Comparative proteomic analysis identifies biomarkers for renal aging

Identification of differentially expressed proteins in young and aged mouse kidneys by comparative analysis of proteomics datasets

To investigate the proteome changes in young (8 weeks, n=6) and aged (96 weeks, n=6) mouse kidneys, we carried out a comprehensive proteomic analysis using Tandem Mass Tag (TMT)-based quantitation. In total, 7,208 proteins were identified from three biological replicates with a tight correlation, indicating that these data were highly reproducible. Our data (labeled as Y) were then compared to those by Gygi et al. (labeled as G-dataset) [16] and Kurschat et al. (labeled as K-dataset) [14], as shown in Venn Diagram in Figure 1A. 6302 proteins were identified in G-dataset [16], from which 5084 proteins (80%) were also present in our dataset (Figure 1A), indicating that about 80% proteins were present in both datasets. Kurschat dataset identified only 1555 proteins, from which 967 proteins (62%) were present also in our dataset. These results demonstrated that more than 60% of proteins were consistently identified in three different laboratories. Scatter plot of any two datasets exhibited a tight correlation, indicating that these data were highly similar (R² = 0.7; Figure 1B).

Figure 1. Differentially expressed proteins (DEPs) in young and aged mouse kidneys. (A) Comparison of three proteomic datasets: this study and (14, 16). (B) Scatter plots showing correlations between our data to G-dataset (left) and to K-dataset (right) (Red dot, Peroxisome; Blue dot, Complement and coagulation cascades; Green dot, Retinol metabolism; Yellow dot, PPAR signaling pathway; Pink dot, Fatty acid degradation). (C) Determination of the cut-off thresholds for DEPs. (D) Volcano diagram of DEPs between aged and young mouse kidneys. (E, F) Venn diagram comparing the up-regulated (E) and down-regulated (F) proteins in this study and (14, 16). (G) Hierarchical clustering of DEPs presented in all three datasets.

The statistical significance threshold cutoff for identifying differentially expressed proteins (DEPs) was determined by population statistics (Figure 1C) [17, 18]. Volcano diagram showed that 224 proteins were upregulated (Aged/Young ratio > 1.5 and FDR adjusted p < 0.05; Supplementary Table 1) and 111 proteins were downregulated (Aged /Young ratio < 0.67 and FDR adjusted p < 0.05; Supplementary Table 2) (Figure 1D). About 64% of the upregulated proteins and 59% of the downregulated proteins were also identified in G-dataset (Figure 1E and 1F). Overall, 49 upregulated proteins and 59 downregulated proteins were present in all three datasets; these proteins were considered as the differentially expressed proteins between young and aged mice. The heat-map showed that among the 108 identified DEPs, IgGs and complement components exhibited a significant increase in aged mouse kidneys (Figure 1G and Supplementary Table 1).

Determination of biomarkers for renal aging by proteomic profiling

Ingenuity Pathway Analysis (IPA) of the 108 DEPs showed that these proteins were highly enriched in oxidative phosphorylation pathway, sirtuin signaling, fatty acid β-oxidation pathway, and bile acid biosynthesis pathway. Specifically, the liver X receptors/retinoid X receptor (LXR/RXR) pathway, production of nitric oxide and ROS in macrophages, and the xenobiotic metabolism CAR signaling pathway were activated, while EIF2 signaling and fatty acid β-oxidation were inactivated (Figure 2A). String analysis showed that multiple proteins associated with oxidative phosphorylation, peroxisomes, retinol metabolism, and cholesterol metabolism were differentially expressed in young and aged mice (Figure 2B). All three datasets showed that peroxisome-associated proteins were downregulated in aged mice (Figure 2C). Peroxisomes contain enzymes that catalyze metabolic reactions, such as oxidation of very long-chain fatty acids, synthesis of ether lipids and bile-acid, and degradation of reactive oxygen and nitrogen species [19]. According to the proteomic results, all enzymes associated with fatty acid oxidation were downregulated; the peroxisomal carnitine O-octanoyltransferase (Crot) was decreased by more than 50%. Unlike peroxisomal proteins, which were uniformly decreased in aged mice, proteins associated with PPAR signaling were either up- or down-regulated in aged mice (Figure 2D). Besides, phosphoenolpyruvate carboxykinase 1 (Pck1) was downregulated by 50%, while CD36 was upregulated. Among proteins related to fatty acid degradation, 3-oxoacyl-ACP synthase (OXSM) was also upregulated (Figure 2E). The other identified differentially expressed proteins, Ahsg, Alb, Apoe, C3, Cat, Clu, Gc, Hpx, Ren, and Slc22a6 are associated with renal damage and renal tubule injury [20–29].

