostic risk groups: favorable, intermediate, and adverse, based on cytogenetic and molecular features.⁵ The adverse- and intermediate-risk groups are associated with poor prognosis, poor overall survival, a higher rate of early mortality, resistance to standard therapies, and frequent relapse after chemotherapy.^6,7 These challenges highlight an urgent need to identify novel therapeutic targets to improve treatment outcomes in AML.

TP53 is a tumor suppressor gene that plays a key role in preventing cancer development. Mutations in TP53 or disruptions of the p53 signaling pathway can drive tumor development.⁸ Somatic TP53 mutations are found in ∼10% of AML.⁹ Notably, a broad range of AML subtypes, including those within the adverse-risk group, retain the wild-type (WT) TP53 gene, suggesting that disruption of the p53 signaling pathway rather than TP53 mutations contributes to leukemogenesis.^10,11 Therefore, identifying regulators of p53 activity in AML with WT TP53 is critical for understanding disease progression and developing novel therapeutic strategies.

The phosphatase of regenerating liver (PRL), also known as protein tyrosine phosphatase 4A (PTP4A) family of proteins, has been implicated in cancer metastasis, invasion, and progression, supporting its potential oncogenic roles.^12,13 The PTP4A family of phosphatases has emerged as a promising therapeutic target for both solid tumors and hematological malignancies.13, 14, 15 Among the 3 PTP4A family members (PTP4A1, PTP4A2, and PTP4A3), PTP4A2/PRL2 has been shown to play important roles in both normal and malignant hematopoiesis.^14,16 Compared with PTP4A1 and PTP4A3, PTP4A2 is highly expressed in hematopoietic stem and progenitors and mature hematopoietic cells.^16,17 Functionally, PTP4A2 acts as a phosphatase that dephosphorylates several substrates in leukemia cells, including PTEN at tyrosine 336 and the E3 ubiquitin ligase Casitas B-lineage Lymphoma (CBL) at tyrosine 371,18, 19, 20 thereby activating downstream pathways and promoting leukemogenesis. Nevertheless, the number of credible substrates identified for PTP4A2 remains limited.

Although previous studies have established the oncogenic role of PTP4A2 in leukemia,¹⁴ the mechanisms by which PTP4A2 promotes leukemogenesis are not fully understood. In this study, we investigated the function of PTP4A2 in AML using the KMT2A-MLLT3–driven AML model. We discovered that PTP4A2 promotes AML by inhibiting the p53 tumor suppressor pathway in LICs.

Methods

Methods not described in detail below are provided in supplemental Methods.

Human cell lines and stem and progenitor cells

The 293T cells and human leukemia cell lines used in this study were purchased from American Type Culture Collection (ATCC) or Leibniz Institute DSMZ-German Collection of Microorganisms and Cell Cultures (DSMZ). Cell lines were cultured according to ATCC or DSMZ handling guidelines, supplemented with 1% antibiotic-antimycotic (anti-anti). KMT2A-MLLT3⁺ human CD34⁺ cells (clones MA9.3 and MA9.4) were obtained from J.C.M. at Cincinnati Children’s Hospital Medical Center.²¹ Human CD34⁺ cells expressing KMT2A-MLLT3 were cultured in Iscove modified Dulbecco medium containing 10% fetal bovine serum, 1% L-glutamine, 1% anti-anti, and 10 ng/mL of each human cytokine, including stem cell factor (SCF), FMS-like tyrosine kinase 3 (FLT3) ligand, thrombopoietin (TPO), interleukin-3, and interleukin-6. The 293T cells were cultured for no more than 10 passages. All the leukemia cell lines were cultured for no longer than 3 months after the initial thawing.

Transplantation assays

Embryonic day 14.5 to 16.5 fetal liver cells were harvested from pregnant female mice. Ptp4a2^+/+ and Ptp4a2^−/− fetal liver cells were transduced with MSCV-KMT2A-MLLT3-IRES–green fluorescent protein (GFP) and sorted for GFP⁺ cells. A total of 200 000 GFP⁺ cells (Ptp4a2^+/+ or Ptp4a2^−/− and CD45.2⁺) were mixed with an equal number of helper bone marrow (BM) cells (CD45.1⁺) and were injected into lethally irradiated (950 cGy) recipient F1 mice (CD45.1⁺CD45.2⁺) via the tail vein. Following primary transplantation, either 100 or 1000 GFP⁺cKit⁺ BM cells (LICs) were isolated from primary recipients and transplanted into sublethally irradiated (500 cGy) secondary recipient mice. For KMT2A-MLLT3⁺ cell transplants involving 4 experimental groups (Ptp4a2^+/+, Ptp4a2^−/−, p53^+/−, and p53^+/−Ptp4a2^−/− fetal liver cells), the transplantation assays were performed in the same manner as above. Competitive BM transplantation assays were performed as described previously.¹⁸

Flow cytometry

Murine hematopoietic stem and progenitor cells (HSPCs) were identified and analyzed by flow cytometry based on their surface marker expression. Before staining, red blood cells were lysed using Red Blood Cell Lysis Buffer (eBioscience). Hematopoietic stem, progenitor, and mature cell populations were defined by the surface markers as follows: long-term HSCs (LT-HSCs; Lin⁻Sca1⁺Kit⁺CD150⁺CD48⁻), short-term HSCs (ST-HSCs; Lin⁻Sca1⁺Kit⁺CD150⁻CD48⁻), multipotent progenitors (MPPs; Lin⁻Sca1⁺Kit⁺CD150⁻CD48⁺), Lin⁻Sca1⁺Kit⁺ cells (LSKs), myeloid progenitors (Lin⁻Sca1⁻Kit⁺), megakaryocyte-erythroid progenitors (MEPs; Lin⁻Sca1⁻Kit⁺CD34⁻CD16/32⁻), common myeloid progenitors (CMPs; Lin⁻Sca1⁻Kit⁺CD34⁺CD16/32^med), granulocyte-monocyte progenitors (GMPs; Lin⁻Sca1⁻Kit⁺CD34⁺CD16/32⁺), myeloid cells (Gr1⁺CD11b⁺), lymphoid cells (Gr1⁻CD11b⁻), B cells (Gr1⁻CD11b⁻B220⁺CD3⁻), and T cells (Gr1⁻CD11b⁻B220⁻CD3⁺). The complete list of antibodies can be found in the study by Chen et al.¹⁸ Apoptosis assays were performed using annexin V in combination with 7-aminoactinomycin D (7-AAD) in the annexin V binding buffer. Flow cytometry experiments were performed using BD Biosciences fluorescence-activated cell sorting Fortessa cytometers and analyzed with FlowJo version 10.9 software.

