Introduction
Diabetic cardiomyopathy (DCM) is a prominent complication of type 2 diabetes mellitus [
1]. The main pathological characteristic features of DCM are cardiac systolic and diastolic dysfunction, which are mainly induced by myocardial hypertrophy and myocardial interstitial fibrosis [
2,
3]. Cardiac microvascular endothelial cells (CMECs) are essential components of heart tissues and are crucial for guaranteeing appropriate microvascular perfusion [
4,
5]. A high-glucose environment and lipid metabolism disorders cause chronic impairment of CMECs and microcirculatory dysfunction and in turn exacerbate cardiac dysfunction and myocardial pathological remodeling [
6,
7]. However, the specific molecular mechanism underlying endothelial dysfunction in DCM remains unclear.
Mitogen-activated protein kinase kinase kinase kinase 4 (MAP4K4), also called Nck-interacting kinase (NIK), is a member of the Ste20 protein kinase family [
8]. MAP4K4 has been widely reported to be a serine/threonine kinase that is closely related to diverse physiological and pathophysiological processes, such as normal embryonic development, atherosclerosis, insulin sensitivity and inflammatory reactions [
9,
10]. Study has shown that endothelial-specific MAP4K4 knockout causes postnatal lethality, highlighting the pivotal role of MAP4K4 in vascular development and homeostasis [
11]. Moreover, MAP4K4 has been suggested to promote angiogenesis under pathological conditions [
12]. Conversely, research has shown that silencing MAP4K4 contributes significantly to promote aortic endothelial cell (EC) activation and improve endothelial permeability [
13]. This discrepancy makes it difficult to identify the regulatory effect of MAP4K4 on cardiac microcirculation in DCM.
Mitochondrial dynamics play important roles in signal transduction and mitochondrial quality control [
14,
15]. Studies have revealed that mitochondrial dynamics disorders play a crucial role in the occurrence and progression of coronary microvascular disorders in DCM patients [
14,
16]. Increased mitochondrial fragmentation has been observed in venous ECs from diabetic patients [
17]. In line with clinical observations, aberrant mitochondrial dynamics in CMECs were found in animal experiments and have been attributed to the pathogenesis of cardiac microvascular disorders. Pathological mitochondrial fission disrupts coronary endothelium-dependent relaxation due to mitochondrial ROS production [
18]. In addition, diabetes-induced Drp1-dependent mitochondrial fission promotes mitochondrial apoptosis in CMECs, thus contributing to capillary degeneration in diabetes [
19]. Moreover, mitochondrial fission impairs vascular permeability, migration, and angiogenesis in DCM [
19]. Drp-1 serves as a critical effector of mitochondrial fission in diabetes [
20]. Phosphorylation and mitochondrial translocation are the classical indices of Drp1 activation, and these processes are believed to be under the control of posttranslational modifications, such as S-nitrosylation, ubiquitination and SUMOylation [
21,
22]. Among these modifications, the S-nitrosylation of Drp1 (SNO-Drp1) is associated with various neuronal diseases [
23]. Only limited evidence has established a link between SNO-Drp1 and cardiovascular diseases [
24]. Whether S-nitrosylation of Drp-1 occurs in diabetes and contributes to cardiac microvascular disorders merits further investigation.
In recent years, emerging evidence has indicated that ferroptosis, a nonapoptotic cell death pattern characterized by iron overload and lipid hydroperoxides, plays a pathophysiological role in the development of DCM [
25]. Diabetes promotes ferroptosis in cardiomyocytes and causes cardiac diastolic dysfunction, which can be reversed by activating the AMPK-NRF2 pathway [
26]. Mitochondrial damage is another feature of ferroptosis. Abnormal mitochondrial structure, rupture of the mitochondrial outer membrane, and reduced mitochondrial membrane potential were observed in the hearts of diabetic mice [
27]. Diabetes promotes endothelial ferroptosis and cardiac microvascular injury in DCM. Our recent study demonstrated that diabetes causes mitochondrial dysfunction and excites mitochondrial iron overload and lipid hydroperoxides in CMECs, whereas maintaining mitochondrial dynamics via MFN2 inhibited endothelial ferroptosis by suppressing the mitochondrial translocation of ACSL4 [
6]. Moreover, the elimination of mitochondrial fragments via AMPKα1-Parkin pathway-dependent mitophagy has been shown to alleviate endothelial ferroptosis and cardiac microvascular disorders in DCM [
7]. However, whether Drp1 and S-nitrosylated Drp1 exacerbate endothelial ferroptosis in DCM remains unknown.