Figure 2. Biomarkers for renal aging. (A) Ingenuity Pathway Analysis (IPA) of DEPs in aged and young mouse kidneys analyzed by proteomics. (B) STRING network analysis based on robust Spearman correlation for DEPs in aged and young mouse kidneys. (C–E) Histogram of peroxisomal proteins (C), proteins associated with PPAR signaling (D), and proteins associated with fatty acid degradation pathway (E). (F) A heat-map showing biomarkers for renal aging; the black line indicates that the corresponding protein was not present in K-dataset. Error bars represent ± SEM.

To identify potential biomarkers for renal aging, we excluded proteins associated with subcellular organelles and abundant plasma proteins. Specifically, we considered proteins that are present in plasma and/or urine, and are highly expressed in renal tissues as biomarkers. Using these criteria, we selected 27 proteins as potential biomarkers for renal aging (Figure 2F and Table 1); expression of 19 proteins was increased, while expression of 8 proteins was decreased in aged kidneys. These proteins participated in 5-hydroxytryptamine degradation, angiogenesis, blood coagulation, pyruvate metabolism, TCA cycle, and vitamin D metabolism (Supplementary Table 3).

Redox proteomics shows high protein glutathionylation in aged mouse kidneys

It has been amply documented that oxidative stress increases in aged population [30, 31]. Our proteomics and western blotting data showed that catalase expression was downregulated in aged kidneys, indicating increased ROS levels (Figure 3A). Furthermore, we found that glutaredoxin-1 (Glrx1) expression was downregulated, using both proteomics and western blotting analyses. Glrx1 is an oxidation repair enzyme that regulates protein mixed disulfides through deglutathionylating an S-glutathionylated substrate [32]. The results proposed that protein glutathionylation was increased in aged mouse kidneys. To confirm it, we performed an analysis of glutathionylated proteins. The experimental procedure to enrich glutathionylated peptides was based on selective deglutathionylation by Glrx1. Briefly, proteins extracted from kidney tissues were treated with iodoacetamide to modify the free sulfhydryl groups, followed by deglutathionylation by Glrx1. The newly generated free sulfhydryl groups were then alkylated with biotin-IAA, enriched by streptavidin affinity column, and analyzed by LC-MS/MS to determine potential glutathionylation sites (Figure 3B). Using this approach, 342 unique glutathionylated peptides from 281 proteins were identified. Interestingly, almost all identified proteins exhibited increased glutathionylation (Figure 3C).

Figure 3. Profiling protein glutathionylation in young and aged mouse kidneys. (A) Western blot analysis of Glrx1 and catalase in kidney tissues. (B) Workflow of profiling cysteine glutathionylation. (C) Volcano plot of glutathionylated proteins in young and aged kidneys; glutathionylated peroxisomal proteins are red-coded. (D) KEGG enrichment analysis of glutathionylated proteins. (E) Histogram of the increased glutathionylation of peroxisomal proteins. (F) Significant enrichment of small amino acids, lysine and arginine against aspartic acid and glutamic acid at the proximal position (±five positions) to glutathionylated cysteine residues. (G) Consensus motif for glutathionylated cysteine residues shows significant enrichment of lysine and arginine across a range of proximal positions. Error bars represent ± SEM.

