3.1 Succinylproteome profiling analysis of intracerebral hemorrhage brain tissues
TLR4-mediated inflammation is the major contribution to the secondary brain injury after ICH (Fang et al. 2013), and we previously showed that inhibiting TLR4 signaling by its antagonist TAK242 significantly reduced the brain injury via downregulation of inflammation response (Wang et al. 2013). Additionally, considering that protein succinylation is closely related to inflammation. Then, we aimed to investigate how neuroinflammation affects protein succinylation in the perihematomal brain tissue by TAK242 treatment of ICH. Brain tissues were collected from mice with ICH treated with TAK242 or vehicle for 3 consecutive days and from Sham group also as control.
Label free LC-MS/MS was performed to detect the succinylproteome of brain tissues with the anti-succinyllysine antibody. The resulting MS/MS data was processed by MaxQuant search engine (v.1.6.15.0), and the mouse SwissProt database (17045 entries) concatenated with reverse decoy database were searched by tandem mass spectra. The accuracy of identification of the levels of spectrum, peptide, and protein is set as FDR < 1% (Fig. 1A). Then, the quality control of succinylproteomics data was confirmed by the peptide lengths distribution (Fig. 1B) and the first-order mass error of spectra (Fig. 1C). PCA analysis identified better quantitative repeatability among brain samples (Fig. 1D). Motif analysis showed that 4 conserved sequences surrounding lysine succinyllyine sites, and a significant abundance in the frequency of Ala, Gly, Ile and Val adjacent to succinylated Lysine residues were identified in the succinylproteome (Fig. 1E, F).
In this study, we identified a concentration of approximately 6700 succinyllysine events and quantified approximately 3500 sites (Fig. 1A). Approximately 36% of succinylproteins in control (Sham) brains were identified with one succinyllysine site, 16.7% with two succinyllysine sites, 11.1% with three succinyllysine sites, and 36.2% with more than three succinyllysine sites, even approximately 10% with more than 10 succinyllysine sites (Fig. 2A), which were comparable to the percentage of succinylproteins in TAK242- or Vehicle- treated ICH brains (Fig. 2A-C). Venn diagram showed that more than 90% of succinyllysine sites (Fig. 2D) and succinylated proteins (Fig. 2E) were identical among the sham, TAK242- and Vehicle-treatment ICH brains. Conversely, 20 succinyllysine sites on 4 succinylated proteins were found to be succinylated exclusively in TAK242-treated ICH but not control and Vehicle-treated brains (Fig. 2D, E), and 44 succinyllysine sites on 13 succinylated proteins were exclusively found in Vehicle-treated but not control and TAK242-treated ICH brains (Fig. 2D, E).
3.2 TAK242 treatment obviously changed the inflammation-associated protein succinylation after ICH
To explore succinylproteome dynamics after TAK242 treatment, we used a stepwise pipeline (Fig. 3A) to define 139 differentially expressed (DE) succinyllysine sites (FDR < 20%, 121 proteins, Supplementary Table 1) in 40 pathways, adapted the weighted gene correlation network analysis (WGCNA) (Langfelder and Horvath 2008) to cluster these DE succinyllysine sites, and interpreted the clusters with protein-protein interaction (PPI) network. Most of these succinyllysine sites were identified into 4 clusters from this analysis (Fig. 3B) except a few numbers of succinyllysine sites (n = 2) (Supplementary Table 2). Cluster 1 (n = 37) displays an increase in the downregulated protein succinylation caused by ICH after TAK242 treatment, including Ccdc51_K159su, Hapln1_K105su and Nefm_K842su (Fig. 3B). These succinylated events are enriched in 24 pathways: β-Alanine metabolism, propanoate metabolism, gap junction, synaptic vesicle cycle, necroptosis, cGMP-PKG signaling pathway, etc. (Fig. 3C). Cluster 2 (n = 30) shows a marked further decrease in the ICH-associated downregulated protein succinylation after TAK242 treatment, including Atp1a3_K349su and Nipsnap2_K276su (Fig. 3B). Cluster 2 is enriched in 16 pathways, especially the following three pathways: mineral absorption, thyroid hormone synthesis and aminoacyl-tRNA biosynthesis (Fig. 3C). In contrast, Cluster 3 (n = 31) shows a marked further increase in the ICH-associated upregulated protein succinylation after TAK242 treatment (Fig. 3B), identifying 12 pathways, such as lipoic acid metabolism, synthesis and degradation of ketone bodies, butanoate metabolism, and valine, leucine and isoleucine degradation (Fig. 3C). Finally, we found that Cluster 4 shows an increase in ICH-associated downregulated protein succinylation, followed by a notable increase after TAK242 treatment (Fig. 3B); and most of these succinylated proteins were enriched metabolism, such as oxidative phosphorylation (OXPHO), fatty acid degradation, glyoxylate and dicarboxylate metabolism, and ferroptosis, etc. (Fig. 3C). Furthermore, we integrated the DE succinylproteins in each Cluster pattern with the PPI network to define 11 functional modules, providing evidence of PPI to support these enrichment pathways (Fig. 3D).
