Acute myeloid leukemia (AML)-derived bone mesenchymal stem cell exosomal METTL14 promotes AML cell growth and glycolysis by HOXA3/WNT7B axis

Abstract

Acute myeloid leukemia (AML)-derived bone mesenchymal stem cell (MSC) exosomes have been confirmed to have a positive effect on AML progression. This study aim to reveal the underlying molecular mechanism by which AML-MSC-derived exosomes promotes AML progression. AML-MSC was isolated from the bone marrow aspirates of AML patients. After incubated with AML-MSC, AML cell functions were analyzed. The expression levels of methyltransferase-like 14 (METTL14), homeobox A3 (HOXA3), WNT family member 7B (WNT7B) and glycolysis-related markers were examined. Exosomes were isolated from AML-MSC and then the obtained exosomes were co-cultured with AML cells. AML-MSC co-culturing could enhance AML cell proliferation and glycolysis, while repress cell apoptosis. METTL14 was upregulated in exosomes from AML-MSC, which could be ingested by AML cells. METTL14 could enhance HOXA3 mRNA stability via promoting its m6A modification. Knockdown of exosomal METTL14 from AML-MSC inhibited AML cell growth and glycolysis, while were reversed by HOXA3. In addition, HOXA3 bound to WNT7B promoter to increase its transcription, and WNT7B overexpression also eliminated si-HOXA3-mediated inhibitory on AML cell growth and glycolysis. Animal study revealed that knockdown of exosomal METTL14 from AML-MSC reduced AML tumorigenesis by decreasing HOXA3 and WNT7B expression. AML-MSC-derived exosomal METTL14 facilitated AML cell growth and glycolysis by activating the HOXA3/WNT7B axis, providing a new mechanism for understanding AML-MSC-derived exosomes to promote AML progression.

Supplementary Information

The online version contains supplementary material available at 10.1007/s10565-026-10143-w.

Introduction

Acute myeloid leukemia (AML) is a malignant clonal blood disease originating from hematopoietic stem and progenitor cells, which is characterized by the arrest of cell development leading to the reduction of normal hematopoietic function (Jager et al. 2021; Qi and Jinhai 2024; Wachter and Pikman 2024). The clinical manifestations are bleeding, anemia, fever and infection, seriously endanger the lives of patients (Shimony et al. 2023). Bone mesenchymal stem cells (MSC) have the function of supporting hematopoiesis. Therefore, the transplantation of hematopoietic stem cells combined with MSC may enhance the effect of transplantation in AML patients (Wu et al. 2024). However, AML-derived MSC (AML-MSC) has been found to lose their vigorous self-renewal ability and maintenance effect on normal hematopoietic stem cell function, and instead support the growth and chemoresistance of AML cells (Liao et al. 2023; Lu et al. 2023). In addition, some studies have shown that AML-MSC can protect AML cells from the killing effect of therapies, enabling leukemia cells to survive in the immunosuppressive microenvironment, thereby promoting the development of AML (Forte et al., 2020; Miari and Williams 2024). Therefore, clarifying the molecular mechanism of AML-MSC affecting AML progression may provide new ideas for alleviating AML process.

Exosome is a kind of vesicle secreted by cells into the extracellular space, which contains rich contents (including RNA, proteins and lipids), and participates in the transmission of molecules between cells (Hu et al. 2020). It has been reported that MSC-derived exosomal miRNA can affect AML cell functions (Cheng et al. 2021). Methyltransferase-like 14 (METTL14) is the core component of m6A methyltransferase complex and participates in m6A modification process (Guan et al. 2022; Zhou et al. 2021). Studies had suggested that METTL14 accelerated AML progression and leukemogenesis (Weng et al. 2018; Zhang et al. 2024). Importantly, previous research indicated that AML-MSC-derived exosomal METTL14 could enhance AML cell proliferation and radioresistance (Wang et al. 2025). Therefore, AML-MSC-derived exosomal METTL14 may be a key regulator of AML progression, and more about its role in AML development need to be further revealed.

In this study, exosomes were isolated from AML-MSC to investigate the effect of exosomal METTL14 on AML cell growth and glycolysis. In addition, through database screening and further analysis, this study pointed out the molecular mechanism by which METTL14 increased homeobox A3 (HOXA3) expression through IGF2BP2-dependent m6A modification, and HOXA3 bound to the WNT family member 7B (WNT7B) promoter to enhance its transcription. The proposed METTL14/HOXA3/WNT7B axis may provide a theoretical basis for AML treatment.

Materials and methods

Samples

Bone marrow samples were collected from 34 AML patients and 34 healthy normal controls (normal bone marrow aspiration results) at Tongji Hospital. Each participant provided written informed consent. Our study was approved by the Ethics Committee of Tongji Hospital and was carried out according to the guidelines of Declaration of Helsinki (TJ-IRB20250301).

Isolation of AML-MSC

Bone marrow aspirates from newly diagnosed AML patients were collected and cultured in DMEM/F12 (Gibco, Grand Island, NY, USA) plus 1% penicillin/streptomycin (Gibco) and 10% FBS (Gibco). After removing the suspension cells and cell debris, AML-MSC that adhered to the flask wall was collected and passaged. AML-MSC at passage 3–6 was used in this study.

