Abstract

   Diabetic cardiomyopathy (DCM), a severe complication of diabetes, is
   characterized by mitochondrial dysfunction, oxidative stress, and DNA
   damage. Despite its severity, the intrinsic factors governing
   cardiomyocyte damage in DCM remain unclear. It is hypothesized that
   impaired iron–sulfur (Fe–S) cluster synthesis plays a crucial role in
   the pathogenesis of DCM. Reduced S‐sulfhydration of cysteine
   desulfurase (NFS1) is a novel mechanism that contributes to
   mitochondrial dysfunction and PARthanatos in DCM. Mechanistically,
   hydrogen sulfide (H[2]S) supplementation restores NFS1 S‐sulfhydration
   at cysteine 383 residue, thereby enhancing Fe–S cluster synthesis,
   improving mitochondrial function, increasing cardiomyocyte viability,
   and alleviating cardiac damage. This study provides novel insights into
   the interplay between Fe–S clusters, mitochondrial dysfunction, and
   PARthanatos, highlighting a promising therapeutic target for DCM and
   paving the way for potential clinical interventions to improve patient
   outcomes.

   Keywords: cysteine desulfurase (NFS1), diabetic cardiomyopathy (DCM),
   hydrogen sulfide (H[2]S), iron–sulfur (Fe–S) cluster, PARthanatos
     __________________________________________________________________

   A reduction in mitochondrial iron‐sulfur (Fe–S) cluster synthesis in
   diabetic cardiomyocytes decreases mitochondrial respiratory chain
   function, leading to oxidative stress and DNA damage. Supplementing
   with hydrogen sulfide (H[2]S) enhances Fe–S cluster synthesis via
   S‐sulfhydration of cysteine desulfurase (NFS1) at C383, preserving
   mitochondrial and nuclear DNA integrity and ameliorating diabetic
   cardiomyopathy (DCM).

   graphic file with name ADVS-12-2406695-g007.jpg

1. Introduction

   Type 2 diabetes mellitus (T2DM) is a chronic metabolic disorder
   characterized by insulin resistance or relative insulin deficiency,
   which leads to elevated blood glucose and lipid levels, and subsequent
   systemic organ dysfunction. Diabetic cardiomyopathy (DCM) is a serious
   complication associated with T2DM. DCM is characterized by the
   prolonged dysregulation of glucose and lipid metabolism that induces
   sustained structural and functional heart damage, thereby significantly
   increasing the risk of cardiovascular diseases.^[ [58]^1 , [59]^2 ,
   [60]^3 , [61]^4 ^]

   Recent evidence has linked DCM development to mitochondrial dysfunction
   and elevated intracellular reactive oxygen species (ROS) levels.^[
   [62]^5 , [63]^6 , [64]^7 , [65]^8 ^] Defective mitochondrial
   respiratory chain complexes disrupt electron transfer, causing electron
   leakage and ROS generation, particularly superoxide^[ [66]^9 , [67]^10
   ^] Furthermore, diabetes increases mitochondrial criticality, rendering
   the cardiac mitochondrial network highly susceptible to minor
   disturbances under intense oxidative stress. This increased sensitivity
   can lead to depolarization and synchronized oscillations throughout the
   myocardial syncytium.^[ [68]^11 ^] Disturbances in mitochondrial
   membrane potential (MMP) further exacerbate ROS formation, whereas the
   diminished antioxidant capacity of DCM cardiomyocytes amplifies ROS
   accumulation, resulting in extensive oxidative damage to proteins, DNA,
   and lipid membranes, thereby intensifying oxidative stress.^[ [69]^12 ,
   [70]^13 , [71]^14 ^]

   Oxidative damage to the nuclear DNA disrupts normal base pairing,
   resulting in genetic mutations.^[ [72]^15 ^] Poly ADP‐ribose polymerase
   1 (PARP1) plays a crucial role in DNA repair by catalyzing the
   generation of poly ADP‐ribose (PAR) to recruit other repair enzymes.^[
   [73]^16 , [74]^17 ^] However, excessive PARP1 expression and activation
   lead to PAR accumulation, inducing translocation of mitochondrial
   apoptosis‐inducing factor (AIF) to the nucleus and causing DNA
   fragmentation‐mediated cell death, a process known as PARthanatos.^[
   [75]^18 , [76]^19 , [77]^20 , [78]^21 ^]

   Iron–sulfur (Fe–S) clusters are crucial protein cofactors involved in
   mitochondrial respiration, citrate cycle, lipid metabolism, and DNA
   damage repair.^[ [79]^22 , [80]^23 , [81]^24 , [82]^25 , [83]^26 ,
   [84]^27 ^] In the mitochondrial matrix, Fe–S clusters are synthesized
   by the Fe–S cluster assembly complex, with cysteine desulfurase (NFS1),
   Fe–S cluster assembly scaffold protein (ISCU), and frataxin (FXN) as
   core components. By using its conserved mobile S‐transfer loop to
   extract sulfur from cysteine, NFS1 provides an essential sulfur source
   for Fe–S cluster synthesis. ISCU accepts sulfur from NFS1 in the form
   of an S‐sulfhydrated intermediate and then binds iron ions to form the
   initial Fe–S clusters. FXN regulates iron supply and modulates NFS1
   activity. The coordinated action of these proteins ensures the
   efficient synthesis of Fe–S clusters.^[ [85]^28 , [86]^29 , [87]^30 ,
   [88]^31 , [89]^32 ^] Fe–S cluster depletion and oxidative stress
   contribute to the degradation of Fe–S proteins.^[ [90]^33 ^]

   Hydrogen sulfide (H[2]S), a gasotransmitter with cardiovascular
   protective properties, primarily regulates target proteins via
   S‐sulfhydration.^[ [91]^34 , [92]^35 , [93]^36 , [94]^37 ^] Previous
   studies indicated that insufficient endogenous H[2]S in DCM exacerbates
   cardiac damage.^[ [95]^38 , [96]^39 ^] Although exogenous H[2]S
   supplementation alleviates cardiac damage, the specific mechanisms and
   interplay between H[2]S and mitochondria are not fully understood.
   H[2]S contributes to electron generation via oxidation by sulfide
   quinone oxidoreductase (SQR), facilitating electron transport in the
   mitochondrial respiratory chain and subsequent ATP synthesis.^[ [97]^40
   ^] Despite these findings, the precise mechanisms by which H[2]S
   influences mitochondrial function in DCM, particularly regarding its
   role in Fe–S cluster synthesis and mitochondrial respiration, remain
   unclear. This gap in our understanding hinders the development of
   targeted therapies to mitigate mitochondrial dysfunction and oxidative
   stress in DCM.

   To address this issue, we utilized leptin receptor gene knockout
   (db/db) mice, a well‐established model of T2DM, and primary
   cardiomyocytes to investigate the molecular mechanisms underlying DCM.
   Our research uncovers a novel mechanism by which H[2]S maintains
   mitochondrial respiration and mitigates PARthanatos. We found that both
   the expression and S‐sulfhydration of NFS1 significantly decreased in
   diabetic cardiomyocytes, which impaired Fe–S cluster synthesis. Fe–S
   cluster deficiency disrupts mitochondrial respiration and leads to DNA
   damage‐induced PARthanatos. H[2]S supplementation enhances NFS1
   S‐sulfhydration at cysteine 383 (C383), thereby promoting Fe–S cluster
   synthesis. Knockdown or mutation of NFS1 at C383 residue abrogates the
   protective effects of H[2]S. Thus, H[2]S ameliorates mitochondrial
   dysfunction and DNA damage in DCM by promoting Fe–S cluster synthesis.
   Given its role in maintaining mitochondrial function, NFS1 may
   represent a promising target for clinical strategies aimed at
   preventing or treating DCM, providing a new theoretical foundation for
   disease intervention.

2. Results

2.1. Cardiac Structure and Function are Impaired in Diabetic Mice

   To investigate cardiac dysfunction in diabetes, db/db mice were used as
   a diabetic model owing to their increased body weight, elevated blood
   glucose levels, and impaired glucose tolerance, which are typical
   manifestations of T2DM (Figures [98]S1A,B, Supporting Information).
   Heterozygous littermates (db/+ mice) served as controls. Results from
   echocardiography revealed decreased cardiac contractile function in
   db/db mice. Specifically, db/db mice showed significant reductions in
   fractional shortening (FS)%, ejection fraction (EF)%, and stroke volume
   (SV) compared to db/+ mice. In addition, db/db mice exhibited increased
   left ventricular posterior wall (LVPW) thickness and left ventricular
   end‐systolic volume (LVESV), with no significant changes in left
   ventricular end‐diastolic volume (LVEDV) (Figure [99]1A). These
   findings indicate impaired cardiac contractility in diabetic mice. The
   heart‐to‐tibia ratio was significantly higher in db/db mice than the
   heart‐to‐tibia ratio in db/+ mice, indicating abnormal cardiac
   enlargement (Figure [100]S1C, Supporting Information). Hematoxylin and
   eosin (H&E) staining and transmission electron microscopy (TEM)
   revealed that the cardiac tissues of db/db mice displayed a
   disorganized arrangement, partial disruption, mitochondrial
   fragmentation, and abnormal alterations in mitochondrial cristae
   structure (Figure [101]1B).

Figure 1.