3.3 Analysis of succinylproteomic data of ICH-associated brain tissues treated with TAK2424
Next, we analyzed the succinylproteomic data between the TAK242- and Vehicle-treated ICH brains to identify how the TLR4-mediated inflammation affects protein succinylation changes in the perihematoma brain tissue by TAK242 treatment of ICH. Firstly, we found that 29 succinyllysine sites of 28 succinylated proteins were increased, and 24 succinyllysine sites on 23 succinylated proteins were downregulated when compared TAK242-treated ICH brains to Vehicle-treated groups (Fig. 4A). GO analysis using A.GO.TOOL (Scholz et al. 2015) showed that, compared to the brain proteome, the brain succinylproteome was overrepresented for GO localization categories of myelin sheath, respiratory chain complex II, succinate dehydrogenase complex, fumarate reductase complex, golgi membrane, T-tubule, plasma membrane, synapse, neuronal cell body membrane, postsynaptic density, etc. (Fig. 4B). Then, we further analyzed the subcellular localization of these significantly changed succinylproteins, and we found that approximately 48% hyposuccinylated proteins were located in the mitochondria and followed by 30.4% in cytoplasm (Fig. 4C). The percentage of succinylproteins located in mitochondria was decreased to about 32% in TAK242-treated ICH brains with concomitant increasing in the percentage of plasma membrane and nucleus to 17.9% and 7.1%, respectively; and peroxisome, cytoskeleton and cytoplasm, nucleus were additionally identified (Fig. 4D).
3.4 Analysis and annotation of differentially succinylated proteins in ICH brains after TAK2424 treatment
We found that ICH-associated hyposuccinylation was comparable with hypersuccinylation in succinylproteins carrying one or two succinyllysine sites per protein after TAK242 treatment when compared to the Vehicle treatment (Fig. 5A). GO cellular component enrichment analysis showed that both hyposuccinylation and hypersuccinylation datasets were significantly enriched for plasma membrane, integral component of membrane and cell periphery (Fig. 5B). Only the hypersuccinylation, but not hyposuccinylation, was significantly linked to structures (e.g., synaptic membrane, presynaptic membrane, presynaptic active zone membrane, presynaptic active zone, postsynaptic specialization, postsynaptic density, and postsynapse), metabolism and its relative enzymes (e.g., succinate dehydrogenase complex, respiratory chain complex II, fumarate reductase complex, and Atpase complex, Atpase dependent transmembrane transport complex), somatodendritic compartment, myelin sheath, cytoplasmic ribonucleoprotein granule, contractile actin filament bundle, actomyosin, etc. (Fig. 5B); whereas hyposuccinylation, but not hypersuccinylation, was enriched for membranes, including vacuolar membrane, trans-golgi network membrane, side of membrane, organelle membrane, membrane, lytic vacuole membrane, lysosomal membrane, golgi membrane, coated vesicle membrane, and clathrin-coated vesicle membrane, coated vesicle and Clathrin-coated vesicle (Fig. 5B). In consistent with the GO localization enrichment results, we found that most of the hyposuccinylated proteins were enriched for metabolism (e.g., sterol metabolic process, secondary alcohol metabolic process, monocarboxylic acid metabolic process, lipid oxidation, cholesterol metabolic process, cellular lipid catabolic process) and transportation (e.g., regulation of anion transport, positive regulation of ion transport, inorganic anion transport, anion transport, anion transmembrane transport) (Fig. 5C). While, only hypersuccinylated proteins were mostly enriched for GO categories linked to synapse working (trans-synaptic signaling, synaptic vesicle transport, synaptic vesicle localization, synapse organization, regulation of synaptic vesicle transport and anterograde