Cell culture, transfection and co-culturing

Human-derived MSC (HMSC), bone marrow stromal cells (HS-5) and AML cell lines (Kasumi-6, NB4 and HL-60) were purchased from Procell (Wuhan, China) or Biovector NTCC (Beijing, China) and grown at HMSC complete medium (CM-H166, Procell), DMEM (Gibco), RPMI-1640 medium (Gibco) and IMDM (Gibco), respectively. The siRNAs against METTL14/HOXA3/IGF2BP2 (si-METTL14/si-HOXA3/si-IGF2BP2), METTL14/HOXA3/WNT7B overexpression vector (OE-METTL14/OE-HOXA3/OE-WNT7B) and negative controls were transfected into AML-MSC, HMSC, NB4 and HL-60 cells using Lipofectamine 3000 (Invitrogen, Carlsbad, CA, USA). For co-culturing, AML-MSC and HMSC were seeded in the lower of transwell chamber, and AML cell lines were planted in the upper chamber. After 48 h, upper AML cells were collected for functional experiments.

Flow cytometry

For detecting the surface markers of AML-MSC and HMSC, AML-MSC or HMSC (1 × 10⁶/mL) was collected and fixed with 2% polyformaldehyde. Then, cells were incubated with FITC-labeled anti-CD34 (ab18227, Abcam, Cambridge, CA, USA), anti-CD45 (ab27287, Abcam), anti-CD29 (ab21845, Abcam), anti-CD90 (ab226, Abcam), and anti-CD105 (ab11415, Abcam), followed by analyzed using FACScalibur flow cytometer (BD Biosciences, San Jose, CA, USA).

For detecting cell apoptosis, co-cultured AML cells were collected (5 × 10⁵ cells) and incubated with Annexin V-FITC and PI (Sigma-Aldrich, St Louis, MO, USA). Cell apoptosis rate was analyzed by FACScalibur flow cytometer (BD Biosciences).

CCK8 assay

Co-cultured AML cells were re-seeded into 96-well plates (2 × 10³ cells/well) for 48 h. After incubated with CCK8 reagent (Sigma-Aldrich), cell viability was detected under a microplate reader at 450 nm.

EdU assay

Basing on EdU In Vitro Kit (RiboBio, Guangzhou, China), co-cultured AML cells (2 × 10⁵ cells/well) were cultured with EdU reagent and DAPI solution in 6-well plates. Finally, EdU positive cell rate was detected under a flow cytometer.

Measurement of cell glycolysis

Glucose consumption, lactate production, and ATP/ADP ratios in cell lysates (2 × 10⁷ cells/mL) were detected using Glucose Assay Kit (ab65333, Abcam), Lactate Assay Kit (ab65331, Abcam), and the ADP/ATP Ratio Assay kit (MAK135, Sigma-Aldrich) basing on kit instructions.

Western blot (WB)

Total protein was isolated, resolved and transferred onto membranes. Membranes were treated with antibodies, and protein blots were detected by ECL reagent. Antibodies (all from Abcam) including anti-METTL14 (1:1000, ab309096), anti-HOXA3 (1:1000, ab230879), anti-WNT7B (1:5000, ab227607), anti-GLUT1 (1:50,000, ab115730), anti-HK2 (1:10,000, ab227198), anti-IGF2BP2 (1:5000, ab124930), anti-CD63 (1:1000, ab271286), anti-CD81 (1:500, ab79559), anti-CD9 (1:1000, ab236630), anti-Calnexin (1:20,000, ab92573), anti-β-catenin (1:1000, ab16051), anti-β-actin (1:1000, ab272085), Goat anti-rabbit IgG (1:50,000, ab205718) or Goat anti-mouse IgG (1:10,000, ab6789).

m6A quantification assay

Total RNA was extracted from co-cultured AML cells by TRIzol reagent (Invitrogen). Basing on the instructions of m6A RNA methylation Detection Kit (ab185912, Abcam), m6A level (%) was analyzed by a microplate reader.

qRT-PCR

After extracted total RNA from cells, exosomes and bone marrow samples using TRIzol reagent, cDNA was obtained by PrimeScript RT Master Mix (Takara, Tokyo, Japan). PCR was performed by mixing specific primers (Table 1), cDNA and SYBR Green (Takara). Relative mRNA expression was calculated by 2^−ΔΔCt methods.

Table 1.

