Discovery of Diagnostic and Therapeutic Targets For Vascular Injury Induced By Methylglyoxal Using Proteomics

Background: Methylglyoxal, a byproduct of diabetes or the consumption of a high-carbohydrate diet, is associated with vascular injury; however, its molecular mechanisms remain unclear. We aimed to systematically characterize molecular proles and offer unique insights into new disease pathways, thereby contributing to understanding the mechanisms and pathogenesis of vascular injury-related cardiovascular diseases. Methods: Cell survival assays were performed to assess DNA damage; oxidative stress was conrmed by colorimetric assays and quantitative uorescence, and cyclooxygenase-2 and the mitogen-activated protein kinase pathways were assessed using ELISA. Differentially expressed proteins were quantitated via TMT-based LC-MS/MS and bioinformatics analysis, and conrmed by parallel reaction monitoring. Results: Vascular injury was assessed through colorimetric assays, quantitative uorescence, ELISA, and survival assays. Of the 4029 proteins identied, 368 were differentially expressed after methylglyoxal treatment, compared with the negative control; 31 were dened as biomarkers or therapeutic targets according to the Gene Ontology Program, Kyoto Encyclopedia of Genes and Genomes, and protein-protein interaction network analyses. Sixteen proteins were signicantly (p<0.05) upregulated (>1.5-fold change) and 15 were dramatically downregulated (<0.667-fold change) and conrmed through parallel reaction monitoring. Conclusions: The 31 proteins identied as biomarkers or therapeutic targets may contribute to vascular dysfunction through DNA damage, oxidative stress, inammation, autophagy, hypertension, endothelial dysfunction, vascular remodeling, and the coagulation cascade. Additionally, new disease pathways involving the Wnt, ErBb, and BMP signaling pathways were identied; all provide scope as potential underlying molecular mechanisms. Therefore, the 31 proteins identied warrant further development as new therapeutic or diagnostic targets for vascular diseases.


Introduction
In humans, methylglyoxal (MGX) is mainly produced in response to a high dietary glycemic load, inducing alterations in glucose and lipid metabolism [1,2]. The progression of certain diseases-such as vascular injury (VI)-is related to MGX, particularly in people with diabetes, or following consumption of a highcarbohydrate diet [2,3]. Vascular injuries (VIs) are the leading cause of human death [4], and although the mechanism of how MGX induces VI has still not been elucidated, the key risk in their development was 2.2 siRNA Knockdown of FANCA HAECs were cultured in Dulbecco's modi ed Eagle's medium with 10% inactivated FBS in a 5% CO 2 atmosphere at 37°C; to repress FANCA gene expression, siRNA-liposome complexes were formed by mixing FANCA siRNA and siRNA-mate for 10 min at room temperature. HAECs were subsequently incubated for 48-96 h in Dulbecco's modi ed Eagle's medium with 10% inactivated FBS at 37°C in a 5% CO 2 atmosphere.

Survival Assays
For the survival assays, HAECs were cultured as described in Sect. 2.2. After 9-12 h, the cells were plated with fresh medium containing a range of indicated concentrations of MGX (0, 5, 10, 15, and 20 µM) daily. Cells were corrected for the concentration of MGX every 24 h and incubated for 8 days to form colonies.
The HAECs were rinsed twice with PBS after treatment, and only colonies with more than 50 cells were counted. Drug lethal dose (LD 50 ) values-determined based on the MGX dose producing 50% cell damagewere obtained from the survival curves.

Page 5/29
We analyzed the changes in the intracellular GSH level to con rm oxidative stress. GSH was measured using a GSH detection kit immediately after treatment with MGX (LD 50 , 10 µM); untreated cells were used as the negative control group. HAECs (1×10 6 ) were brie y lysed via ultrasound after treatment, mixed with both GSH (1 mM) and reagent 1 (according to the manufacturer's instructions), and incubated at 37°C for 5 min. The results were measured using an absorbance microplate reader (420 nm; Molecular Devices, California, USA) after adding a reaction mixture of reagents 2-5 to the samples and incubating at 37°C for 15 min.