trans-synaptic signaling) and vesicle functions (e.g., vesicle organization, vesicle docking, regulation of secretion by cell, regulation of secretion, anion transport) (Fig. 5C). The GO molecular function enrichment analysis showed that both hyposuccinylation and hypersuccinylation datasets were significantly enriched for protein kinase binding and kinase binding (Fig. 5D). However, only hypersuccinylation, but not hyposuccinylation, was significantly linked to succinate dehydrogenase activity, steroid hormone binding, nitric-oxide synthase binding, G-protein coupled receptor binding, chaperone binding, exopeptidase activity, dopamine receptor binding, cytoskeletal protein binding, etc. (Fig. 5D); whereas hyposuccinylation, but not hypersuccinylation, was enriched for symporter activity, solute: cation symporter activity, clathrin binding, carbohydrate binding, C-acyltransferase activity, anion transmembrane transporter activity, and Acetyl-coa c-acyltransferase activity (Fig. 5D). Additionally, the KEGG analysis of these significantly changed succinylated proteins showed that hypersuccinylation was enriched for carbon metabolism, synaptic vesicle cycle, citrate cycle (TCA cycle), carbohydrate digestion and absorption, arginine biosynthesis, SNARE interactions in vesicular transport, etc. (Fig. 6A), while hyposuccinylation was significantly related to fatty acid elongation, fatty acid degradation, fatty acid metabolism, and lysosome. (Fig. 6B).
3.5 Combined transcriptome and succinylproteome analysis in ICH brains treated by TAK242
Furthermore, we performed a combination analysis of a previously reported RNA-seq data of neural cells (Zhang et al. 2014) with our succinylproteome data to reveal the cellular information of ICH-associated significantly changed protein succinylations after TAK242 treatment. Given that neurons, astrocytes, microglia, and endothelial cells are the four kinds of most studied neural cell types in stroke research, therefore, we chose the four kinds of neural cells for this joint analysis. The relatively high expression genes of neural cells were defined as that the gene expression in one of the four neural cell types is higher than all the other three cellular types.
After matching the relatively high expression genes of neural cells with the obviously changed succinylproteomic data, we found that most of the succinylproteins were relatively high expressions in neurons, astrocytes and endothelial cells (Fig. 7A-C). Furthermore, we found that 37% of hypersuccinylated proteins were identified in neurons, 22% in astrocytes, 16% in endothelial cells, and 7% in microglia (Fig. 7B); while, the percentage of hyposuccinylated proteins was decreased to 32% in neurons with concomitant increasing in the percentage of endothelial cells (Fig. 7C). As the specificity expression level of these proteins in nerve cells varies greatly, for example, the ratios of the matched proteins in neurons and astrocyte exceeds 5. Therefore, we further analyzed the information of ratio distribution specificity of these matched succinylated proteins. The pie charts show that the specificity ratios of 7 proteins (Ina, Atp1a3, Cntn1, Vsnl1, Dnm1, Cend1 and Stxbp1) in neurons exceed 10 than other neural cells, and 5 proteins (Mapt, Uchl1, Sncb, Slc25a22, Stx1b) exceed 5 (Fig. 7D). For the astrocytes, there were also 3 proteins’ specificity ratio exceed 10, including Atp1a2, Slc1a3, and Slc6a11, and 1 protein (Acsl6) exceed 5 (Fig. 7E). In contrast, the specificity ratios of all the matched succinylated proteins in the microglia (Fig. 7F) and endothelial cells (Fig. 7G) were under 3. Moreover, we found that the specificity ratios of most of these succinylated proteins in the four neural cells varied from 1 to 3 (Fig. 7D-G), especially for endothelial cells because of 86% protein specificity ratios were under 2 (Fig. 7G).