Primer sequences used for qRT-PCR

Name		Primers for PCR (5’−3’)
METTL14	Forward	GTAGCACAGACGGGGACTTC
	Reverse	GCCAGCCTGGTCGAATTGTA
METTL3	Forward Reverse	TTGTCTCCAACCTTCCGTAGT CCAGATCAGAGAGGTGGTGTAG
FTO	Forward Reverse	ACTTGGCTCCCTTATCTGACC TGTGCAGTGTGAGAAAGGCTT
WTAP	Forward Reverse	CTTCCCAAGAAGGTTCGATTGA TCAGACTCTCTTAGGCCAGTTAC
HOXA3	Forward	TGAGTTCGGGCTTGGGTTTT
	Reverse	GGCGCGGGTGATTTATGAAC
WNT7B	Forward	CGACAGACGGACGGAGGT
	Reverse	ACAGTAGAAGCATGGTGGGC
β-actin	Forward	CTTCGCGGGCGACGAT
	Reverse	CCACATAGGAATCCTTCTGACC

Exosome isolation and identification

The culture medium of HMSC and AML-MSC was collected and centrifuged. After filtrated by a 0.22 μm filter, the supernatant was centrifuged at 100,000 × g for 15 min. The obtained exosomes precipitates were collected and stored in PBS. The morphology of HMSC-exo and AML-MSC-exo was validated by TEM. The exosome biomarkers (CD63, CD81 and CD9) and negative control (Calnexin) in AML-MSC-exo were verified by WB. The size distribution and concentration were analyzed by NTA. The exosomes derived from AML-MSC transfected with si-NC or si-METTL14 were named as si-NC-AML-MSC-exo or si-METTL14-AML-MSC-exo.

Co-cultured system

AML cells were incubated with 10 μg of HMSC-exo or AML-MSC-exo for 24 h, and then the cells were collected to detect METTL14 expression. In addition, AML cells were incubated with si-NC-AML-MSC-exo/si-METTL14-AML-MSC-exo and transfected with OE-HOXA3 for 24 h to measure cell functions.

Exosome uptake assay

The isolated AML-MSC-exo and HMSC-exo was incubated with 4 mg/mL Dil (Invitrogen) for 30 min. Then, HL-60 cells were incubated with Dil-labeled AML-MSC-exo (10 μg) or HMSC-exo (10 μg) for 24 h. After fixed with 4% paraformaldehyde and counterstained with DAPI solution, the uptake of Dil-labeled exosomes was observed under a fluorescence microscope.

MeRIP assay

Total RNA was collected from si-NC/si-METTL14-transfected AML cells. After sheared into fragments, fragmented RNA was incubated with magnetic beads pre-coated with anti-m6A and anti-IgG using Magna MeRIP Kit (Millipore, Billerica, MA, USA). After eluted and quantified, the m6A level of HOXA3 was detected with qRT-PCR.

RIP assay

Transfected AML cells were lysed, and cell lysates were treated by Protein A/G magnetic beads coated with anti-METTL14, anti-IGF2BP2 or anti-IgG at 4 °C overnight with rotation. In immunoprecipitated RNA, HOXA3 enrichment was determined with qRT-PCR.

RNA pull-down assay

Biotin-labeled HOXA3 probe and NC probe (Genepharma, Shanghai, China) were incubated with AML cell lysates for 1 h at 4 °C with rotation. Then, cell lysates were incubated with streptavidin agarose beads (Invitrogen). Protein samples were eluted from the beads to detect METTL14 signal using WB.

Dual-luciferase reporter assay

The wild-type and mutant-type transcript of HOXA3 or WNT7B was inserted into pmirGlo vector to construct HOXA3-wt/mut or WT/MUT-WNT7B. AML cells were co-transfected with si-NC/si-METTL14 and HOXA3-wt/mut vectors, as well as si-NC/si-HOXA3 and WT/MUT-WNT7B vectors. After 48 h, luciferase activity was evaluated to assess interaction.

Actinomycin D (Act D) treatment

The si-NC/si-METTL14/si-IGF2BP2-transfected AML cells were treated with 10 μg/mL Act D (MCE, Monmouth Junction, NJ, USA). At each time points (0, 4 and 8 h), HOXA3 mRNA expression was determined using qRT-PCR.

ChIP assay

AML cells were cross-linked, lysed and sonicated to DNA fragments. Cell fragments were immunoprecipitated with Protein A/G magnetic beads coated with anti-HOXA3 or anti-IgG overnight. The enrichment of WNT7B promoter was then detected by qRT-PCR.

Mice xenograft models

BALB/c nude mice (8-weeks, 20–24 g; SPF Biotechnology Co., Ltd., Beijing, China) were subcutaneously injected with NB4 or HL-60 cells transfected with or without lentiviral OE-HOXA3 vector (n = 5/group). When tumor volume reached about 100 mm³, mice were intratumorally injected with 10 μg exosomes derived from AML-MSC cells transfected with sh-NC or sh-METTL14 (sh-NC-AML-MSC-exo/sh-METTL14-AML-MSC-exo) once every two days. After 24 days, mice were euthanatized by cervical dislocation under 2% isoflurane anesthesia, and tumor samples were collected. Additionally, tumor samples were used to prepare the paraffin sections for HE and immunohistochemical (IHC) analysis. Animal experiments were approved by the Animal Research Committee of Tongji Hospital (TJ-IRB20250301) and performed in compliance with the ARRIVE guidelines and the Basel Declaration.

Statistical analysis

Data are shown as mean ± SD by GraphPad Prism 8.0 software. The differences were analyzed using Student’s t-test or ANOVA. P < 0.05 was considered as statistically significant.

