Background The mechanisms by which diabetes increases atherosclerosis and cardiovascular disease risk even after glucose normalisation remains unknown. We hypothesised that:

1. hyperglycaemia alters cellular metabolism;

2. these changes drive pro–inflammatory responses, which remain altered after glucose normalisation and

3. these increase atherosclerosis in vivo.

Methods and results Hyperglycaemia alters monocyte, macrophage and hematopoietic stem cell (HSC) metabolism, significantly increasing glycolysis (FDR=0.02, human monocyte non-targeted metabolomics screen). In vitro, hyperglycaemia increased pro-inflammatory macrophage gene expression upon LPS +IFNy stimulation (IL-6, p<0.001) and both monocyte adherence to activated endothelium and macrophage uptake of modified lipid (p<0.001); all responses were normalised by the glycolytic inhibitor dichloroacetate (DCA) or 2-deoxy-glucose (2DG). Bone marrow derived macrophages (BMDM) from diabetic mice, grown in physiological glucose retained heightened pro-inflammatory responses, indicating hyperglycaemic memory in the HSC niche as well, as differentiated cells. To understand if diabetic HSC memory has a role in driving disease in vivo, bone marrow from diabetic mice (vs. wild type control) was transplanted into LDLR^-/- mice. After 12 weeks, plaque burden in the aortic root (p=0.036) and plaque lipid content (p=0.0076), were greater in the mice receiving cell from the diabetic donor, confirming a memory effect. To investigate the mechanism underlying hyperglycaemic memory, ATAC-seq analysis was performed on diabetic and wild-type (WT) HSCs. Differential peak analyses indicated that cells from diabetic mice had an altered chromatin structure, potentially mediated through the increased histone modifications H3K27ac and H3K4me3 (p<0.01). These histone modifications are normalised by DCA. Motif analysis revealed that binding sites for the transcription factors PU.1, CTCF and RUNX1 are significantly enriched in peaks differentially present in diabetic HSC.

In conclusion Diabetic hyperglycaemia alters HSC and macrophage metabolism to induce epigenetic changes which increases their pro-inflammatory responses and drives atherosclerotic disease in vivo. PU.1, CTCF and RUNX1 have been previously associated to chromatin priming elements. This novel demonstration of immunological memory may help to explain why targeting elevated glucose is often ineffective in reducing cardiovascular risk in diabetes.