Virus packaging and transduction

Lentiviral short hairpin RNA (shRNA) plasmids (pLB) targeting human PTP4A2 (shPTP4A2) or carrying a control luciferase sequence were generated previously.²² Lentiviral particles were produced using polyethyleneimine transfection with a third-generation packaging system (pMDL, pMD2.G, and pRSV-Rev). Retroviral plasmids used in this study included MSCV-KMT2A-MLLT3-IRES-GFP, MSCV-RUNX1-RUNX1T1-9a-IRES-GFP, and MIGR1-based vectors encoding WT-p53, p53-S392A, or p53-S392D. Retroviral particles were generated by polyethyleneimine transfection with the packaging plasmids pCL-Eco for mouse cells and pCL-Ampho for human cells. Retrovirus and lentivirus production and transduction were performed as described previously.¹⁸ GFP⁺ cells were sorted by fluorescence-activated cell sorting 48 hours after transduction.

Statistical analysis

The animal sample size was determined by power analysis; χ² analysis indicated that a group size of 7 mice provided 80% power to detect a difference with 95% confidence. A P value of <.05 was considered statistically significant, with calculations performed using GraphPad Prism 10 or R and statistical significance denoted as not significant, P > .05, ∗P < .05, ∗∗P < .01, ∗∗∗P < .001, and ∗∗∗∗P < .0001. Results are presented as the mean ± standard error of the mean from at least 3 independent experiments. The statistical methods used for analysis are detailed in the respective figure legends.

All experimental procedures related to animal studies were approved by the Institutional Animal Care and Use Committees of Indiana University School of Medicine and Northwestern University Feinberg School of Medicine.

Results

PTP4A2 is highly expressed in AML

To determine the role of PRL2, encoded by PTP4A2, in AML pathogenesis, we first analyzed its expression in patient samples from the Beat AML database.^23,24 PTP4A2 (PRL2) expression is significantly higher in AML samples than in healthy CD34⁺ cells, and its progressive increase across risk categories suggests a positive association with AML prognosis (Figure 1A). To further validate this observation in an independent data set, we analyzed PTP4A2 expression in the St. Jude AML cohort. Consistent with the Beat AML data set, PTP4A2 expression was significantly elevated in the adverse- and intermediate-risk groups compared with that in the favorable-risk group (supplemental Figure 1A). To investigate the role of PTP4A2 in AML, we performed gene set enrichment analysis (GSEA) on 250 patients with adverse-risk AML in the Beat AML database.²⁴ Patients were stratified into PTP4A2^high and PTP4A2^low expression groups based on the median PTP4A2 expression levels within the adverse-risk cohort. Cell cycle and DNA damage repair–related pathways are enriched in patients with adverse-risk AML with lower PTP4A2 expression (Figure 1B-C).

Figure 1.

PTP4A2 is highly expressed in AML. (A) PTP4A2 (PRL2) expression level in adverse-, intermediate-, and favorable-risk AML (ELN2017), and healthy CD34⁺ cells from Beat AML data sets. (B) Dot plot of selected pathways ranked by normalized enrichment score (NES). Positive NES represents the pathways enriched in adverse-risk group patients with low PTP4A2 expression, whereas negative NES represents the pathways enriched in adverse-risk group patients with high PTP4A2 expression (n = 125). (C) Specific GSEA plot showing the enrichment of the DNA repair genes, cell cycle checkpoints, DNA repair, G0 and early G1, and apoptosis by CDKN1A via TP53 pathways in adverse-risk group patients with low PTP4A2 expression compared with those with high PTP4A2 expression. (D) Dot plot of selected pathways ranked by NES. Positive NES represents the pathways enriched in patients with KMT2A-rearranged AML with low PTP4A2 expression, whereas negative NES represents the pathways enriched in patients with KMT2A-rearranged AML with high PTP4A2 expression (n = 20). (E) Specific GSEA plot showing the enrichment of the apoptosis by CDKN1A via TP53, G0 and early G1, cell cycle checkpoints, and DNA repair pathways in patients with KMT2A-rearranged AML with low PTP4A2 expression compared with those with high PTP4A2 expression. (F) Western blot analysis of PTP4A2 expression in various AML cell lines (Molm-13, THP-1, MV4-11, RS4-11, SKNO-1, Kasumi-1, U937, and HL-60) compared with that in CB cells. (G) Western blot analysis of PTP4A2 expression in HL-60 and U937 cells overexpressing the vector control or KMT2A-MLLT3. Statistical significance: not significant (ns), P > .05; ∗P < .05; ∗∗∗∗P < .0001. CB, cord blood; Ctrl, control; ERK, extracellular signal-regulated kinase; ELN2017, European LeukemiaNet 2017; IL6, interleukin-6; PI3K, phosphatidylinositol 3-kinase; RPKM, reads per kilobase of transcript per million mapped reads.

Mixed lineage leukemia (MLL) gene (KMT2A) rearrangement represents one of the most frequent chromosomal translocations in AML, occurring in ∼15% to 20% of all AML.^6,25 KMT2A-rearranged AMLs are typically classified within the intermediate- to adverse-risk groups, depending on their C-terminal fusion partners.^3,4 Therefore, we also performed GSEA in the cohort of 40 patients with KMT2A-rearranged AML. The KMT2A-rearranged cohort exhibited a similar enrichment pattern to that observed in the adverse-risk group (Figure 1D-E), suggesting shared signaling pathways between the 2 cohorts.