Results

AML-MSC promoted AML cell growth and glycolysis

By detecting the phenotypic markers, our study revealed that AML-MSC and HMSC were negative for CD34 and CD45, while positive for CD29, CD90 and CD105 (Fig. 1A and Supplementary Fig. 1), confirming that the isolated AML-MSC and HMSC was successful. Following, HMSC or AML-MSC was co-cultured with AML cell lines (NB4 and HL-60). Compared to HMSC, AML-MSC enhanced the viability, increased EdU positive cells and repressed apoptosis rate in NB4 and HL-60 cells (Fig. 1B-D). Moreover, AML-MSC promoted glucose consumption, lactate production, ATP/ADP ratios and glycolysis-related proteins (GLUT1 and HK2) in NB4 and HL-60 cells (Fig. 1E-I). Above data showed that AML-MSC could accelerate AML progression.

Fig. 1.

Effect of AML-MSC on AML cell proliferation and glycolysis. A Flow cytometry was used to measure the phenotypic markers of AML-MSC. (B-I) AML cell lines (NB4 and HL-60) were co-cultured with HMSC or AML-MSC (n = 3). Cell proliferation and apoptosis were detected by CCK8 assay (B), EdU assay (C) and flow cytometry (D). E–G Cell glycolysis was assessed by detecting glucose consumption, lactate production and ATP/ADP ratios. (H-I) GLUT1 and HK2 protein levels were examined by WB. **P < 0.01, ***P < 0.001

AML-MSC enhanced m6A methylation and METTL14 expression in AML cells

After co-cultured with AML-MSC, the m6A level in AML cell lines (NB4 and HL-60) was markedly enhanced (Fig. 2A). Further analysis showed that AML-MSC could significantly promote the expression of 4 methylases, especially METTL14 (Fig. 2B). Therefore, METTL14 was selected in this study. After exosomes were isolated from HMSC and AML-MSC, TEM was used to observe exosome micromorphology (Fig. 2C). In addition, the detection of exosome marker proteins (CD63, CD81 and CD9) confirmed that AML-MSC-exo extraction was successful (Fig. 2D). In addition, NTA revealed the size and concentration of AML-MSC-derived exosomes (Supplementary Fig. 2). Further detection revealed that METTL14 expression was higher in AML-MSC-exo than that in HMSC-exo (Fig. 2E). After co-incubation of Dil-labeled AML-MSC-exo with AML cells, AML cells successfully took up exosomes as observed by fluorescence microscopy (Fig. 2F). Also, WB results confirmed that METTL14 protein expression was markedly enhanced in AML cells after co-cultured with AML-MSC-exo (Fig. 2G). Besides, METTL14 protein level was reduced in AML cells co-cultured with the exosomes from AML-MSC treated with GW4869 (exosome secretion inhibitor) (Fig. 2H), suggesting that METTL14 was derived from the exosomes of AML-MSC. METTL14 mRNA was upregulated in AML patients (Fig. 2I), and its protein level was elevated in 3 AML cell lines (Kasumi-6, NB-4 and HL-60) (Fig. 2J).

Fig. 2.

METTL14 expression in AML-MSC-exo and AML patients. A m6A level was examined by m6A RNA methylation Detection Kit in NB4 and HL-60 cells co-cultured with HMSC/AML-MSC (n = 3). B The mRNA expression of METTL3, METTL14, FTO and WTAP in HMSC/AML-MSC was detected by qRT-PCR (n = 3). C The microstructure of exosomes from HMSC/AML-MSC was observed under TEM. D Exosome marker proteins (CD63, CD81 and CD9) in AML-MSC-exo were measured by WB. E METTL14 mRNA expression in HMSC-exo and AML-MSC-exo was examined by qRT-PCR (n = 3). F Fluorescence microscope was used to observe Dil-labeled Exo in HL-60 cells co-cultured with AML-MSC-exo or HMSC-exo (n = 3). G METTL14 protein expression was determined using WB in NB4 and HL-60 cells co-cultured with HMSC-exo/AML-MSC-exo (n = 3). H METTL14 protein level was tested by WB in NB4 and HL-60 cells co-cultured with exosomes from AML-MSC treated with DMSO/GW4896 (n = 3). I METTL14 mRNA expression was assessed by qRT-PCR in the bone marrow samples of AML patients (n = 34) and normal controls (n = 34). J WB was used to detect METTL14 protein expression in HS-5 cells and 3 AML cell lines (n = 3). *P < 0.05, **P < 0.01, ***P < 0.001