Here, our data demonstrated that MAP4K4 plays a crucial role in microcirculatory disturbance in DCM. In diabetes, upregulated MAP4K4 dramatically interferes with mitochondrial morphology and function, inhibits GPX4 expression, and promotes endothelial ferroptosis by stimulating SNO-Drp1, ultimately promoting cardiac microvascular injury. Thus, our data identify MAP4K4 as a key regulator of SNO-Drp1, which is a potential therapeutic target for DCM in diabetes.
Materials and methods
Animals
Adult male db/db mice and littermate nondiabetic male db/m mice were purchased from Shanghai SLAC Laboratory Animal Co., Ltd. (China). Throughout the experiment, the mice were kept in standard housing with free access to food and drinking water. All animal experiments were ethically approved by the Animal Experimental Ethics Committee and performed according to the animal experiment guidelines of Nanjing Medical University and Fudan University.
Four-week-old male db/m and db/db mice were transfected with Tie2-enhanced adeno-associated virus (AAV9) carrying shMAP4K4, wild-type Drp1 (Drp1-WT) or Cys650-mutated Drp1 (Drp1-C650A) via the tail vein to achieve EC-specific overexpression or knockdown of these proteins, and a total of 6 × 10
11 vector genomes were injected into db/db mice via the tail vein every 8 weeks for 24 weeks. The EC transfection efficiency and specificity of AAV9 were confirmed by western blotting and immunofluorescence. For MAP4K4 inhibitor treatments, DMX-5804 (MCE, USA) was orally administered at a dose of 3 mg/kg three times per week at 4 weeks of age for 24 weeks [
28].
CMEC culture and treatment
The extraction and culture of primary CMECs from cardiac samples were performed as previously described [
2]. After dissecting the coronary arteries, endocardium and epicardium, ventricular tissue was harvested to prepare cell suspensions. Then, anti-CD31 magnetic beads (Thermo Fisher Scientific) were added and incubated with the cell suspension at 4 °C with low-speed rotation for 30 min. The primary CMECs were purified by magnetic isolation and then seeded in confocal dishes for fluorescence imaging or lysed for protein extraction.
For in vitro experiments, human CMECs (HCMECs, Lonza Bioscience) were incubated and cultured in EC medium (ECM, ScienCell Research Laboratories). HCMECs were cultured to 90% confluence before treatment with high glucose (HG, 25 mmol/L) and free fatty acids (FFAs, 0.5 mmol/L) for 72 h. The mixed FFA solutions were prepared as described previously [
6]. CMECs were transfected with lentiviruses (LVs) encoding shMAP4K4, MAP4K4, PDI, CBR1, GPX4, shGPX4 or their negative controls at the appropriate multiplicity of infection (MOI) according to the manufacturers’ instructions. The transfection efficiencies were confirmed by western blot analysis. For in vitro testing, the MAP4K4 inhibitor DMX-5804 (5–15 μM) or L-NAME (50 μM, MCE, USA) was used for 72 h [
29].
Genomic mutations were introduced into cells using the CRISPR–Cas9 system. Single-guide RNAs (sgRNAs) were designed by Genomeditech (China, Shanghai) to target the genomic area adjacent to mutation sites in C505 or C644. After reaching 60% confluence, the cells were cotransfected with sgRNAs (0.5 μg) and single-stranded donor oligonucleotides (ssODNs) as templates to introduce mutations. Twenty-four hours after transfection, the cells were trypsinized, diluted to obtain single cells and seeded into 96-well plates. Genomic DNA was extracted, followed by sequencing of the PCR products spanning the mutation sites.
Echocardiography
Cardiac function was evaluated by two-dimensional echocardiography. The left ventricular ejection fraction (LVEF), left ventricular fractional shortening (LVFS), left ventricular end-diastolic dimension (LVEDD), and E/A ratio were calculated. After the echocardiography, the mice were humanely euthanized for serum and myocardial tissue collection. Heart weight and tibia length were measured to calculate the heart hypertrophy index.
