Epigenetic regulation by HDAC11-driven STAT3 activation promotes pro-inflammatory cytokine production and endothelial dysfunction upon bacterial infection in diabetes

TL;DR

This study shows that HDAC11 promotes inflammation and vascular issues in diabetic patients during bacterial infections, suggesting it could be a new treatment target.

Contribution

The study identifies HDAC11 as a novel epigenetic regulator linking diabetes, inflammation, and endothelial dysfunction during bacterial infections.

Findings

HDAC11 expression is significantly higher in diabetic endothelial cells infected with E. coli.

HDAC11 inhibition reduces pro-inflammatory cytokines and restores endothelial function in diabetic cells.

HDAC11 interacts with STAT3 to sustain inflammation in diabetic endothelial cells.

Abstract

Diabetes mellitus is associated with low-grade inflammation, resulting in susceptibility to infections and related complications. Histone deacetylase 11 (HDAC11) regulates host immune responses upon infections including fungal and gram-negative bacterial infections. Here, we hypothesise that bacterial infection may influence epigenetic regulation via HDAC11, resulting in the exacerbation of the inflammation responses.

Induced pluripotent stem cells (iPSCs) from non-diabetic (ND) and diabetic (DB) donors were differentiated into endothelial cells (ECs). The iPSCs-derived ECs (iPS-ECs) were infected with Escherichia coli (E. coli) to mimic sepsis. Bulk RNA sequencing was performed to validate the gene expression in transcriptomic level. qRT-PCR, ELISA, and western blot were conducted to assess gene expression levels. Immunocytochemistry (ICC) staining was used to visualise the protein expression, and functional tests were performed to assess the iPS-ECs’ response to infection.

This study revealed that HDAC11 expression is significantly elevated in iPS-ECs derived from DB donors when infected with E. coli. HDAC11 upregulation was associated with increased production of pro-inflammatory cytokines and vascular dysfunction. Administration of a HDAC11 inhibitor effectively suppressed pro-inflammatory cytokine expression and restored endothelial function. Mechanistic analyses demonstrated that interaction of HDAC11 with the signal transducer and activator of transcription 3 (STAT3) sustained the inflammatory response in iPS-ECs derived from DB donors.

Our findings highlight the role of HDAC11 to mediate inflammation-driven vascular impairment in diabetes, suggesting HDAC11 is a promising therapeutic target to mitigate endothelial dysfunction and inflammation, improving endothelial health in people with diabetes.

The online version contains supplementary material available at 10.1186/s12964-026-02696-4.

Figures

D). In comparison to their iPSC counterparts, iPS-ECs from both ND and DB donors showed high expression of these endothelial markers while lacking expression of pluripotency markers such as Oct4, Lin28, and NANOG (Supplementary Figs. 2 A–F). This confirmed successful endothelial differentiation and the loss of stemness characteristics.

Inflammation-related hallmarks are significantly elevated in *E. coli*-infected diabetic endothelial cells. (A) Bubble plots displayed the enriched hallmarks for each comparison condition. GSEA plots for inflammation-related hallmarks include (B) HALLMARK_IL6_JAK_STAT3_SIGNALING, (C) HALLMARK_TNFA_SIGNALING_VIA_NFκB, and (D) HALLMARK_INFLAMMATORY_RESPONSE. The hallmark gene datasets were sourced from MSigDB. Bar plots beneath each GSEA plot showed the leading-edge genes contributing to each Hallmark. The normalized enrichment score (NES) and corresponding p-value were indicated on each GSEA pl

HDAC11 regulates inflammation in endothelial cells, particularly in the context of diabetes. (A-B) Knockdown and overexpression of HDAC11 alter global HDACs and HATs activity. (C-E) At the baseline level, the iPS-ECs derived from diabetic donors exhibited an upregulation of HDAC11. (F) The STRING database analysis ([https://string-db.org](https://string-db.org)) revealed functional and physical protein interactions of HDAC11 with IL1 $\documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \use

Inhibition of HDAC11 significantly reduces STAT3 signalling cascade leading to reduced inflammation. (A) Administration of SIS17 reduced HDAC11 activity in both the non-diabetic (ND) and diabetic (DB) groups. (B-C) Correspondingly, the reduction in HDAC11 activity led to decreased HDAC11 protein expression. (D) Western blot images from the drug inhibitor-treated group. Treatment with SIS17 significantly reduced both total STAT3 (E) and phosphorylated STAT3. (F) Furthermore, VCAM-1 expression, was downregulated in the SIS17-treated group, both at the mRNA (G) and protein levels (H-I). Furthermo

Inhibition of HDAC11 restores endothelial cell function following *E. coli* infection. Inhibition of HDAC11 significantly reduced ROS (A) and LDH (B) production in iPS-ECs with *E. coli* infection. In vitro tube formation assays in both control and HDAC11 inhibitor-treated groups were performed. (C-E) Endothelial cell function was impaired after *E. coli* infection, with the effect being particularly pronounced in diabetic donors. However, treatment with SIS17 significantly improved endothelial cell function, as evidenced by an increase in meshed area and longer tube length. Moreover, SIS17 tr

Interaction of HDAC11 and STAT3 is required for sustained inflammatory response in iPS-ECs derived from DB donors with *E. coli* infection. (A) Efficient STAT3 knockdown and successful HDAC11 overexpression were achieved in iPS-ECs derived from DB donors. (B) STAT3 knockdown significantly reduced CXCL6 expression. HDAC11 overexpression increased the transcription of CCL2, CCL5, CXCL6, and CXCL10; however, STAT3 knockdown prior to HDAC11 overexpression attenuated this inflammatory response. (C) Western blot analysis confirmed STAT3 depletion and HDAC11 overexpression. HDAC11 overexpression led

Full text

Background

Diabetes mellitus leads to vascular dysfunction and degeneration, contributing to both macrovascular (cardiovascular disease, peripheral artery disease, and stroke) and microvascular (retinopathy, nephropathy, neuropathy) complications [1]. Compared to healthy individuals, patients with diabetes mellitus also experience increased susceptibility to a wide range of bacterial infections [2]. For example, E. coli is a prevalent gram-negative pathogen responsible for diabetes-related infections, including urinary tract infection (UTI), diabetic foot ulcer (DFU), and sepsis [3–5]. Moreover, E. coli bacteraemia among diabetes patients is associated with higher risk of sepsis and septic shock, suggesting that chronic hyperglycaemia contributes to impaired immune function during bacteraemia [5–7].

Sepsis is a life-threatening condition caused by dysregulation of the host immune response to infection; it is marked by a cytokine storm leading to a systemic inflammatory response syndrome (SIRS), oxidative stress, microcirculatory dysfunction, and endothelial barrier dysfunction [8]. Endothelial cells (ECs) form a monolayer located in the inner surface of blood and lymphatic vessels. ECs are major contributor of sepsis as they amplify the immune response. During sepsis, ECs are both the target and source of inflammation linking local and systemic immune responses. Based on their location, ECs are among the initial cells in circulation to identify microbial components, acting similarly to cells of the innate immune system [8]. ECs express several pattern recognition receptors (PRRs) that recognise pathogen-associated molecular patterns (PAMPs), including Toll-like receptor-4 (TLR-4) that recognise lipopolysaccharide (LPS), CD-48 that recognise type 1 fimbriae (FimH), gp-96 and integrin that recognise outer membrane protein A (OmpA). For example, exposing ECs to E. coli activates TLR-4 signalling leading to the production of proinflammatory cytokines, such as TNFα and IL1β [9]. Moreover, clinical E. coli strains isolated from sepsis patients bind to ECs via the interaction of bacterial OmpA with integrin receptors, resulting in disruption of the EC barrier integrity [10]. Another study reported that E. coli infection also promotes increased oxidative stress in an obesity mouse model [11].

