Downregulated calmodulin expression contributes to endothelial cell impairment in diabetes

Abstract

Endothelial dysfunction, a central hallmark of cardiovascular pathogenesis in diabetes mellitus, is characterized by impaired endothelial nitric oxide synthase (eNOS) and NO bioavailability. However, the underlying mechanisms remain unclear. Here in this study, we aimed to identify the role of calmodulin (CaM) in diabetic eNOS dysfunction. Human umbilical vein endothelial cells and murine endothelial progenitor cells (EPCs) treated with high glucose (HG) exhibited downregulated CaM mRNA/protein and vascular endothelial growth factor (VEGF) expression with impeded eNOS phosphorylation and cell migration/tube formation. These perturbations were reduplicated in CALM1-knockdown cells but prevented in CALM1-overexpressing cells. EPCs from type 2 diabetes animals behaved similarly to HG-treated normal EPCs, which could be rescued by CALM1-gene transduction. Consistently, diabetic animals displayed impaired eNOS phosphorylation, endothelium-dependent dilation, and CaM expression in the aorta, as well as deficient physical interaction of CaM and eNOS in the gastrocnemius. Local CALM1 gene delivery into a diabetic mouse ischemic hindlimb improved the blunted limb blood perfusion and gastrocnemius angiogenesis, and foot injuries. Diabetic patients showed insufficient foot microvascular autoregulation, eNOS phosphorylation, and NO production with downregulated CaM expression in the arterial endothelium, and abnormal CALM1 transcription in genome-wide sequencing analysis. Therefore, our findings demonstrated that downregulated CaM expression is responsible for endothelium dysfunction and angiogenesis impairment in diabetes, and provided a novel mechanism and target to protect against diabetic endothelial injury.

Introduction

Diabetes mellitus is a chronic metabolic disorder, affecting more than 536.6 million people worldwide [1]. Vascular endothelial dysfunction and injury, characterized by defective endothelial nitric oxide synthase (eNOS) and NO production, and impaired vasodilation, are the hallmarks and predictors of diabetic cardiovascular diseases [2, 3]. These disturbances gradually progress and eventually lead to tissue ischemia and cardiovascular complications such as diabetic kidney disease, heart disease, and peripheral arterial diseases, accounting largely for the high rates of disability and mortality in diabetes patients [2–4]. Although concerted efforts have been made to develop treatments such as hypoglycemia as well as lipid-lowering, anti-oxidation, anti-inflammation, and anti-thrombosis therapies to prevent and treat diabetes complications, the mortality and mobility of diabetes complications are still increasing at a dangerous speed. Therefore, the molecular mechanism(s) underlying vascular endothelial injury and more effective treatment for diabetic cardiovascular complications are urgently needed.

In addition to the recognition that oxidation and inflammation play a fundamental role in eNOS dysfunction and endothelial impairment [3–6], alterations in several signaling pathways have been observed in association with eNOS dysfunction in diabetes, which is the initial step of vascular endothelial injury. These include: (1) high-glucose-induced protein kinase C activation, leading to insulin receptor substrate 1 phosphorylation, and reduced activation of phosphatidylinositol 3-kinase/Akt/eNOS [4, 7]; (2) oxidative stress and dysregulated lipid metabolism under hyperglycemia impair the activity of AMP-activated protein kinase, thereby inducing eNOS uncoupling [8, 9]; (3) increased Wnt5a signaling and JNK activity contribute to eNOS dysfunction [10]; (4) promoted expression of protein phosphatase 2 A decreases the activity of calmodulin kinase II (CaMKII) by reducing its phosphorylation at threonine 286 (Thr286), and subsequently eNOS dephosphorylation in vessels of type 2 diabetic rats [11, 12]. On the contrary, enhanced CaMKII activation due to highly activated oxidation reactions has been observed in diabetic heart, islets, and liver, and it is involved in metabolic disorders and related tissue injuries [13–15]. Physiologically, CaMKII activation depends on CaM binding with intracellular Ca²⁺. Additionally, by interacting with eNOS, activated CaM directly controls ~80% of eNOS phosphorylation at serine 1177 (S1177) and NO production in animals and humans [16–18]. Moreover, Ca²⁺/CaM signaling is closely associated with the activation of hypoxia-inducible factor-1 (HIF-1), and consequently the expression of pro-angiogenic factors such as vascular endothelial growth factor (VEGF) [19–21]. Therefore, by cross-talking with multiple signaling pathways, CaM plays a pivotal role in maintaining physiological homeostasis of the vascular endothelium and angiogenesis.