Measurement of the Mitochondrial Membrane Potential
HAECs were treated with MGX (LD 50 , 10 µM) or without MGX (control), and the effect of MGX on mitochondrial function was determined using a mitochondrial membrane potential assay kit. Cells were brie y washed and seeded onto a 96-well plate with 100 µl buffer/well. A working solution of the carbocyanine dye JC-1 (100 µl/well) was added after treatment, followed by incubation in the dark at 37°C for 10 min. The results were detected using an F-2500 uorescence spectrophotometer (Hitachi High-Technologies Corporation, Tokyo, Japan) at λ ex = 529 nm and λ em = 590 nm.

Measurement of Mitogen-Activated Protein Kinase (MAPK) Pathways and COX-2 using Cell ELISA
To determine the effect of MGX, we assessed the changes in MAPK and COX-2 levels using a cell ELISA kit. First, 100 µL of 10,000 cells were seeded into each well and incubated overnight. HAECs were treated with MGX (LD 50 , 10 µM), and untreated cells were used as negative controls. Cells were washed and xed with 100 µL of xing solution and incubated at room temperature for 20 min. Next, quenching buffer (200 µL) was added for 20 min and washed; the blocking buffer (200 µL) was then added, and the cells were incubated at 37°C for 1 h.Third, primary antibodies (ab126425, ab176660, ab176645, ab267646) were added to corresponding wells, while HRP-conjugated secondary antibodies were incubated at room temperature for 2 h and 1 h, respectively. Finally, they were washed; 100 µL of 3,3′,5,5′-tetramethylbenzidine was then incubated and shaken at room temperature for 30 min under light protection; stop solution (50 µl) was then added, and the OD 450 nm of each well was immediately read.

Peptide Fractionation with Strong Cation Exchange Chromatography
Strong cation exchange chromatography was brie y performed using the AKTA Puri er system (GE Healthcare, Illinois, USA). Gradient elution involving 0-10%, 10-20%, 20-45%, and 50-100% of buffer B (500 mM KCl and 10 mM KH2PO 4 in 25% acetonitrile, pH 2.7) was used to elute the peptides at a ow rate of 1 ml/min for 2, 25, 5, and 5 min, respectively; peptides were collected according to the manufacturer's instructions. All fractions were pooled and desalted with C18 standard density cartridges (Empore SPE, 7 mm/3 mL; Sigma-Aldrich, Missouri, USA), vacuum centrifugation, and reconstituted peptides with tri uoroacetic acid. 2.7.3 Liquid chromatography (LC) -electrospray ionization MS/MS analysis using a fusion orbitrap Brie y, 5 µl of each fraction was analyzed using the Q Exactive HF-X Orbitrap Mass Spectrometer with the Easy nLC system (Thermo Fisher Scienti c, Massachusetts, USA), and precursor ions were selected from the survey scan (300-1800 m/z) to obtain MS data. Predictive automatic gain control was used to identify the target value at a resolution of 70000 at m/z 200.

Bioinformatics
Gene ontology (GO) annotated histograms were determined using the Blast2GO program (https://www.blast2go.com/), while the Kyoto Encyclopedia of Genes and Genomes (KEGG) database was used for protein pathway analysis according to the KAAS program (http://www.genome.jp/kaasbin/kaas_main). Protein-protein interaction (PPI) networks were analyzed using the publicly available program STRING (http://string-db.org/); the minimum required interaction score was set to 0.400. Four sources-genomic context, high-throughput experiments, coexpression, and previous knowledge-were used for direct or indirect PPIs according to STRING.