To validate the elevated PTP4A2 expression in AML, we examined PTP4A2 levels in various human leukemia cell lines by western blot. PTP4A2 was highly expressed across all leukemia cell lines compared with that in healthy cord blood cells, and its expression was particularly elevated in cell lines derived from patients with KMT2A-rearranged AML, such as Molm-13 and MV4-11 (Figure 1F). Furthermore, PTP4A2 expression was markedly increased in HL-60 and U937 cells transduced with KMT2A-MLLT3 compared with that in the vector control (Figure 1G). All these results demonstrate that PTP4A2 is highly expressed in AML cells and that its expression can be further upregulated by KMT2A-MLLT3. Importantly, these findings do not imply that PTP4A2 overexpression is specific to KMT2A-rearranged AML; rather, KMT2A-MLLT3 is used here as a well-established experimental model of aggressive, high-risk AML to investigate the functional contribution of PTP4A2 to LICs.

Ptp4a2 is critical for KMT2A-MLLT3–driven leukemogenesis

To determine the role of Ptp4a2 in the initiation and maintenance of AML, we used a well-established mouse model of human AML induced by KMT2A-MLLT3.^26,27 We introduced KMT2A-MLLT3 into Ptp4a2^+/+ and Ptp4a2^−/− fetal liver cells, which are enriched for HSPCs, using a retrovirus carrying complementary DNA that encodes KMT2A-MLLT3 (MSCV-KMT2A-MLLT3-IRES-GFP) and then performed in vitro and in vivo studies using transduced cells (GFP⁺) (Figure 2A). We found that the loss of Ptp4a2 decreases the proliferation (supplemental Figure 1B) and replating potential (Figure 2B) of HSPCs expressing KMT2A-MLLT3. The vector control exhibited a modest effect on the colony formation of both Ptp4a2^+/+ and Ptp4a2^−/− fetal liver cells (supplemental Figure 1C). We then transplanted transduced cells (GFP⁺) into lethally irradiated recipient mice. The development of leukemia in hosts that received Ptp4a2^+/+ HSPCs transduced with KMT2A-MLLT3 retrovirus was rapid, with all control animals succumbing to the disease and dying 7 weeks following transplantation. In contrast, animals repopulated with Ptp4a2^−/− HSPCs showed extended survival (Figure 2C). To determine the role of Ptp4a2 in LIC maintenance, we performed limiting dilution transplantation assays and transplanted 100 or 1000 LICs (Kit⁺GFP⁺) purified from the BM of primary recipients that had developed leukemia into sublethally irradiated recipient mice. Although most recipient mice repopulated with Ptp4a2^+/+ LICs developed leukemia and died within 30 days, the recipient mice repopulated with Ptp4a2^−/− LICs showed extended survival (Figure 2D). Given that Ptp4a2 is important for HSPC proliferation and the loss of Ptp4a2 impairs HSPC engraftment in recipient mice,^17,18 these findings suggest that Ptp4a2 deficiency may decrease the number or the proliferation of LICs that can reinitiate leukemia in secondary recipient mice.

Figure 2.

Ptp4a2 promotes leukemogenesis by inhibiting p53 activity in LICs. (A) Experimental schematic of KMT2A-MLLT3⁺ fetal liver cell transplantation. (B) Serial replating of 5000 Ptp4a2^+/+ and Ptp4a2^−/− fetal liver cells transduced with KMT2A-MLLT3 (n = 3). (C) Survival of primary transplantation recipient mice that underwent transplantation with KMT2A-MLLT3⁺Ptp4a2^+/+ or Ptp4a2^−/− fetal liver cells. (D) Survival of secondary transplantation recipient mice that underwent transplantation with KMT2A-MLLT3⁺Ptp4a2^+/+ Kit⁺ or KMT2A-MLLT3⁺Ptp4a2^−/−Kit⁺ BM cells from primary transplantation recipient mice. (E) Dot plot of selected pathways ranked by NES. Positive NES represents the pathways enriched in the KMT2A-MLLT3⁺Ptp4a2^−/− group, whereas negative NES represents the pathways enriched in the KMT2A-MLLT3⁺Ptp4a2^+/+ group (n = 2). (F) Specific GSEA plot showing the enrichment of the regulation of TP53 activity pathway in KMT2A-MLLT3⁺Ptp4a2^−/− LICs compared with that in the Ptp4a2^+/+ group. (G) Specific GSEA plot showing the enrichment of the regulation of TP53 activity through phosphorylation pathway in KMT2A-MLLT3⁺Ptp4a2^−/− LICs compared with that in the Ptp4a2^+/+ group. (H) Quantitative polymerase chain reaction (qPCR) analysis of p53, p21, and Ptp4a2 in Ptp4a2^+/+ and Ptp4a2^−/− fetal liver cells expressing KMT2A-MLLT3 (n = 4). Statistical significance: ns, P > .05; ∗∗P < .01; ∗∗∗P < .001; ∗∗∗∗P < .0001. FDR, false discovery rate; LSCs, leukemic stem cells; NOM, nominal.

Ptp4a2^−/− LICs show enhanced p53 activity

To determine the mechanisms by which Ptp4a2 enhances LIC maintenance, we performed RNA sequencing (RNA-seq) analysis to compare gene expression in Ptp4a2^+/+ and Ptp4a2^−/− LICs expressing KMT2A-MLLT3. GSEA revealed significant enrichment of several p53-related pathways in Ptp4a2^−/− LICs expressing KMT2A-MLLT3 (Figure 2E). Notably, the regulation of TP53 activity, regulating TP53 activity through phosphorylation, cell cycle arrest, and TP53-mediated transcription of cell cycle gene signatures, was significantly enriched in Ptp4a2^−/− LICs compared with that in Ptp4a2^+/+ LICs (Figure 2F-G; supplemental Figure 1D-E). p21 is a key target of p53 in hematopoietic cells.^28,29 We found that the TP53 target gene p21 is significantly upregulated in Ptp4a2^−/− LICs (Figure 2H), suggesting that the p53 pathway is activated in Ptp4a2^−/− LICs.