   Figure 1
   [102]Open in a new tab

   Cardiac structure and function are impaired in diabetic mice. A)
   Representative left ventricular M‐mode echocardiographic tracings and
   quantification of fractional shortening (FS) %, ejection fraction (EF)
   %, left ventricular posterior wall (LVPW), left ventricular
   end‐diastolic volume (LVEDV), stroke volume (SV) and left ventricular
   end‐systolic volume (LVESV) of mice (n = 4). B) Representative
   photographs of mouse hearts (scale bars, 2 mm), H&E staining sections
   of the left ventricle (scale bars, 25 µm), and TEM images of left
   ventricular sections (scale bars, 2 µm). The blue arrows indicate
   abnormal cardiac structures. The red arrows indicate damaged
   mitochondria. C) Volcano plot of DEGs between db/db and db/+ mouse
   hearts acquired from RNA‐Seq. D) Pie chart representing the results of
   cellular component enrichment analysis based on DEGs. E) KEGG pathway
   enrichment analysis of DEGs. F) Molecular functions of GO enrichment of
   DEGs. G) Biological processes of GO enrichment of DEGs. H) GSEA of
   DEGs. I) Heatmap of DEGs involved in mitochondrial respiratory
   complexes. CI‐IV, complex I–IV. J) Heatmap of DEGs associated with Fe–S
   clusters. K) Tissue‐specificity of NFS1 in humans. All quantitative
   data are presented as mean ± SD from independent experiments. ^** P <
   0.01, ^*** P < 0.001 versus db/+ by unpaired t test.

   To elucidate the molecular mechanisms responsible for cardiac damage in
   diabetes, we conducted a comprehensive analysis of the cardiac
   transcriptomes of db/db mice and db/+ mice (Figure [103]1C). A
   substantial proportion of proteins encoded by the differentially
   expressed genes (DEGs) between these two groups (18.89%) were localized
   in the mitochondria (Figure [104]1D). Kyoto Encyclopedia of Genes and
   Genomes (KEGG) pathway analysis revealed significant enrichment of DEGs
   in pathways associated with DCM and cardiac muscle contraction
   (Figure [105]1E). Gene Ontology (GO) and Gene Set Enrichment Analyses
   (GSEA) showed significant enrichment of genes related to the
   mitochondrial respiratory chain, iron ion homeostasis, sulfur compound
   biosynthesis, and Fe–S cluster binding/assembly (Figure [106]1F–H).

   To further confirm the expression changes in Fe–S cluster‐related genes
   in diabetic hearts, we analyzed single‐cell RNA sequencing data
   ([107]GSE213337) of cardiac specimens collected from healthy mice
   (Ctrl) and mice with diabetes mellitus (DM) from the public Gene
   Expression Omnibus (GEO) database.^[ [108]^41 ^] Analysis of DEGs in
   cardiomyocyte subpopulations revealed enrichment of glucose/lipid
   metabolism, mitochondrial oxidative phosphorylation, (ROS metabolic
   processes, and Fe–S cluster biosynthesis (Figure [109]S1D, Supporting
   Information). Furthermore, GSEA indicated pronounced enrichment of DEGs
   in pathways involving mitochondrial electron transport and respiratory
   chain complex assembly (Figure [110]S1E, Supporting Information).
   Notably, the downregulated DEGs induced the genes responsible for
   encoding proteins associated with mitochondrial respiration (Figure
   [111]S1F, Supporting Information). These observations further confirm
   that Fe–S cluster synthesis disruption and mitochondrial respiratory
   dysfunction occur in the cardiomyocytes of diabetic mice.

   Among the genes associated with the mitochondrial respiratory chain,
   multiple genes involved in the formation of complexes I, II, III, and
   IV were downregulated in the hearts of db/db mice (Figure [112]1I).
   Furthermore, among the genes involved in de novo synthesis of Fe–S
   clusters, Nfs1, Iscu, and frataxin (Fxn) exhibited reduced expression
   in db/db mice (Figure [113]1J). NFS1 is a highly conserved protein
   abundantly expressed in the human heart and other parenchymal organs
   (Figure [114]1K). Taken together, these data suggest that mitochondrial
   dysfunction, especially damage to the Fe–S cluster assembly, emerges in
   DCM.

2.2. Disruptions of Fe–S Cluster Synthesis in Diabetic Cardiomyopathy

   To understand how diabetes affects Fe–S cluster synthesis, we examined
   the key proteins involved in this process. NFS1 transfers sulfur to
   ISCU through the formation of S‐sulfhydrated intermediates,
   facilitating Fe–S cluster synthesis.^[ [115]^32 ^] We observed
   significantly reduced NFS1, ISCU, and FXN protein levels in db/db mice
   (Figure [116]2A). To visualize the S‐sulfhydration levels of these
   target proteins, we used biotin switch assays, which exploit the
   affinity of biotin for free thiol groups (Figure [117]S2A, Supporting
   Information). The findings of this method revealed significantly
   reduced S‐sulfhydration levels of NFS1 and ISCU in db/db mice
   (Figure [118]2B). Mitochondrial aconitase (mAco/Aco2) enzyme involved
   in the citrate cycle and Xeroderma pigmentosum Group D (XPD) protein
   involved in nuclear DNA repair contain Fe–S clusters.^[ [119]^22 ,
   [120]^42 ^] In our study, db/db mice showed reduced XPD expression
   along with decreased protein levels and mAco activity, indicating the
   loss of Fe–S clusters (Figure [121]2C,D).

Figure 2.

   Figure 2
   [122]Open in a new tab

   Disruptions of Fe–S cluster synthesis in diabetic cardiomyopathy. A)
   Protein levels of NFS1, ISCU, and FXN in cardiac tissues (n = 6). B)
   S‐sulfhydration levels of NFS1 and ISCU in cardiac tissues (n = 4). C)
   Protein levels of XPD (n = 6) and mAco (n = 7) in cardiac tissues. D)
   Activity of mAco in cardiac tissues (n = 4). E) Protein levels of NFS1,
   ISCU, and FXN in cardiomyocytes (n = 5). F) Protein levels of XPD and
   mAco in cardiomyocytes (n = 4). G) Citrate concentration in
   cardiomyocytes (n = 4). H) Workflow for Nfs1‐KD cardiomyocyte cell line
   establishment. I) Principal component analysis (PCA) plot of Nfs1‐KD
   and control groups. J) Volcano plot of DEPs in the proteomic analysis.
   K) Subcellular localization of the top 100 DEPs. L) Enrichment of
   molecular functions, biological processes, and KEGG pathways of DEPs.
   M) Heatmap of proteins related to Fe–S clusters and iron homeostasis.
   N) Heatmap of proteins related to the mitochondrial respiratory chain
   and cysteine metabolism. All quantitative data are presented as mean ±
   SD from independent experiments. ^* P < 0.05, ^** P < 0.01, ^*** P <
   0.001 versus db/+ or Ctrl by unpaired t test.

   To mimic the hyperglycemic and hyperlipidemic environment in vitro,
   primary neonatal mouse cardiomyocytes were exposed to high glucose (HG,
   40 mm) and palmitate (PA, 200 µm). This treatment led to similar
   alterations in Fe–S cluster assembly proteins and Fe–S proteins as
   observed in vivo (Figure [123]2E,F). Additionally, the accumulation of
   citrate, a substrate for mAco‐mediated enzymatic reactions, indicated
   the loss of mAco activity in the HG/PA‐treated groups (Figure [124]2G).
   Furthermore, we found a significant accumulation of intracellular iron
   ions in HG/PA‐treated cardiomyocytes, which resulted from decreased
   intracellular iron utilization (Figure [125]S2B, Supporting
   Information). These findings suggest a reduction in Fe–S cluster
   synthesis in diabetic cardiomyocytes.

   To further explore the role of impaired Fe–S cluster synthesis in DCM,
   we employed CRISPR/Cas9 gene editing to knock down the Nfs1 gene
   (Nfs1‐KD) in cardiomyocytes. Of the three knockdown strategies tested,
   Y28627 exhibited the most stable effects (Figure [126]S2C, Supporting
   Information). We established an Nfs1‐KD cardiomyocyte cell line and
   performed proteomic analyses (Figure [127]2H,I). The results revealed
   that NFS1 depletion led to significant alterations in the expression
   profiles of 668 proteins, with 217 upregulated and 451 downregulated
   proteins (Figure [128]2J). Notably, ≈17% of the top 100 differentially
   expressed proteins (DEPs) were localized within the mitochondria
   (Figure [129]2K). Functional enrichment analysis showed that the
   depletion of NFS1 affected several physiological processes, including
   diabetic cardiomyopathy (including alterations in energy metabolism,
   mitochondrial dysfunction, and cardiac structural changes),
   cysteine/methionine metabolism, Fe–S cluster binding and damaged DNA
   binding (Figure [130]2L). Moreover, we observed that the knockdown of
   Nfs1 led to the downregulation of proteins involved in Fe–S cluster
   biosynthesis and the mitochondrial respiratory chain
   (Figure [131]2M,N). These results suggest a potential link between
   disruptions in Fe–S cluster synthesis and the development of DCM, as
   well as mitochondrial and DNA damage.

2.3. Impaired Fe–S Cluster Synthesis Induces Mitochondrial PARthanatos and
Cardiomyocyte Death in Diabetic Cardiomyopathy

   Fe–S clusters are critical components of mitochondrial respiratory
   complexes, such as NDUFS1/2/3 in complex I (NADH ubiquinone
   oxidoreductase) and SDHB in complex II (succinate dehydrogenase).^[
   [132]^43 , [133]^44 , [134]^45 ^] In Nfs1‐KD group, the protein levels
   of NDUFS3, SDHB, and COX4 (complex IV) were diminished, consistent with
   the proteomic analysis results (Figure [135]3A), implying that NFS1
   deficiency impairs mitochondrial respiration. Similar reductions in the
   levels of these respiratory complex proteins were observed in
   HG/PA‐treated cardiomyocytes (Figure [136]3B). Additionally, Nfs1‐KD
   led to decreased mAco activity, which is consistent with the reduction
   observed in cardiac tissues of db/db mice, as shown in Figure [137]2D
   (Figure [138]3C).