CaM, containing four EF-hand Ca²⁺-binding sites, participates in virtually all Ca²⁺-dependent intracellular processes [22, 23]. Upon Ca²⁺ binding to the EF hands, CaM undergoes a conformational change that exposes hydrophobic patches on each domain, which facilitates its interaction with target proteins [18, 23]. Recently, reduced CaM protein expression has been observed in obese diabetic mouse liver; CaM overexpression ameliorated hyperglycemia and fatty liver in diabetic animals [24, 25], providing a clue for CaM involvement in diabetic metabolism. Herein, we investigated whether CaM-associated regulation is involved in diabetic endothelial dysfunction and angiogenesis impairment. In vivo, ex vivo, and in vitro studies demonstrated that downregulated CaM expression was responsible, at least partly, for the impairments in diabetic vascular endothelium and endothelial progenitor cells (EPCs).

Materials and methods

Human tissue samples

Human colonic arteries were collected from the abandoned paracancerous tissue of diabetic patients and non-diabetic patients after surgical excision in the Cancer Hospital, Chinese Academy of Medical Sciences (Beijing, China). Volunteers of healthy subjects and hospitalized type 2 diabetes patients were recruited from Friendship Hospital, Affiliated Capital Medical University for the evaluation of post occlusive reactive hyperemia. The sample’ usage was consented by the patients and approved by the Research Ethics Committee of the Cancer Hospital (2021010711291302) and Capital Medical University (AEEI-2015-045).

Animals

Male C57BL/6 J (C57), C57BLKS/J lar-m/lepr^db (db/m) and C57BLKS/J lar-Lepr^db/Lepr^db (db/db) mice (all at 10–12-week-old) were purchased from Peking University Health Science Center (Beijing, China) and Beijing Vital River Laboratory Animal Technology Co., Ltd (Beijing, China). Male Goto-Kakizaki (GK) and Wistar rats (12-week-old) were purchased from Shanghai SLAC Laboratory Animal Co., Ltd (Shanghai, China). All animal handling procedures were approved by Animal Care and Use Committee of Capital Medical University (AEEI-2015-045), and in accordance with the Guide for the Care and Use of Laboratory Animals published by the US National Institutes of Health (NIH Publication No. 8523, revised 2011). Animals were raised in the Experiment Animal Center, Capital Medical University (Grade SPF), and kept in a 12-h light-dark cycle room with temperature control. To avoid sex impact, male mice were used and counted for evaluation and statistical analysis in this study.

Post occlusive reactive hyperemia (PORH)

Microvascular autoregulation was assessed by the means of PORH using single-point laser Doppler flowmetry (Periflux System 5000, 780 nm laser diode, Perimed, Järfälla, Sweden) [26]. An appropriate size cuff was placed at the patients’ big toe joint of right lower extremity. After the rest flow (RF) at baseline for three-minute time period recording, blood flow was occluded for 3 min by inflating an appropriately sized blood pressure cuff placed around the big toe to 200 mmHg to completely block the blood flow. The cuff was then released suddenly, and the data were collected by the PeriSoft 2.5.5 program. The maximum microvascular perfusion value was defined as peak flow [PF]. The time span between cuff deflation and peak perfusion was defined as time to peak [TM]. Recovery time [TR] was characterized as time span between peak perfusion and reacquisition of mean rest flow perfusion values, and respective time spans were determined using the device-related software.

Statistical analysis

All data are expressed as the mean ± SD. For unpaired observations, independent two-tailed Student’s t test was performed, and one-way ANOVA followed by post-hoc Tukey’s multiple comparison test (normally distributed) or Kruskal-Wallis followed by Dunn’s multiple comparison’s test (non-normally distributed) were used for multiple groups. Statistically significant differences between groups were defined as P < 0.05.

Results

Downregulated CaM expression contributes to dysfunctional eNOS phosphorylation and tube formation in high-glucose-treated HUVECs