These results are consistent with those from our RNA-seq analysis of samples from patients with AML from the Beat AML database, in which gene signatures associated with p21-mediated apoptosis via p53, G0/G1 phase of cell cycle, cell cycle checkpoints, and DNA repair pathways were significantly enriched in patients with AML with low PTP4A2 expression (Figure 1B-E; supplemental Figure 1F-G). Consistent with previous studies showing that most patients with KMT2A-rearranged leukemia retain WT TP53,^10,11 our analysis of the Beat AML and The Cancer Genome Atlas databases further supported these findings. Among 54 patients with KMT2A-rearranged AML, only 1 harbored a TP53 mutation (supplemental Figure 1H). Notably, the analysis of adverse-risk AML in the Beat AML cohort revealed no significant difference in PTP4A2 expression between WT TP53 and mutated samples (supplemental Figure 1I). All these findings suggest that PTP4A2 contributes to leukemogenesis by inhibiting the p53 tumor suppressor pathway.

Loss of Ptp4a2 activates p53 in HSPCs

To determine whether the loss of Ptp4a2 activates p53 is a general mechanism that extends to normal HSPCs, we analyzed Ptp4a2^+/+ and Ptp4a2^−/− BM Kit⁺ cells following irradiation (supplemental Figure 2A). Quantitative polymerase chain reaction analysis showed significantly higher p21 in Ptp4a2^−/− Kit⁺ cells than in Ptp4a2^+/+ controls (Figure 3A). Western blot further revealed increased p21 protein level in the Ptp4a2^−/− group, whereas p53 level remained unchanged (Figure 3B-C). These results suggest that the loss of Ptp4a2 may enhance p53 activity, thereby upregulating its downstream target p21 in HSPCs.

Figure 3.

Loss of Ptp4a2 activates p53 and decreases the repopulating potential of p53^+/− HSPCs. (A) qPCR analysis of p53 and p21 in Kit⁺ BM cells isolated from Ptp4a2^+/+ and Ptp4a2^−/− mice following 650 cGy irradiation (n = 3). (B) Western blot analysis of p53, p21, and PTP4A2 in Kit⁺ BM cells isolated from Ptp4a2^+/+ and Ptp4a2^−/− mice following 650 cGy irradiation. (C) Quantification of p21, p53, and PTP4A2 levels in Ptp4a2^+/+ and Ptp4a2^−/− Kit⁺ BM cells based on western blot analysis in panel B (n = 3). (D-F) Absolute numbers of LT-HSCs (Lin⁻Sca1⁺Kit⁺CD150⁺CD48⁻), ST-HSCs (Lin⁻Sca1⁺Kit⁺CD150⁻CD48⁻), and MPPs (Lin⁻Sca1⁺Kit⁺CD150⁻CD48⁺) in the BM of Ptp4a2^+/+, Ptp4a2^−/−, p53^+/−, and p53^+/−Ptp4a2^−/− mice (n = 8 mice per group). (G) Colony-forming assays of 20 000 BM mononuclear cells (BMMNCs) from Ptp4a2^+/+, Ptp4a2^−/−, p53^+/−, or p53^+/−Ptp4a2^−/− mice after the first week of culture (n = 3). (H) Serial replating of 20 000 BMMNCs from Ptp4a2^+/+, Ptp4a2^−/−, p53^+/−, or p53^+/−Ptp4a2^−/− mice over 3 weeks (n = 3). (I) Frequency of donor-derived cells (CD45.2⁺) in the PB of recipient mice repopulated with Ptp4a2^+/+, Ptp4a2^−/−, p53^+/−, or p53^+/−Ptp4a2^−/− BM cells over 16 weeks (n = 9-10 mice per group). (J) Frequency of donor-derived cells in the BM of transplantation recipient mice. (K-M) Absolute number of donor-derived LT-HSCs, ST-HSCs, and MPPs in the BM of transplantation recipient mice (n = 9-10 mice per group). Statistical significance: ns, P > .05; ∗P < .05; ∗∗P < .01; ∗∗∗P < .001; ∗∗∗∗P < .0001. G, granulocyte; GEMM, granulocyte, erythroid, macrophage, and megakaryocyte; GM, granulocyte-macrophage; M, macrophage.

Loss of Ptp4a2 decreases the repopulating potential of p53^+/− HSPCs

To determine the interaction between p53 and Ptp4a2 in hematopoiesis and leukemogenesis, we have generated p53^+/−Ptp4a2^−/− mice. We were unable to generate p53^−/−Ptp4a2^−/− double-knockout mice because of embryonic lethality. We isolated BM mononuclear cells from Ptp4a2^+/+, Ptp4a2^−/−, p53^+/−, and p53^+/−Ptp4a2^−/− mice and then examined the number of HSPCs by flow cytometry analysis. In steady-state hematopoiesis, Ptp4a2^−/− mice exhibited a significant reduction of absolute number in LT-HSCs and ST-HSCs, whereas MPPs were unchanged (Figure 3D-F). We also observed a decreased frequency of ST-HSCs and lineage negative (Lin⁻) cells in the BM of Ptp4a2^−/− mice (supplemental Figure 2C,F), whereas the frequency of LT-HSCs, MPPs, and LSKs was unchanged (supplemental Figure 2B,D-E). Given that the spleen serves as the site of extramedullary hematopoiesis,³⁰ we also examined the frequency of HSPCs in the spleens of these 4 groups of mice. We observed decreased spleen size and numbers of splenocytes in Ptp4a2^−/− and p53^+/−Ptp4a2^−/− mice compared with those in Ptp4a2^+/+ and p53^+/− mice (supplemental Figure 2G-H); however, both Ptp4a2^−/− and p53^+/−Ptp4a2^−/− mice have a normal frequency of LT-HSCs and ST-HSCs in the spleen (supplemental Figure 2I-J).