The induced pluripotent stem cells (iPSCs) are obtained with minimal patient invasion, resulting in large quantities of cells, unlike primary endothelial cells, which often require tissue biopsy, have limited expansion capacity, and rapidly undergo donor-dependent senescence. While human vascular endothelial cells (HUVECs) provide a useful standardised endothelial model, they are non-patient-specific, derived from a single developmental origin, and lack disease-associated epigenetic priming, limiting their ability to capture inter-individual variation in complex disorders. In contrast, iPSC-based cellular models can be produced indefinitely and in large quantities, while retaining the genetic background and epigenetic memory of the donor [12], enabling mechanistic interrogation of patient-specific disease pathways that may not be preserved in primary cells or present in HUVECs. Moreover, unlike both primary endothelial cultures and HUVECs, iPSCs exhibit multilineage differentiation potential, allowing generation of diverse disease-relevant cell types to model pathological mechanisms in a system-wide, donor-matched manner [13]. This versatility supports personalised disease modelling, high-throughput drug screening, toxicity testing, and evaluation of emerging therapies, providing broader translational value compared to traditional primary or HUVEC-based systems [14]. ECs-derived induced pluripotent stem cells (iPS-ECs) have emerged as a modality for studying mechanisms of the disease and identifying potential drug targets [15, 16], as they recapitulate fundamental functions of ECs, such as immunological signalling, transport, haematological, and mechanical response [14].

Histone deacetylases (HDACs) remove the acetyl/acyl groups from histone and non-histone proteins. HDACs consist of 4 classes: class I HDACs (HDAC1, 2, 3, and 8), class IIa HDACs (4, 5, 7, 9), class IIb (6 and 10), class III (SIRT1-7), and class IV (HDAC11). HDAC11 is the only member of class IV and shares structural features with both class I and class II HDACs [17]. Unlike class I and II, HDAC11 interacts with HDAC6 in vivo but does not interact with classical protein complexes, such as mSin3A, RbAp48, NCoR, or SMART [18]. HDAC11 regulates metabolic inflammation and immune function [19]. Additionally, several studies have also reported that HDAC11 is involved in numerous pathological conditions such as ischemia injury [20], endothelial dysfunction [21, 22], atherosclerosis [23], and metabolic disease [24, 25]. Moreover, ablation of HDAC11 in a high-fat diet mouse model improves cholesterol and triglyceride levels, enhances insulin sensitivity and glucose tolerance, and reduces liver damage [26]. HDAC11 also plays a crucial role in regulating inflammation and immunological process [27–29]. HDAC11 is significantly upregulated in response to cytokines and microbial infections, leading to enhanced inflammatory responses in ECs and macrophages [22, 30]. Therefore, HDAC11 regulates both metabolic and immune functions in several conditions in a cell-type specific manner.

Even though the role of HDAC11 in immune cells is well-documented, its function in bacterial infections involving ECs from diabetic people has not been investigated. Here, we report that HDAC11 amplifies inflammatory responses in iPS-ECs derived from DB donors, highlighting its potential contribution to diabetes-related vascular dysfunction upon infections.

Methods

Experimental design

The iPSCs were obtained from mononuclear cells (MNCs) of non-diabetic (ND) and diabetic (DB) donors according to our previous methods [31]. The iPSCs from different donors were obtained from a large patient cohort, as described [32]. The ND donors exhibited normotension, body mass index (BMI)≤30 and normoglycaemia condition. Type-2 diabetes mellitus (T2DM) donors were included in this study with 10-15-year disease duration and BMI ≥30 indicating T2DM with obesity condition [32]. Donors used in this study is presented in the Supplementary Table 1. The iPSCs at passage 15–20 were differentiated into endothelial cells (iPS-ECs) subsequently the iPS-ECs between passage 3–6 were used in this study.

Biological materials and ethical consideration

Ethical approval (expediency number REC 14/NI/1109) was granted by the Office for Research Ethics Committees of Northern Ireland (ORECNI).

iPS-ECs generation

The iPSCs reaching 80–90% confluency was used for iPS-ECs differentiation. Cells were incubated with passaging solution (StemMACS, Cat. No. 130-104-688) for 5–7 min at the 37 °C. Then, the cells were collected in the aggregation medium containing KnockOut DMEM F12 (Thermofisher Scientific, Cat. No. 12660012), KnockOut serum replacement (Gibco, Cat. No. 10828-028), Penicillin-Streptomycin (Thermofisher Scientific, Cat. No. 15140122), non-essential amino acids (Gibco, Cat. No. 11140-050), Glutamax (Gibco, Cat. No. 35050-061), and β-mercaptoethanol (Gibco, Cat. No. 31350-010) and centrifuged at 200 xg for 5 min. The cell pellet was resuspended in aggregation medium containing ROCK (Rho-associated kinase) inhibitor (Generon, HY-10583) before seeding onto ultra-low attachment plates (Costar, Cat. No. 3471) [33]. ROCK inhibitor was used in this study to reduce cytoskeletal tension and inhibit anoikis, thereby increasing cell viability and promoting junctional stability [34, 35]. The next day, the medium was changed into N2B27 containing DMEM F/12 (Gibco, 31330-038), N2 supplement (Thermofisher Scientific, Cat. No. 17502048), B27 supplement (Thermofisher Scientific, Cat. No. 12587010), Neurobasal medium (Gibco, Cat. No. 21103-049) supplemented with 12 µM Chir99021 (ApexBio Tech, Cat. No. B5779) and 30 ng/mL of BMP4 (Biotechne, Cat. No. 314-BP) to induce mesodermal formation. On day 3, the medium was refreshed into N2B27 supplemented with 2 µM of forskolin (Biotechne, Cat. No. 1099/10) and 100 ng/mL human VEGF-A (Biotechne, Cat. No. NBC1-21277) for 2 days to induce budding of the ECs. On day 5, the embryoid bodies (EBs) were mechanically broken down into 100–200 μm in size and seeded onto gelatine-coated dishes, followed by incubation with StemPro34 medium (Thermofisher Scientific, Cat. No. 10639011), Penicillin-Streptomycin, Fetal Bovine Serum (Gibco, Cat. No. 10270-106), Glutamax and supplemented with 100 ng/mL of VEGF-A and 100 ng/mL FGF2 (Biotechne, Cat. No. NBP2-34921-1MG) to obtain 2D ECs cultures. The medium was changed every other day.

On the day 10, T25 flasks were coated with Collagen-IV (Biotechne, Cat. No. 3410-010-02) for 2 h. In parallel, the cells were washed and passaged with TryplE (Thermofisher Scientific, Cat. No. 12604013). TryplE was used in this study as a substitution of trypsin. The activity of TryplE is gentler than trypsin thus it gives higher viability and suits for sensitive cells such as iPS-ECs. The cells were then collected with the EGM2-MV medium (Lonza, Cat. No. CC-3202/6), counted, and filtered using filter mesh (Fisher Scientific, Cat. No. 22-363-548). Cells were subsequently transferred into a new 15 mL corning tube, spun for 5 min at 500 xg. and the supernatant was discarded. The cell pellet was resuspended with 100 µL of the selection buffer and 40 µL of CD144 magnetic beads (Miltenyi Biotec, Cat. No. 130-097-857) for 15 min and sorted using magnetic-activated cells sorting (MACS). Pure EC populations were seeded on top of the collagen-IV-coated flasks.