METTL14 promoted HOXA3 mRNA stability by m6A modification

GEO database (GSE97443) found that silencing of METTL14 in NB4 cells could significantly inhibit the expression of WNT7B and HOXA3 (Fig. 3A). Following, si-METTL14 was constructed to reduce METTL14 protein expression in NB4 and HL-60 cells (Fig. 3B). The detection of HOXA3 expression suggested that METTL14 knockdown markedly decreased HOXA3 protein level (Fig. 3C). Multiple websites (RMbase V3.0, RMvar, RMDisease V2.0) predicted that there were m6A modification sites in HOXA3 (Fig. 3D), and the m6A sites of HOXA3 predicted by SRAMP database were shown in Fig. 3E. HOXA3 had increased expression in the bone marrow samples of AML patients (Fig. 3F), and its level was correlated with METTL14 positively (Fig. 3G). MeRIP assay suggested that METTL14 knockdown markedly reduced the m6A level of HOXA3 (Fig. 3H-I), and RIP assay indicated that HOXA3 enrichment could be increased by METTL14 antibody (Fig. 3J). Besides, RNA–Protein Interaction Prediction database predicts that METTL14 can interact with the mRNA of HOXA3 (Fig. 3K). Through RNA pull-down assay, METTL14 could be pulled down by HOXA3 probe (Fig. 3L). Moreover, METTL14 silencing markedly inhibited the luciferase activity of HOXA3-wt vector (Fig. 3M-N). To further evaluate whether METTL14 methylated HOXA3 mRNA at the m6A site, site-directed mutagenesis experiments were carried put. In HL-60 cells co-transfected with HOXA3-wt/mut and si-NC/si-METTL14, METTL14 knockdown markedly decreased HOXA3 protein level in cells transfected with HOXA3-wt vector, but had no significant effect on that in cells transfected with HOXA3-mut vector (Supplementary Fig. 3). Additionally, downregulation of METTL14 could repress the mRNA stability of HOXA3 (Fig. 3O-P). All data showed that METTL14 enhanced HOXA3 expression by m6A modification.

Fig. 3.

METTL14 regulated HOXA3 through m6A modification. A GSE97443 database analyzed the differentially expressed genes in NB4 cells transfected with sh-NC/sh-METTL14. B-C METTL14 and HOXA3 protein levels were detected by WB in NB4 and HL60 cells transfected with si-NC/si-METTL14 (n = 3). D RMbase V3.0, RMvar, RMDisease V2.0 websites predicted the m6A modification sites in HOXA3. E The m6A sites of HOXA3 was predicted by SRAMP database. F HOXA3 mRNA expression was detected by qRT-PCR in the bone marrow samples of AML patients (n = 34) and normal controls (n = 34). G Pearson correlation analysis was used to measure the correlation between HOXA3 and METTL14 expression in AML patients. MeRIP assay (H-I) and RIP assay (J) were used to assess the regulation of METTL14 on HOXA3 (n = 3). K RNA–Protein Interaction Prediction database predicted the interaction between METTL14 and HOXA3 mRNA. RNA pull-down assay (L) and dual-luciferase reporter assay (M–N) were performed to confirm the interaction between METTL14 and HOXA3 (n = 3). O-P Act D treatment assay was used to evaluate the regulation of si-METTL14 on the mRNA stability of HOXA3 (n = 3). **P < 0.01, ***P < 0.001

IGF2BP2 recognized METTL14-mediated m6A modification of HOXA3

ENCORI database predicted that m6A reader IGF2BP2 might interact with HOXA3 (Fig. 4A). Moreover, ENCORI and GEPIA database showed that IGF2BP2 expression was positively correlated with HOXA3 in AML patients (Fig. 4B-C). RIP assay indicated that HOXA3 enrichment was markedly increased by IGF2BP2 antibody, and METTL14 knockdown markedly inhibited the interaction between IGF2BP2 and HOXA3 (Fig. 4D-E). After knockdown of IGF2BP2 protein expression by si-IGF2BP2 transfection in NB4 and HL-60 cells (Fig. 4F), the detection of HOXA3 expression confirmed that IGF2BP2 silencing could repress HOXA3 mRNA and protein levels (Fig. 4G-H). Also, IGF2BP2 knockdown significantly decreased HOXA3 mRNA stability (Fig. 4I-J). Above data verified that IGF2BP2 could recognize the m6A modification of HOXA3.

Fig. 4.

The regulation of IGF2BP2 on HOXA3 expression. A ENCORI database predicted the interaction between IGF2BP2 and HOXA3. ENCORI database (B) and GEPIA database (C) analyzed the correlation between IGF2BP2 and HOXA3 expression in AML patients. D-E RIP assay was used to analyze the regulation of si-METTL14 on the interaction between IGF2BP2 and HOXA3 (n = 3). F WB was used to confirm the transfection efficiency of si-IGF2BP2 (n = 3). G-H HOXA3 mRNA and protein levels were detected by qRT-PCR and WB in NB4 and HL-60 cells transfected with si-NC/si-IGF2BP2 (n = 3). I-J The regulation of si-IGF2BP2 on the mRNA stability of HOXA3 was assessed by Act D treatment assay (n = 3). *P < 0.05, **P < 0.01, ***P < 0.001