Subsequently, we performed serial replating assays using Ptp4a2^+/+, Ptp4a2^−/−, p53^+/−, and p53^+/−Ptp4a2^−/− BM cells. p53^+/− BM cells displayed enhanced colony-forming potential, with a higher proportion of granulocyte, erythroid, macrophage, and megakaryocyte (GEMM) colonies in week 1 (Figure 3G), whereas p53^+/−Ptp4a2^−/− BM cells showed decreased replating potential compared with that of p53^+/− BM cells (Figure 3H). We next performed competitive BM transplantation assays to further assess the functional impact of p53 on Ptp4a2^−/− HSPCs. We transplanted 500 000 Ptp4a2^+/+, Ptp4a2^−/−, p53^+/−, or p53^+/−Ptp4a2^−/− BM cells (CD45.2⁺) together with 500 000 WT competitor BM cells (CD45.1⁺) into lethally irradiated (9.5 Gy) recipient mice (CD45.1⁺CD45.2⁺). Peripheral blood (PB) was collected from recipient mice every 4 weeks for a chimerism analysis of donor-derived cells (CD45.2⁺). As expected, p53^+/− BM cells showed increased engraftment compared with Ptp4a2^+/+ and Ptp4a2^−/− BM cells at 16 weeks following transplantation, whereas the loss of Ptp4a2 decreased the repopulating potential of p53^+/− BM cells (Figure 3I), suggesting that Ptp4a2 deficiency may impair the reconstitutive capacity of p53^+/− BM cells. To determine whether the loss of Ptp4a2 affects HSC function under transplantation-induced stress, the recipient mice were euthanized at 16 weeks following transplantation, and we examined the frequency of donor-derived HSPCs in their BM and spleen. The loss of Ptp4a2 decreased the engraftment of p53^+/− cells in the BM (Figure 3J). However, both the absolute number and frequency of donor-derived LT-HSCs, ST-HSCs, and MPPs in recipient BM were comparable between the Ptp4a2^+/+ and Ptp4a2^−/− groups (Figure 3K-M; supplemental Figure 2K-M). Although p53^+/− recipients showed increased absolute numbers of ST-HSCs and MPPs, these were significantly reduced by the loss of Ptp4a2 (Figure 3L-M). These results indicate that Ptp4a2 deficiency does not impair the long-term maintenance of HSCs following transplantation stress. Notably, recipient mice repopulated with p53^+/− BM had splenomegaly (supplemental Figure 2N-O) and an increased frequency of LT-HSCs, ST-HSCs, MPPs, and LSKs compared with the spleens of other groups (supplemental Figure 2P-S). However, Ptp4a2 deficiency in the p53^+/− background significantly decreased the number of ST-HSCs, MPPs, and LSKs in the spleen of recipient mice (supplemental Figure 2Q-S). In addition, the frequency of donor-derived LT-HSCs, ST-HSCs, MPPs, and LSKs in the recipient spleen was comparable between the Ptp4a2^+/+ and Ptp4a2^−/− groups (supplemental Figure 2P-S). All these findings demonstrate that Ptp4a2 deficiency impairs the functions of p53^+/− HSPCs both in vitro and in vivo while having minimal impact on long-term HSC maintenance after transplantation.

The extended survival of recipient mice repopulated with Ptp4a2^−/− LICs depends on p53

To determine the role of Ptp4a2 and p53 in KMT2A-MLLT3–driven leukemia, we introduced KMT2A-MLLT3 into Ptp4a2^+/+, Ptp4a2^−/−, p53^+/−, and p53^+/−Ptp4a2^−/− fetal liver cells using retroviruses and then used transduced cells (GFP⁺) to perform serial replating assays. We found that the loss of Ptp4a2 significantly decreases the replating potential of p53^+/− LICs (Figure 4A). We then performed transplantation assays using Ptp4a2^+/+, Ptp4a2^−/−, p53^+/−, and p53^+/−Ptp4a2^−/− HSPCs expressing KMT2A-MLLT3. PB was collected from recipient mice at week 4 following transplantation for chimerism and complete blood count analysis. Compared with other groups, recipient mice repopulated with Ptp4a2^−/− LICs show decreased engraftment in PB and reduced white blood cell count (Figure 4B-C). We monitored survival and disease development in recipient mice. Although the survival of recipient mice repopulated with Ptp4a2^+/+ or p53^+/− LICs is comparable, the loss of 1 allele of p53 in the Ptp4a2^−/− background significantly reduced the survival of recipient mice (Figure 4D). Importantly, the loss of Ptp4a2 did not alter the number of LIC-enriched fractions (GFP⁺Kit⁺), HSC-like (LSK CD150/CD48), and GMP-like populations in the BM of recipient mice that underwent transplantation with fetal liver cells expressing KMT2A-MLLT3 (Figure 4E-I; supplemental Figure 3A). Compared with control recipients, recipient mice that underwent transplantation with Ptp4a2^−/− LICs exhibited significantly prolonged survival (Figure 4D), indicating that Ptp4a2 deficiency compromises LIC function. A pathological analysis of the BM, spleen, and liver of recipient mice revealed acute leukemia in all groups (Figure 4J; supplemental Figure 3B).

Figure 4.