Histopathologic staining
The cardiac tissues were fixed in 4% paraformaldehyde, dehydrated, embedded in paraffin, and sectioned into 4-mm sections. Myocardial fibrosis and collagen content were evaluated by Masson’s trichrome staining, and cardiac hypertrophy was assessed using wheat germ agglutinin (WGA) staining. The degree of fibrosis and the cross-sectional area of the cardiomyocytes were calculated using ImageJ software (version 1.53c, NIH, USA).
Fluorescence staining of cells and tissues
For myocardial tissue microvascular perfusion, 100 μL of FITC-conjugated lectin (1 mg/mL) was injected intravenously into the mice through the tail vein to assess myocardial perfusion [
30,
31]. Ten minutes after lectin injection, the mouse heart samples were frozen, made into 5 μm sections, and further processed through immunostaining with an anti-CD31 antibody. The perfused microvascular density is expressed as the ratio of lectin-FITC-labeled microvessels to CD31-highlighted microvessels.
To observe mitochondrial morphology, CMEC mitochondria were stained with MitoTracker Red (Invitrogen) according to the manufacturer’s instructions. Total intracellular reactive oxygen species (ROS) and mitochondrial ROS (mitoROS) generation were detected using a Reactive Oxygen Species Assay Kit (Beyotime, China) and mitoSOX superoxide indicator (Invitrogen), respectively, after 30 min of incubation. Fluorescence images were obtained using a laser confocal microscope (FV3000, Olympus, Japan) and qualified using ImageJ software.
Cell viability and LDH detection
Cell viability was examined using a Cell Counting Kit-8 (CCK-8, Epizyme, China) based on the manufacturer’s instructions. Cells were seeded in 96-well plates prior to the addition of CCK-8 working solution (100 μL) to each well. Then, the plates were incubated for 4 h at 37 °C in the dark. The absorbance was read at 450 nm by a plate reader (Thermo Scientific, USA). LDH release was measured using an LDH Cytotoxicity Assay Kit (Beyotime) according to the manufacturer’s instructions, and the absorbance was detected at 490 nm.
NO content
Nitric oxide (NO) content in myocardial tissues and NO release from CMECs were assessed using an NO assay kit (Beyotime). For measurement of NO release from CMECs, the cell culture medium was directly assessed according to the manufacturer’s instructions, and the value was normalized to the cell number (5 × 105). For myocardial tissue, the samples were homogenized and centrifuged (12,000×g, 15 min) to collect the supernatant. The protein concentration was quantified using a Bradford protein assay kit (Solarbio, China). Then, the NO content in cardiac tissue was normalized to the protein concentration.
Tissue and cellular monolayer permeability measurements
Heart tissue permeability was quantified by the Evans blue (EB) method according to previously reported protocol [
32]. Mice were administered EB dye (10 mg/ml) via the tail vein for 30 min before sacrifice. Then, the hearts were removed, dissected, and weighed. Afterward, the samples were placed into formamide solution (500 μl), fully ground, incubated at 65–70 °C overnight, and then centrifuged at 10,000×
g for 40 min to collect the supernatant. The absorbance of the supernatant and standards was measured using a plate reader (Thermo Scientific, USA) at 620 nm.
To evaluate cellular monolayer permeability in vitro, ten thousand HCMECs were inoculated in the upper chamber of a Transwell insert (0.4 μm pore size, Corning, USA) for 3 days to confirm stable monolayer junction formation. Then, 1 mg/mL FITC-dextran (100 μL, Solarbio) was added to the upper chamber. The amount of FITC-dextran that penetrated the lower chamber through the paracellular space was determined according to the FITC-dextran fluorescence intensities [
2].
Transendothelial electrical resistance (TEER) was measured to assess junctional function. CMECs were seeded onto fibronectin-coated inserts. After the cells reached confluence and were subjected to the above treatments, a MilliCell ERS-2 system (Millipore, USA) was used to measure the TEER [
19].