HOXA3 overexpression reversed the effect of exosomal METTL14 knockdown

To confirm the specific role of exosomal METTL14, the effect of HMSC-derived exosomes with METTL14 overexpression on AML cell progression was explored. The results showed that HMSC-derived METTL14 overexpression enhanced the viability, EdU positive cell rate, glucose consumption, lactate production, and ATP/ADP ratios in NB4 and HL-60 cells (Supplementary Fig. 4A-E). To explore whether exosomal METTL14 regulated HOXA3 expression to mediate AML cell progression, the recue experiments were carried out. After transfected with si-METTL14 into AML-MSC, METTL14 protein expression was markedly reduced (Fig. 5A). Following, NB4 and HL-60 cells were transfected with OE-HOXA3 and co-cultured with the exosomes from AML-MSC transfected with si-NC/si-METTL14. As shown in Fig. 5B, OE-HOXA3 eliminated exosomal si-METTL14-mediated the decreasing on HOXA3 protein level. The si-METTL14-AML-MSC-exo treatment significantly inhibited AML cell viability, decreased EdU positive cell rate, and enhanced apoptosis rate, which were abolished by OE-HOXA3 (Fig. 5C-F). Also, HOXA3 upregulation eliminated the suppressing effect of si-METTL14-AML-MSC-exo on glucose consumption, lactate production, ATP/ADP ratios and glycolysis-related proteins (GLUT1 and HK2) in NB4 and HL-60 cells (Fig. 5G-K). Thus, these data suggested that AML-MSC-derived exosomal METTL14 might promote AML cell progression by enhancing HOXA3 expression.

Fig. 5.

Effects of exosomal si-METTL14 and OE-HOXA3 on AML cell progression. A METTL14 protein expression was detected by WB in AML-MSC transfected with si-NC/si-METTL14 (n = 3). B-K NB4 and HL-60 cells were transfected with OE-HOXA3 and co-cultured with the exosomes from AML-MSC transfected with si-NC/si-METTL14 (n = 3). B HOXA3 protein expression was determined using WB. CCK8 assay (C), EdU assay (D-E) and flow cytometry (F) were used to detect cell proliferation and apoptosis. G-I Glucose consumption, lactate production and ATP/ADP ratios were detected to assess cell glycolysis. J-K WB was used to test GLUT1 and HK2 protein levels. *P < 0.05, **P < 0.01, ***P < 0.001

HOXA3 activated the transcription of WNT7B

Consistent with the results of GSE97443 database, silencing of METTL14 could decrease WNT7B protein expression in NB4 and HL-60 cells (Fig. 6A). Furthermore, the underlying molecular mechanisms were analyzed. LinkedOmics database and GEPIA database predict a significant positive association between HOXA3 and WNT7B expression (Fig. 6B-C). qRT-PCR detected that WNT7B was upregulated in AML patients (Fig. 6D), and there had positive correlation between WNT7B and HOXA3 (Fig. 6E). Jaspar website discovered that HOXA3 had binding sites in WNT7B promoter regions (Fig. 6F). Following, si-HOXA3 was transfected into NB4 and HL-60 cells to decrease HOXA3 protein expression (Fig. 6G). HOXA3 knockdown reduced the luciferase activity of WT-WNT7B vector (Fig. 6H-I). Moreover, the enrichment of WNT7B promoter could be enhanced by HOXA3 antibody (Fig. 6J). Also, HOXA3 silencing markedly decreased the mRNA and protein expression of WNT7B in AML cells (Fig. 6K-L).

Fig. 6.

The regulation of HOXA3 on the transcription of WNT7B. A WNT7B protein expression was detected by WB in NB4 and HL-60 cells transfected with si-NC/si-METTL14 (n = 3). LinkedOmics database (B) and GEPIA database (C) predicted the association between HOXA3 and WNT7B expression. D WNT7B mRNA expression in the bone marrow samples of AML patients (n = 34) and normal controls (n = 34) was detected by qRT-PCR. (E) The correlation between WNT7B and HOXA3 expression in AML patients was analyzed by Pearson correlation analysis. F Jaspar website predicted the binding sites of HOXA3 in WNT7B promoter regions. G The transfection efficiency of si-HOXA3 was confirmed by WB (n = 3). Dual-luciferase reporter assay H-I and ChIP assay J were used to confirm the interaction between HOXA3 and WNT7B promoter (n = 3). K-L WNT7B mRNA and protein levels were examined by qRT-PCR and WB in NB4 and HL-60 cells transfected with si-NC/si-HOXA3 (n = 3). **P < 0.01, ***P < 0.001

Overexpressed WNT7B abolished the regulation of si-HOXA3

To investigate whether HOXA3 regulated AML cell progression by mediating WNT7B, the following experiments were carried out. OE-WNT7B was transfected into NB4 and HL-60 cells to increase WNT7B protein level (Fig. 7A). HOXA3 knockdown reduced cell viability, EdU positive cell rate and increased apoptosis rate in AML cells, while OE-WNT7B abolished above effects (Fig. 7B-E). Furthermore, the inhibitory effect of si-HOXA3 on glucose consumption, lactate production, ATP/ADP ratios and glycolysis-related proteins (GLUT1 and HK2) were reversed by OE-WNT7B (Fig. 7F-J). It has been reported that WNT7B regulates GLUT1 expression to mediate cancer progression through β-catenin (Jiang et al. 2025; Zhang et al. 2025). Our results showed that si-HOXA3 could reduce β-catenin protein level, and this effect could be reversed by OE-WNT7B (Supplementary Fig. 5). Therefore, HOXA3 facilitated AML cell growth and glycolysis by upregulating WNT7B/β-catenin/GLUT1 pathway.