The extended survival of recipient mice repopulated with Ptp4a2^−/− LICs depends on p53. (A) Serial replating of 5000 KMT2A-MLLT3⁺GFP⁺Ptp4a2^+/+, Ptp4a2^−/−, p53^+/−, or p53^+/−Ptp4a2^−/− fetal liver cells over 8 weeks (n = 3). (B) Frequency of donor-derived cells in the PB of KMT2A-MLLT3⁺ fetal liver cells transplanted primary recipient mice at week 4. (C) White blood cell (WBC) count (10³/μL) of primary recipient mice that underwent transplantation with KMT2A-MLLT3⁺GFP⁺Ptp4a2^+/+, Ptp4a2^−/−, p53^+/−, or p53^+/−Ptp4a2^−/− fetal liver cells at week 4 (n = 11-12 mice per group). (D) Survival of primary transplantation recipient mice that underwent transplantation with KMT2A-MLLT3⁺ fetal liver cells (n = 11-12 mice per group). (E-F) Absolute number of donor-derived GFP⁺Kit⁺ and GFP⁺ leukemic L-GMP cells in the BM of primary transplantation recipient mice euthanized at week 7 (n = 4 mice per group). (G-I) Absolute number of donor-derived GFP⁺ LT-HSC–like, ST-HSC–like, and MPP-like LICs in the BM of primary transplantation recipient mice euthanized at week 7 (n = 4 mice per group). (J) Hematoxylin and eosin staining images (×20) of the BM, spleen, and liver of primary transplantation recipient mice (Ptp4a2^+/+, Ptp4a2^−/−, p53^+/−, and p53^+/−Ptp4a2^−/−) euthanized at week 7. Scale bar, 150 μm. (K) Frequency of donor-derived cells in the PB of secondary recipient mice at week 4. (L) WBC count (10³/μL) of secondary recipient mice at week 4 (n = 8-15 mice per group). (M) Survival of secondary transplantation recipient mice that underwent transplantation with KMT2A-MLLT3⁺Kit⁺ BM cells from primary transplantation recipient mice (n = 8-15 mice per group). Statistical significance: ns, P > .05; ∗P < .05; ∗∗P < .01; ∗∗∗P < .001; ∗∗∗∗P < .0001. FL, fetal liver; L-GMP, leukemic granulocyte-monocyte progenitor.

To determine the impact of p53 deficiency on LIC maintenance, we performed secondary transplantation using LICs (GFP⁺Kit⁺) isolated from the BM of primary recipients. Recipient mice repopulated with Ptp4a2^−/− LICs show decreased engraftment in PB and reduced white blood cell count compared with those in other groups at 4 weeks after transplantation (Figure 4K-L). Previous studies have shown that recipient mice repopulated with GMPs expressing KMT2A-MLLT3 predominantly develop AML.^26,31 A flow cytometry analysis of BM cells from deceased mice revealed a substantial accumulation of myeloid blasts, along with a significant reduction in B and T cells (supplemental Figure 3C-D). Complete blood count analysis of secondary transplantation recipient mice at 4-week posttransplantation revealed lymphopenia and myeloid expansion, particularly in monocytes and neutrophils, along with a significant reduction in platelet count compared with that in normal transplantation recipients (supplemental Figure 3E-O). Recipient mice repopulated with p53^+/− LICs show reduced survival compared with those repopulated with Ptp4a2^+/+ LICs. Consistent with primary transplantation, compared with mice repopulated with Ptp4a2^−/− cells, recipient mice that underwent transplantation with p53^+/−Ptp4a2^−/− cells displayed significantly reduced survival (Figure 4M). These findings suggest that p53 tumor suppressor activity contributes to the improved survival observed in recipient mice repopulated with Ptp4a2^−/− LICs.

Ptp4a2^−/− LICs show enhanced apoptosis and senescence that is p53-dependent

Our RNA-seq data in LICs indicate that p53-dependent apoptosis and senescence gene signatures are enriched in Ptp4a2^−/− LICs expressing KMT2A-MLLT3 (Figure 2., Figure 5.E and 5A-B). To validate these findings, we examined apoptosis and senescence in Ptp4a2^+/+, Ptp4a2^−/−, p53^+/−, and p53^+/−Ptp4a2^−/− LICs expressing KMT2A-MLLT3. Compared with Ptp4a2^+/+ LICs, Ptp4a2^−/− LICs showed increased apoptosis, whereas the loss of 1 allele of p53 in the Ptp4a2^−/− background brought apoptosis back to normal (Figure 5C-D). Oncogene-induced cellular senescence is a tumor suppressive mechanism triggered by oncogene activation.^32,33 p53/p21 is one of the important pathways regulating cellular senescence.³⁴ We hypothesized that p53-dependent senescence contributes to the extended survival observed in recipient mice repopulated with Ptp4a2-deficient LICs. Indeed, Ptp4a2^−/− LICs exhibited increased senescence, whereas the loss 1 allele of p53 results in reduced senescence (Figure 5E-F), indicating that enhanced senescence observed in Ptp4a2^−/− LICs is p53-dependent. Thus, we demonstrate that PTP4A2 impairs p53 activation, thereby inhibiting LIC apoptosis and senescence to promote leukemogenesis.

Figure 5.

Ptp4a2^−/− LICs show enhanced apoptosis and senescence that is p53 dependent. (A) Specific GSEA plot showing the enrichment of the apoptosis by CDKN1A via TP53 pathway in KMT2A-MLLT3⁺Ptp4a2^−/− LICs compared with that in the Ptp4a2^+/+ group. (B) Specific GSEA plot showing the enrichment of the cellular senescence pathway in KMT2A-MLLT3⁺Ptp4a2^−/− LICs compared with that in the Ptp4a2^+/+ group. (C) Flow cytometry analysis of apoptosis in KMT2A-MLLT3⁺ fetal liver cells from Ptp4a2^+/+, Ptp4a2^−/−, p53^+/−, and p53^+/−Ptp4a2^−/− groups. (D) Quantification of apoptotic cell frequency (annexin V⁺7-AAD⁻) in KMT2A-MLLT3⁺ fetal liver cells (n = 4). (E) Images of senescence-associated β-galactosidase assay (Leica microscope 4×) in KMT2A-MLLT3⁺Ptp4a2^+/+, Ptp4a2^−/−, p53^+/−, and p53^+/−Ptp4a2^−/− fetal liver cells. Senescent cells appear blue, and nonsenescent cells appear unstained. Scale bar, 20 μm. (F) Quantification of senescent cells percentage in KMT2A-MLLT3⁺ fetal liver cells from Ptp4a2^+/+, Ptp4a2^−/−, p53^+/−, and p53^+/−Ptp4a2^−/− groups (n = 4). One-way analysis of variance followed by the Tukey multiple comparisons test. Statistical significance: ns, P > .05; ∗P < .05; ∗∗∗P < .001; ∗∗∗∗P < .0001. FDR, false discovery rate; FL, fetal liver; NOM, nominal.