Figure 3.

   Figure 3
   [139]Open in a new tab

   Impaired Fe–S cluster synthesis induces mitochondrial PARthanatos and
   cardiomyocyte death in diabetic cardiomyopathy. A) Protein levels of
   NDUFS3 (n = 4), SDHB (n = 4), COX4 (n = 7), and XPD (n = 6) in
   cardiomyocytes. B) Protein levels of NDUFS3, SDHB, and COX4 in
   cardiomyocytes (n = 4). C) Activity of mAco in cardiomyocytes (n = 5).
   D) Normalized OCR of cardiomyocytes (n = 5). E) Quantification of OCR
   in D (n = 5). F) Flow cytometry results showing MitoSOX staining of
   cardiomyocytes (n = 4). G) γ‐H2AX foci in cardiomyocytes (n = 3). Scale
   bars, 50 µm. H) Protein levels of PARP1 (n = 4), AIF (n = 3), and PAR
   (n = 4) in cardiomyocytes. I) Confocal microscopy showing AIF (red)
   co‐localization with nuclei (blue) in cardiomyocytes (n = 5). Scale
   bars, 20 µm. J) Viability of cardiomyocytes (n = 5). K) Protein levels
   of NDUFS3, SDHB, and COX4 in cardiac tissues (n = 4). L) Activity of
   mitochondrial complexes I, II, and IV in cardiac tissues (n = 3). M)
   Concentrations of hydroxyl radicals (n = 3) and hydrogen peroxide (n =
   4) in cardiac tissues. N) Quantification of 8‐OHdG levels in cardiac
   tissues (n = 3). O) PARP1 and PAR protein levels in cardiac tissues (n
   = 3). P) Interaction between PAR and AIF in cardiac tissues. Q) Protein
   levels of nuclear AIF (nAIF, n = 3) and mitochondrial AIF (mAIF, n = 4)
   in cardiac tissues. MT, mitochondrion. All quantitative data are
   presented as mean ± SD from independent experiments. ^* P < 0.05, ^** P
   < 0.01, ^*** P < 0.001 versus Ctrl or db/+; ^## P < 0.01, ^### P <
   0.001 versus HG/PA by unpaired t ‐test or ordinary one‐way ANOVA.

   To further investigate mitochondrial function, we monitored changes in
   mitochondrial oxygen consumption rate (OCR) in cardiomyocytes. HG/PA
   treatment disrupted mitochondrial respiration, as evidenced by impaired
   basal respiration, maximal respiration, and spare respiratory capacity
   (Figure [140]3D,E). The mitochondrial respiratory chain is the primary
   source of ROS, that cause cellular oxidative stress. ROS are a general
   term for a large family of oxidants and mainly include hydroxyl
   radicals, hydrogen peroxide, and superoxide.^[ [141]^46 ^]
   Mitochondria‐derived ROS (superoxide) were detected using MitoSOX
   probes and flow cytometry, which revealed a significant increase in
   mitochondrial ROS levels in cardiomyocytes treated with HG/PA
   (Figure [142]3F). Direct measurement revealed a significant increase in
   superoxide levels in the HG/PA‐treated group (Figure [143]S2D,
   Supporting Information). Analysis of dihydroethidium (DHE) staining
   confirmed elevated cellular ROS levels in cardiomyocytes treated with
   HG/PA (Figure [144]S2E, Supporting Information). The analysis of
   Tetramethyl‐ rhodamine ethyl ester (TMRE) staining revealed
   demonstrated mitochondrial membrane potential (MMP) depolarization in
   the HG/PA‐treated groups, indicating abnormal electron transfer in the
   mitochondria (Figure [145]S2F, Supporting Information).

   XPD, a DNA helicase essential for nucleotide excision repair, was
   downregulated in Nfs1‐KD groups (Figure [146]3A). As a biomarker of DNA
   damage, the percentage of γ‐H2AX‐positive cardiomyocytes was
   significantly higher in HG/PA‐treated group, indicating compromised
   nuclear DNA integrity (Figure [147]3G). The initial step in the DNA
   damage repair response is the expression of poly ADP‐ribose polymerase
   (PARP1). Excessive poly ADP‐ribose (PAR) produced by PARP1 binds to
   mitochondrial apoptosis‐inducing factor (AIF), leading to its release
   and subsequent nuclear translocation, causing DNA fragmentation and
   cell death, in a process known as PARthanatos.^[ [148]^17 , [149]^18 ,
   [150]^19 , [151]^21 ^] Nuclear PARP1 and AIF levels were significantly
   elevated in the HG/PA‐treated group, accompanied by PAR accumulation
   (Figure [152]3H). Confocal microscopy analysis revealed noticeable AIF
   nuclear translocation under HG/PA conditions, which could be reversed
   by the PARP1 inhibitor ABT888 (Figure [153]3I). Hyperglycemic and
   hyperlipidemic environments induced cardiomyocyte death, which was
   partially ameliorated by ABT888 treatment (Figure [154]3J).

   The in vivo expression and activity of mitochondrial complexes were
   downregulated in db/db mice (Figure [155]3K,L). Hydroxyl radical and
   hydrogen peroxide levels were significantly increased in the db/db mice
   (Figure [156]3M). Moreover, there was a significant increase in the DNA
   oxidative damage marker 8‐hydroxy‐2′‐deoxyguanosine (8‐OHdG) levels in
   db/db mice (Figure [157]3N). These mice also exhibited PARP1 and PAR
   upregulation (Figure [158]3O). The findings of Co‐immunoprecipitation
   (Co‐IP) experiments confirmed the enhanced interaction between AIF and
   PAR in db/db mice (Figure [159]3P). These changes caused the
   translocation of AIF from the mitochondria to the nucleus
   (Figure [160]3Q). These results suggest that impaired Fe–S cluster
   synthesis leads to ROS accumulation and severe mitochondrial damage,
   inducing DNA damage and subsequent PARthanatos in diabetic
   cardiomyocytes.

2.4. H[2]S Supplementation Improves Cardiac Function in Diabetic
Cardiomyopathy

   Sulfur compound biosynthesis and cysteine metabolism were found to be
   affected in the hearts of db/db mice and Nfs1‐KD group cardiomyocytes,
   respectively. Hydrogen sulfide (H[2]S) is an endogenous gaseous
   signaling molecule and an inorganic sulfur‐containing compound that
   modulates the stability or activity of proteins through S‐sulfhydration
   of cysteine residues.^[ [161]^37 ^] The primary enzymes responsible for
   endogenous H[2]S production are cystathionine γ‐lyase (CSE),
   cystathionine β‐synthase (CBS), and 3‐mercaptopyruvate
   sulfurtransferase (3‐MST).^[ [162]^40 , [163]^47 ^] Our results
   revealed a significant reduction in CSE expression in the cardiac
   tissues of db/db mice, whereas CBS and 3‐MST expression levels remained
   relatively unchanged (Figure [164]4A,B; Figure [165]S3A, Supporting
   Information). Correspondingly, serum H[2]S concentration was markedly
   decreased in db/db mice (Figure [166]4C).

Figure 4.

   Figure 4
   [167]Open in a new tab

   H[2]S supplementation improves cardiac function in diabetic
   cardiomyopathy. A) Representative immunohistochemistry (IHC) staining
   images of CSE, CBS, and 3‐MST in cardiac tissues. Scale bars, 25 µm. B)
   Protein levels of CSE, CBS, and 3‐MST in cardiac tissues (n = 4). C)
   Serum H[2]S levels in mice (n = 3). D) Representative M‐mode
   echocardiography images of mouse hearts. E) Quantification of FS%, EF%,
   LVPW, SV, and LVESV (n = 4). F) Representative images of the hearts and
   quantification of heart‐to‐tibia values in mice (n = 4). Scale bars,
   5 mm. G) Representative photographs of H&E‐stained sections of the left
   ventricle (scale bars, 25 µm) and TEM images of left ventricular
   sections (scale bars, 2 µm). The blue arrows indicate abnormal cardiac
   structures. The red arrows indicate damaged mitochondria. H) Protein
   levels of NDUFS3, SDHB, and COX4 in cardiac tissues (n = 3). I) XPD
   protein levels in cardiac tissues (n = 3). All quantitative data are
   presented as mean ± SD from independent experiments. ^** P < 0.01, ^***
   P < 0.001 versus db/+; ^# P < 0.05, ^## P < 0.01, ^### P < 0.001 versus
   db/db by unpaired t test or ordinary one‐way ANOVA.

   To address the deficit in H[2]S, we treated diabetic models with two
   exogenous H[2]S donors: NaHS (a classic H[2]S donor) and GYY4137 (a
   slow‐releasing H[2]S donor). Intraperitoneal injections of NaHS or
   GYY4137 provided exogenous H[2]S supplementation to db/db mice,
   effectively increasing serum H[2]S levels (Figure [168]4C). Similar
   patterns of CSE, CBS, and 3‐MST expression were observed in vitro
   (Figure [169]S3B, Supporting Information). A significant decrease in
   H[2]S levels was observed in the HG/PA‐treated group, which was
   restored by treatment with NaHS or GYY4137 (Figure [170]S3C, Supporting
   Information).