Bacterial culture

E. coli K1 was a clinical isolated given by Prof. Krasnodembskaya used in the experiments. For each experiment, bacterial colonies were seeded from frozen stocks and grown overnight at 37 °C in liquid Luria-Bertani (LB) medium (Sigma, Cat. No. L3022-1KG) with agitation (200 rpm). Before each experiment, the bacterial cells were washed with warm PBS, and the optical density (OD at λ = 600 nm) of the suspension was measured. The number of colony forming unit (CFU) was calculated according to the following equation: OD600 = 0.9-1 corresponds to 3 × 108 CFU/mL.

E. coli infection

ECs were seeded in a 12-well plate with a seeding density of 100,000 cells per well and incubated at 37°C with 5% CO2 incubator until they reached 70-80% confluency. At this point, cells were infected with E. coli at a multiplicity of infection (MOI) of 20 bacteria per cell. After 2 hours of infection, the cells were washed thrice with warm media, followed by incubation with an experimental medium (Promocell, Cat. No. C-22220) containing gentamicin (100 µg/ml) for 1 hour at 37°C. The medium then replaced with EGM2-MV medium (Lonza, Cat. No. CC-3202/6), and the cells were incubated for 48 hours, at which time were harvested for RNA and protein extraction. The supernatant was collected and stored at -80°C for further experiments.

Quantitative PCR

The RNA was extracted using an RNA extraction kit (QiagenTM, Cat. No. 74104) followed by cDNA synthesis (Thermofisher Scientific, Cat. No. 4374966). 10ng of cDNA was amplified and used to perform PCR with SYBR Green (Thermofisher Scientific, Cat. No. 4368708). The primer sequences for inflammatory regulators, cytokines, and pluripotency markers are listed in Supplementary Table 2 whereas the PCR conditions are shown in Supplementary Table 3.

Immunoblotting

25 µg of protein were separated by 8% SDS-PAGE and electro transferred to polyvinylidene (PVDF) membranes (GE Healthcare Life Sciences, Cat. No. 10600021). Membranes were blocked with 5% bovine serum albumin (BSA) for 1 h at room temperature and subsequently with primary antibodies overnight (Supplementary Table 4). The next day, the membranes were washed and incubated with secondary antibodies. Finally, the membranes were visualised by chemiluminescence (BioRad, Cat. No. 170–5061) using the G: Box gel documentation system (Syngene, Serial number DR404/1513).

Immunofluorescence staining

Cells were seeded in the 8-well chamber (Thermofisher scientific, Cat. No. 154534PK) and fixed with 4% paraformaldehyde (PFA) (Thermofisher scientific, Cat. No. 28906). This is followed by cell permeabilization using 0.1% Triton-X and blocking with 5% goat serum for 30 min at room temperature. Then, cells were incubated with the specified antibody (Supplementary Table 3) for 1 h in 37 °C and then incubated with secondary antibodies followed by staining with 4,6-diamino-2-phenylindole dihydrochloride (DAPI) (Sigma, Cat. No. 62248). The slides were mounted using Vectashield mounting medium (VECTOR Laboratories, Cat. No. H-1000) and visualised with a STELLARIS confocal microscope (Leica microsystems).

Human cytokine array

Supernatants collected at 48 hours post-infection were analysed using the human cytokine protein array (RayBiotech, AAH-CYT-8). Each membrane was incubated with blocking buffer followed by incubation with supernatant aliquots for 2 hours at room temperature in the shaker. Then, the membranes were washed prior to incubation with the biotinylated antibody cocktail and HRP-Streptavidin. Dot blot images were captured using G:Box gel documentation and quantified using Protein Array Analyzer for ImageJ (http://image.bio.methods.free.fr/ImageJ/?Protein-Array-Analyzer-for-ImageJ.html).

Enzyme-linked immunosorbent assay (ELISA)

Supernatants collected at 48 h post-infection were spun at 10,000 rotations per motion (rpm) for 10 min at 4 °C to generate cell-free supernatant. The concentration of TNFα, CCL2, and CXCL6 levels in the medium was determined by ELISA kit (Biotechne, Cat. No. DTA00D, DCP00, and DC00 respectively) according to the manufacturer’s instructions.

Measurement of reactive oxygen species (ROS) production

The cells were seeded with 10,000 in a 96-well plate (Corning, Cat. No. 3904) and the total ROS from E. coli-infected and uninfected iPS-ECs was measured simultaneously. Then, 500X ROS assay stain (Invitrogen, Cat. No. 88-05930-74) was prepared accordingly by adding 40 µl DMSO into the vial of ROS assay stain concentrate. The 500X ROS assay stain was mixed with 1 mL ROS assay buffer and 100 µl of this mix was added into each well. After 60 min incubation at the 37 °C with 5% CO2, the plates were read using a FlexStation3 microtiter plate-reading fluorometer at the excitation of 480 nm and the emission of 520 nm.

Lactate dehydrogenase (LDH) assay

Cell lysis was investigated using cell supernatants collected at 48 h post-infection. As a 100% lysis control, iPS-ECs were incubated with 10% Triton-X. 50µL of cell supernatants and lysis control were transferred into a 96-well plate and LDH was measured according to the manufacturer’s instructions (Sigma Aldrich, Cat. No. 04744926001).

FITC dextran permeability assay

iPS-ECs from DB donors were seeded onto trans-well inserts at a density of 15,000 cells per well. Subsequently, iPS-ECs were infected with E. coli and treated with 1 mg/mL FITC-dextran (Sigma Aldrich, Cat. No. 46944) for 20 min at room temperature. Fluorescence intensity was measured at 485 nm excitation and 520 nm emission and the relative percentage diffusion was calculated to determine ECs permeability.

RNA-seq library preparation and sequencing, differentially expressed genes, network and gene set enrichment analysis (GSEA)

RNA sequencing libraries were prepared using the NextSeq500 platform (Illumina) and aligned to the human reference genome (Hg38, UCSC Genome Browser). The sequencing data were then normalised, and differentially expressed genes (DEGs) across experimental conditions were identified using the DESeq2 package in R (version 4.2.1) and the top DEGs represented by heatmap. Additionally, downstream functional analysis of the transcriptome was performed based on the identified DEGs. Following DEG identification, gene set enrichment analysis (GSEA) was conducted. A false discovery rate (FDR) threshold of < 0.05 was applied to classify upregulated and downregulated genes. These genes were subsequently used for pre-ranked GSEA, performed via the fgsea R package with MSigDB datasets. Significant positive and negative hallmarks of the dataset were visualised through bubble plots and GSEA plots, providing insights into the differences between the infection and control groups.

Inhibition of HDAC11 using ShRNA and drug inhibitor

iPS-ECs were seeded in a 12-well plate at 1 × 10^5^ cells/well and allowed to reach 70–80% confluency. The medium was removed, and the cells were treated with 25 mM SIS-17 (Adooq, Cat. No. A18463), an HDAC11 inhibitor, for 6 h prior to infection and followed up to 48 h. Additionally, the expression of HDAC11 was manipulated using shRNA (CCCGACGTGGTGGTATACAAT).