To determine whether this phenotype extends beyond KMT2A-rearranged AML, we examined LICs expressing the RUNX1–RUNX1T1-9a fusion.³⁵ The loss of Ptp4a2 similarly impaired the colony-forming capacity (supplemental Figure 4A) and increased senescence (supplemental Figure 4B-C) of LICs expressing RUNX1–RUNX1T1-9a. However, in contrast to the KMT2A-MLLT3 model, RUNX1–RUNX1T1-9a–induced senescence of Ptp4a2^−/− LICs was p53-independent (supplemental Figure 4B-C). We also found that compared with Ptp4a2^+/+ mouse embryonic fibroblasts, Ptp4a2^−/− mouse embryonic fibroblasts show significantly increased senescence (supplemental Figure 4D-E).

PTP4A2 interacts with p53 and dephosphorylates it at serine 392 to inhibit its stability and activity

To determine the role of PTP4A2 in human LICs, we transduced human CD34⁺ cells expressing KMT2A-MLLT3 with lentiviruses expressing a control shRNA or 2 shRNAs targeting different regions of human PTP4A2, respectively (supplemental Figure 5A-B).^18,21 We performed colony-forming unit (CFU) and proliferation assays using transduced cells (GFP^high) and found that the knockdown of PTP4A2 decreased the colony formation and proliferation of human CD34⁺ cells expressing KMT2A-MLLT3, including both MA9.3 and MA9.4 clones (Figure 6A; supplemental Figure 5C-E). Consistent with our findings in murine LICs expressing KMT2A-MLLT3 (Figure 2H), the knockdown of PTP4A2 significantly increases p21 expression in human CD34⁺ cells expressing KMT2A-MLLT3 (Figure 6B), suggesting that PTP4A2 deficiency activates p53 in human LICs.

Figure 6.

PTP4A2 interacts with p53 and dephosphorylates it at serine 392 to inhibit its stability and activity. (A) Colony-forming assays of 5000 KMT2A-MLLT3⁺ human CD34⁺ cells (MA9.3) transduced with lentivirus carrying shControl (luciferase) or 2 different PTP4A2 shRNAs (shPTP4A2#1 and shPTP4A2#2) (n = 3). Scale bar, 250 μm. (B) qPCR analysis of p53, p21, and PTP4A2 in steady-state KMT2A-MLLT3⁺ human CD34⁺ cells expressing shControl, shPTP4A2#1, and shPTP4A2#2 (n = 4). (C) Western blot analysis of p53 half-life in steady-state KMT2A-MLLT3⁺ human CD34⁺ cells expressing shControl, or shPTP4A2. Cells were treated with 100 μg/mL of the protein synthesis inhibitor cycloheximide (CHX) for 0, 0.5, 1, or 2 hours. (D) Quantification of p53 half-life based on western blot analysis from panel C (n = 3). (E) GST pull-down assays in KMT2A-MLLT3⁺ human CD34⁺ cells (MA9.3) and MV4-11 cells. Lysates from MA9.3 and MV4-11 cells were incubated with purified GST, GST-PTP4A2, and GST-PTP4A2-CSDA, followed by immunoblotting for p53 and PTEN. (F) Phosphorylation of p53. (G) Western blot analysis of phosphorylated p53 at Ser15, Ser20, and Ser392 in Ptp4a2^+/+ or Ptp4a2^−/− Kit⁺ mouse BM cells following 500 cGy irradiation. (H) Quantification of phosphorylated p53 (Ser392) levels in Ptp4a2^+/+ or Ptp4a2^−/− Kit⁺ mouse BM cells based on western blot analysis in panel G (n = 3). (I) Western blot analysis of phosphorylated p53 at Ser15, Ser20, Thr81, Ser46, and Ser392 in KMT2A-MLLT3⁺ human CD34⁺ cells expressing shControl, shPTP4A2 #1, or shPTP4A2#2 following 500 cGy irradiation. (J) Quantification of phosphorylated p53 (Ser392) levels in KMT2A-MLLT3⁺ human CD34⁺ cells expressing shControl, shPTP4A2#1, or shPTP4A2#2 based on western blot analysis in panel I (n = 3). (K) Western blot analysis of p53 half-life in steady-state THP-1 cells overexpressing MIGR1–WT-p53, MIGR1–p53-S392A, or MIGR1–p53-S392D. Cells were treated with 100 μg/mL of the protein synthesis inhibitor CHX for 0, 4, 6, or 8 hours. Statistical significance: ns, P > .05; ∗P < .05; ∗∗P < .01; ∗∗∗∗P < .0001. Ctrl, control; GAPDH, glyceraldehyde-3-phosphate dehydrogenase; PRD, proline-rich domain; REG, regulatory domain; TAD, transactivation domain; TET, tetramerization domain.

To determine the half-life of WT p53 in leukemia cells, we treated human CD34⁺ cells expressing KMT2A-MLLT3 with cycloheximide and collected cells at 0, 0.5, 1, and 2 hours after treatment. We found that the knockdown of PTP4A2 significantly increased p53 half-life in human LICs (Figure 6C-D), suggesting that PTP4A2 plays a crucial role in regulating p53 turnover in LICs.