Human umbilical vein endothelial cells (HUVECs) cultured under high glucose (HG) conditions for 48 h were used to assess endothelial cell dysfunction due to diabetes. These cells demonstrated reduced eNOS phosphorylation at S1177 (p-S1177), S116, and S633 (all positive regulators of eNOS), and enhanced phosphorylation at threonine 495 (p-T495, a negative regulator of eNOS). Angiogenesis-related proteins’ expression including VEGF, VEGFR2, p-Akt, CaMKII-pT286, and CD34, and functional performance such as NO level and cell migration/tube formation were all dampened in HUVECs exposed to HG, compared with isotonic treatment (Fig. 1a–e). Noticeably, HG exposure significantly lowered the expression of CaM mRNA and protein, compared with the control (Fig. 1a–c). Furthermore, interference with CaM expression or function by CaM-siRNA or N-(6-aminohexyl)-5-chloro-1-naphthalenesulfonamide (W7), a CaM inhibitor, mimicked the changes induced by HG. In contrast, introduction of human CALM1 gene (CaM-wt) via plasmid transfection prevented the downregulation of eNOS-p-S1177 and VEGF and the functional disorders that occurred in HG-treated HUVECs (Fig. 2 and Supplementary Fig. S1). These beneficial effects of CaM-wt were blocked by pretreatment of the cells with the NOS inhibitor L-NAME (Supplementary Fig. S2), suggesting an eNOS-dependent mechanism.

Fig. 1. High-glucose treatment induced downregulated eNOS phosphorylation and CaM expression with impaired function in HUVECs.

HUVECs were cultured in medium containing normal glucose (NG), NG+mannitol (as isotonic control, Isot) or HG (33 mM) for 48 h. The expression levels of CaM, VEGF, VEGFR2, CD31, CaMKII-pT286, CaMKII, p-Akt, Akt, p-eNOS (p-S1177, p-S116, p-S633 and p-T495), and eNOS were detected by specific antibody (a). Quantification analysis of phosphorylated eNOS levels in relative to eNOS, and CaM, VEGF, VEGFR2, CD34 and p-Akt expression was normalized by β-actin, and then normalized to the value in NG group (b). Downregulated CaM mRNA expression and NO generation due to HG treatment were detected by real-time PCR and NO assay kit, respectively (c). HUVEC function, represented by cell migration and tube formation, were evaluated using scratch-wound migration and a Matrigel-based capillary tube formation, respectively. The distance between two sides of scratch, and the numbers of tubes were quantified after 24 and 48 h culture as indicated (d and e). All scale bars: 200 μm. Data are mean ± SD from independent experiments indicated with n in each panel. *P < 0.05, **P < 0.01 vs. Isot with Student’s t test performance as indicated.

Fig. 2. High-glucose treatment and knocking down CaM induced downregulated eNOS phosphorylation and CaM expression with impaired function in HUVECs.

HUVECs in NG medium were transfected with CaM-siRNA or Ctrl-siRNA (a select negative control) for 6 h, and further cultured in NG medium for 24 h. Cell lysates were used to detect the expression of CaM, VEGF, and phosphorylation levels of CaM at Tyr138, CaMKII and eNOS-S1177, and the phosphorylation levels of eNOS, CaMKII and CaM normalized by eNOS, CaMKII and CaM, respectively, as well as the expression of VEGF and CaM in relative to β-actin. All data were normalized to the Ctrl-siRNA group (a). Tube formation was assessed by Matrigel angiogenesis assay in HUVECs transfected with CaM-siRNA at 4 and 24 h culture as indicated (b). HUVECs were transfected with a plasmid carrying human CALM1 gene (CaM-wt) or vector for 4 h, and then cultured either in NG or HG medium for 48 h. The expression of CaM, VEGF, and phosphorylation levels of CaMKII and eNOS-S1177 were detected. Quantification analysis of eNOS and CaMKII phosphorylation, and VEGF and CaM expression was normalized to the Ctrl-vector (c). Tube formation was assessed in HUVECs with different treatment as indicated, and tube numbers were quantified at 48 h after transfection (d). All scale bars: 200 μm. Data are mean ± SD from independent experiments indicated with n in each panel. *P < 0.05, **P < 0.01 vs. CaM-siRNA with Student’s t test performance; or vs. CaM-wt with one-way ANOVA test performance, as indicated.

Downregulated CaM expression contributes to impaired eNOS phosphorylation and tube formation in high-glucose-treated EPCs

EPCs have the potential to proliferate and migrate into a disrupted endothelium and differentiate into endothelial cells to maintain endothelium integrity and function upon endothelial injury [27, 28]. They also repair damaged tissues by promoting angiogenesis under ischemic conditions [27–29]. However, pathological conditions such as high glucose, high lipid, and oxidation significantly affect EPC quantity and function, but the underlying mechanism remains unknown. In this study, we isolated EPCs from bone marrow of normal and diabetic mice and rats. The basic characteristics of EPC cultures that indicated a successful EPC isolation are: spindly peripheral cells after 3 days, spindle-shaped adherent early EPCs after 7 days, and late EPC cobblestone-like morphology by 14 days in culture, as well as ~95% cells positive for CD34, CD31, or VEGFR2 immunostaining (Supplementary Fig. S3).