   In db/db mice, NaHS or GYY4137 treatment significantly improved cardiac
   function (Figure [171]4D). Enhanced echocardiographic parameters,
   including EF%, FS%, SV, LVPW, and LVESV, were observed in H[2]S‐treated
   db/db mice (Figure [172]4E). H[2]S supplementation also alleviated
   abnormal cardiac enlargement in these mice (Figure [173]4F).
   Histological analyses with H&E staining and TEM demonstrated that
   exogenous H2S alleviated the disordered arrangement of myocardial
   fibers and restored mitochondrial morphology in db/db mouse hearts
   (Figure [174]4G).

   Furthermore, we examined the effects of H[2]S supplementation on Fe–S
   proteins. H[2]S treatment preserved the levels of key Fe–S proteins,
   including NDUFS3, SDHB, and XPD, in db/db mice (Figure [175]4H,I).
   Collectively, these findings suggest that H[2]S supplementation
   effectively enhances cardiac function and increases the Fe–S protein
   levels in db/db mice.

2.5. H[2]S Supplementation Abrogates Mitochondrial Dysfunction and
PARthanatos in Diabetic Cardiomyocytes

   H[2]S is known for its role in reducing oxidative stress. In the
   present study, we investigated the effects of H[2]S supplementation on
   mitochondrial dysfunction and PARthanatos in diabetic cardiomyocytes.
   By elevating the levels of NDUFS3, SDHB, and COX4 proteins, exogenous
   H[2]S increased the activity of mitochondrial complexes in db/db mice
   (Figure [176]5A). This was accompanied by a reduction in ROS
   accumulation in cardiac tissues, demonstrating the antioxidative
   properties of H[2]S (Figure [177]5B). Given its dual effects on XPD and
   ROS, H[2]S effectively reduced the levels of 8‐OHdG, a marker of
   oxidative DNA damage, in the hearts of db/db mice (Figure [178]5C).
   Further investigation revealed that H[2]S supplementation decreased
   PARP1 levels and PAR production (Figure [179]5D,F). Co‐IP analysis
   confirmed a decreased interaction between PAR and AIF in db/db mice
   treated with H[2]S donors (Figure [180]5F). Consequently, the nuclear
   translocation of AIF, a hallmark of PARthanatos, was reduced by H[2]S
   supplementation (Figure [181]5G).

Figure 5.

   Figure 5
   [182]Open in a new tab

   H[2]S supplementation abrogates mitochondrial dysfunction and
   PARthanatos in diabetic cardiomyocytes. A) Activity of mitochondrial
   complexes I, II, and IV in cardiac tissues (n = 4). B) Concentration of
   hydroxyl radicals and hydrogen peroxide in cardiac tissues (n = 3). C)
   Quantification of 8‐OHdG levels in cardiac tissues (n = 3). D) Protein
   levels of PARP1 and PAR in cardiac tissues. E) Quantification of (D) (n
   = 3). F) Interaction between PAR and AIF in cardiac tissues. G) Protein
   levels of nAIF and mAIF in cardiac tissues (n = 3). H) Protein levels
   of NDUFS3, SDHB, and COX4 in cardiomyocytes (n = 3). I) Normalized OCR
   of cardiomyocytes (n = 4). J) Superoxide concentration in
   cardiomyocytes (n = 4). K) γ‐H2AX foci in cardiomyocytes (n = 3). Scale
   bars, 50 µm. L) Protein levels of PARP1, AIF, and PAR in
   cardiomyocytes. M) Quantification of L (n = 3). N) Confocal microscopy
   showing AIF (red) co‐localization with nuclei (blue) in cardiomyocytes
   (n = 4). Scale bars, 20 µm. O) Viability of cardiomyocytes (n = 4). All
   quantitative data are presented as mean ± SD from independent
   experiments. ^* P < 0.05, ^** P < 0.01, ^*** P < 0.001 versus db/+ or
   Ctrl; ^#P < 0.05, ^##P < 0.01, ^###P < 0.001 versus db/db or HG/PA by
   ordinary one‐way ANOVA.

   AP39, a mitochondrion‐targeted H[2]S donor, plays a protective role by
   preserving the activity of mitochondrial complexes.^[ [183]^48 ^]
   Similar to NaHS and GYY4137, AP39 effectively upregulated the subunits
   of respiratory complexes in HG/PA‐treated cardiomyocytes in vitro
   (Figure [184]5H). Therefore, H[2]S supplementation prevented the HG/PA
   induced impairment of mitochondrial respiration (Figure [185]5I).
   Additionally, H[2]S supplementation partially eliminated mitochondrial
   ROS in cardiomyocytes and aided in the restoration of MMP, as
   demonstrated by the results of TMRE staining (Figure [186]5J; Figure
   [187]S3D,E, Supporting Information).

   Exogenous H[2]S also alleviated the accumulation of γ‐H2AX in diabetic
   cardiomyocytes (Figure [188]5K). Both H[2]S supplementation and PARP1
   inhibition (via ABT888) decreased PAR production and nuclear
   translocation of AIF (Figure [189]5L–N). The viability of the
   cardiomyocytes in the HG/PA‐treated groups was significantly improved
   by H[2]S supplementation and ABT888 treatment (Figure [190]5O). These
   findings demonstrate that H[2]S supplementation can safeguard
   mitochondrial function and prevent PARP1 overactivation‐induced cell
   death in diabetic cardiomyocytes.

2.6. H[2]S Promotes Fe–S Cluster Synthesis by S‐Sulfhydration Independent of
Cysteine

   H[2]S supplementation has been demonstrated to enhance the expression
   of Fe–S cluster‐containing proteins, such as NDUFS3, SDHB, and XPD.
   This prompted us to investigate whether H[2]S directly influences the
   Fe–S cluster synthesis system. Our results revealed that NFS1 and ISCU
   protein levels were significantly increased in db/db mice treated with
   exogenous H[2]S (Figure [191]6A). Increasing evidence suggests that
   H[2]S functions as a modulator of S‐sulfhydration.^[ [192]^37 ^] To
   further verify the effects of H[2]S supplementation on S‐sulfhydration,
   we conducted liquid chromatography‐tandem mass spectrometry (LC‐MS/MS)
   analysis on cardiac tissues from db/db mice treated with or without
   GYY4137. The analysis revealed that proteins associated with Fe–S
   cluster biosynthesis, iron ion regulation, and sulfur compound
   metabolism were significantly enriched among the differential
   S‐sulfhydrated proteins (Figure [193]6B). Proteins modified by H[2]S
   were also implicated in pathways related to DCM, including glucose and
   lipid metabolism, citrate cycle, and oxidative phosphorylation (Figure
   [194]S4A, Supporting Information). Indeed, results of biotin switch
   assays revealed that treatment with exogenous H[2]S restored
   S‐sulfhydration of NFS1 and ISCU, indicating the transfer of sulfur
   between these two proteins (Figure [195]6C). In addition, the restored
   mAco activity further confirmed the role of H[2]S in promoting Fe–S
   cluster synthesis (Figure [196]6D).

Figure 6.

   Figure 6
   [197]Open in a new tab

   H[2]S promotes Fe–S cluster synthesis by S‐sulfhydration independent of
   cysteine. A) Protein levels of NFS1 and ISCU in cardiac tissues (n =
   3). B) Functional enrichment of differential S‐sulfhydrated proteins in
   cardiac tissues by LC‐MS/MS analysis. C) S‐sulfhydration levels of NFS1
   and ISCU in cardiac tissues (n = 3). D) Activity of mAco in cardiac
   tissues (n = 4). E) Protein levels of NFS1 and ISCU in cardiomyocytes
   (n = 3). F) Citrate concentration in cardiomyocytes (n = 4). G)
   Activity of mAco in cardiomyocytes (n = 4). mNFS1^WT, wild‐type mouse
   NFS1. H) S‐sulfhydration levels of NFS1 and ISCU in cardiomyocytes (n =
   4). I) Cysteine concentration in cardiac tissues (n = 4). J) Cysteine
   concentration in cardiomyocytes (n = 3). K) S‐sulfhydration levels of
   NFS1 and ISCU in cardiomyocytes (n = 3). CC‐, cysteine free. L)
   Activity of mAco in cardiomyocytes (n = 3). All quantitative data are
   presented as mean ± SD from independent experiments. ^* P < 0.05, ^** P
   < 0.01, ^*** P < 0.001 versus db/+ or Ctrl; ^# P < 0.05, ^## P < 0.01,
   ^### P < 0.001 versus db/db, HG/PA, mNFS1^WT+HG/PA or CC‐ by ordinary
   one‐way ANOVA.

   Similarly, exogenous H[2]S upregulated NFS1 and ISCU expression, and
   decreased cellular citrate levels in HG/PA‐treated cardiomyocytes
   (Figure [198]6E,F). Based on these results, we inferred that NaHS and
   GYY4137 play identical roles in the S‐sulfhydration of NFS1 as H[2]S
   donors.

   To directly supplement the loss of NFS1 in diabetic models, we
   constructed a vector overexpressing wild‐type mouse NFS1 (mNFS1^WT) and
   transfected it into cardiomyocytes. The combined treatment with
   mNFS1^WT and exogenous H[2]S rescued mAco activity and NFS1
   S‐sulfhydration under HG/PA conditions (Figure [199]6G,H). However, the
   expression of mNFS1^WT alone did not elicit significant changes in mAco
   activity or NFS1 S‐sulfhydration (Figure [200]6G,H). These findings
   suggest that H[2]S upregulates NFS1 expression and elevates NFS1
   S‐sulfhydration levels, thereby promoting Fe–S cluster synthesis.