Fig. 7.

Effects of si-HOXA3 and OE-WNT7B on AML cell progression. A WNT7B protein expression was detected by WB in NB4 and HL-60 cells transfected with OE-NC/OE-WNT7B (n = 3). B-J NB4 and HL-60 cells were transfected with si-NC, si-HOXA3 and OE-WNT7B (n = 3). CCK8 assay (B), EdU assay (C-D) and flow cytometry (E) were performed to assess cell proliferation and apoptosis. F–H Glucose consumption, lactate production and ATP/ADP ratios were measured to evaluate cell glycolysis. (I-J) The protein levels of GLUT1 and HK2 were detected using WB. *P < 0.05, **P < 0.01, ***P < 0.001

Knockdown of exosomal METTL14 from AML-MSC reduced AML tumor growth

To further confirm the role of exosomal METTL14 form AML-MSC on AML progression, the animal experiments were carried out. The injection of NB4 and HL-60 cells with sh-METTL14-AML-MSC-exo markedly reduced AML tumor volume and weight, while this effect could be abolished by HOXA3 overexpression (Fig. 8A-B and F-G). Moreover, HOXA3 and WNT7B protein levels were decreased in sh-METTL14-AML-MSC-exo group, which was reversed by HOXA3 overexpression (Fig. 8C-D and H-I). In addition, cells were arranged disorderly and varied in size and shape, as well as Ki67 and HOXA3 positive cells were reduced in sh-METTL14-AML-MSC-exo group, while HOXA3 overexpression eliminated this effect (Fig. 8E and J).

Fig. 8.

Effects of exosomal METTL14 and HOXA3 on AML tumor growth. Nude mice were injected with NB4 or HL-60 cells transfected with lentivirus OE-HOXA3 and injected with sh-NC/sh-METTL14-AML-MSC-exo (n = 5). Tumor volume (A and F) and weight (B and G) were measured. C-D and H-I HOXA3 and WNT7B protein levels were examined by WB. E and J HE and IHC staining was performed. **P < 0.01, ***P < 0.001

Discussion

Many studies have confirmed that MSC promotes AML cell growth and therapy resistance (Miari and Williams 2024). Besides, Forte et al. revealed that bone MSC support AML cell chemoresistance by enhancing antioxidant defense (Forte et al., 2020). Importantly, it had been reported that AML-derived MSC could promote chemoresistance and epithelial-mesenchymal transition in AML cells (Lu et al. 2023), which supported the conclusion that AML-MSC contributed to the malignant progression of AML. This study investigated the effects of AML-MSC on AML cell proliferation, apoptosis and glycolysis. The results showed that after co-cultured with AML-MSC, AML cell proliferation and glycolysis abilities were enhanced, while apoptosis ability was inhibited, suggesting that AML-MSC accelerated AML cell growth and glycolysis.

Many previous studies have shown that MSC-derived exosomal RNA is involved in the regulation of AML progression. For example, MSC-derived exosomal miR-34c-5p promoted eradication of leukemia stem cells to impede AML development (Wen et al. 2023). Therefore, the AML-MSC-derived exosomal components were investigated in our study. This study found that AML-MSC co-incubation could increase the m6A level in AML cells, and AML-MSC had the strongest promoting effect on METTL14 expression by detecting 4 m6A-related enzymes. Furthermore, exosomes derived from AML-MSC were isolated, and our study verified that METTL14 was transferred from AML-MSC to AML cells through exosomes. Previous study has suggested that METTL14 may serve as an oncogene in AML progression (Weng et al. 2018; Zhang et al. 2024). Through loss-of-functional experiments, this study showed that knockdown of exosomal METTL14 from AML-MSC repressed AML cell growth, glycolysis and reduced tumorigenesis. Consistent with the findings of Wang et al. (Wang et al. 2025), our study also confirmed the positive role of exosomal METTL14 derived from AML-MSC in AML progression, verifying that AML-MSC accelerated AML malignant behavior via shuttling exosomal METTL14.

HOXA3 belongs to the HOX gene family and plays a vital role in embryonic development, which main function is to regulate the expression of other genes as a transcription factor (Gordon 2018; Yang et al. 2023). Studies have revealed that HOXA3 plays oncogenic role in many tumors, including gastric cancer (Lu et al. 2020) and glioblastoma (Yang et al. 2023). In AML, HOXA3 has been confirmed to be overexpressed in AML patients (Xuan et al. 2025), and can serve as a potential leukemia stem cell marker (Mohd Amin et al. 2023). Besides, upregulation of HOXA3 promoted chemotherapy resistance in AML, and its level was correlated with chemotherapy response in NPM1 mutation-negative leukemia specimens (Allen et al. 2024). In this study, the GEO database found that silencing of METTL14 in NB4 cells could inhibit HOXA3 expression, and multiple databases analysis revealed that HOXA3 existed m6A methylation modification sites. Through experimental verification, our results determined that METTL14 promoted HOXA3 mRNA stability and protein expression through m6A modification, and this process could be recognized by IGF2BP2. Further assay showed that HOXA3 knockdown inhibited AML cell growth and glycolysis, and its upregulation eliminated the repressing effect of si-METTL14-AML-MSC-exo on AML cell functions. These data support the conclusion that exosomal METT14 from AML-MSC promoted HOXA3 expression through IGF2BP2-dependent m6A modification, thereby accelerating AML malignant progression.