Like the HUVECs, normal mouse and rat EPCs treated with HG for 48 h demonstrated eNOS dephosphorylation at S1177, downregulation of CaM expression and CaMKII phosphorylation, and disordered tube formation (Fig. 3a–d). CaM expression knockdown using CaM-siRNA in rat and mouse EPCs exhibited similar alterations as HG treatment (Fig. 3e–g, Supplementary Fig. S4). Therefore, these findings demonstrated that reduced CaM expression was associated with dysfunctional eNOS and angiogenesis in HG-treated cells.

Fig. 3. High-glucose treatment and knocking down CaM induced downregulated eNOS phosphorylation and CaM expression with impaired function in normal endothelial progenitor cell (EPCs).

EPCs were isolated from bone marrow of normal mice and rats. The adherent EPCs were cultured in NG+mannitol or HG for 48 h. The expression of eNOS-pS1177, CaMKII-pT286, eNOS, CaMKII and CaM was detected by Western blot, and quantification analysis of the phosphorylation levels of eNOS and CaMKII was normalized to their basal protein, and then normalized to the level of NG group in mouse or rat EPCs (a and c). Tube formation was assessed by Matrigel angiogenesis assay in different EPC groups as indicated, and the numbers of tubes were quantified after 48 h culture (b and d). Rat EPCs in NG medium were transfected with CaM-siRNA or Ctrl-siRNA for 6 h, and further cultured in NG medium for 24 h. Cell lysates were used to detect the expression of CaM, Akt and VEGF, and phosphorylation levels of CaMKII, Akt and eNOS-S1177 (e). Quantification analysis of p-eNOS, p-Akt and CaMKII-pT286 levels relative to eNOS, Akt and CaMKII, respectively, and the expression of VEGF, Akt, eNOS and CaM normalized by β-actin, and then all data were normalized by the Ctrl-siRNA group (f). Angiogenic function of EPCs was assessed by a Matrigel-based capillary tube formation (g). Scale bars: 200 μm. Data are mean ± SD from independent experiments indicated with n in each panel. *P < 0.05, **P < 0.01 vs. HG or CaM-siRNA with Student’s t test performance.

Downregulated CaM expression contributes to endothelial cell impairment in diabetic animals

To verify whether the CaM change was associated with dysfunctional diabetic endothelial cells in vivo, we used two genetic animal models of type 2 diabetes; obese db/db mice and non-obese Goto-Kakizaki (GK) rats (Supplementary Table S1). Immunohistochemistry of animal aortas with anti-CD31, and double-immunostaining with anti-CD34 and anti-CaM antibodies revealed intermittent intima defects with decreased expression of CD31, CD34, and CaM in both diabetic aortas, compared with the respective controls (Ctrl); wild db/m mice and Wistar rats (Fig. 4a–d). The co-immunoprecipitation assay exhibited reduced eNOS interaction with CaM in diabetic gastrocnemius (Fig. 4e, f). Accordingly, NO formation and both the aortic tension natural decline and acetylcholine-dependent vasodilation after constriction induced by phenylephrine (PE) were blunted, which was accompanied by a significant increase in eNOS phosphorylation at T495 and a decrease at p-S1177, p-S633, and p-S116 in db/db mice and GK rats, compared with their respective controls (Supplementary Fig. S5). Similar to the HG-treated normal EPCs (Fig. 3), the isolated EPCs from db/db and GK animals exhibited impaired eNOS phosphorylation and tube formation with decreased expression of CaM mRNA/protein and angiogenesis-related proteins. Furthermore, transfection of CaM-wt into cultured diabetic EPCs recovered eNOS phosphorylation, NO production, VEGF and VEGFR2 expression, and tube formation (Fig. 5). To identify the role of CaM in neovascularization, we performed the ex vivo matrigel plug assay, which mainly detects capillary-like blood vessels and can calculate the relative neovascularization level [30]. After planning for 14 days in a normal mouse back hypodermic skin, the matrigel plugs filled with adenovirus vectors (vector-Ad), or adenoviruses carrying CaM-shRNA or trifluoperazine (TFP), a CaM inhibitor at a concentration of 10 μM, displayed less neovascularization inside the plugs with CaM-shRNA and TFP than those in the controls. In contrast, intravenous injection of adeno-associated virus-9 (CaM-AAV, 1 × 10¹² vg/mL) into the db/db mouse tail induced more neovascularization, compared with vector-AAV (Supplementary Fig. S6).

