Hyperglycemia-triggered lipid peroxidation destabilizes STAT4 and impairs anti-viral Th1 responses in type 2 diabetes

Type 2 diabetes (T2D) patients are more susceptible to severe respiratory viral infections, but the underlying mechanisms remain elusive. Here, we show that both COVID-19-infected T2D patients and influenza-infected T2D mice exhibit a defective Th1 response, which is an essential component of anti-viral immunity. This defect stems from intrinsic metabolic perturbations in CD4+T cells driven by hyperglycemia. Mechanistically, hyperglycemia triggers mitochondrial dysfunction and excessive fatty acid synthesis, leading to elevated oxidative stress and aberrant lipid accumulation within CD4+T cells. These abnormalities promote lipid peroxidation (LPO), which drives the carbonylation of STAT4, a crucial Th1 lineage-determining factor. Carbonylated STAT4 undergoes rapid degradation, causing reduced T-bet induction and diminished Th1 differentiation. LPO scavenger ameliorates these Th1 defects in T2D patients with poor glycemic control, and restores viral control in T2D mice. Thus, this hyperglycemia-LPO-STAT4 axis underpins reduced Th1 activity in T2D hosts, with important implications for managing T2D-related viral complications. PBMCs were isolated from healthy donors, well-controlled T2D, and poor-controlled T2D patients and stimulated with aCD3/aCD28 antibodies (biolegend) in T cell medium (RPMI 1640 medium supplemented with 10% FBS, 50M -mercapteoethanol, 1M HEPES, 1X non-essential amino acids (NEAA), 100mM sodium pyruvate, and 1% penicillin-streptomycin) for 48 hours. Untreated and stimulated PBMCs were submitted for scRNA-seq sequencing with Single Cell Gene Expression Flex (10X Genomics). Alive naï4+ T cells (CD45RA+CCR7+CD4+) were sorted from healthy donors, well-controlled T2D, and poor-controlled T2D patients through FACs with Influx (BD Biosciences). Their RNA was extracted using RNeasy Mini Kit (Qiagen) and submitted for paired-end RNA sequencing. Reads were mapped to UCSC human genome GRCh38/hg38 using STAR (version 2.6.1) and gene counts were quantified by featureCounts (version 1.6.4).