Immunoprecipitation

The cell pellet was lysed with lysis buffer (Thermofisher Scientific, Cat. No. 87787) followed by incubation for 5 min on ice with periodic mixing. Then, the tubes were spun in 13,000 x g for 10 min at 4 °C. The cell lysate was transferred to a new Eppendorf tube. The protein concentration was adjusted to 1 µg/µL and 10 µL of agarose A/G beads (Santacruz, sc-2003) were added to 500 µl of the lysate to preabsorb for 1-hour at 4 °C on a rocker platform or rotating device. Afterwards, primary antibodies (2 µg/mg of protein) were added and incubated overnight at 4 °C on a rocker platform. The next day, 20 µL of agarose A/G beads (Santacruz, sc-2003) were added and incubated for 2 h at 4 °C on a rocker platform or rotating device. Finally, the mixture was spun at 1,000 xg for 5 min at 4 °C to obtain the pellet and eluted with sample buffer prior to western blot.

Statistical analysis

Data were presented as Mean ± SD. The data were analysed using student’s t-tests when comparing two conditions and one-way ANOVA or Kruskal-Wallis when comparing more than two conditions. GraphPad Prism (version 9.4.1 (458)) was used for statistical software, and p < 0.05 was used to indicate significance among the groups.

Results

Reproducible generation of iPS-ECs derived from non-diabetic (ND) and diabetic (DB) donors

iPSCs derived from different donors [32] were used to generate iPS-ECs. Six independent iPSC donor lines—three ND (1 female, 2 male) and three DB (1 female, 2 male)—were randomly selected. Using different donors allowed us to capture biological variability. Each donor represented a true biological replicate, whereas comparisons among successive differentiations per donor represented biological replicates of within-donor variation.

We developed an efficient and reproducible protocol for deriving iPS-ECs from ND and DB donors, as evidenced by the fact that 90–98% of the cells exhibited EC characteristics compared to the negative staining (see Materials and Methods and Supplementary Figs. 1 A–C). Furthermore, the iPS-ECs expressed EC markers, including CD144, CD31, KDR, and ZO-1 (Supplementary Fig. 1D). In comparison to their iPSC counterparts, iPS-ECs from both ND and DB donors showed high expression of these endothelial markers while lacking expression of pluripotency markers such as Oct4, Lin28, and NANOG (Supplementary Figs. 2 A–F). This confirmed successful endothelial differentiation and the loss of stemness characteristics.

Functionally, iPS-ECs derived from ND donors successfully formed endothelial tube structures in an in vitro Matrigel assay. In contrast, iPS-ECs from DB donors exhibited impaired tube formation, suggesting compromised endothelial function consistent with DB-associated vascular defects (Supplementary Fig. 2G).

iPS-ECs exhibit vascular inflammation upon bacterial infection

The DB donor-derived iPS-ECs exhibited higher inflammatory responses than ND donors after E. coli infection for 48 h, as determined by the mRNA expression of IL-6, TNF, CCL2 and CCL5 (Fig. 1A). These findings were further supported by cytokine array analysis of the cell culture supernatants, which revealed increased levels of pro-inflammatory cytokines in E. coli infected-iPS-ECs (Fig. 1B). The quantification of the cytokine array results confirmed that several cytokines and chemokines, including IL6, TNFα, CCL5, CXCL6, and CXCL10, were significantly elevated in DB iPS-ECs with E. coli infection (DBE) (Fig. 1C). Moreover, the bacterial infection activated the ECs, as indicated by the increased expression of vascular cell adhesion molecules-1 (VCAM-1), a glycoprotein that plays a key role in mediating the adhesion of immune cells to the endothelium (Fig. 1D). At the protein level, VCAM-1 production was significantly higher in the DBE compared to the ND-derived iPS-ECs infected with E. coli (NDE group) (Figs. 1E-F).Fig. 1E. coli infection promotes increased production of inflammatory cytokines. (A) A schematic diagram illustrates the generation of iPSC-derived endothelial cells (iPS-ECs) from human donor blood. Mononuclear cells (MNCs) were isolated, reprogrammed into induced pluripotent stem cells (iPSCs) and subsequently differentiated into iPS-ECs for experimental use. (B) Forty-eight hours post-E. coli infection, several pro-inflammatory cytokines were significantly upregulated. (C) Cytokine-specific antibody arrays were incubated with cell-free supernatant, with representative figures of the protein array shown. (D) Semi-quantitative analysis of protein arrays was conducted using ImageJ. (E) VCAM-1 mRNA expression was significantly upregulated in DBE compared to NDE and DB. (F) A representative western blot image of VCAM-1 expression. (G) Densitometric analysis of VCAM-1 protein expression. Statistical Significance: Statistical analysis was tested by Kruskal-Wallis or One-Way ANOVA, ✻ = p < 0.05 vs. ND; ✻✻ = p < 0.01 vs. ND; ✻✻✻ = p < 0.001 vs. ND; ✻✻✻✻ = p < 0.0001 vs. ND. ND = non-diabetic, DB = diabetic. • = ND = non-diabetic, ■ = NDE = non-diabetic with E. coli infection, ○ = DB = diabetic, □ = DBE = diabetic with E. coli infection

iPS-EC from donors with diabetes show altered responses to E. coli infection

Our previous research demonstrated a considerable increase in ROS levels within the vascular organoids derived from donors with diabetes [36]. Here, we observed a notable increase in ROS expression levels following E. coli infection which was particularly evident in the DBE group (Fig. 2A). While ROS production is essential to facilitate bacterial clearance, activation of the inflammatory signalling cascades, and immune responses [37], prolonged ROS formation, as observed in the DBE group, might be implicated in endothelial dysfunction.

Fig. 2E. coli-infected iPS-ECs have increased ROS production and LDH release leading to endothelial dysfunction. (A)** The total ROS levels were higher in the diabetic group compared to the non-diabetic group, with E. coli infection further amplifying ROS production in both donor groups. (B) The LDH assay revealed elevated levels in the E. coli-infected iPS-ECs groups. (C-D) The tube formation assay demonstrated that E. coli infection resulted in fewer and shorter capillary-like structures compared to the control groups, as indicated by shorter tube branch lengths and smaller meshed areas. Imaging Details: Magnification: 5X. Statistical Significance: Statistical analysis was tested by Kruskal-Wallis or One-Way ANOVA, ✻=p < 0.05 vs. ND; ✻✻ =p < 0.01 vs. ND; ✻✻✻ =p < 0.001 vs. ND; ✻✻✻✻ =p < 0.0001 vs. ND. ND = non-diabetic, DB=diabetic. •=ND = non-diabetic, ■ =ND + E.coli = non-diabetic with E. coli infection, ○=DB=diabetic, □=DB + E.coli=diabetic with E. coli infection

Lactate dehydrogenase (LDH), an essential glycolytic enzyme, is closely associated with metabolic oxidative stress. Under physiological conditions, LDH remains intracellular, but it is released to the extracellular space upon cell damage or loss of cell permeability [38]. Our results indicated a significant increase in LDH release in the infection group (Fig. 2B), suggesting EC damage.

Additionally, the E. coli-infected iPS-ECs demonstrated increased endothelial dysfunction, as evidenced by tube length and a failure to form tubular structures (Fig. 2C-D). Our previous study indicated that the iPS-ECs derived from DB donors exhibited endothelial dysfunction, and this dysfunction was more pronounced in the DBE group. Together, our results suggested that the iPS-ECs exhibit higher ROS and LDH release as well as impaired EC angiogenesis. These conditions are further exacerbated and become more evident in the DBE group during E. coli infection.