Fig. 4. Aortic endothelial injury with reduced CaM expression in diabetic animals.

Tissue sections from thoracic aorta of db/db mice and GK rats, and counterpart control db/m mice and Wistar rats (Ctrl), were stained with antibody against CD31. The representative images showed obvious deficient CD31 in diabetic aortic intima (a and b). Aorta sections were double stained with antibodies against CD34 (red) or vWF (red) and CaM (green) as indicated, and the representative images exhibited less expression of CD34 and CaM in diabetic intima compared to those in counterpart controls (c and d). Scale bars: 50 μm, and n = 5 animals for each group. Gastrocnemius tissue lysates were immunoprecipitated with antibody against eNOS, and less CaM binding with eNOS was found in both db/db mice and GK rats compared to respective controls (e and f). Number per group: 4–6 animals.

Fig. 5. CALM1-gene transduction improved diabetic endothelial progenitor cell (EPCs) function via activating eNOS/NO/VEGF pathway.

Isolated EPCs from bone marrows of db/m and db/db mice, and Wistar and GK rats were transfected with vector or human CaM-wt for 4 h, then further cultured for 48 h. The expression levels of eNOS phosphorylation at S1177, S116 and S633, eNOS, p-Akt, Akt, CaMKII-pT286, CaMKII, CaM, VEGF and VEGFR2 were detected by Western blot. Quantitative analysis of p-eNOS in relative to eNOS, CaM and other proteins in relative to β-actin was normalized to those in control group (Ctrl) (a, b, g and h). NO levels in medium from different groups of EPCs were measured by nitrate reductase method (c and i). CaM mRNA expression was determined by real-time PCR (d and j). Angiogenesic function of EPCs was assessed by Matrigel tube assay, and the numbers of formed tubes were compared between different groups (e, f, k and l). Scale bars: 200 μm. Data are mean ± SD from independent experiments indicated with n in each panel. *P < 0.05, **P < 0.01 compared among groups as indicated with one-way ANOVA test performance.

Taken together, these results suggest that downregulation of CaM expression is likely responsible for the injuries of arterial endothelium and EPCs, two characteristics of diabetic vascular pathologies [2–4, 16, 31, 32].

CALM1 gene therapy improves the injured ischemic limb in diabetic mice

To confirm the association of CaM deficiency with impaired diabetic angiogenesis, the effect of CALM1 gene expression on a diabetic foot model that directly manifests ischemic pathologies [29, 30] was evaluated. CaM-AAV was injected into the gastrocnemius one week before a complete ligation of the femoral artery in db/db mice. The GFP florescence detected in the vessels of gastrocnemius muscle indicated a successful transfection of AAV (Supplementary Fig. S7). Diabetes mice displayed no difference in general blood examinations including glucose and lipid levels, and blood pressure, except for a partial recovery of the NO level compared with the vector (Ctrl) mice (Supplementary Table S2).

Blood perfusion examination showed a slower and incomplete blood flow recovery in diabetic ischemic limb compared with that of non-diabetic ischemic limb. Notably, local CaM-AAV transduction significantly enhanced the blood flow in ischemic diabetic foot, which was comparable to the recovery in normal mouse ischemic limb, whereas mice transfected with vector-AAV had much slower recovery (Fig. 6a, b). Additionally, the ischemic damage and dysfunction, including ulcer/necrosis and lameness, were ameliorated in CaM-AAV treatment group, compared with the vector group (Fig. 6c, d, and Video 1). The ischemic gastrocnemius atrophy, characterized by decreased weight, reduced contractible area, and muscle degeneration with increased fibrosis and decreased dystrophin areas, was also markedly rescued by CALM1 gene delivery (Fig. 6e, f, and Supplementary Fig. S8). Importantly, CaM-AAV treatment significantly improved the impaired eNOS-S1177 phosphorylation and NO formation, and the reduced CaM, VEGF, VEGFR2, CD34, and CD31 expression, along with an increase in the capillary density in ischemic diabetic muscle (Fig. 7a, b, and Supplementary Fig. S9). However, CaM expression in the diabetic liver, kidney, and aorta was unaltered by the injection of CaM-AAV into the gastrocnemius (Supplementary Fig. S10). Additionally, the impaired interaction of CaM and eNOS was recovered by CaM-AAV delivery into ischemic diabetic gastrocnemius compared with that in the vector-treated muscle (Fig. 7c).

