Metformin Slows Expansion of Mutant Blood Stem Cells Linked to Cancer Risk
A landmark Nature study shows metformin can suppress the clonal advantage of DNMT3A-mutant blood stem cells, potentially preventing leukemia progression.
Summary
Clonal haematopoiesis occurs when mutant blood stem cells outcompete normal ones, raising risks of blood cancers and inflammatory disease. The most common driver mutation, DNMT3A R882, was studied in a mouse model carrying the equivalent Dnmt3a R878H mutation. Researchers found these mutant stem cells rely on elevated mitochondrial respiration (OXPHOS) for their competitive edge. Metformin, a diabetes drug that inhibits mitochondrial complex I, reduced this advantage both in vitro and in vivo over seven months. Multi-omics analysis revealed metformin restores methylation potential and reverses aberrant DNA and histone methylation patterns in mutant cells. The effect was also demonstrated in human DNMT3A R882H stem cells generated by prime editing, providing strong preclinical support for investigating metformin as a preventive therapy against clonal haematopoiesis.
Detailed Summary
Clonal haematopoiesis is an age-related condition where a single mutant blood stem cell expands to dominate the bone marrow, increasing risk of haematologic malignancies and cardiovascular or inflammatory disease. Mutations in DNMT3A—particularly at arginine 882 (R882)—are the most frequent drivers, found in 50–60% of clonal haematopoiesis carriers. R882 mutations are heterozygous missense changes that reduce DNMT3A methyltransferase activity and dominantly suppress the wild-type allele, causing widespread CpG hypomethylation. Despite this prevalence, no pharmacologic intervention had been identified to suppress mutant clone expansion.
The research team used a Dnmt3a R878H/+ knock-in mouse model—the murine equivalent of human DNMT3A R882H—to characterize metabolic differences between mutant and wild-type haematopoietic stem and progenitor cells (HSPCs). Through extracellular flux analysis, they demonstrated that mutant lineage-negative, KIT-positive (LK) cells exhibit higher basal and maximal oxygen consumption rates, elevated mitochondrial ROS, and increased mitochondrial membrane potential relative to mass—collectively indicating upregulated oxidative phosphorylation (OXPHOS). This metabolic phenotype was validated by reanalysis of published RNA-seq datasets from primary AML samples and single-cell RNA-seq data from human clonal haematopoiesis, both showing elevated OXPHOS gene expression specifically in DNMT3A R882-mutated cells.
To test whether this metabolic reprogramming is required for competitive advantage, the team performed in vitro competition assays and used shRNA-mediated knockdown of ETC complex I (Ndufv1) and complex IV (Cox15) subunits, which reduced both oxygen consumption and the competitive edge of mutant cells. Metformin at a clinically relevant concentration (50 µM) phenocopied this effect, and rescue with NDI1—a metformin-resistant yeast complex I analogue—confirmed on-target complex I inhibition as the mechanism. In a competitive bone marrow transplantation model, metformin administered in drinking water (5 mg/mL) for seven months significantly reduced the expansion of Dnmt3a R878H/+ donor cells in peripheral blood, bone marrow HSCs, and myeloid progenitor compartments, with effects in both myeloid and lymphoid lineages.
To dissect the mechanism beyond metabolic suppression, the team performed multi-omics profiling—including single-cell RNA-seq, metabolomics, reduced representation bisulfite sequencing (RRBS), CUT&RUN for H3K27me3, and ATAC-seq. They found that Dnmt3a R878H/+ HSPCs have an elevated SAM-to-SAH ratio (enhanced methylation potential) following metformin treatment, which was associated with reversal of aberrant CpG hypomethylation at specific differentially methylated regions and normalization of H3K27 trimethylation profiles. These epigenetic corrections were linked to reduced chromatin accessibility at loci associated with HSPC self-renewal, suggesting metformin recalibrates the epigenetic landscape of mutant cells.
Critically, the team extended findings to human biology by using prime editing to introduce the DNMT3A R882H mutation into human CD34+ HSPCs. These edited cells also showed a competitive advantage in xenograft assays that was significantly reduced by metformin treatment, closely mirroring the mouse data. Taken together, these results establish that DNMT3A R882 mutant HSPCs are metabolically and epigenetically distinct from wild-type counterparts and are selectively vulnerable to metformin. The study provides compelling preclinical rationale for clinical trials evaluating metformin as a preventive strategy against DNMT3A R882-driven clonal haematopoiesis and its downstream consequences.
Key Findings
- DNMT3A R882-mutant HSPCs exhibit elevated mitochondrial OXPHOS, which is required for their competitive expansion over wild-type cells.
- Metformin at clinically relevant doses suppressed the competitive advantage of Dnmt3a R878H/+ HSCs in vivo over 7 months in mice.
- Metformin increased SAM/SAH methylation potential and reversed aberrant CpG hypomethylation and H3K27me3 profiles in mutant HSPCs.
- Prime-edited human DNMT3A R882H HSPCs recapitulated the competitive advantage, which was also reduced by metformin treatment.
- NDI1 rescue confirmed metformin's effect is specifically mediated through mitochondrial complex I inhibition, not off-target toxicity.
Methodology
The study used a Dnmt3a R878H/+ knock-in mouse model with in vitro and in vivo competitive repopulation assays, shRNA knockdowns, and an NDI1 rescue experiment to establish mechanistic causality. Multi-omics profiling—including scRNA-seq with HSPC antibody barcoding, RRBS, CUT&RUN, ATAC-seq, and metabolomics—was applied to bone marrow cells from treated and untreated transplant recipients. Human relevance was validated using prime-edited DNMT3A R882H CD34+ HSPCs in xenograft competition assays.
Study Limitations
All mouse experiments used a transplantation model rather than endogenous clonal haematopoiesis, which may not fully replicate the natural bone marrow niche environment or aging context. Human data relied on prime-edited cells rather than patient-derived samples with naturally occurring mutations, and long-term clinical safety and efficacy in human clonal haematopoiesis remain untested. The study focused exclusively on the R882 hotspot; whether findings extend to non-R882 DNMT3A mutations or other clonal haematopoiesis drivers (e.g., TET2, ASXL1) is unknown.
Enjoyed this summary?
Get the latest longevity research delivered to your inbox every week.