To understand how PTP4A2 modulates p53 activity in leukemia cells, we performed substrate trapping assays in human CD34⁺ cells expressing KMT2A-MLLT3, MV-4-11, and Molm-13 cells using purified glutathione S-transferase (GST), GST-PTP4A2, and a substrate-trapping mutant GST-PTP4A2-CSDA (C101S/D69A), which permanently binds to its substrate because of the loss of catalyzing activity. Compared with PTP4A2, the PTP4A2-CSDA mutant associates strongly with its substrates.^18,19 We found that, compared with PTP4A2, the PTP4A2-CSDA mutant shows increased association with p53 and PTEN in KMT2A-MLLT3⁺CD34⁺ cells, MV-4-11 and Molm-13 cells (Figure 6E; supplemental Figure 5F). Considering that PTEN is a known PTP4A2 substrate,^20,36 these findings suggest that p53 is a potential PTP4A2 substrate in leukemia cells.

p53 is a sequence-specific transcription factor whose activity is dependent on posttranslational modifications such as phosphorylation (Figure 6F). A key component of p53 activity is its first transactivation domain, which is regulated by phosphorylation.^37,38 To determine whether PTP4A2, as a phosphatase, affects p53 phosphorylation, we examined Ptp4a2^+/+ and Ptp4a2^−/− Kit⁺ BM cells as well as KMT2A-MLLT3⁺ human CD34⁺ cells expressing control or PTP4A2 shRNAs. Western blot was performed to detect p53 phosphorylation at serine 15, 20, 46, and 392 and threonine 81, along with total p53. Although PTP4A2 deficiency did not alter p53 phosphorylation at serine 15, 20, and 46 as well as threonine 81, the loss of PTP4A2 significantly increased p53 phosphorylation at serine 392 in both Ptp4a2^−/− Kit⁺ BM cells (Figure 6G-H) and KMT2A-MLLT3⁺ human CD34⁺ cells expressing PTP4A2 shRNAs (Figure 6I-J).

The phosphorylation of serine 392 stabilizes the tetramer formation of p53 and enhances p53 activity and stability in cancer cells following different stresses.^39,40 To determine whether serine 392 phosphorylation regulates p53 stability in leukemia cells, we generated p53 mutants in which serine 392 was substituted with either alanine (S392A) to prevent phosphorylation or with aspartic acid (S392D) to mimic constitutive phosphorylation. Empty vector control, WT-p53, p53-S392A, or p53-S392D constructs were overexpressed in THP-1 cells, which are known to lack functional p53. Western blot confirmed the expression of total and phosphorylated p53 (Ser392) in the corresponding transduced cells (supplemental Figure 5G). Because overexpression of exogenous WT or mutant p53 saturated the p53 signal, cells were treated with cycloheximide for extended periods (0, 4, 6, and 8 hours) to assess p53 half-life. Notably, compared with those expressing WT-p53 or p53-S392A, cells expressing p53-S392D exhibited a more prolonged p53 half-life (Figure 6K), indicating that phosphorylation at serine 392 enhances p53 stability.

To assess whether PTP4A2 is regulated by p53, we overexpressed WT p53 in p53-deficient mouse BM cells and human AML cell lines (THP-1, HL-60, and U937). The overexpression of WT p53 did not alter PTP4A2 protein levels (supplemental Figure 5H-I), indicating that PTP4A2 is not regulated by p53. All these results demonstrate that PTP4A2 regulates p53 phosphorylation and stability in leukemia cells.

Discussion

Although PTP4A2 is upregulated in human AML cells, its substrates in leukemia cells are largely unknown. We demonstrate that PTP4A2 can downregulate PTEN protein through dephosphorylating PTEN at tyrosine 336.²⁰ Indeed, we observed decreased AKT phosphorylation in Ptp4a2-null HSPCs.^18,19 We also identified PTP4A2 as a key mediator of oncogenic FLT3 and KIT signaling in AML by dephosphorylating E3 ubiquitin ligase CBL at tyrosine 371, thereby promoting leukemogenesis.^18,19 In this study, we discovered that PTP4A2 inhibits the p53 tumor suppressor pathway to maintain LIC fitness. The loss of PTP4A2 increased p53 activity and stability, resulting in upregulated p21 and apoptosis and senescence induction in LICs. The stability and activity of p53 are tightly regulated by various posttranslational modifications such as phosphorylation.^37,38 We found that PTP4A2 directly interacts with p53 and dephosphorylates it at serine 392, thereby decreasing p53 stability and activity in leukemia cells. The phosphorylation of serine 392 enhances p53 tetramerization, activity, and stability in cancer cells^39,40 as well as enhances p53-dependent apoptosis.^41,42 Therefore, our findings reveal a previously unrecognized mechanism by which PTP4A2 negatively regulates p53 signaling in AML.

An important question raised by these findings is whether the observed function of PTP4A2 is restricted to KMT2A-rearranged AML or extends to other AML subtypes. Using a RUNX1–RUNX1T1-9a–driven AML model, we found that the loss of Ptp4a2 similarly impaired colony-forming capacity and increased senescence in LICs. However, in contrast to KMT2A-MLLT3–driven LICs, senescence in the RUNX1–RUNX1T1-9a model was less pronounced and not rescued by p53 haploinsufficiency, indicating a p53-independent mechanism. One possibility is that KMT2A fusion–driven leukemias strongly activate HOX/MEIS gene programs,^43,44 creating higher oncogenic stress that engages p53-dependent checkpoints, making senescence more sensitive to PTP4A2 loss. However, RUNX1-RUNX1T1–driven AML may regulate senescence through alternative pathways.

The stability and activity of p53 are also regulated by acetylation. The acetylation of p53 is not only required for structural changes for DNA binding activity but also for increasing p53 protein stability.^45,46 Defective p53 acetylation has been implicated in leukemia development.⁴⁵ In the future, we will examine the impact of PTP4A2 deficiency on p53 acetylation in LICs and AML cells. We expect that Ptp4a2^−/− LICs will show an increased p53 acetylation compared with that in Ptp4a2^+/+ LICs.

In summary, we have uncovered a novel mechanism by which PTP4A2 counteracts tumor suppressor p53 in LICs to promote leukemogenesis. Our study also identified PTP4A2 as a key regulator of p53 activity in leukemia cells.

Conflict-of-interest disclosure: The authors declare no competing financial interests.