Assessment of mitochondrial function
Mitochondrial respiration activity was measured according to oxygen consumption rates (OCRs), which were assessed via a Seahorse XF96 Extracellular Flux Analyzer (Seahorse Bioscience, USA) as reported previously [
33]. HCEMCs were seeded into XF24 cell culture plates at 5 × 10
4 cells/well. Following HG/FFA exposure, the HCMECs were serum starved for 6 h, and then oligomycin, FCCP, rotenone, and antimycin A were sequentially added during the assay according to the manufacturer’s protocols to determine the OCR profiles. The basal OCR, ATP-linked OCR, maximal OCR and spare respiratory OCR were calculated as previously described [
34].
For detection of the mitochondrial membrane potential (MMP), CMECs were seeded in a 96-well black plate (Corning, USA) and incubated with 5 μM TMRM (Invitrogen, USA) at 37 °C for 20 min in the dark. The MMP was calculated as the fluorescence intensity (red) divided by the total cell number.
Coenzyme Q10 (CoQ10) concentrations were detected via a CoQ10 ELISA Kit (CUSABIO, China) according to the product instructions. After treatment, the HCMECs were harvested, homogenized, and centrifuged (10,000×g, 10 min) to collect the supernatants. The obtained supernatant was mixed and incubated with HRP-conjugate at 37 °C for 40 min. Following sufficient washing, TMB substrate was added, and the mixture was incubated for 25 min at 37 °C in the dark. After adding the termination solution, the absorbance values of each group were obtained at 450 nm.
Lipid peroxidation and ROS production
The lipid peroxide (LPO) content and MDA level were detected with an LPO Content Assay Kit (Solarbio) and an MDA assay kit (Beyotime), respectively, following the manufacturer’s instructions. The protein contents of the sample lysates were quantified by a BCA assay immediately after the intervention. The LPO and MDA levels were calculated according to the basis of total protein content. Iron levels were examined with an Iron Assay Kit (Abcam, USA). ROS levels were measured using a reactive oxygen species assay kit (Nanjing Jiancheng Bioengineering Institute, China), and H2O2 content was evaluated by Amplex Red (Beyotime). All the above experiments were performed according to the manufacturers’ instructions.
Transwell assay
CMECs were collected in serum-free medium and seeded in the upper chambers of Corning Transwell chambers (8 µm pores). ECM containing 10% serum was added to the lower compartment to induce cell migration. After 24 h of migration, the migrated cells were fixed with 4% PFA and stained with crystal violet (Solarbio). For each sample, five microscopic visual fields were randomly selected, and the average cell number of these fields was calculated with ImageJ (version 1.53c, NIH, USA).
RNA extraction and real-time PCR
Prior to PCR, total RNA was extracted using a Total RNA Extraction Kit (Solarbio) according to the manufacturer’s instructions. Then, the isolated mRNA was treated with DNase I (Takara, Japan) and reverse-transcribed into cDNA using a cDNA reverse transcription kit (Invitrogen, USA). Real-time quantitative PCR was carried out using SYBR Green I (TSINGKE, China) on a Bio-Rad CFX96 Real-Time PCR system. The primers used are listed in Additional file
6: Table S1. The results of qPCR are shown as the 2
−ΔΔCt method.
Biotin switch assay of S-nitrosylated proteins
A biotin-switch assay was carried out using an S-nitrosylated protein detection kit (Cayman Chemical, USA) as previously described [
35]. Briefly, whole-cell lysates, cytosolic fractions and mitochondria were collected and incubated in blocking reagent. Thereafter, the supernatant was collected and incubated with ice-cold acetone at − 20 °C. The reduction of S-nitrosothiol groups was carried out with reducing buffer to yield free thiols. Thereafter, the samples were covalently labeled with biotin in labeling buffer, and the biotinylated proteins were purified by streptavidin-coupling beads with agitation. The S-nitrosylation of Drp1 was quantified by SDS‒PAGE.
Western blot analysis and immunoprecipitation (IP)
Mitochondria were extracted using a mitochondria isolation kit (Beyotime, China) according to the manufacturer’s instructions. Proteins were extracted using RIPA lysis buffer (Millipore, USA) supplemented with protease inhibitors and then centrifuged at 12,000 rpm (20 min, 4 °C). The protein samples were quantified via a BCA assay, separated via SDS‒PAGE, and transferred electrophoretically onto polyvinylidene difluoride (PVDF) membranes. Then, the membranes were blocked, incubated overnight with primary antibodies at 4 °C, and incubated for 1 h with conjugated secondary antibodies at room temperature. The protein bands were detected with electrochemiluminescence western blotting substrate (Thermo Fisher, USA), and the luminescence signals were measured with ImageJ software. The primary antibodies used are listed in Additional file
6: Table S2.