RNA sequencing validates the iPS-ECs response to infection

To further investigate the global effect of E. coli infection in iPS-ECs at the transcriptomic level, we performed GSEA analysis based on the DEGs among the infected and non-infected groups. The FDR < 0.05 and a normalised enrichment score (NES) > 1.2 were used to determine the enrichment. Heatmaps represented top 40 upregulated DEGs highlighting that E. coli infection promoted inflammation in NDE and DBE groups as it was represented by red colour (Supplementary Fig. 3).

We used gene ontology (GO) analysis with the GSEA to identify significant altered pathways. This study revealed significant enrichment of several hallmarks in DBE group, including TNFα signalling via NFκB (p = 0.035, NES 1.789) and inflammatory response (p = 0.041, NES 1.799), compared to NDE group. The enrichment of these pathways suggested a pronounced pro-inflammatory phenotype in the DBE group. Moreover, the IL6_JAK_STAT3 signalling (p = 0.018, NES 2.217) was significantly enriched in the DBE compared to DB groups. Further analysis identified several genes with notably enriched expression associated with these pathways (Fig. 3). The IL6_JAK_STAT3 signalling pathway has been shown to play a critical role in regulating inflammation by mediating the production of the cytokines and chemokines [39, 40]. These findings highlighted an exaggerated inflammatory response in DBE, as evidenced by the enrichment of key immune-related pathways.

Fig. 3. Inflammation-related hallmarks are significantly elevated in E. coli-infected diabetic endothelial cells.** (A)** Bubble plots displayed the enriched hallmarks for each comparison condition. GSEA plots for inflammation-related hallmarks include (B) HALLMARK_IL6_JAK_STAT3_SIGNALING, (C) HALLMARK_TNFA_SIGNALING_VIA_NFκB, and (D) HALLMARK_INFLAMMATORY_RESPONSE. The hallmark gene datasets were sourced from MSigDB. Bar plots beneath each GSEA plot showed the leading-edge genes contributing to each Hallmark. The normalized enrichment score (NES) and corresponding p-value were indicated on each GSEA plot. Log2 FC refers to the Log2 Fold Change. DB_E = Diabetic endothelial cell infected with E. coli; ND_E = Non-diabetic endothelial cell infected with E. coli; DB = Diabetic endothelial cell; ND = Non-diabetic endothelial cell

We also found that HDAC11 positively regulated multiple-inflammation-associated genes that contributed to the inflammatory response of the cells (Supplementary Fig. 4A). STAT3, a downstream target of HDAC11, mediated the inflammatory signalling. Moreover, the causal network analysis revealed that STAT3 played a central role in regulating the expression of various inflammatory cytokines and chemokines (Supplementary Fig. 4B). STAT3, known to be activated by ligands such as IL6, acts as a transcription factor that regulates immune responses and inflammation [39, 40]. Our data demonstrated that STAT3 was essential for regulating inflammatory chemokines (CCL2, CCL5, CXCL6, CXCL20, CXCL11 and etc.), transcription factors (NFKB2, JUN, FOS), metabolic genes (PDK4, PPARA), and apoptotic regulators (BCL2, CASP1). These effects were more pronounced in DBE group, indicating that the HDAC11/STAT3 axis plays a central role regulating inflammation in ECs derived from DB donors.

HDAC11 inhibition reduces inflammation in the DB-derived iPS-ECs donors through STAT3

Among all HDAC family members, HDAC11 is the only HDAC that shows upregulation in response to E. coli infection in iPS-ECs derived from DB donors (Supplementary Fig. 5). In contrast, the other HDACs were normally expressed or only slightly downregulated. Concurrently, the DBE group showed a significant downregulation of KAT2A, KAT5, and KAT8, suggesting a decrease in acetyltransferase ability (Supplementary Fig. 5). Therefore, HDAC11 is specifically induced under pathogen stimuli in DB iPS-ECs.

Moreover, knockdown and overexpression of HDAC11 altered global HDAC and histone acetyltransferase (HAT) activities and was accompanied by significant changes in histone H3 and H4 acetylation activity levels (Fig. 4A–B). Knocking down HDAC11 lowered overall HDAC activity and raised global HAT activity, which was associated with higher acetylation of both H3 and H4. In contrast, overexpressing HDAC11 increased global HDAC activity and reduced HAT activity as shown by reduction of H3 and H4 activity.

Fig. 4HDAC11 regulates inflammation in endothelial cells, particularly in the context of diabetes. (A-B) Knockdown and overexpression of HDAC11 alter global HDACs and HATs activity. (C-E) At the baseline level, the iPS-ECs derived from diabetic donors exhibited an upregulation of HDAC11. (F) The STRING database analysis (https://string-db.org) revealed functional and physical protein interactions of HDAC11 with IL1 \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\:{\upbeta\:}$$\end{document} , ISG15, IL6, NFKB1, TNFα, STAT3, CCL2, and RELB. (G-H) The representative figure and semi-quantitative analysis of HDAC11 expression in the response of E. coli. (I) HDAC11 was predominantly expressed in the nucleus of iPS-ECs. (J-L) Following E. coli infection, STAT3 was activated, and this effect was obviously seen in the diabetic donors. Statistical Significance: Statistical analysis was tested by Kruskal-Wallis or One-Way ANOVA, ✻=p < 0.05 vs. ND; ✻✻ =p < 0.01 vs. ND; ✻✻✻ =p < 0.001 vs. ND; ✻✻✻✻ =p < 0.0001 vs. ND. ND = non-diabetic, DB = diabetic. • = ND = non-diabetic, ■ = NDE = non-diabetic with E. coli infection, ○ = DB = diabetic, □ = DBE = diabetic with E. coli infection

Therefore, our findings suggest that HDAC11 is a significant modulator of inflammation in the DB donors (Fig. 4C-E). Additionally, protein-protein interaction prediction using the String database indicated that HDAC11 interacted with several transcription factors such as NFkB1, NFkB2, and STAT3, implicating a possible role in regulating inflammatory pathways (Fig. 4F). In addition, HDAC11 expression was strongly induced by E. coli infection, an effect that was most pronounced in the DBE group (Fig. 4G-H). Previous studies indicated that HDAC11 is localised in the nucleus [17, 18] and the cytoplasm [41]. Immunostaining of iPS-ECs confirmed that HDAC11 was expressed in both nucleus and cytoplasmic (Fig. 4I).

STAT3 is crucial for the survival of ECs, modulating inflammatory responses, and protecting ECs against apoptosis. We observed that E. coli infection led to STAT3 activation, as evidenced by increased phosphorylation of STAT3 at tyrosine residue 705 (pSTAT3 Y705) in both E. coli infected groups. However, the expression of pSTAT3 Y705 was higher in the DBE group compared to the NDE group (Fig. 4J-L). This finding may explain the exaggerated production of inflammatory cytokines and chemokines observed in the DBE group.

Based on these findings, the subsequent experiments focused on the iPS-ECs derived from DB donors to further dissect the molecular mechanisms responsible for the exaggerated inflammatory response to E. coli infection. This approach allowed us to get a deeper understanding of how diabetes-associated cellular changes, particularly those involving the HDAC11/STAT3 axis modulates the inflammatory behaviour of ECs under pathogen stimulation.