Fig. 6. CALM1-gene transduction improved blood perfusion, ischemic limb necrosis and claudication in db/db mice with HLI.

C57BL/6 (Ctrl) and db/db mice were administered with intromuscular injection of vector (Ctrl-AAV) or CaM-AAV for 6 days and then subjected to HLI. The plantar perfusion was monitored at before (0) and 1, 7, 14, and 21 days after HLI using a Pericam Perfusion Speckle Imager. Images and line graph show the time course of blood flow recovery (calculated as the ratio between ischemic (IS) and non-ischemic foot blood flow (NI) in Ctrl and db/db mice, respectively (a and b). Representative images indicate the scoring generated for the degree of planta pedis with ulcers, and the mean limb damage scores and incidences were assessed using this semi-quantitative scoring system (c). Therapeutic effect of CALM1 transfer on ambulatory impairment compared to vector-treated diabetic mice was presented using the semi-quantitative scoring system for lameness (d). Representative images of congenic contralateral non-ischemic limb (R, NI) and ischemic limb (L, IS) gastrocnemius muscle in WT (Ctrl) and db/db mice (e). Quantification analysis of the weight and superficial area (calculated as the ratios between ischemic muscle weight/cross section area (IS) and contralateral non-ischemic muscle weight/cross section area (NI) in WT mice and db/db mice (f). Data are mean ± SD from independent experiments indicated with n in each panel. *P < 0.05, **P < 0.01 compared among groups as indicated with one-way ANOVA test performance.

Fig. 7. CALM1-gene transduction enhanced eNOS/NO/VEGF activity in db/db mice with HLI.

C57BL/6 (Ctrl) and db/db mice were administered with intromuscular injection of vector (Ctrl-AAV) or CaM-AAV for 6 days and then subjected to HLI. The expression levels of HIF-1α, CaM, VEGF, VEGFR2, CD31, CD34, CaMKII-pT286, CaMKII, eNOS, and the eNOS-pS1177 in gastrocnemius of C57BL/6 and db/db mice were detected by Western blot (a). Quantitative analysis of these protein level relative to GAPDH was normalized to Ctrl-Sham, and the quantitative analysis of p-eNOS expression relative to eNOS level was normalized to the Ctrl-Sham group (b). The gastrocnemius NO levels were evaluated by nitrate reductase assay (b). CaM level in vascular endothelium was detected by immunoprecipitation using anti-eNOS antibody in gastrocnemius lysates, and quantification of CaM relative to eNOS was normalized to Ctrl-Sham (c). Data are mean ± SD from independent experiments indicated with n in each panel. *P < 0.05, **P < 0.01 compared among groups as indicated with one-way ANOVA test performance.

Endothelial dysfunction and decreased CaM expression in diabetic patients

Finally, we assessed the function of the vascular endothelium in hospitalized volunteer patients suffering from type 2 diabetes in Friendship Hospital, Capital Medical University, and healthy volunteer subjects by measuring their foot microvascular autoregulation using post occlusive reactive hyperemia (PORH), an approach representing a regulated vascular tone by fluid shear stress on the endothelium [33, 34]. Their general serum parameters are presented in Supplementary Table S3. Diabetic patients showed a blunted perfusion recovery, represented by the reduced peak of blood flow and extended recovery time, compared with normal subjects (Fig. 8a, b). Moreover, colonic arteries from paracancerous tissues of diabetic patients exhibited decreased eNOS-p-S1177 and serum NO levels compared with those in non-diabetic arteries (Fig. 8c–e). Immunostaining evaluations showed hypertrophic and defective intima with reduced CaM expression in diabetic arteries, compared with non-diabetic arteries (Fig. 8f, g). Searching for CALM gene changes in human diabetes in the National Center for Biotechnology Databases (https://www.ncbi.nlm.nih.gov/), we found that both CALM1 and CALM2 genes were downregulated, but CALM3 gene transcription was upregulated in type 1 and type 2 diabetic patients, compared with healthy subjects (Supplementary Fig. S11).

Fig. 8. Diabetes impaired vascular endothelial-associated relaxation and caused reduced CaM expression in patient artery.