Parallel Reaction Monitoring (PRM)
PRM analysis was used to verify different protein expression levels. The TMT protocol (2.4.1 Protein Preparation and TMT Labeling) was brie y used to prepare the target peptide, and a standard internal reference was prepared using the stable isotope AQUA peptide. Next, 5-35% acetonitrile gradients were used to perform liquid chromatography for 45 min; Q Exactive Plus MS was used for PRM analysis. The energy of collision, state of charge, and retention time of the target peptides were processed using an optimized protocol. Raw data were subsequently analyzed using Skyline (MacCoss Lab, Seattle, USA), and protein expression levels were determined and compared with those of the control.

Statistical Analysis
The MS data were analyzed using MaxQuant software version 1.6.5.0 (http://maxquant.org/). Functional annotations were performed by Database for Annotation, and PPI analysis was performed using the STRING database (https://string-db.org/). Data are presented as the mean ± standard deviation from the results of three to ve repetitions. The Student's t-test was used to evaluate the statistical differences for paired data, and statistical analyses were performed using SPSS software (version 22.0, SPSS Inc., Illinois, USA); p < 0.05 was considered statistically signi cant. HAECs were treated with either MGX (LD50, 10 µM) or without (control) to detect its effect on the DNA damage, disruption of GSH, and depletion of the mitochondrial membrane potential, MAPK pathways, and COX-2 expression.. To determine whether MGX induced DNA interstrand crosslinks in HAECs, cells were exposed to MGX in the culture medium. HAECs de cient in the FANC pathway (LD 50 = 9 µM) were moderately sensitive to MGX, compared with HAECs con rmed to be positive for the FANC pathway (LD 50 = 10 µM; Fig. 1A); therefore, we established DNA damage to be one of the mechanisms of VI; MGX treatment (LD50,10 µM) was performed to determine the level of GSH (62% relative to the negative control group; The activation levels of the MAPK pathway members, including P-ERK (185%), P-JNK (298%), and P-P38 (157%), increased immediately after MGX treatment (Fig. 1E), thus contributing to the vascular remodeling, disintegrity and dysfunction of HAECs and autophagy.  Table 1), of which 228 were upregulated and 141 were downregulated. Statistical analysis (p-value < 0.05) and fold change (FC) cutoff values (FC > 1.5, or < 0.667) referred to the quantitative protein concentration ratios between the MGX treatment and negative control groups. Cluster analysis showed that the protein expression pattern in the treatment group was signi cantly different from that in the negative control group (Fig. 2).

Differentially Expressed Proteins Analyzed through KEGG Pathway Annotation and Enrichment
KEGG pathway enrichment analysis was used to determine the functions of the 369 differentially expressed proteins. The top 10 annotated KEGG pathways are shown in Fig. 3. Genetic information processing (e.g., ribosomes, and DNA damage and repair, such as nucleotide excision repair, protein processing in the endoplasmic reticulum, and ubiquitin-mediated proteolysis) was most signi cantly affected, followed by metabolism (e.g., biosynthesis of unsaturated fatty acids). KEGG pathway annotation analysis indicated that proteins signi cantly different from those identi ed during MGX treatment (LD 50 , 10 µM) correlated with biological functions including DNA damage and repair, oxidative stress, abnormal lipid metabolism, signaling (such as MAPK) pathways, thyroid hormone signaling pathways, autophagy, immune and platelet activation, and so on. More importantly, we formulated unique insights into new disease pathways involving the Wnt, ErBb, and BMP signaling pathways (Supplementary Table 2). .

GO Annotation Enrichment Analysis
GO analysis includes three different aspects: cellular components, molecular function, and biological processes. Among the biological processes, the term "biological process" forms the highest percentage, followed by the term "regulation of biological process" (Fig. 4A). Cellular component GO analysis revealed that the terms "intracellular part" and "cytoplasm" were the most crucial factors (Fig. 4B). Accordingly, protein function indicated that protein, RNA, receptor, or enzyme binding were highly enriched, suggesting that MGX has a strong in uence on molecular function classi cation (Fig. 4C). All biological behaviors coordinate with each other and depend on different proteins; thus, we manually analyzed every record in Supplementary Table 1 and classi ed every differentially expressed protein to further understand the potential underlying molecular mechanisms and offer potential biomarkers or therapeutic targets (Table 1). Table 1 Classi cation of each differentially expressed protein according to the GO analysis Note: Bold-type represents nodes that connect more than two different nodes; numbers in parentheses represent the number of different nodes connected by the same node.