Inhibition of HDAC11 reduces inflammation

To further investigate how HDAC11 regulates inflammation, we treated iPS-ECs derived from DB donors with SIS17 (25 µM), a selective HDAC11 inhibitor, for 6 h prior to E. coli infection. This treatment significantly reduced HDAC11 activity in the SIS17-treated group compared to the DMSO group (Fig. 5A). SIS17 was selected as it is a selective small-molecule inhibitor of HDAC11 [17, 42]. Mechanistically, it acts by binding to the catalytic pocket of the enzyme and blocking its Zn²⁺-dependent deacetylase activity, preventing substrate access and hydrolysis [42]. Unlike many classes I/II HDAC inhibitors that primarily act on histone deacetylation, HDAC11 has stronger fatty-acid deacylase activity (e.g., defatty-acylation of lysine residues on metabolic and immune-regulatory proteins) [42]. SIS17 inhibits this enzymatic activity, leading to accumulation of lysine acyl marks on HDAC11 substrates, which functionally impacts pathways such as cellular metabolism and immune responses [17, 42]. Interestingly, our study elicited that the protein level of HDAC11 was significantly lower in the SIS17-treated group than in the DMSO-treated group (Fig. 5B-C). However, this mechanism remains elusive and needs further investigations.

Fig. 5. Inhibition of HDAC11 significantly reduces STAT3 signalling cascade leading to reduced inflammation. (A) Administration of SIS17 reduced HDAC11 activity in both the non-diabetic (ND) and diabetic (DB) groups. (B-C) Correspondingly, the reduction in HDAC11 activity led to decreased HDAC11 protein expression. (D) Western blot images from the drug inhibitor-treated group. Treatment with SIS17 significantly reduced both total STAT3 (E) and phosphorylated STAT3. (F) Furthermore, VCAM-1 expression, was downregulated in the SIS17-treated group, both at the mRNA (G) and protein levels (H-I). Furthermore, the mRNA expression of IL6, TNFα, CCL2, CCL5, CXCL6, and CXCL10, was significantly decreased in the SIS17-treated groups (J-O). The reduction of TNFα and CCL2 in the SIS17-treated group was also evident at the protein level **(P-Q). **Statistical Significance: Statistical analysis was tested by Kruskal-Wallis or One-Way ANOVA, ✻ = p < 0.05 vs. DB+DMSO; ✻✻ = p < 0.01 DB+DMSO; ✻✻✻ = p < 0.001 DB+DMSO; ✻✻✻✻ = p < 0.0001 DB+DMSO. ND = non-diabetic, DB = diabetic. • = ND = non-diabetic, ■ = NDE = non-diabetic with E. coli infection, ○ = DB = diabetic, □ = DBE = diabetic with E. coli infection

Since HDAC11 inhibition also affected the expression of total STAT3 (Fig. 5D-F), we normalised the expression of phosphorylated STAT3 to a housekeeping gene. This semi-quantitative analysis revealed that E. coli infected cells pre-treated with SIS17 had a significant reduction in both total STAT3 and the pSTAT3 Y705, which was associated with lower inflammatory cytokines and chemokines production.

Furthermore, while ECs were activated in response to E. coli, as shown by the elevated VCAM-1 expression in the DMSO groups, VCAM-1 expression was significantly reduced in the iPS-ECs treated with HDAC11 inhibitor (Fig. 5G-I). At the mRNA level, the expression of the pro-inflammatory cytokines, including IL1β, CCL5, TNFα, CXCL10, CXCL6, IL6, and CCL2, was also significantly downregulated in the SIS17-treated group (Fig. 5J-O). At the protein level, E. coli infection led to increased production of pro-inflammatory cytokines such as TNFα and CCL2 (Fig. 5P-Q). These findings suggest that pharmacological inhibition of HDAC11 attenuated the inflammatory response induced by E. coli infection.

Silencing HDAC11 by shRNA recapitulated the effects observed with pharmacological inhibition. The shRNA-mediated knockdown of HDAC11 led to a significant reduction in both mRNA and protein levels of HDAC11. HDAC11 expression was higher in the non-targeting control (shNT) group but was substantially diminished in the HDAC11-silenced (shHDAC11) group (Supplementary Fig. 6A-C). In the presence of E. coli infection, knockdown of HDAC11 also decreased the expression of STAT3 and pSTAT3 Y705, thereby diminishing the production of pro-inflammatory cytokines and chemokines. Meanwhile, there was no difference in VCAM-1 expression between the shNT and shHDAC11 groups. However, upon E. coli infection, VCAM-1 expression was decreased in the shHDAC11 compared to shNT group (Supplementary Fig. 6D-F). Similarly, the mRNA expression of IL1β, CCL5, TNFα, CXCL10, CXCL6, IL6, and CCL2 was significantly downregulated in the shHDAC11 compared to the shNT group (Supplementary Fig. 6G-L).

Together, these results underscore the pivotal role of HDAC11 as a regulator of ECs inflammation during E. coli infection. Both genetic silencing and drug inhibition of HDAC11 effectively dampened the inflammatory response, particularly in iPS-ECs derived from DB donors.

Moreover, we investigated next whether HDAC11 inhibition improves EC function during E. coli infection. The results showed that the HDAC11 inhibition significantly reduced ROS generation in both SIS17-treated groups (Fig. 6A). Also, LDH release was significantly reduced in E. coli-infected groups following HDAC11 inhibition (Fig. 6B). HDAC11 inhibition also restored ECs function at 48 h post-infection, regardless of the presence of infection (Fig. 6C-E). In vitro Matrigel assay demonstrated that SIS17 treatment significantly improved EC function as indicated by increase meshed area and tube length in both control and infection group. SIS17 treatment also reduced the permeability in both control and infection group (Fig. 6F), suggesting a protective effect of HDAC11 inhibition on EC barrier integrity. Although permeability showed a slight upward trend in the infected condition, the difference was not statistically significant. Overall, these results indicate that HDAC11 inhibition improves EC function upon E. coli infection by reducing LDH release, lowering ROS production, and stabilising endothelial barrier function.

Fig. 6. Inhibition of HDAC11 restores endothelial cell function following E. coli infection. Inhibition of HDAC11 significantly reduced ROS (A) and LDH (B) production in iPS-ECs with E. coli infection. In vitro tube formation assays in both control and HDAC11 inhibitor-treated groups were performed. (C-E) Endothelial cell function was impaired after E. coli infection, with the effect being particularly pronounced in diabetic donors. However, treatment with SIS17 significantly improved endothelial cell function, as evidenced by an increase in meshed area and longer tube length. Moreover, SIS17 treatment reduced endothelial permeability in both DB and DBE group as evidenced by less FITC dextran permeation (F). Imaging Details: Magnification: 5×. Statistical Significance: Statistical analysis was tested by Kruskal-Wallis or One-Way ANOVA, ✻ = p < 0.05 DB+DMSO; ✻✻ = p < 0.01 DB+DMSO; ✻✻✻ = p < 0.001 DB+DMSO; ✻✻✻✻ = p < 0.0001 DB+DMSO. DB = diabetic, DBE = diabetic with E. coli infection

HDAC11 regulates inflammation through STAT3 in E. coli-infected iPS-ECs

To better explore the interaction between HDAC11 and STAT3, we performed experiments in silencing STAT3 and overexpressing HDAC11. Knockdown of STAT3 by shRNA resulted in significant reduction of STAT3 expression without affecting HDAC11 expression, while overexpression of HDAC11 led to increase in STAT3 and pSTAT3 Y705. This indicated that HDAC11 may acts upstream of STAT3 positively regulating STAT3 expression and contributing to the amplification of the inflammatory response (Fig. 7A). Indeed, after HDAC11 overexpression, various pro-inflammatory cytokines, including IL6, CCL2, CCL5, CXCL6, and CXCL10 were elevated, while STAT3 knockdown significantly reduced the mRNA expression of CXCL6 (Fig. 7B). The reduction of CXCL6 protein expression in knockdown of STAT3 was confirmed using ELISA (Fig. 7D). Elevated total STAT3 and pSTAT3 Y705 was observed in OE HDAC11 whereas knockdown of STAT3 decreased both proteins (Fig. 7C). Moreover, the knockdown of STAT3, combined with overexpression of HDAC11, failed to induce either STAT3 or pSTAT3 Y705 (Fig. 7C), resulting in lower secretion of the CCL2, CXCL6, and CXCL10 in comparison to the OE HDAC11. Collectively, these data suggest that HDAC11 promotes the activation of the STAT3 pathway either by enhancing STAT3 transcription or by stabilising and facilitating its phosphorylation.