To perform a coimmunoprecipitation (Co-IP) assay, the cells were lysed using Cell Lysis Buffer for IP (Beyotime) and incubated with Ig-A/G-magnetic beads (BioLinkedIn, China) preconjugated with antibodies against GPX4 (Santa Cruz, sc-166570) at 4 °C overnight. After extensive washing, the immunocomplexes were mixed with loading buffer and boiled for 10 min to elute the target proteins.
Statistical analysis
The data were analyzed using GraphPad Prism software (version 10.0, GraphPad Software, USA). All the data are presented as the mean ± SEM. Statistical analyses were performed using Student’s t test, one-way ANOVA with Tukey’s post hoc test, and two-way ANOVA with Bonferroni correction. Differences were considered significant when *P < 0.05, **P < 0.01, or ***P < 0.001.
Discussion
Our study revealed that MAP4K4 regulates SNO-Drp1 by interfering with GPX4 in endothelial cells. In addition, human C644 (mouse C650) was identified as the core site of SNO-Drp1 in diabetic injury. Knockdown of MAP4K4 or treatment with a MAP4K4 inhibitor (DMX-5804) inhibited SNO-Drp1 and protected endothelial cells and cardiac microcirculation against diabetes by enhancing mitochondrial functions and suppressing oxidative stress injury and ferroptosis. Similarly, the human C644A (mouse C650A) mutation negated SNO-Drp1 and abolished the adverse effects of Drp1 in promoting endothelial ferroptosis and cardiac microvascular dysfunction. These data reveal that MAP4K4 and SNO-Drp1 promote the development of cardiac microvascular dysfunction in diabetes and provide theoretical evidence that the MAP4K4-GPX4-Drp1 signaling pathway is a novel therapeutic target for DCM.
MAP4K4 is a member of the Ste20 family of kinases, which have been broadly reported to participate in multiple cardiovascular diseases [
9,
42]. MAP4K4 is activated in failing human hearts and induces oxidative stress to promote cell death in myocardial infarction models [
43]. In contrast, MAP4K4 inhibition rescues mitochondrial function in cardiomyocytes, ameliorates apoptosis and reduces infarction size [
43‐
45]. Pathological cardiac hypertrophy or amyotrophy precedes heart failure [
8,
31]. Silencing of MAP4K4 inhibits the prohypertrophic factor NFAT in angiotensin II-treated hearts, and the inhibition of MAP4K4 preserves cardiomyocyte function after doxorubicin treatment [
8,
46]. In addition to its role in cardiomyocytes, MAP4K4 is highly expressed in endothelial cells and participates in the inflammatory response [
9]. EC-specific MAP4K4 deletion alleviates aortic atherosclerosis by decreasing macrophage permeation and lipid accumulation [
13,
47]. More importantly, one study demonstrated that animals lacking endothelial MAP4K4 were protected from skeletal muscle microvascular rarefaction via the suppression of endothelial senescence and increased metabolic capacity in obese mice [
48]. However, the role and intrinsic mechanism of MAP4K4 in diabetes-induced microvascular dysfunction have not been explored. The present study demonstrated that silencing or inhibiting MAP4K4 (DMX-5804) in DCM significantly improved microvascular density, angiogenesis, and endothelial-dependent microvascular perfusion by activating the VEGF and eNOS signaling pathways. Chronic myocardial interstitial edema and the inflammatory response contribute to myocardial fibrosis, cardiac remodeling, and heart dysfunction [
2,
6,
49]. Inhibiting MAP4K4 significantly improved endothelial barrier function and endothelium-related inflammatory responses, ultimately improving cardiac remodeling and dysfunction. These data imply that cardiac or EC-specific MAP4K4 inhibition confers cardiovascular protection against long-term diabetes.