PPI Network for Differentially Expressed Proteins
To further understand the biological functions of the differentially expressed proteins, a PPI network was mapped according to the STRING interaction scores (Fig. 5).
PPIs are essential for all biological behaviors; differentially expressed proteins may have the same or better biological function, and proteins with high connectivity may be biomarkers or therapeutic targets affecting the entire biological system. We manually analyzed the biological function of each differentially expressed protein and counted the number of connected nodes that connect to the same node (Supplementary Table 3 and Table 1). .

Validation of Proteins by PRM
According to GO annotation enrichment analysis and the number of connected nodes that connect to the same node, we manually classi ed and arranged some key points as biomarkers or therapeutic targets, we used PRM analysis to con rm the FC 1 ( Table 2).

Discussion
Exogenous MGX is present in foods and is an important pollutant, released into the environment through activities such as smoking [12]; however, high levels of endogenous MGX have been found in the plasma of diabetic patients as the main dicarbonyls [3]. MGX causes VI through DNA damage (Fig. 1A) [13], oxidative stress and mitochondrial dysfunction (Fig. 1B-C) [8], in ammation (Fig. 1D) [14], autophagy (Fig. 1A-E) [3], and hypertension (Fig. 1A-E) [13]. These markers were measured using, making it di cult to re ect the true complexity of VI based on speci c pathophysiological concepts. To date, traditional technology has failed to dramatically promote the development of clinical medicine due to high labor and economic costs; these are essential to measure the large amounts of proteins, as well as analyze the bulky, di cult-to-explain clinical results.
To the best of our knowledge, the current study is the rst to employ large-scale proteomics to determine the following information after MGX treatment of HAECs; rst, the constructed mechanisms of these important molecular pro les ( Finally, we determined proteins that were dramatically up-or downregulated as a biomarker or drug target based on either the FC (such as accession numbers H7C2W5 or B7Z553), proteins with high connectivity (such as protein accession number A0A2R8Y7V5), or both the FC and proteins with high connectivity, indicating that all biomarkers or therapeutic targets were medically effective.. For example, compared with the negative control, the ratio of protein accession numbers A0A2R8Y7V5 and E9PPA5 were signi cantly downregulated; this indicated that they were injured and could not work after MGX treatment, resulting in a better therapeutic effect if the level of injured proteins is increased. A few proteins (Table 2) have been listed as diagnostic biomarkers in other papers such as protein accession number P01034; however, these studies either focused on other diseases (such as acute aortic dissection and not VI), or did not include MGX treatment [15].
Our results showed that the FANC pathway contributes to the attenuation of interstrand crosslinks induced by MGX (Fig. 1A); however, to our knowledge, no studies have identi ed the key proteins involved in DNA serine/threonine-protein kinase Chk1 (protein accession number E9PPA5), which weakens DNA damage signaling on autosomes [17]; protein accession number P16104-referring to histone H2AX-which is related to DNA double-strand breaks and helps to recruit DNA repair factors [18]; and protein accession number B4DWZ4, which refers to FEN1, a lethal, synthetic gene resulting in both BRCA1 and BRCA2 loss-offunction [19]. Compared with the negative control, the ratio of protein accession numbers A0A2R8Y7V5 and E9PPA5 were signi cantly downregulated, indicating that they were injured after MGX treatment; conversely, upregulation of protein accession numbers P16104 and B4DWZ4 means that they worked to repair DNA damage.