Fig. 7. Interaction of HDAC11 and STAT3 is required for sustained inflammatory response in iPS-ECs derived from DB donors with E. coli infection.** (A)** Efficient STAT3 knockdown and successful HDAC11 overexpression were achieved in iPS-ECs derived from DB donors. (B) STAT3 knockdown significantly reduced CXCL6 expression. HDAC11 overexpression increased the transcription of CCL2, CCL5, CXCL6, and CXCL10; however, STAT3 knockdown prior to HDAC11 overexpression attenuated this inflammatory response. (C) Western blot analysis confirmed STAT3 depletion and HDAC11 overexpression. HDAC11 overexpression led to increased total STAT3 and phosphorylated STAT3 at tyrosine 705 (pSTAT3 Y705). In contrast, STAT3 inhibition followed by HDAC11 overexpression resulted in reduced STAT3 and pSTAT3 Y705 levels. (D) STAT3 inhibition significantly decreased CXCL6 expression, whereas HDAC11 overexpression enhanced CXCL6 levels. However, STAT3 knockdown followed by HDAC11 overexpression failed to induce CXCL6 production, further indicating that STAT3 is required for HDAC11-mediated inflammatory activation. (E) Under physiological conditions, STAT3 is localized in both the nucleus and cytoplasm. HDAC11 overexpression promoted nuclear translocation of STAT3, coinciding with elevated HDAC11 levels. Imaging Details: Magnification: 63×; Scale bar: 100 μm. Statistical Significance: Statistical analysis was tested by Kruskal-Wallis or One-Way ANOVA, ✻ p < 0.05 vs. Ctrl; ✻✻ p < 0.01 vs. Ctrl; ✻✻✻✻ p < 0.0001 vs. Ctrl. Symbols: shSTAT3: Knockdown of STAT3 using shRNA.; OE HDAC11: Overexpression of HDAC11

HDAC11 has been shown to interact with STAT3 in bone marrow derived macrophages (BMDMs) and 293T cells [30]. In agreement with these previous observations, immunocytochemistry staining indicated that HDAC11 might interact with STAT3 in the nucleus in response to E. coli infection (Supplementary Fig. 7A) and overexpression of HDAC11 (Fig. 7E). The interaction between HDAC11 and STAT3 in DBE was also confirmed using immunoprecipitation in which STAT3 pulldown demonstrated an interaction with HDAC11 (Supplementary Fig. 7B). The reciprocal pulldown using HDAC11 antibody revealed binding to both STAT3 and pSTAT3 Y705 (Supplementary Fig. 8C). These results indicate that HDAC11 not only interacts with total STAT3 but also with its activated phosphorylated form.

Complementary experiments using STAT3 knockdown (Fig. 8A-E) and pharmacological inhibition of pSTAT3 (Supplementary Fig. 8A-B) showed that STAT3 plays an essential role in mediating the inflammatory response induced by HDAC11. Furthermore, HDAC11 overexpression significantly enhanced the expression of the inflammatory mediators accompanied with an elevation of pSTAT3 level. However, silencing STAT3 using shRNA and inhibiting pSTAT3 pharmacologically revealed attenuation of the inflammatory response. Collectively, these findings suggest that HDAC11 functions upstream of STAT3 by positively regulating STAT3 activation, which is in turn required to produce pro-inflammatory chemokines particularly upon E. coli infection of iPS-ECs from diabetic donors (Supplementary Fig. 9).

Discussion

This study sought to elucidate the contribution of HDAC11 to the inflammatory response observed in in iPS-ECs derived from DB donors subjected to E. coli infection. Using iPS-ECs from DB and ND donors, we demonstrate that exposure to E. coli elicits a significantly increased inflammatory phenotype in DB cells. This hyperinflammatory state was accompanied by a pronounced upregulation of HDAC11, which was further amplified by the infection. Mechanistically, we identify STAT3 as a central downstream effector of HDAC11, mediating sustained inflammatory signalling in iPS-ECs derived from DB donors. Together, these findings reveal that HDAC11 is a central regulator of infection-induced ECs inflammation in diabetes and highlight its potential as a therapeutic target to mitigate infection-driven vascular dysfunction.

iPS-ECs obtained from specific donors retain the genetic and epigenetic imprint of those individuals, providing a relevant platform to investigate human disease mechanisms. In this study, we demonstrated a substantial differential gene expression profile between control and infection groups. The iPS-ECs demonstrated broadly similar responses to E. coli infection. However, the iPS-ECs derived from DB donors showed high expression of TLRs, which are key sensors of bacterial components, and a subsequently amplified production of pro-inflammatory cytokines and chemokines. It is well-established that people with diabetes are prone to suffer from any infections due to suppression of the immune system and hyperglycaemia [43]. The exaggerated release of inflammatory cytokines and chemokines triggered by E. coli not only highlights the pro-inflammatory predisposition in diabetes but also suggests a potential link to the onset and progression of diabetes-associated vascular complications. The excessive cytokines production as observed in DBE group mirrors signatures associated with septic shock, a condition with disproportionately high mortality rates in people with diabetes.

Endothelial activation represented another key outcome of E. coli exposure, as evidenced by significant induction of VCAM-1, intracellular cell adhesion molecule (ICAM-1), and Selectins. Hyperglycaemia significantly increases the expression of VCAM-1, ICAM-1, and E-selectin resulting an increase the adherence of the neutrophils and leukocytes to the endothelial surface [44, 45]. Additionally, in lymphatic microvascular EC (LEC), the expression of VCAM-1 and ICAM-1 was notably increased after exposure to 100 ng/mL of LPS for 12 h, a response that become even more pronounced at higher LPS concentrations (10 µg/mL for 24 h), mediated by NFκB activation [46]. Comorbidities such as diabetes with infection, also exacerbating EC activation, might contribute to multiple organ failure and sepsis. Upon uncontrolled inflammation, as observed in sepsis, VCAM-1 is highly expressed and associated with in-hospital mortality [47]. This study showed that E. coli infection enhanced the activation of the EC, as obviously observed in the DBE. Besides, EC activation is associated with an increase in vascular permeability, allowing the infiltration of molecules like proteins into the interstitial space, resulting oedema and delaying wound healing [48].

Oxidative stress is a hallmark of diabetic vasculopathy. Activation of the polyol pathway increases nicotinamide adenine dinucleotide phosphate (NADPH) consumption and elevates ROS production, contributing to endothelial injury [49, 50]. Previous studies suggest that vascular organoids and iPS-ECs derived from DB donors exhibit increased ROS production and EC dysfunction [36, 51]. Our present work shows that infection with E. coli results in elevated production of ROS. Since DB already causes elevated ROS levels, the additional ROS produced during infection was significantly higher in the DBE iPS-ECs compared to those of the NDE and DB groups. Prolonged ROS generation results in sustained inflammation and tissue damage, which may also contribute to complications such as chronic wound healing issues and sepsis. A study by Joshi et al., shows that ROS level is produced after an hour post-infection (hpi) of uropathogenic E. coli in uroepithelial cells and reaches the highest peak at 3 hpi [52]. Furthermore, LPS-infected HUVEC exhibits higher ROS level at 24-hour post LPS treatment, as measured by mean fluorescence intensity [53]. Collectively, our findings suggested that persistent ROS generation drives inflammation and tissue damage, exacerbating the endothelial dysfunction seen in chronic diabetic complications.