The roles of Drp1 in mitochondrial homeostasis, myocardial microcirculatory disturbance and DCM have been widely investigated, with a major focus on the phosphorylation of Drp1 at Ser616 and Ser637 [
50,
51]. In addition to being phosphorylated, Drp1 can undergo multiple posttranslational modifications, such as palmitoylation, SUMOylation, acetylation, and ubiquitination [
21,
52]. Importantly, accumulating evidence has demonstrated the involvement of SNO-Drp1 in facilitating mitochondrial fission, promoting mitochondrial dysfunction, and triggering mitochondria-dependent cell death [
29,
53]. In addition, S-nitrosylation disturbs the phosphorylation of Drp1 between Ser616 and Ser637 [
29]. Nine cysteine residues of Drp1 are potential S-nitrosylation sites, among which human C505 and C644 have been verified [
23]. The mouse C650A (human C644A) mutation inhibits SNO-Drp1 in isoprenaline-induced heart failure[
24]. However, little is known about the role of SNO-Drp1 in diabetes and endothelial injury. Here, our findings demonstrated that diabetes promoted SNO-Drp1, and human C644, rather than C505, was identified as the cysteine site of Drp1 for S-nitrosylation. Additionally, S-nitrosylation of Drp1 at C644 was revealed as a downstream signaling pathway for MAP4K4 in diabetes and may be a potential target for cardiovascular protection.
As a classic GSH peroxidase-related protein, GPX4 has been suggested to participate in a variety of biological functions, such as the regulation of oxidative stress and cell death, via its intrinsic role in modulating the GSH/GSSG balance [
54,
55]. In addition, GSH levels are closely related to S-nitrosylation in a variety of proteins [
56,
57]. The present data revealed that MAP4K4 could regulate multiple proteins that modulate S-nitrosylation, among which only GPX4 inhibited SNO-Drp1 in diabetic injury. GPX4 is widely known to inhibit ferroptosis [
58]. Interestingly, Drp1 has recently been reported to modulate the occurrence and development of glioma ferroptosis [
59]. Additionally, Drp1 oligomerization can promote ferroptosis, which in turn suppresses hepatocellular carcinoma cell growth [
60]. However, it is not clear whether SNO-Drp1 can promote ferroptosis in endothelial injury in diabetes. The present study demonstrated that diabetes promoted SNO-Drp1 and promoted ferroptotic injury in endothelial cells. In contrast, MAP4K4 silencing, DMX-5804 treatment or the human Drp1 C644A mutation inhibited SNO-Drp1 and alleviated endothelial ferroptotic injury in diabetes, thereby improving cardiac microvascular disorders. Therefore, in addition to affecting apoptosis, the present study identified MAP4K4 as a crucial regulator of ferroptosis in diabetes, and this process involves the GPX4-dependent GSH/GSSG balance and the S-nitrosylation of Drp1.
Even though the present study highlights novel findings that explain the potential of the MAP4K4-GPX4-Drp1 pathway in facilitating cardiac microvascular disorder in diabetes, several potential biases and limitations should be considered. The present work used db/db mice to verify our assumptions, which were not reconfirmed by other preclinical models of metabolic and cardiovascular diseases, such as ob/ob, Apoe
−/− or Ldlr
−/− mouse models. Previous reports identified MAP4K4 as an upstream signal for the AMPK-mTOR pathway, Hippo pathway, and JNK pathway [
9]. However, the mechanisms by which diabetes stimulates MAP4K4 expression have not yet been investigated. The present work suggested that MAP4K4 modulates S-nitrosylation and endothelial ferroptosis by balancing GSH/GSSH from GPX4; however, the intermediary steps were not identified, and other complex signaling networks were not ruled out. Similarly, the source of NO and the direct effects of MAP4K4, which participates in the S-nitrosylation of Drp1, were not confirmed, considering eNOS and NO are reduced by diabetes. Studies aimed at solving the above issues in DCM are warranted.
To summarize, our study clearly presents evidence that MAP4K4 promotes SNO-Drp1 by suppressing the expression of GPX4, which induces endothelial ferroptosis and dysfunction and ultimately leads to cardiac microvascular disorders in DCM. This research therefore underscores novel potential roles of MAP4K4 and SNO-Drp1 in DCM and suggests that MAP4K4 or SNO-Drp1 may be novel therapeutic targets for endothelial ferroptosis.
Publisher's Note
Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.