WNT7B belongs to the Wnt family and participates in the signal transmission process between cells mainly through its encoded proteins (Li et al. 2024). Abnormal expression or dysfunction of WNT7B is associated with a variety of diseases. For example, WNT7B upregulation promoted pancreatic cancer cell stemness and chemoresistance (Zhang et al. 2021). Also, WNT7B knockdown could inhibit colorectal cancer cell metastasis and proliferation (Chen et al. 2023). Teng et al. revealed that UNC5B repressed AML cell bone marrow adhesion and proliferation by downregulating WNT7B (Teng et al. 2024), suggesting that WNT7B might contribute to AML malignant behavior. In this study, database analysis showed a positively correlation between HOXA3 and WNT7B expression, and there was a binding site between HOXA3 and WNT7B promoter. Further analysis revealed that HOXA3 enhanced the transcription of WNT7B via binding to its promoter region. The reversal effect of WNT7B on si-HOXA3-mediated inhibition on AML cell growth and glycolysis confirmed that HOXA3 facilitated AML cell process by upregulating WNT7B. More important, the present study confirmed that METTL14 knockdown inhibited WNT7B expression in vitro, and interfering of exosomal METTL14 from AML-MSC reduced WNT7B levels in vivo, which refined our speculation on the METTL14/HOXA3/WNT7B axis.

At present, several limitations of this study should be acknowledged. First, the experiments were primarily conducted using AML cell lines (NB4 and HL-60) and xenograft mouse models, which may not fully recapitulate the complexity of patient-specific heterogeneity. Future studies should validate these mechanisms in primary AML patient samples and patient-derived xenograft models to enhance clinical relevance. Second, although the ChIP experiments demonstrate HOXA3 binding to the WNT7B promoter, but the possibility that HOXA3 indirectly regulates WNT7B expression through other intermediate molecules cannot be excluded. In the further, RNA-seq can be performed to assess genome-wide transcriptional changes following HOXA3 knockdown to confirm whether WNT7B is a direct transcriptional target of HOXA3.

In conclusion, this study points to a potential molecular mechanism by which AML-MSC-derived exosomes contribute to the malignant progression of AML. This study indicated that exosomal METTL14 from AML-MSC promoted AML cell growth and glycolysis through mediating the transcriptional activation of WNT7B by promoting the m6A modification of HOXA3 (Fig. 9). Therefore, targeted inhibition of exosomal METTL14 expression in AML-MSC may be a promising approach to alleviate AML progression. However, translating this concept into clinical practice requires addressing several operational challenges: 1. Specificity of targeting: Strategies to selectively inhibit METTL14 in AML-MSC-derived exosomes without affecting normal hematopoietic support functions of MSCs are crucial. Potential approaches include developing MSC-specific nanocarriers for siRNA delivery or small-molecule inhibitors that disrupt METTL14 packaging into exosomes. 2. Off-target effects: METTL14 is involved in global m6A methylation; systemic inhibition may disrupt normal RNA metabolism in healthy tissues. Therefore, localized delivery or transient inhibition strategies should be explored to minimize adverse effects. 3. Synergy with existing therapies: Combining exosomal METTL14 inhibition with conventional chemotherapy or targeted agents may enhance therapeutic efficacy and overcome chemoresistance. 4. Biomarker potential: Detecting exosomal METTL14 levels in patient bone marrow or peripheral blood could serve as a non-invasive biomarker for AML progression or treatment response. Clinical validation in longitudinal cohorts is needed to assess its prognostic utility. Addressing the above challenges will be critical for developing exosome-targeted therapies that improve outcomes in AML patients.

Fig. 9.

Summary of the mechanisms of this study. Exosomal METTL14 from AML-MSC promoted AML cell proliferation, glycolysis and inhibited apoptosis by HOXA3/WNT7B pathway

Supplementary Information

Below is the link to the electronic supplementary material.

Author contributions

W.L. and X.M. performed the experiments and drafted the manuscript. J.W. and S.Y. collected and contributed the methodology. S.S. analyzed the data and prepared figures. R.Z. and Y.W. operated software and edited the manuscript. Y.X. designed and supervised the study. All authors reviewed the manuscript.

Funding

None.

Data availability

The data are available from the corresponding author upon request.

Declarations

Ethics approval

Our study was approved by the Ethics Committee of Tongji Hospital and was carried out according to the guidelines of Declaration of Helsinki (TJ-IRB20250301).

Animal experiments were approved by the Animal Research Committee of Tongji Hospital (TJ-IRB20250301). Animal studies were performed in compliance with the ARRIVE guidelines and the Basel Declaration.

Conflict of interests

The authors declare no competing interests.