Representative images of lower limbs skin microvascular flow measured by PORH in type 2 diabetic patients and healthy subjects (a), and quantification of the values [peak flow (PF) - resting flow (RF)]/RF × 100% and the recovery time (b) was described in Methods. Western blot analysis shows decreased eNOS-pS1177 level in colonic arteries of type 2 diabetic patients compared with non-diabetic subjects (c and d). NO levels in plasma of type 2 diabetic patients and non-diabetic subjects were analyzed using nitrate reductase assay (e). Representative images of colonic artery sections stained with H&E (f), or double stained with vWF (red) and CaM (green) antibodies (g). Nucleus were detected by DAPI (blue). Two-head arrows indicate intima hyperplasia, the white arrows indicate detectable CaM distribution in intima of non-diabetic artery, while the red arrows indicate disappearance of CaM distribution in diabetes artery. Scale bars: 200 μm (H&E) and 100 μm (IF), n = 5 samples for both groups. Data are mean ± SD from independent experiments indicated with n in each panel. *P < 0.05, **P < 0.01, ***P < 0.001 between two groups as indicated with Student’s t test performance.

Discussion

Dysfunctional eNOS and NO deficiencies are the hallmark traits and predictors of vascular endothelium disorder and injury in diabetes. Protection of eNOS activity and NO level is one of the important therapeutic strategies against diabetes disability and mortality [4, 5, 28, 30]. In the present study, reduced expression of CaM mRNA and protein was associated with the dysregulated eNOS phosphorylation/NO production, and with dysfunctional endothelial cells and angiogenesis in diabetes (Figs. 1–5, and Supplementary Figs. S1−S5). Transduction of the human CALM1 gene significantly improved diabetic endothelial cell function, ischemic limb blood flow, angiogenesis, and tissue injuries along with enhanced eNOS phosphorylation, NO formation, and VEGF expression (Figs. 6 and 7, and Supplementary Figs. S6−S10). Notably, diabetic patients also exhibited reduced CaM expression in the arteries and lowered CALM1 and CALM2 gene levels compared with those in healthy subjects (Fig. 8 and Supplementary Fig. S11), suggesting the clinical relevance of this study. Interestingly, the CALM3 gene level is significantly higher in diabetic patients than in normal subjects (Supplementary Fig. S11). This subtype’s transcripts are higher than the CALM1 and CALM2 gene levels in human heart [35, 36]. As Ca²⁺/CaMKII is upregulated and significantly contributes to diabetic cardiomyopathy [14, 15, 37], it is possible that the upregulated CALM3 gene is likely associated with diabetic myocardium injury.

CaM, an important Ca²⁺-binding signaling protein, plays an essential role in regulating various cellular functions [17–19, 22–25]. In the vascular endothelium, CaM promotes eNOS phosphorylation and ~80% of NO generation. Loss of bioavailable NO not only leads to vasodilation impairment, but also causes augmented oxidative stress owning to decreased NO-dependent free radical purging and antioxidant/antiinflammatory functions, which accounts largely for diabetic vasopathology [3, 6, 16, 38]. Additionally, CaM positively modulates angiogenesis through direct and indirect promotion of HIF-1 and VEGF functions [19–21, 38]. Therefore, CaM as a signal nexus plays an important role in keeping endothelium and angiogenesis homeostasis through crosstalk with multiple signaling pathways [5, 8, 17–20]. It has been noticed that the total amount of CaM in cells is lower than that of its target proteins, making it a limiting factor for downstream functions [39, 40], thereby lower CaM expression than physiological requirements may pose a high risk in the pathogenesis of the endothelium and vessels. Recently, a number of studies have displayed a reduced level of CaM in vascular smooth muscles and liver in diabetic animals [24, 25, 41], with beneficial effects of CALM1 gene expression on insulin resistance, hyperglycemia, and fat liver in type 2 diabetes [24, 25]. However, the relationship between CaM changes and dysfunctional vascular endothelium, as well as the circulating EPC impairments in mobilization, migration, and homing in the context of diabetes, remains unclear [7, 14, 16, 42–44]. In this study, EPCs under diabetic or HG conditions exhibited substantial downregulated CaM mRNA/protein expression and impaired downstream signaling molecules including eNOS and CaMKII phosphorylation and VEGF expression, as found in HG-treated HUVECs. Exogenous delivery of the CALM1 gene to cultured diabetic EPCs reversed these pathological changes; in particular, obvious therapeutic effects of CALM1 gene expression were observed in the local ischemic tissue by boosting angiogenesis, restoring blood flow, and reducing tissue necrosis in the diabetic foot. Therefore, these results demonstrated that the reduced CaM expression most likely contributes to the angiogenesis impairment in diabetic ischemic injuries. Thus, local delivery of the CALM1 gene into the ischemic limb may be a potential choice for the treatment of a diabetic foot.

