Human FOXP3⁺ regulatory T (T_reg) cells are central to immune tolerance. However, their heterogeneity and differentiation remain incompletely understood. Here we use single-cell RNA and T cell receptor sequencing to resolve T_reg cells from healthy individuals and patients with or without acute graft-versus-host disease (aGVHD) who undergo stem cell transplantation. These analyses, combined with functional assays, separate T_reg cells into naïve, activated, and effector stages, and resolve the HLA-DR^hi, LIMS1^hi, highly suppressive FOXP3^hi, and highly proliferative MKI67^hi effector subsets. Trajectory analysis assembles T_reg subsets into two differentiation paths (I/II) with distinctive phenotypic and functional programs, ending with the FOXP3^hi and MKI67^hi subsets, respectively. Transcription factors FOXP3 and SUB1 contribute to some Path I and Path II phenotypes, respectively. These FOXP3^hi and MKI67^hi subsets and two differentiation pathways are conserved in transplanted patients, despite having functional and migratory impairments under aGVHD. These findings expand the understanding of T_reg cell heterogeneity and differentiation and provide a single-cell atlas for the dissection of T_reg complexity in health and disease.

Interaction between stromal cells and acute myeloid leukemia (AML) cells in bone marrow (BM) is known to contribute importantly to chemoresistance and disease recurrence. Therefore, disruption of a crosstalk between AML cells and BM microenvironment may offer a promising therapeutic strategy for AML treatment. Here, we demonstrate that in a niche-like co-culture system, AML cells took up functional mitochondria from bone marrow stromal cells (BMSCs) and inhibition of such mitochondrial transfer by metformin, the most commonly prescribed drug for type 2 diabetes mellitus, significantly enhanced the chemosensitivity of AML cells co-cultured with BMSCs. The chemo-sensitizing effect of metformin was acted through reducing the mitochondrial transfer and mitochondrial oxidative phosphorylation (OXPHOS) in the recipient AML cells. In addition, metformin potentiated the antitumor efficacy of cytarabine (Ara-C) in vivo in an NCG immunodeficient mouse xenograft model by inhibiting the mitochondrial transfer and OXPHOS activity in the engrafted human AML cells. Altogether, this study identifies a potential application of metformin in sensitizing AML cells to chemotherapy and unveils a novel mechanism by which metformin executes such effect via blocking the mitochondrial transfer from stromal cells to AML cells.

Background: Bone marrow stromal cells (BMSCs) are known to promote chemoresistance in acute myeloid leukemia (AML) cells. However, the molecular basis for BMSC-associated AML chemoresistance remains largely unexplored. Methods: The mitochondrial oxidative phosphorylation (OXPHOS) levels of AML cells were measured by a Seahorse XFe24 cell metabolic analyzer. The activity of total or mitochondrial signal transducer and transcription activator 3 (STAT3) in AML cells was explored by flow cytometry and Western blotting. Real-time quantitative PCR, Western blotting and enzyme-linked immunosorbent assay (ELISA) were used to analyze expression of interleukin 6 (IL-6) in the human BMSC line HS-5, and IL-6 was knocked out in HS-5 cells by CRISPR/Cas9 system. Results: In this study, we observed that co-culturing with BMSCs heightened OXPHOS levels in AML cells, thus promoting chemoresistance in these cells. HS-5 cell-induced upregulation of OXPHOS is dependent on the activation of STAT3, especially on that of mitochondrial serine phosphorylated STAT3 (pS-STAT3) in AML cells. The relationship among pS-STAT3, OXPHOS, and chemosensitivity of AML cells induced by BMSCs was demonstrated by the STAT3 activator and inhibitor, which upregulated and downregulated the levels of mitochondrial pS-STAT3 and OXPHOS, respectively. Intriguingly, AML cells remodeled HS-5 cells to secrete more IL-6, which augmented mitochondrial OXPHOS in AML cells and stimulated their chemoresistance. IL-6 knockout in HS-5 cells impaired the ability of these cells to activate STAT3, to increase OXPHOS, or to promote chemoresistance in AML cells. Conclusions: BMSCs promoted chemoresistance in AML cells via the activation of the IL-6/STAT3/OXPHOS pathway. These findings exhibit a novel mechanism of chemoresistance in AML cells in the bone marrow microenvironment from a metabolic perspective.

Adipogenesis of bone marrow mesenchymal stem cells (MSCs) promotes chemoresistance of acute myeloid leukaemia (AML) cells. MSCs from AML patients (AML‐MSCs) display enhanced adipogenesis compared with bone marrow MSCs from healthy donors. However, the precise molecular mechanism by which adipogenesis of MSCs from AML marrow differs from normal counterparts remains obscure. We found that METTL3 significantly inhibits MSC adipogenesis. Here, we aimed to identify the molecular mechanism linking METTL3 and MSC adipogenesis. Analysis of m ⁶ A epigenetic changes in MSCs determined via RIP‐qPCR and MeRIP‐qPCR indicated that METTL3 affects AKT protein expression in MSCs by mediating m ⁶ A modification of AKT1‐mRNA . Downregulated METTL3 expression in AML‐MSCs induced an increase in AKT protein, resulting in enhanced MSC adipogenesis, thereby contributing to chemoresistance in AML cells. Therefore, targeting AKT regulation by mRNA modification in MSC adipogenesis might provide a novel therapeutic strategy to overcome AML chemoresistance.

The anti‐CD44 monoclonal antibody (mAb) A3D8 induces differentiation or apoptosis in vitro in various subtypes of acute myeloid leukemia (AML) via p27^Kip1 upregulation. Bone marrow (BM) stromal cells play a vital role in the development of chemoresistance in AML cells attached to the stroma. To investigate the effect of BM stroma adhesion induced AML resistance to A3D8, we developed a co‐culture system composed of an AML‐derived cell line (NB4) cultured with either a human BM stroma cell line (HS‐5) or mesenchymal stem cells (MSCs). We found that NB4 cells adhered to HS‐5 cells or MSCs developed resistance against the anti‐proliferative effects of A3D8, and this action is caused by the activation of PI3K/Akt signaling following p27^Kip1 down‐regulation and cytoplasmic re‐localization. The stromal co‐culture‐induced resistance can be partially abolished by inhibiting the PI3K/Akt signaling pathway. Such findings were confirmed in two additional AML‐derived cell lines as well as in primary AML cells. Our results suggest that BM stroma can induce A3D8 resistance in part via the PI3K/Akt–p27^Kip1 axis, and blocking PI3K/Akt pathway maybe necessary for anti‐CD44 treatment on AML in BM microenvironment. © 2014 The Authors. Molecular Carcinogenesis published by Wiley Periodicals, Inc.