Oxidative stress is a key factor that damages HAECs via MGX. Protein accession number A0A2R8Y7V5 refers to a Fanconi anemia subtype D2 protein (BRCA1 or FANC D 2 ), which may be used in a potential strategy to inhibit energy metabolism by reprogramming mitochondrial metabolism [20]. The cystatin C protein (protein accession number P01034) can inhibit the NF-κB signaling pathway to reduce oxidative stress [21], while the E3 ubiquitin-protein ligase RBX1 protein (protein accession number P62877)-an essential component of the Cul1-Rbx1-Skp1-F boxSkp2 SCF complex-mediates ubiquitination [22]. All three proteins were downregulated, indicating that oxidative stress dramatically damaged HAECs (Table 2); GSH detection results con rmed this conclusion (Fig. 1B).
In addition to producing ATP, the mitochondria are also related to oxidative stress, since the mitochondrial respiratory chain increases the production of reactive oxygen species. The 28S ribosomal protein S2, mitochondrial (protein accession number: Q9Y399) is a necessary component to maintain the structural and essential function of the mitochondria [23], while the APP gene (protein accession number: H7C0V9) relates to age-associated mitochondrial dysfunction in APP/PS1 transgenic mice [24]. The EIF5A gene (protein accession number: I3L397), as well as polyamines regulates mitochondrial substrate oxidation through hypusination of EIF5A [25]. Characterization of protein accession number H7C2W5 variants in wild and spontaneously hypertensive rats indicated the potential pathogenesis of hypertension [26]. Compared with the negative control, protein accession number H7C0V9 was downregulated, indicating that mitochondria were injured after MGX treatment; this was veri ed by the detection of depleting mitochondrial membrane potential (Fig. 1C). Upregulation of 28S ribosomal protein S2 and eukaryotic translation initiation factor 5A thus, they work to maintain mitochondrial function..
Additionally, we identi ed several proteins related to the MAPK pathway and measured the activation of these pathways (Fig. 1E) to con rm their biomarkers or therapeutic targets after MGX treatment. E3 ubiquitin-protein ligase RBX1 (protein accession number: P62877) is important for RING-type E3 ligases [27], whereas protein accession number P40306, a proteasome subunit beta type-10, activates the MAPK pathway, enhancing PI resistance by increasing E3 ubiquitin-protein ligase RBX1 capacity [28]. Additionally, ubiquitin-60S ribosomal protein L40 (protein accession number: P62987) and several associated pathways, including MAPK/ERK, might be involved in the development of diffuse large B cell lymphoma [29].
Compared with the negative control, protein accession number P62877 was signi cantly downregulated, indicating damage after MGX treatment; conversely, protein accession numbers P40306 and P62987 were upregulated, indicating that they worked to activate the MAPK pathway.
Hypertension was con rmed based on the results of the following two proteins: cDNA FLJ58696, highly similar to endoglin (protein accession number: B7Z6Y5), which can both damage and protect vascular function by up-or downregulating the level of protein accession number B7Z6Y5, respectively [30]; and cDNA FLJ58394 (protein accession number: B7Z8Y6), highly similar to platelet endothelial cell adhesion molecule and an attractive target for treating vascular dysfunction through upregulation due to its low expression level [31].
Vascular remodeling was con rmed since the activation of MAPK pathways, DNA damage, oxidative stress, and arterial dysfunction can directly induce vascular structural changes. Our proteomics results also indicated new biomarkers or therapeutic targets. Mutant Apo B 100 (protein accession number: E1A689), which is downregulated after MGX treatment, can decrease lipid transporter activity and promote arterial stiffness [32], while upregulation (FC = 3.15) of proteasome subunit beta type-4 (protein accession number: B4DFL3) can promote apoptosis by inhibiting cell proliferation [33]. Additionally, uncharacterized protein LRP2 (protein accession number: Q4ZG84) upregulation (FC = 1.90) can promote underlying cases of brain vascular malformation [34], and the biological function of vitronectin (protein accession number: B7Z553) promoted vascular remodeling, such as cell adhesion, migration, or proliferation [35].