KEGG analysis revealed that proteins with increased glutathionylation were associated with carbon metabolism, amino acid metabolism, and peroxisome pathways in aged mice (Figure 3D). Quantitative proteomics showed that peroxisomal proteins were downregulated, and glutathionylome analysis revealed that peroxisomal proteins were highly glutathionylated (Figure 3E). Further analysis revealed that lysine and arginine residues were highly enriched against aspartic acid and glutamic acid at the proximal positions (±5) to glutathionylated cysteine residues. Consensus sequence for glutathionylation motif showed enrichment of arginine against aspartic acid and glutamic acid across a range of proximal positions (Figure 3F and 3G). We also found that glycine (G), alanine (A), valine (V), and leucine (I), which are small and nonpolar amino acids, frequently appeared in the proximity of glutathionylated residues.

NMN rescues proteostasis in aged mouse kidneys

Recent studies have shown that nicotinamide mononucleotide (NMN) supplementation may ameliorate age-associated organ dysfunction [33, 34]. To examine whether NMN treatment could rescue aged-impaired proteostasis, we treated aged mice (96 weeks) with NMN (500 mg/kg body weight, every two days for 4 weeks), and analyzed the kidney tissues of untreated and NMN-treated mice by quantitative proteomics. NMN treatment upregulated 190 proteins, while it downregulated 134 proteins in kidneys (Figure 4A). In addition, our results showed that NMN reversed proteostasis compared to untreated aged mice. IPA analysis showed that proteins associated with oxidative phosphorylation, sirtuin signaling, fatty acid β-oxidation, and bile acid biosynthesis pathway were upregulated in NMN-treated mice (Figure 4B). Moreover, we found that peroxisomal proteins, NAD(H)-related proteins, and proteins associated with fatty acid β-oxidation pathway were increased after NMN treatment (Figures 4C–4E).

Figure 4. NMN restores proteostasis in aged kidneys. (A) Volcano plot of DEPs in untreated and NMN-treated kidneys. (B) IPA analysis of DEPs in untreated and NMN-treated kidneys (p < 0.05). (C) Expression of NAD-dependent oxidoreductase. (D–E) Expression of proteins associated with peroxisomes (D) and fatty acid oxidation pathways (E). (F) Electron micrograph showing part of the mouse kidneys stained for catalase by DAB. The electron-dense reaction product of DAB is observed in the matrix of all peroxisomes, while mitochondria and other components of the cytoplasm are not stained. (Mag. ×1,050). (G) The number of peroxisomes in young, aged and NMN-treated kidneys based on electron micrographs. Data were analyzed using Student’s t-test. *p < 0.05, **p < 0.01, and ***p < 0.001. p < 0.05 was considered statistically significant. Error bars represent ± SEM.

To determine whether NMN increases peroxisome biogenesis, we analyzed mouse kidneys by electron microscopy, using specific peroxisome labeling by diaminobenzidine (DAB) (Figure 4F). Our results showed that the number of peroxisomes significantly decreased in aged mouse kidneys, but NMN efficiently increased the number of peroxisomes (Figure 4G). NMN supplementation also decreased IgGs and complement protein levels, suggesting that NMN might reduce kidney injuries (data not shown).

Animals and animal care

C57BL/6J mice were housed at the Laboratory Animal Research Center, Tsinghua University. Mice were intraperitoneally injected with nicotinamide mononucleotide (NMN) at 500 mg/kg body weight every two days for 4 weeks. Untreated and NMN-treated mice, 8- and 96-weeks-old, were sacrificed for further analyses. Mouse renal tissues were collected and washed in cold PBS, and immediately snapped frozen in liquid nitrogen for the following study including the biochemical, proteomics and metabolomics analyses. The remaining tissues were fixed in 4% paraformaldehyde for further histopathological examination.

Quantitative proteomics analysis

Proteins were extracted from kidney tissues by 8 M urea containing 1% protease inhibitor mix. Protein (200 μg) disulfides were reduced by 5 mM Tris (2-carboxyethyl) phosphine (TCEP) for 10 min at room temperature. Then, free cysteine residues were alkylated by 10 mM iodoacetamide for 30 min in the dark. After dilution with PBS, proteins were digested with sequencing-grade trypsin for 16 h at 37 °C at a substrate/enzyme ratio of 50:1 (w:w). Peptides were desalted using Sep-Pak C18 Cartridge (Waters, 186004619), vacuum dried, and labeled with Tandem Mass Tag (TMT, Thermo, Waltham, MA). Mixed peptides were fractionated using the high-pH reversed phase chromatography followed by nano LC-MS/MS analysis.