In addition to the CaM/eNOS pathway alteration, dephosphorylated Akt and CaMKII, and suppressed HIF-1α, VEGF, and VEGFR2 expression have been observed in HG-treated cells and in diabetic animal EPCs [4–7, 11, 12]. CALM1 gene transduction significantly counteracted all these changes, except for the suppressed Akt expression and hypophosphorylation in HG-treated cells and diabetic EPCs (Fig. 5). Inhibition of HIF activity by its inhibitor significantly affected Akt expression and its phosphorylation in normal HUVECs, which is contrary to the unaltered phosphorylation observed in the treatment with W7 (Supplementary Fig. S12f). Thus, the diabetes related Akt alterations likely resulted from other signaling pathways’ alterations rather than CaM. Upon binding with Ca²⁺, CaM promotes the activity of CaMKII by increasing the phosphorylation level of T286 in various tissues including the endothelium and EPCs, which may suggest that CaM deficiency, at least partly, contributes to dephosphorylation of CaMKII-T286 [11, 12]. In addition to T286, CaMKII can also be phosphorylated on T305/306, which located at the CaM-binding element of the autoinhibitory domain. Phosphorylation of T305/306 was found to result in inhibition of CaMKII by preventing Ca²⁺-CaM binding. However, in diabetic vascular endothelium, studies are mainly focused on T286 phosphorylation, while the specific role of T305/306 phosphorylation is not so clear as T286 [45–47].

Multiple studies have demonstrated that HIF-1 is destabilized and inhibited by hyperglycemia under hypoxic conditions in diabetes [48, 49], resulting in a reduced expression of HIF-1-targeted genes such as VEGF, eNOS, SDF-1, and CXCR4, impaired responses upon cellular hypoxia, and delayed wound healing in diabetes [50]. CaM interacts with HIF-1 and positively regulates HIF-1-induced VEGF expression in many types of cells [19–21, 51]. Our results support these findings by exhibiting downregulated HIF-1 and VEGF in HG-treated HUVECs and diabetic ischemic gastrocnemius, which were counteracted by CALM1 gene therapy (Figs. 1, 5 and 7, and Supplementary Fig. S12). Accordingly, inhibition of CaM with W7 suppressed HIF-1 and VEGF expression and reduced eNOS phosphorylation, while inhibition of HIF-1 blocked VEGF but not CaM expression along with eNOS dephosphorylation in normal HUVECs (Supplementary Fig. 12). Our results suggest that CaM activation is the critical signaling nexus through cross-talking with the eNOS, HIF-1, and VEGF pathways in the maintenance of vascular endothelium and angiogenesis homeostasis (Fig. 9). The cause of downregulated CaM expression in diabetes is still unclear, hence further detailed studies are needed.

Fig. 9. The carton shows the mechanism underpinning the downregulated CaM expression involved in dysfunctional and injured vascular endothelial cells and EPCs in diabetes.

Diabetes induces downregulation of CaM expression, causing reductions in eNOS phosphorylation, NO production and VEGF expression. In addition, the downregulated CaM also affects HIF-1-induced VEGF expression and angiogenesis. Therefore, CaM is an important modulator in maintenance of vascular endothelium and angiogenesis homeostasis through cross talks between eNOS and HIF-1, which are disturbed because of diabetes.

In conclusion, our study demonstrated that reduced CaM expression occurs in diabetic vascular endothelial cells and EPCs, contributing to diabetes-related disorders in HIF/eNOS/NO/VEGF signaling activities and coupling, and the subsequent impaired vascular endothelium and angiogenesis. Local delivery of the CALM1 gene into diabetic ischemic limb dramatically improved the limb dysfunction and tissue injuries, thus providing a novel mechanism for vascular endothelium disorders and a potential therapeutic target for enhancing the endothelial repair capacity and tissue angiogenesis in diabetes.

Author contributions

TTL, HHX, and ZJL performed the research and wrote the manuscript. HPZ, ZXZ and HTZ prepared patient samples and collected information. ZQW, JYX, QL, YM and HJY provided research material and techniques and contributed to data interpretation. DLL directed the project, wrote, reviewed, and edited the manuscript. DLL is the guarantor of this project and, as such, has full access to all the data in the study and take responsibility for the integrity of the data and the accuracy of the data analysis.

Competing interests

The authors declare no competing interests.

Supplementary information

The online version contains supplementary material available at 10.1038/s41401-023-01127-1.