   Considering that cysteine serves as a substrate for NFS1, we further
   investigated whether modification of NFS1 by H[2]S involves changes in
   cysteine levels. Under hyperglycemic and hyperlipidemic conditions,
   decreased cellular cysteine levels were reversed by exogenous H[2]S,
   revealing that NFS1 substrate availability was restored
   (Figure [201]6I,J). Consistent with this, H[2]S treatment also restored
   expression of the cystine transporter xCT, explaining the recuperative
   effect of H[2]S on cysteine levels in diabetic models (Figure [202]S4B,
   Supporting Information). However, we propose that H[2]S modulation of
   Fe–S cluster biosynthesis extends beyond its effect on cysteine
   concentrations.

   To validate this hypothesis, cardiomyocytes were cultured in a
   cysteine‐deficient medium (CC‐). Supplementation of the HG/PA or CC‐
   groups with cysteine restored NFS1 S‐sulfhydration (Figure [203]6K).
   Interestingly, H[2]S also enhanced NFS1 S‐sulfhydration even in the
   absence of cysteine (Figure [204]6K). Under cysteine depletion, both
   the activity and protein levels of mAco significantly decreased,
   whereas NFS1 expression remained relatively unaffected (Figure [205]6L;
   Figure [206]S4C, Supporting Information). Interestingly, the
   introduction of exogenous H[2]S effectively mitigated the impairment of
   mAco caused by cysteine depletion (Figure [207]6L; Figure [208]S4C,
   Supporting Information). These findings indicate that H[2]S directly
   provides sulfur atoms necessary for the synthesis of Fe–S clusters
   through NFS1 S‐sulfhydration, independent of its cellular substrate
   cysteine (Figure [209]S4D, Supporting Information).

2.7. H[2]S Facilitates NFS1 S‐Sulfhydration Dependent on Cysteine 383 Residue

   To determine whether NFS1 S‐sulfhydration is involved in the effect of
   H[2]S on DCM, the mass spectrum was obtained through LC‐MS/MS analysis
   using a time‐of‐flight (TOF) mass analyzer. The analysis revealed that
   NFS1 is highly S‐sulfhydrated at the C383 residue after GYY4137
   treatment in db/db mouse hearts (Figure [210]7A). Additionally, using
   the OxiMouse database, we identified cysteine residues in cardiac NFS1
   of mice susceptible to oxidation when exposed to ROS.^[ [211]^49 ^]
   Among the oxidized cysteine sites, C383 was the most significantly
   modified (Table [212]1 ). This cysteine residue is located within the
   mobile S‐transfer loop domain of NFS1, and is essential for sulfur
   transfer. The active site of NFS1 is highly conserved across different
   species, indicating that its crucial functional role has been
   maintained throughout evolution (Figure [213]S5A, Supporting
   Information). To determine the importance of this site in sulfur
   transfer, we introduced a mutation at C383 in mouse NFS1 and changed it
   to alanine. We constructed a vector expressing this mutated form
   (mNFS1^C383A). We transfected FLAG‐tagged mNFS1^WT or FLAG‐tagged
   mNFS1^C383A into constructed Nfs1 knockdown cardiomyocytes (Figure
   [214]S5B, Supporting Information). The results showed that mNFS1^C383A
   did not undergo S‐sulfhydration (Figure [215]7B). Furthermore, Nfs1
   knockdown disrupted the expression of mitochondrial complex subunits,
   impaired maximal mitochondrial respiration, and facilitated AIF nuclear
   translocation. This impairment was restored by overexpressing mNFS1^WT
   but not mNFS1^C383A (Figure [216]7C–E). These data suggest that C383 is
   crucial for NFS1 function.

Figure 7.

   Figure 7
   [217]Open in a new tab

   H[2]S facilitates NFS1 S‐sulfhydration dependent on cysteine 383
   residue. A) S‐sulfhydration of NFS1 detected by LC‐MS/MS analysis. B)
   S‐sulfhydration level of NFS1 in cardiomyocytes. C) Protein levels of
   NFS1, mAco, NDUFS3, and SDHB in cardiomyocytes (n = 3). D) Normalized
   OCR of cardiomyocytes (n = 5). E) Confocal microscopy showing AIF (red)
   co‐localization with nuclei (blue) in cardiomyocytes (n = 4). Scale
   bars, 20 µm. F) Quantification of EF%, FS%, LVPW, LVEDV, SV, and LVESV
   in mice (n = 4). G) S‐sulfhydration levels of NFS1 in cardiac tissues.
   H) Activity of mAco in cardiac tissues (n = 3). I) Protein levels of
   NDUFS3 and SDHB in cardiac tissues (n = 3). J) Activity of
   mitochondrial complexes I, II, and IV in mouse hearts (n = 3). K)
   Concentrations of hydroxyl radicals (n = 3) and hydrogen peroxide (n =
   4) in cardiac tissues. L) Protein levels of PARP1 and AIF in nuclei of
   cardiac tissues (n = 4). Scale bars, 20 µm. All quantitative data are
   presented as mean ± SD from independent experiments. ^* P < 0.05, ^** P
   < 0.01, ^*** P < 0.001 versus Ctrl; ^# P < 0.05, ^## P < 0.01, ^### P <
   0.001 versus Nfs1‐KD or db/db by ordinary one‐way ANOVA.

Table 1.

   Oxidized cysteine sites in mouse NFS1.
   Protein (Unitprot ID)    Site Oxi‐percent (heart)
   Mouse NFS1 ([218]Q9Z1J3) C224     6.25 ± 1.30
   Mouse NFS1 ([219]Q9Z1J3) C310         NA
   Mouse NFS1 ([220]Q9Z1J3) C383    30.29 ± 1.07
   Mouse NFS1 ([221]Q9Z1J3) C428    11.22 ± 0.30
   [222]Open in a new tab

   The percentage of oxidized cysteine sites (Oxi‐percent) in mouse NFS1
   was retrieved from the OxiMouse database
   ([223]https://oximouse.hms.harvard.edu).

   We hypothesized that the C383 of NFS1 is the target of H[2]S, which
   restores the function of NFS1 under oxidative stress. To test this
   hypothesis, mNFS1^C383A was expressed in db/db mice (Figure [224]S5C,D,
   Supporting Information). Assessment of cardiac function revealed that
   in db/db mice expressing mNFS1^C383A, supplementation with H[2]S did
   not improve impaired cardiac contractility (Figure [225]7F). This
   result indicates that the efficacy of H[2]S in promoting cardiac
   function recovery is diminished upon mutation of the C383 site of NFS1.

   In mice expressing the mNFS1^C383A mutant, H[2]S did not elevate NFS1
   S‐sulfhydration levels, thus directly affecting NFS1 function
   (Figure [226]7G). The NFS1 mutation disrupted the beneficial effects of
   H[2]S on mAco activity, suggesting dysregulation of Fe–S cluster
   synthesis (Figure [227]7H). While H[2]S interacts with wild‐type NFS1
   to optimize mitochondrial respiratory function and support energy
   production, the protein levels and activity of mitochondrial complexes
   were significantly compromised in mice expressing mNFS1^C383A
   (Figure [228]7I,J). Furthermore, after the expression of mNFS1^C383A,
   H[2]S failed to decrease ROS accumulation, PARP1 expression, or AIF
   nuclear accumulation, indicating that the protective mechanism of H[2]S
   was blocked by the mutation (Figure [229]7K,L). Altogether, C383 of
   NFS1 is the target of H[2]S modification to modulate Fe–S cluster
   synthesis.

3. Discussion

   Cardiomyocytes require mitochondria for energy production to sustain
   cardiac contractility. Individuals with DCM exhibit abnormal
   mitochondrial respiration and oxidative phosphorylation, which are
   critical for maintaining cardiac function.^[ [230]^1 ^] However, the
   precise mechanisms underlying these abnormalities require further
   investigation. In the present study, we provide new perspectives on the
   mechanisms that contribute to mitochondrial dysfunction in DCM. The
   findings of this study revealed reduced protein levels and activity of
   the components involved in the respiratory chain, citrate cycle, and
   Fe–S cluster assembly complex.

   Fe–S clusters serve as essential cofactors for various proteins, and
   impaired Fe–S cluster synthesis adversely affects the components of
   mitochondrial respiration, citrate cycle, and DNA repair.^[ [231]^22 ,
   [232]^23 , [233]^42 , [234]^50 ^] The core machinery for the de novo
   biosynthesis of Fe–S clusters is located in the mitochondrial matrix
   and relies on sulfur transfer to NFS1 and ISCU through
   S‐sulfhydration.^[ [235]^32 , [236]^51 , [237]^52 , [238]^53 ^] NFS1 is
   a conserved cysteine desulfurase that provides sulfur to nascent Fe–S
   clusters.^[ [239]^31 , [240]^43 ^] Our study demonstrates that Nfs1
   knockdown is detrimental to cellular metabolism. Interestingly,
   diabetes or HG/PA treatment led to decreased S‐sulfhydration levels of
   NFS1 and ISCU in cardiomyocytes, thereby reducing de novo synthesis of
   Fe–S clusters. Furthermore, Fe–S cluster deficiency leads to
   mitochondrial dysfunction, oxidative stress, iron overload, and DNA
   damage. Recognizing Fe–S clusters as regulatory factors in DCM provides
   new insights into potential therapeutic targets related to Fe–S cluster
   metabolism.