Exposure to MGX induced HAEC integrity and dysfunction, causing lipid-especially low-density lipoprotein -movement to the artery wall by passive ltration across a compromised endothelium of high permeability.
We have shown that some related biomarkers or therapeutic targets, such as arterial dysfunctional protein accession numbers B7Z6Y5 and B7Z8Y6, can damage vascular functions [30,31]. Upregulation (FC = 9.07) of protein accession number B1ARF3, a type II inositol 1,4,5-trisphosphate 5phosphatase, can control cell migration, adhesion, and polarity by in uencing various aspects of cytoskeletal organization [39].
Exposure to MGX-induced oxidative stress or mitochondrial dysfunction are important factors related to lipid metabolism; for example, MGX causes low-density lipoproteins to react much more actively to the free radical oxidation of lipids [40]. The Fanconi anemia subtype D2 protein (BRCA1) (protein accession number: A0A2R8Y7V5) may potentially be used to inhibit energy metabolism by reprogramming mitochondrial metabolism [20]; ABPP protein (protein accession number: H7C0V9), and age-associated mitochondrial dysfunction in APP/PS1 transgenic mice [24]. The relationship between VI and dyslipidemia is currently of high interest.
Exposure to MGX induced immunity or autophagy, and was con rmed by upregulating the expression of COX-2 ( Fig. 1D)  Up-or downregulation of the following proteins ( increases sumoylation of nuclear c-Myb, which is degraded in response to Wnt-1 signaling [51]. Wnt signaling is increasingly essential in amyloid precursor protein (APP) processing (protein accession number: H7C0V9) [52], as well as the series of molecular signals initiated by binding of a Wnt protein to a frizzled family receptor through protein accession number P62987 [53]; additionally, proteasome subunit beta type-10 (protein accession number: P40306) can modulate the Wnt signaling pathway in MC3T3-E1 cells [54].
Our results also reveal the close relationship between the ErbB or BMP signaling pathways and changes in proteins following MGX treatment. Alzheimer's disease was found to be related to protein accession number H7C0V9, which regulates development from distinct origins, including ErbB signaling pathways [55]. PDK1 (protein accession number: H3BQ10) can cooperate with ERBB-1 through the EGFR-PI3K-PDK1 pathway, regulating ErbB signaling [56], and PRKCA (protein accession number: A0A169TED2) was shown to be associated with neuroactive ligand-receptor interactions and activation of the ErbB signaling pathway [57]. Similarly, the BMP signaling pathway has never been reported to relate to MGX damage. Estrogen receptor alpha (ER) + cells exposed to estradiol induced immediate LRP2 (protein accession number: Q4ZG84; FC = 1.90) activation [58], and TRIM33 (protein accession number: A0A024R0F6; FC = 1.78) was essential for osteoblast proliferation and differentiation via the BMP pathway [59].
To the best our knowledge, this is the rst study to offer three new signaling pathways while simultaneously classifying a large number of biomarkers or therapeutic targets (Table 2) to improve both treatment and prevention of MGX injuries, especially in patients with diabetes. However, this study has several limitations; rst, 31 proteins were de ned as biomarkers or therapeutic targets. These were determined using HAECs, never con rmed with clinical samples, but can guide to con rm all targets were medical effectiveness with clinical specimens. Second, although new disease pathways involving the Wnt, ErBb, and BMP signaling pathways were determined, further information such as how they contribute to the pathogenesis of VI remains unclear. Finally, several proteins (Table 2) have been listed as diagnostic biomarkers in other papers such as protein accession number P01034; however, they focused on diseases such as acute aortic dissection, rather than MGX induced VI [15].    Using the STRING database to determine protein-protein interactions between differentially expressed proteins.Nodes represent the proteins and lines represent protein-protein interactions.