The MS/MS spectra were searched against the UniProt mus musculus database (June 24, 2017; 59,594 sequences) processed by the Sequest HT search engine using Proteome Discoverer (PD) 2.1 software. The following parameters were used for searching: fixed modifications of carbamidomethylation on cysteine and TMT 6-plex on lysine and peptide N-terminus; variable modification of oxidation on methionine; two trypsin missed cleavages were permitted; the tolerances of precursor and fragment mass were 10 ppm and 0.02 Da; at least two unique peptides for identification of proteins; and 1% FDR (false discovery rate) tolerance at peptide-spectrum match (PSM) level.

Glutathionylation enrichment assay

Tissue samples were homogenized and lysed by 8 M urea containing 1% protease inhibitor mix; 10 mg protein samples were used for enrichment of glutathionylated peptides. Briefly, the lysates were alkylated by iodoacetamide without reduction, and digested with sequencing-grade trypsin for 16 h at 37 °C. After digestion, samples were incubated with Glrx1 in the presence of GSH, NADPH, and GR (glutaredoxin reductase, Sigma). Biotin-IAA (Thermo Fisher) was used to alkylate the newly generated sulfhydryl groups. The products were incubated with streptavidin beads (Thermo Fisher), eluted by 1% trifluoroacetic acid solution, and labeled by Tandem Mass Tag (TMT).

Western blotting analysis

Tissue samples were lysed in RIPA buffer (Solarbio, Beijing, China) containing 1% Protease Inhibitor Cocktail (Selleck, Shanghai, China). The lysates were centrifuged at 12,000 rpm for 20 min; the protein concentration was determined by the BCA protein assay (Solarbio, Beijing, China). Equal amounts of protein were separated by SDS-PAGE, and transferred to PVDF membranes. After 1 h blocking in TBST containing 5% skim milk, primary and secondary antibodies were used to detect the target proteins. Immunoblots were processed through the enhanced chemiluminescence (ECL); the images were quantified using ImageLab software. Actin was used as an internal loading control. Primary and secondary antibodies were the following: anti-Glrx1 antibody was purchased from Abcam (1:1000 diluted, ab187507); anti-actin antibody was from Cell Signaling Technology (1:50000 diluted, 4970); anti-rabbit-IgG antibody was purchased from Cell Signaling Technology (1:2000 diluted, 7074S).

Cytochemical detection of peroxisomes by transmission electron microscopy

Mice were sacrificed and one of the kidneys was removed and fixed by glutaraldehyde-formaldehyde overnight at 4 °C. After fixation, the tissue was rinsed in PBS and sectioned into 200 μm slices by the Oxford Vibratome (Foster City, CA, USA). The 200 μm sections were pre-incubated (1 h, room temperature) in DAB medium (Proteintech, Chicago, US) without hydrogen peroxide, followed by incubation (1 h) in a complete medium containing 0.15% hydrogen peroxide. After incubation, the sections were rinsed with PBS, and re-fixed in glutaraldehyde-formaldehyde fixative. After dehydration in chilled graded ethanol solutions and embedding in Epon by propylene oxide for 12-16 h, ultrathin sections were cut by an ultra-microtome, and examined by electron microscope.

Statistics

GraphPad Prism (version 6.0) and R (version 3.5.2) programs were used for statistical analyses. All experiments were repeated three times and statistical differences were assessed by two-tailed unpaired Student’s t test and considered significant at p ≤ 0.05 (*, p < 0.05; **, p < 0.01; ***, p < 0.001). Data were expressed as the mean ± SEM (standard error of mean). For further evaluation, differentially expressed proteins in proteomics study were analyzed by Ingenuity Pathway Analysis (IPA, QIAGEN) to reveal their potential relationships and intracellular pathways. Gene Ontology (GO) and Kyoto Encyclopedia of Genes and Genomes (KEGG) analyses for the identified proteins were performed to construct a network with related proteins.