   The mitochondrial respiratory chain is the primary source of cellular
   ROS. The mitochondrial redox systems, including thioredoxin 2 (Trx2)
   and peroxiredoxin 3 (Prx3), protect against DCM by regulating
   mitochondrial oxidative stress.^[ [241]^54 , [242]^55 ^] Fe–S clusters,
   which are essential components of the mitochondrial respiratory chain
   and citrate cycle, play crucial roles in maintaining mitochondrial
   function and redox balance.^[ [243]^43 , [244]^44 , [245]^45 ^] In this
   study, Fe–S cluster deficiency impaired electron transport, inducing
   ROS accumulation and mitochondrial dysfunction, which were restored by
   H2S supplementation.

   We believe that iron overload is a consequence of insufficient Fe–S
   cluster synthesis. Previous studies have demonstrated that the levels
   of the iron transport proteins transferrin receptor 1 (TFR1, cell
   membrane) and mitoferrin 1 (MFRN1, mitochondrial membrane) are
   increased in the cardiomyocytes of diabetic mice, which is the direct
   cause of iron overload in the cytoplasm and mitochondria.^[ [246]^39 ^]
   Excessive levels of iron ions further exacerbate oxidative stress, and
   the inhibition of the GSH/GPX4 pathway induces ferroptosis in
   cardiomyocytes, which is another mechanism underlying cardiac damage in
   DCM.^[ [247]^39 ^]

   One of the most significant findings of our study was the role of H2S
   in promoting Fe–S cluster production. H2S, a gasotransmitter known for
   its protective effects on the cardiovascular system,^[ [248]^34 ,
   [249]^35 , [250]^36 ^] enhances Fe–S cluster synthesis. We present
   evidence that this effect is mediated by direct S‐sulfhydration of NFS1
   at C383, independent of intracellular cysteine levels. By enhancing the
   sulfur availability and supporting the functionality of NFS1, H2S
   facilitates the de novo synthesis of Fe–S clusters, thereby improving
   mitochondrial function in DCM.

   This study also highlights the protective role of H2S in preventing
   PARthanatos, a cell death mechanism associated with the excessive
   expression and activation of PARP1.^[ [251]^18 , [252]^19 , [253]^20 ,
   [254]^21 ^] Under hyperglycemic and hyperlipidemic conditions, PARP1 is
   upregulated, leading to nuclear translocation of PAR‐associated AIF. A
   recent study showed that augmented nuclear localization of
   truncated/oxidized AIF is responsible for DNA laddering and cell death
   in arrhythmogenic cardiomyopathy.^[ [255]^56 ^] In the present study,
   H2S was found to mitigate PARthanatos by preventing oxidative DNA
   damage and promoting DNA repair. The ability of H2S to preserve cell
   viability under hyperglycemic and hyperlipidemic conditions provides
   further evidence for its therapeutic potential in DCM. Nevertheless,
   additional research is required to fully understand the intricate
   interplay between these processes.

   The present study has some limitations. First, the experiments explored
   multiple pathways related to Fe–S cluster synthesis deficiency in DCM;
   however, Nfs1 knockout animal models were not utilized, which could
   provide more definitive views on the role of NFS1. Second, although we
   demonstrated that H2S increased the protein levels of NFS1 and ISCU,
   the precise mechanisms by which H2S enhances their expression remain
   unresolved. Understanding the underlying pathways may further elucidate
   the therapeutic potential of DCM.

4. Conclusion

   In our study, we demonstrated that impaired NFS1 expression and
   function lead to mitochondrial dysfunction and PARthanatos in DCM.
   H[2]S supplementation can enhance Fe–S cluster synthesis by promoting
   the S‐sulfhydration of NFS1 at C383, thus maintaining mitochondrial
   function and DNA integrity. These findings elucidate new mechanisms
   underlying cardiac dysfunction in DCM and underscores the therapeutic
   potential of H[2]S for alleviating these abnormalities.

5. Experimental Section

Animal Experiment

   The Harbin Medical University Medical Science Ethics Committee approved
   all animal experiments performed in this study (HMUIRB2022015). The
   T2DM model was established using female homozygous leptin
   receptor‐deficient (db/db) mice, while heterozygous (db/+) mice served
   as the control group. The mice were procured from the Animal Laboratory
   Center of Nanjing University (China). The mice were housed in
   pathogen‐free facilities, maintained under controlled conditions of
   22 °C, 55% humidity, and a 12/12 h day/night cycle. Two‐thirds of db/db
   mice were randomly chosen, with one‐third administered intraperitoneal
   injections of NaHS (40 µmol kg^−1) and the other third receiving
   intraperitoneal injections of GYY4137 (133 µmol kg^−1), both starting
   at 8 weeks of age.

Echocardiograph

   An echocardiography system (GE VIVID7 10S, St. CT., Fairfield) was used
   to evaluate the cardiac function of mice. Prior to the examination, the
   mice were administered Avertin (240 mg kg^−1) to induce sedation and
   chest hair was removed as a preconditioning measure. The mice were
   subsequently placed in a supine position on an animal handling platform
   for parameter measurement. Notably, all procedures were executed under
   double‐blind conditions with regard to both genotype and treatment.

Hematoxylin and Eosin (H&E) Staining

   The cardiac tissues were fixed in 4% paraformaldehyde overnight and
   dehydrated in graded ethanol. After clearing in xylene, samples were
   embedded in paraffin, and cut into 5–10 µm‐thick sections. Following
   the deparaffinization, the samples were rehydrated using immersing in
   xylene and alcohol. For H&E staining, the sections were stained with
   hematoxylin for 10 min and eosin for 2 min. After washed, samples were
   dehydrated and observed.

Transmission Electron Microscopy (TEM)

   The cardiac tissues from the left ventricular hearts of mice were
   prepared for semi‐thin sections. The samples were fixed in 2.5%
   glutaraldehyde at 4 °C for 12 h and treated with 1% osmium tetroxide
   for 2.5 h. After being embedded in Epon 812 (Electron Microscopy
   Sciences, USA), the sections were stained with uranyl acetate and
   detected under a Zeiss Axiophot microscope.

RNA‐Seq and Analysis

   For RNA sequencing, total RNA underwent poly‐T oligo‐attached magnetic
   bead purification, followed by fragmentation using divalent cations.
   First‐strand cDNA synthesis utilized random hexamer primers, and
   second‐strand cDNA synthesis was performed with DNA Polymerase I and
   RNase H. Blunt ends were generated, and adenylation of 3′ ends was
   carried out. Ligation of NEBNext Adaptor with a hairpin loop structure
   facilitated hybridization. Library fragments were size‐selected
   (≈240 bp) and purified with the AMPure XP system. Enzyme‐treated,
   adaptor‐ligated cDNA underwent PCR with Phusion High‐Fidelity DNA
   polymerase and Universal PCR primers. PCR products were purified, and
   library quality was assessed. Clustering of index‐coded samples was
   performed with the TruSeq PE Cluster Kit v4‐cBot‐HS (Illumina) on a
   cBot Cluster Generation System. Differential expression analysis
   employed DESeq2, with adjusted P‐values < 0.05 indicating differential
   expression. Gene Ontology (GO) enrichment analysis of DEGs utilized the
   GOseq R package, addressing gene length bias through the Wallenius
   non‐central hyper‐geometric distribution.

Enrichment Analysis of Differentially Expressed Genes (DEGs) in DM Hearts

   Single‐cell RNA sequencing data of mouse hearts were retrieved from the
   NCBI Gene Expression Omnibus (GEO) repository under accession Series
   [256]GSE213337. The data was processed using Seurat (Version 4.3.0),
   and cells with high ribosomal expression (Percent.ribo < 10) were
   filtered out. Subsequently, the cells were annotated, and differential
   analysis between DM and Ctrl was performed specifically on
   cardiomyocytes (Tnni3, Fabp3, Tnnc1, Mb, Actc1). A total of 864 DEGs
   (94 up‐regulated and 770 down‐regulated, Log2FoldChange > 0.25) were
   subjected to enrichment analysis for GO and KEGG pathways, as well as
   GSEA analysis using clusterProfiler (Version 4.6.2) with a P‐value
   cutoff of 0.5 and a q‐value cutoff of 0.5.

Isolation of Primary Cardiomyocytes

   Primary cardiomyocytes were isolated from neonatal mice using a
   standardized procedure. Briefly, the neonatal mice (1–3 days old) were
   euthanized following approved ethical guidelines. The hearts were
   rapidly excised, rinsed in ice‐cold PBS to remove blood, and then
   digested in pancreatin for 20 min at 37 °C. After being minced into
   small pieces, the tissues were subjected to enzymatic digestion using
   collagenase at 37 °C to dissociate the cells. Following digestion, the
   suspension was mixed with DMEM to terminate the reaction. After several
   digestion cycles, the cell suspension was centrifuged at 1500 rpm for
   10 min to collect the cells. The isolated cells were pre‐plated and
   incubated for 2 h to reduce unwanted contamination. After allowing
   fibroblasts to attach, the supernatant containing the cardiomyocytes
   was carefully filtered and collected to a new culture dish for further
   cultivation under controlled conditions.

Cellular Experiment

   Primary cardiomyocytes were cultured in DMEM (with 10% FBS) and
   incubated at 37 °C in a humidified incubator with 5% CO[2].
   Cardiomyocytes were randomly divided into following groups: Ctrl, HG/PA
   (glucose 40 mm, palmitate 200 µm, Sigma), HG/PA + NaHS (sodium
   hydrosulphide 100 µm, Sigma), HG/PA+GYY4137 (100 µm, Sigma), HG/PA+AP39
   (100 nm, MCE), HG/PA+ABT888 (10 µm, MCE), HG/PA+Cysteine (1 µm, MCE),
   CC‐ (cysteine and cystine‐free DMEM, Gibco), CC‐+GYY4137. Drugs were
   added to the cultured medium directly. HG and PA were used to mimic
   hyperglycemia and hyperlipidemia condition.