Lactate dehydrogenase (LDH) is released into the extracellular fluid and is a classical marker of membrane damage and cellular stress [53]. Elevated serum LDH level is commonly observed in individuals with DB, and this level correlates with BMI [54], glycated albumin [55], and insulin antibodies [55]. Furthermore, elevated LDH levels are associated with DB complications, including diabetic kidney disease (DKD) [56]. In our study revealed that the LDH level was higher in iPS-ECs derived from DB than in ND donors, with the difference being more pronounced in the DBE compared to both DB and NDE groups. Exposure of EC to LPS and cytokines increased the release of LDH which might indicate EC damage. An in vitro study using HUVECs exposed to 40ng/mL of TNFα indicated an upregulation of LDH in parallel with the production of pro-inflammatory cytokines such as IL1β, IL6, and IL18 [22]. Furthermore, HUVECs treated with different concentration of LPS (0.5, 1, 2, and 4 µg/mL) for 24 h exhibited a significant increase in LDH secretion [53]. Excessive cytokine production and ROS generation likely create a feed-forward loop that exacerbates LDH release and endothelial injury.

Endothelial dysfunction, characterised by impaired angiogenesis, reduced NO bioavailability, and increased vasoconstriction, is a major contributor to diabetic vascular complications [57, 58]. Consistent with previous research, iPS-ECs derived from DB donors exhibited an inability to form tube-like structures and reduced tube length. Infection further worsened these defects, driven in part by inflammatory cytokines such as IL-6 and TNF, which disrupt endothelial vasodilation and promote oxidative stress [59]. Ultimately, these results illustrate how chronic inflammation, oxidative stress, and endothelial activation converge to drive severe dysfunction in the diabetic vasculature.

HDAC11 regulates immune cells [28, 29, 60] and metabolic diseases [24]. However, the mechanism of HDAC11 regulating inflammation remains elusive. Upon exposure to cytokine stimulation, ECs show a significant upregulation of HDAC11, which is followed by an increase in pro-inflammatory cytokines production [22]. This study shows that E. coli stimulation significantly enhanced HDAC11 expression, particularly in the DBE group. Inhibition of HDAC11, either via shRNA or the selective inhibitor SIS17, suppressed inflammatory cytokine production, reduced LDH and ROS levels, and improved endothelial function.

Our data also indicate that HDAC11 contributes to the global balance between histone acetylation and deacetylation. HDAC11 contains functional deacetylation motifs which are similar to those in class I and class II HDACs [18]. HDAC11 knockdown reduced overall HDAC activity and increased HAT activity, while HDAC11 overexpression produced the opposite effect. These findings agree with previous reports showing increased H3 and H4 acetylation after HDAC11 loss [28, 30]. Moreover, increased HDAC11 expression was associated with a global reduction of H3K27ac in retinal pigmented epithelial (RPE) cells from age-related macular degeneration (AMD) [61]. These epigenetic shifts agree with studies demonstrating that altered histone acetylation promotes inflammatory cytokine production and vascular dysfunction in diabetes [62, 63].

The HDAC11 upregulation was associated with increased levels of STAT3, a transcription factor known to play a critical role in EC proliferation and maintaining endothelial barrier integrity. STAT3 is activated by phosphorylation at the tyrosine 705 (pSTAT3 Y705) or serine 727 (pSTAT3 S727), enabling it to perform its functions. We found that the pSTAT3 Y705 significantly increased in the DBE group, which coincides with increased inflammation. STAT3 regulates the transcription of genes involved in inflammation [64]. After phosphorylation, pSTAT3 dimerizes and translocate to the nucleus activating the transcription of multiple genes involved in inflammatory responses. Previous research has documented the interaction between HDAC11 and STAT3 in BMDM infected with Candida albicans [30]. We found that inhibition of HDAC11 using drug inhibitor and gene manipulation results in the reduction of STAT3 and pSTAT3 Y705 in the iPS-ECs derived from DB donors stimulated with E. coli, suggesting the interaction of HDAC11 between STAT3 and pSTAT3 Y705 in iPS-ECs derived from DB donors.

A major mechanistic insight from this study is the identification of STAT3 as a direct effector of HDAC11. Prior to this study, it was known that HDAC11 can repress negative regulators such as PU.1 [60]. Second, HDAC11 may reinforce a positive feedback loop involving STAT3 signalling, which would sustain inflammation by inhibiting NOS2 [30]. Our evidence also shows that HDAC11 also interacts with both STAT3 and pSTAT3, potentially stabilising its active form and prolonging the inflammatory response. Together, we propose a hierarchical regulatory model in which HDAC11 enhances inflammatory responses by upregulating and activating STAT3, thus linking epigenetic memory to inflammation.

These findings also complement the broader concept of epigenetic memory in diabetic vasculature. Diabetic vascular organoids exposed to chronic hyperglycemia exhibit persistent elevation of inflammatory cytokines and anti-angiogenic factors, even after glycaemia normalisation [36], reflecting stable epigenetic programming. Our data suggest that HDAC11-mediated alterations in STAT3 activation contribute to this inflammatory imprint.

From a translational viewpoint, the HDAC11–STAT3 axis represents a promising therapeutic target. Changing epigenetic pathways has become a promising technique to treat diseases, including bromodomain and extra-terminal domain (BET) inhibition. Blocking BET protein family has been shown to improve human aortic endothelial cells (HAECs) function by reducing anti-angiogenic molecules, restoring VEGFA expression, and lowering ROS production [62]. Targeting BETs improves cardiac function in human pluripotent stem cells-derived cardiac organoid (hCO) exposed to SARS-CoV-2. Phosphoproteome analysis revealed increased STAT1 S727 phosphorylation and BRD4 activation in SARS-CoV-2-infected hCOs, suggesting that these epigenetic and transcriptional regulators contribute to the cytokine storm and inflammatory damage which can be restored by administration of BETs inhibitor [63].

Despite these advances, further work is needed to fully define how HDAC11 shapes endothelial epigenetic landscapes during infection. Time-resolved studies such as Assay for Transposase-Accessible Chromatin using sequencing would help delineate HDAC11-dependent regulatory changes. Additionally, in vivo diabetic models are essential for evaluating how HDAC11 influences endothelial–immune interactions, cytokine dynamics, and vascular health under physiologically relevant conditions. These models could also provide mechanistic insight into how targeting HDAC11 ameliorates endothelial dysfunction, reduces inflammatory cytokine production, and improves vascular health during pathogenic challenges.

In summary, diabetic ECs exhibit enhanced sensitivity to E. coli infection, driven by HDAC11 upregulation, altered acetylation balance, and STAT3 hyperactivation. Inhibition of HDAC11 effectively reduces oxidative stress, cytokine production, and endothelial dysfunction. These findings establish HDAC11 as a key epigenetic regulator linking diabetes to infection-induced vascular inflammation and highlight its potential as a therapeutic target to protect the diabetic vasculature during infectious challenges.

Supplementary Information

Supplementary Material 1.

Supplementary Material 2.