Immunoblotting

   Cardiac tissues or cells were collected and lysed using RIPA Lysis
   Buffer (Beyotime) or Cell Lysis Buffer for Western and IP (Beyotime),
   supplemented with PMSF (Beyotime). The protein extract was separated by
   electrophoresis on a 10% or 12% SDS polyacrylamide gel and transferred
   to a polyvinylidene fluoride (PVDF) membrane (BioTrace) using the wet
   transfer method. The PVDF membrane was then subjected to an 8‐h
   incubation at 4 °C with anti‐NDUFS3 (Proteintech, 1:1000), anti‐SDHB
   (Proteintech, 1:1000) anti‐UQCRQ (Proteintech, 1:1000), anti‐COX4
   (Proteintech, 1:1000), anti‐mAco (Proteintech, 1:1000), anti‐AIF
   (Proteintech, 1:1000), anti‐PAR (Enzo, 1:800), anti‐PARP1 (Proteintech,
   1:1000), anti‐NFS1 (Proteintech, 1:1000), anti‐ISCU (Proteintech,
   1:1000), anti‐FXN (Proteintech, 1:1000), anti‐XPD (Proteintech,
   1:1000), anti‐CSE (Proteintech, 1:1000), anti‐CBS (Proteintech,
   1:1000), anti‐3‐MST (Abclonal, 1:1000), anti‐FLAG (Proteintech,
   1:1000), anti‐Histone 3 (Proteintech, 1:1000), anti‐VDAC1 (Proteintech,
   1:1000), or anti‐TUBB (Proteintech, 1:1000) followed by the application
   of a secondary HPR conjugated antibody for 1.5 h at room temperature.
   After washing with TBST for 20 min, the signal was detected using
   chemiluminescence. Densitometry analysis was performed using the image
   processing and analysis program Image J.

Biotin Switch Assays

   The biotin switch assay was conducted following established
   procedures.^[ [257]^57 ^] Briefly, cardiac tissues or cardiomyocytes
   were homogenized in 250 mm HEN buffer (250 mm HEPES, 1 mm EDTA, 0.1 mm
   neocuproine, and 100 µm deferoxamine) and centrifuged at 13 000 g for
   30 min at 4 °C. Subsequently, 500 µg of lysates were mixed with
   blocking buffer (HEN buffer with 2.5% SDS and 20 mm MMTS) at 50 °C for
   25 min with frequent vortex. To eliminate MMTS and precipitate
   proteins, cold acetone was added. Samples were precipitated at −20 °C
   for 1 h and proteins were re‐suspended in HEN buffer containing 1% SDS.
   Then, 4 mm biotin‐HPDP was added to the suspension. Following a 3‐h
   incubation at 25 °C, biotinylated proteins were captured using
   streptavidin‐agarose beads and subsequently washed with HEN buffer. The
   biotinylated proteins were analyzed through immunoblotting.

Aconitase Activity Measurement

   The determination of aconitase activity was preformed using Aconitase
   Activity Kit (Solarbio). Briefly, the cardiac tissues or cardiomyocytes
   were homogenized on ice with a homogenizer or mortar pulp. The
   homogenate was sonicated and centrifuged at 11 000 g for 15 min at
   4 °C. The absorbance of the supernatant was measured at 240 nm using a
   Uvmini‐1240 spectrophotometer (Shimadzu), and the activity was
   calculated according to the instructions.

Citrate Measurement

   The concentration of citrate in the cardiac tissues and cardiomyocytes
   was detected using the Citrate Content Kit (Solarbio) according to the
   instructions. cardiomyocytes were collected and thoroughly grinded on
   ice, followed by centrifuging at 11 000 g at 4 °C for 10 min. The
   absorbance was measured at 545 nm.

Detection of Fe^2+

   The Fe^2+ was detected using BioTracker 575 Red Fe^2+ Dye
   (Sigma–Aldrich). Briefly, the cardiomyocytes were incubated with
   serum‐free media containing Fe^2+ Dye (5 µm) for 1 h under 37 °C (in
   dark). After incubation, cardiomyocytes were washed by PBS carefully.
   The fluorescence intensity was detected by a fluorescence microscope.

Isolation of Nucleus

   To separate the nucleus of cardiac tissues and cardiomyocytes, a
   Nucl‐Cyto‐Mem Preparation Kit (Applygen, China) was used following the
   manufacturer's instructions. Nuclear fractions were then detected for
   protein concentration with the BCA Protein Assay kit (Solarbio). Then,
   equal amounts of protein concentration were analyzed by western blot.
   Histone 3 (H3) was used as a nuclear marker.

CRISPR/Cas9‐Mediated NFS1 Deletion

   The experiment utilized a lentiviral vector carrying the CRISPR/Cas9
   system to knock down the Nfs1 gene in cardiomyocytes. Guide RNAs
   (gRNAs) targeting Nfs1 were designed, synthesized, and incorporated
   into a Cas9 expression lentiviral vector. Following the transduction of
   cells with the lentiviral particles, Cas9 induced double‐strand breaks
   in the NFS1 gene. The edited cells were selected using antibiotics.
   Validation involved western blot and functional assays.

Proteomic Analyses

   Proteomic analyses were conducted by Jingjie PTM BioLab Co. Ltd.
   (Hangzhou, China). Cells were lysed and the supernatant was collected
   after centrifugation. Proteins were precipitated with acetone,
   resuspended in TEAB, and digested with trypsin overnight.
   Dithiothreitol (DTT) was used for reduction, followed by iodoacetamide
   (IAA) for alkylation. Peptides were separated using the EASY‐nLC 1200
   system with a gradient of formic acid and acetonitrile and then
   analyzed on an Orbitrap Exploris 480 mass spectrometer. The mass
   spectrometry data were acquired in data‐independent acquisition (DIA)
   mode with high‐resolution scans, and fragmentation was performed in the
   HCD collision cell.

Isolation of Mitochondria

   Tissue Mitochondria Isolation kit (Beyotime) and Cell Mitochondria
   Isolation kit (Beyotime) were used to isolate mitochondria from cardiac
   tissues and cardiomyocytes following the manufacturer's instructions,
   respectively. The final precipitate of mitochondria was resuspended in
   mitochondrial lysis buffer. The protein concentration was determined
   using a BCA Protein Assay kit (Solarbio).

Mitochondrial Complex Activity

   The activity of mitochondrial complexes I, II, and IV was assessed
   using the Complex Activity Assay Kit (Cayman) following the
   manufacturer's protocol. Briefly, isolated mitochondria from mice were
   combined with the provided reagents in a well‐mixed manner.
   Subsequently, the prepared mixture was added to a 96‐well plate, and
   the absorbance was measured at 3‐min intervals for a total duration of
   15 min at 25 °C.

Mitochondrial Oxygen Consumption Rate (OCR) Measurement

   To measure cellular mitochondrial OCR, Seahorse XF Cell Mitochondrial
   Stress Test Kit (Agilent) was utilized. Cardiomyocytes were initially
   seeded onto a XF96 microplate at an optimized density of 5000
   cells/well and incubated overnight at 37 °C in a humidified incubator
   with 5% CO[2]. The culture medium was replaced with an assay medium on
   the day of the experiment. The microplate was then incubated in a
   non‐CO[2] chamber to equilibrate. During the assay, the Seahorse XF96
   analyzer sequentially injected 1.5 µm Oligomycin,1.0 µm FCCP, and
   0.5 µm rotenone/antimycin A, respectively. The Seahorse XF96 analyzer
   continuously monitored the OCR changes in real‐time.

Measurement of Hydrogen Peroxide, Hydroxyl Radicals, and Superoxide

   Hydrogen peroxide was detected using a Hydrogen Peroxide Assay Kit
   (Beyotime) following the manufacturer's protocol. Briefly, 5 mg cardiac
   tissues were homogenized in 100 µL lysis buffer and centrifuged at
   12 000 g for 5 min at 4 °C. The supernatant was collected for
   subsequent assays. In a 96‐well plate, 50 µL of sample or 10 mm
   hydrogen peroxide hydrogen peroxide standard solution and 100 µL of
   assay reagent were added to each well. After 30 min, the absorbance at
   570 nm was then measured.

   Hydroxyl radicals were determined using a Hydroxyl Free Radical Assay
   Kit (Nanjing Jiancheng). Cardiac tissue homogenates were diluted to a
   10% concentration using physiological saline. A volume of 20 µL of
   homogenate was then subjected to the measurements outlined in the
   tissue determination protocol.

   Superoxide was measured using a Superoxide Assay Kit (Beyotime)
   following the manufacturer's instructions. 5000 cardiomyocytes were
   seeded in each well of a 96‐well plate. Upon reaching a cell fusion
   rate of 80–90%, the culture medium was aspirated, and the cells were
   washed once with PBS. Subsequently, 200 µL of superoxide detection
   working solution was added to each well, followed by incubation at
   37 °C for 3 min. The treatment factors were introduced into the
   detection wells for 8 h. Absorbance was measured at 450 nm.

Flow Cytometry Analysis of Mitochondrial ROS

   For Fow cytometric measurement, the cardiomyocytes were harvested from
   the DMEM and washed once in warmed PBS. As many as 0.5 × 10^6 cells are
   suspended in 0.5 mL of DMEM in an Eppendorf tube containing 2.5 µm
   MitoSOX (Invitrogen) at 37 °C for 20 min protected from light. After
   incubation, the cells are then washed gently three times with warmed
   PBS, and then resuspended in warm PBS for detection. Then the Eppendorf
   tubes were transferred to a BD FACSCelesta cell analyzer for recording
   the percentage of stained cells and mean fluorescence intensity of
   MitoSOX.

Dihydroethidium (DHE) and MitoSOX Staining

   In cultured cardiomyocytes, the cytosolic ROS was detected by 25 µm DHE
   (Beyotime) at 37 °C for 30 min, and the mitochondrial ROS was detected
   by 5 µm MitoSOX (Invitrogen) at 37 °C for 20 min. After incubation,
   cardiomyocytes were washed by PBS three times. Then the fluorescence
   intensity was viewed with the fluorescence microscope.

Assessment of Mitochondrial Membrane Potential (MMP)

   The MMP was measured by the Mitochondrial Membrane Potential Assay Kit
   with TMRE (Beyotime). The cardiomyocytes were incubated with TMRE
   (500 nm) for 30 min at 37 °C. Then, cells were washed with PBS for
   three times. The fluorescence intensity was detected by a fluorescence
   microscope.

The 8‐Hydroxy‐2′‐deoxyguanosine (8‐OHdG) Measurement

   The content of 8‐OHdG in mice was measured using the Mouse 8‐OHdG ELISA
   Assay Kit (Mlbio, China) following the manufacturer's instructions for
   incubation with samples and standards. After washing, an enzyme‐labeled
   secondary antibody is added to form a sandwich complex. Subsequent
   washing is performed, and a substrate solution is added to initiate a
   color reaction. The absorbance of each well is measured, and a standard
   curve is generated to calculate the concentration of 8‐OHdG in the
   samples.

Co‐Immunoprecipitation (Co‐IP)

   Cardiac tissues or cardiomyocytes were lysed and diluted at the
   concentration of 2 µg µL^−1. 320 µg protein in each sample was used for
   the Co‐IP detection. The lysates mixed with antibodies (15 µg/500 µg
   protein) were incubated with Protein A/G magnetic beads (Bimake, USA)
   for 12 h at 4 °C with gentle rotation. The beads were washed three
   times with lysis buffer after harvested and then were eluted into SDS
   Loading Buffer (Beyotime) for western blot.

Immunofluorescence (IF)

   Cardiomyocytes were seeded on a glass coverslips pretreated with TC
   (Solarbio). After treatment for 24 h, cells were washed three times
   with PBS and fixed with 4% formaldehyde for 1 h. The cardiac tissues
   were subjected to dehydration and embedded in OCT after fixed with 4%
   paraformaldehyde. The tissue blocks were rapidly frozen using liquid
   nitrogen, and cut into 10–20‐micrometer‐thick sections. Subsequently,
   the sections were collected on glass slides. Then, samples were
   permeabilized with 1% Triton‐X100. After blocking, the samples were
   stained with anti‐γ‐H2AX in DNA Damage Assay Kit (Beyotime), anti‐AIF
   (Proteintech, 1:100) or anti‐CSE (Proteintech, 1:100) for 1 h at room
   temperature. After washing, samples were stained with secondary
   antibody and DAPI. The coverslips were mounted on glass slides.
   Confocal imaging was performed on ZEISS LSM 900 microscope with
   confocal modality. The co‐localization analysis was performed using
   Image‐Pro Plus 6.0 software.

Measurement of Cell Viability

   The CCK‐8 assay was used to detect the cell viability. Cardiomyocytes
   were seeded in a 96‐well plate, and the CCK‐8 solution was added
   post‐treatment incubation. The plate was incubated at 37 °C in the dark
   for 1.5 h to allow viable cells to convert CCK‐8 into formazan, which
   was then quantified by measuring absorbance at 450 nm.

Immunohistochemistry (IHC)

   Cardiac tissues were collected and fixed in 4% paraformaldehyde.
   Dehydration and embedding in paraffin were performed. The tissues were
   sectioned into 3–5 µm‐thick slices, affixed to glass slides and
   subjected to dewaxing. The samples were washed with water three times,
   each for 10 min. To repair proteins in the tissues, heat‐induced
   antigen retrieval was employed. Blocking was carried out using a
   blocking solution, followed by overnight incubation with primary
   antibodies at 4 °C. Subsequently, the samples were washed with water
   three times, each for 10 min. Incubation with a universal secondary
   antibody was conducted at room temperature for 1 h. The samples were
   washed with water three times, each for 10 min, and the process of
   chromogenic detection was then initiated.

Detection of H[2]S

   The measurement of serum H[2]S was performed using a H[2]S content
   Assay Kit (Solarbio) following the manufacturer's instructions. The
   methylene blue assay quantified H[2]S by reacting sample H[2]S with
   N,N‐dimethyl‐p‐phenylenediamine sulfate in acidic conditions,
   facilitated by ferric chloride, to form methylene blue. Samples were
   pre‐treated with zinc acetate and then mixed with the reactants. After
   incubation at room temperature for 10 min, the absorbance was measured
   at 665 nm.

   The H[2]S levels in cultured cardiomyocytes were analyzed using a
   selective H[2]S probe, 7‐azido‐4‐methylcoumarin (C‐7Az, Sigma–Aldrich).
   The cells were exposed to PBS containing 50 µm C‐7Az at 37 °C in the
   absence of light for 30 min, followed by two washes with PBS.
   Subsequently, fluorescence microscopy (Olympus, XSZ‐D2) was employed to
   visualize the resulting fluorescence.

Overexpression of NFS1

   The overexpression of wild‐type mouse NFS1 was achieved through plasmid
   transfection using Lipofectamine 3000 Transfection Kit (Invitrogen).
   The cardiomyocytes were cultured and pre‐treated before being
   transfected with the NFS1‐expressing plasmid using Lipofectamine 3000.
   Post‐transfection, the cells were allowed to grow, and NFS1 protein
   expression was analyzed using western blot.

Cysteine Measurement

   The concentration of cysteine in cardiac tissues or cardiomyocytes was
   detected using Cysteine Content Kit (Solarbio) according to the
   instructions. Briefly, cardiomyocytes were collected and ultrasonically
   disrupted. After centrifuging at 11 000 rpm at 4 °C for 10 min, the
   absorbance was measured at 600 nm.

Liquid Chromatography‐Tandem Mass Spectrometry (LC‐MS/MS) Analysis

   LC‐MS/MS analysis was performed by Jingjie PTM BioLab Co. Ltd.
   (Hangzhou, China). Cardiac tissues were trypsin‐digested, and the
   tryptic peptides were separated using a NanoElute UHPLC system with a
   gradient of solvents A (0.1% formic acid, 2% acetonitrile in water) and
   B (0.1% formic acid in acetonitrile). Peptides were then analyzed on a
   timsTOF Pro mass spectrometer in PASEF mode, and MS/MS data were
   processed with MaxQuant software. S‐sulfhydration was identified with a
   false discovery rate (FDR) set to <1%.

Mutation of NFS1

   Adenoviruses expressing GFP and NFS1‐GFP were obtained from Cyagen
   Biosciences Inc. A pM vector harboring full‐length mouse NFS1 with a
   single C383A mutation and GFP cDNA was constructed by inserting it
   between the Kozak and T2A sites. Cardiomyocytes were transfected with
   the adenovirus and incubated for 6–12 h prior to being treated with
   various reagents. Following treatment, protein expression was assessed
   by western blot analysis.

Statistical Analysis

   All statistical analyses were performed with Prism 9.0 (GraphPad). All
   data were expressed as the mean ± standard deviation (SD). Comparisons
   of distributions were made by unpaired t test (between 2 groups) or
   ordinary one‐way ANOVA (between 3 or more groups) followed by Tukey's
   multiple comparisons test. P‐values less than 0.05 were considered
   significant.

Ethical Statement

   Ethical considerations were observed in conducting all animal
   experiments, in line with the Guide for the Care and Use of Laboratory
   Animals approved by the Animal Care Committees of Harbin Medical
   University and International Guiding Principles for Biomedical Research
   Involving Animals (HMUIRB2022015).

Conflict of Interest

   The authors declare no conflict of interest.

Author Contributions

   M.W., S.Z., J.T., and F.Y. contributed equally to this work. M.W.
   contributed to investigation, methodology, visualization, writing
   (original draft), software, data curation, formal analysis, and
   validation. S.Z. was involved in investigation, methodology,
   visualization, writing (original draft), software, formal analysis, and
   validation. J.T. participated in investigation, methodology,
   visualization, writing (original draft), formal analysis, and
   validation. F.Y. contributed to investigation, methodology,
   visualization, formal analysis, and validation. H.C. was involved in
   investigation, methodology, and visualization. S.B. participated in
   investigation and formal analysis. J.K. contributed to investigation
   and visualization. K.P. was involved in investigation and
   visualization. J.H. participated in investigation and visualization.
   M.D. contributed to investigation, visualization, software, and data
   curation. S.D., Z.T., and S.F. contributed to investigation. H.F.
   provided supervision, writing (review & editing), and project
   administration. F.L., B.Y., and S.L. were involved in
   conceptualization, supervision, project administration, resources, with
   S.L. also contributing to writing (review & editing) and validation;
   and W.Z. participated in conceptualization, supervision, writing
   (review & editing), data curation, funding acquisition, project
   administration, resources, and validation.

Supporting information

   Supporting Information
   [258]ADVS-12-2406695-s001.docx^ (1.7MB, docx)

Acknowledgements