Mouse podocyte clone 5 (MPC5) used in the cell experiment was purchased from Shanghai Kwaisai Biotechnology Co., Ltd. Cells were cultured in T25 cell culture flasks with medium consisting of 1% penicillin, 10% fetal bovine serum, and 90% low-glycemic DMEM (Shanghai Xiaopeng Biological Technology Co., Ltd). All cells were cultured at 37°C, 5% CO₂.

Glycogen synthase kinase 3β knockdown (CRISPR Cas9 technology) with flag tag and lentiviral vector with flag were purchased from Shanghai Jikai Gene Chemical Technology Co., Ltd. In addition, all viral vectors are resistant to puromycin. The multiplicity of infection (MOI) and the optimal infection conditions of the lentivirus for MPC5 infection were determined by experiments, and finally, an MOI of 15 was selected on the basis of the company’s instructions. First, we spread two six-well plates with 100,000 cells per well. Next, we added knockout GSK-3β lentivirus and control lentivirus in line with the viral vector solution and the MOI value obtained in the preliminary experiment. In addition, we carried out preliminary experiments to determine the best working concentration according to the instructions of puromycin (Dalian Meilun Biotechnology Co., Ltd). Twenty-one concentration gradients were set from 0 to 10 μg/ml, and after 48 h of observation, it was confirmed that 99% of cells without puromycin resistance could be killed at a concentration of 2 μg/ml. After the virus was transfected into the cells for 48 h, diluted puromycin was added to the six-well plate at a concentration of 2 μg/ml. Finally, we removed the culture solution with puromycin and added a culture medium to expand the cells after 48 h of puromycin treatment.

Total protein was extracted from transfected cells using radioimmunoprecipitation assay (RIPA) lysis buffer (Beijing Soleibao Technology Co., Ltd). Protein concentration was quantified using a BCA kit (Shanghai Biyuntian Biotechnology Co., Ltd). All of the cell lysates were denatured by boiling in loading buffer. Ten percent separating gel and 6% concentrated gel were configured according to the instructions of the gel making kit (Shanghai Biyuntian Biotechnology Co., Ltd). Then, we added 20 to 30 μg of protein per well to the channel. The concentrated gel was electrophoresed at 80 V, and the separation gel was electrophoresed at 120 V constant pressure electrophoresis. Then, proteins were separated by sodium dodecyl sulfate-polyacrylamide gel electrophoresis (SDS-PAGE) and transferred to a polyvinylidene fluoride (PVDF) membrane at a voltage of 250 mA for 90 min. Blocking was completed using 5% skim milk/TBST. The film was washed with TBST three times for 5 min, and the primary antibody was added at a ratio of 1:1,000 (the antibody diluent was purchased from Shanghai Biyuntian Biotechnology Co., Ltd) and incubated overnight at 4°C. The next day, after washing the membrane three times with TBST, the membrane was incubated with the secondary antibody at room temperature for 2 h. Finally, protein bands were visualized with electrochemiluminescence (ECL) (Dalian Meilun Biotechnology Co., Ltd) on the AI600 machine (Thermo Fisher Scientific).

The validated successfully transfected cells were expanded and cultured at 37°C and 5% CO₂. We washed the cells with precooled PBS solution several times before preparing the samples, collected the cells with a cell scraper, and placed them in an EP tube. Samples were taken at -80°C. Then, four times the volume of lysis buffer (8 M urea, 1% protease inhibitor, 1% phosphatase inhibitor) was added and analyzed by ultrasonic testing. We centrifuged the sample at 12,000 × g for 10 min at 4°C to remove cell debris, transferred the supernatant to a new centrifuge tube, and determined the protein concentration using the BCA kit.

For digestion, the protein solution was reduced with 5 mM dithiothreitol for 30 min at 56°C and alkylated with 11 mM iodoacetamide for 15 min at room temperature in darkness. The protein samples were then diluted by adding 100 mM TEAB to a level of urea below 2 M. Finally, trypsin was added at a trypsin–protein mass ratio of 1:50 for the first overnight digestion and a trypsin–protein mass ratio of 1:100 for a second 4-h digestion.

The peptides digested by trypsin were desalted with Strata X C18 (Phenomenex) and then freeze-dried in vacuo. The peptide was dissolved with 0.5 M TEAB and labeled according to the tandem mass tag (TMT) kit operating instructions. Then, one unit of TMT/iTRAQ reagent was thawed and reconstituted in acetonitrile. The peptide mixtures were then incubated for 2 h at room temperature and pooled, desalted, and dried by vacuum centrifugation.

The peptides were fractionated by high pH reversed-phase high-performance liquid chromatography (HPLC) using a Thermo Betasil C18 column (5 μm particles, 10 mm ID, 250 mm length). In light of the following operation, the peptide segment has a step gradient of 8 to 32% acetonitrile, pH 9.0, 60 min time separation, and 60 components. Then, the peptides were combined into six fractions and dried by vacuum centrifugation.

The peptides were dissolved in mobile phase A of liquid chromatography containing 0.1% formic acid and 2% acetonitrile and then separated using the EASY-nLC 1000 ultrahigh-performance liquid system. Liquid phase B was composed of 0.1% formic acid and 90% acetonitrile. The liquid gradient settings were as follows: 0–20 min, 8∼22% B; 20–33 min, 22∼35% B; 33–37 min, 35∼80% B; and 37–40 min, 80% B. The flow rate was maintained at 550.00 nl/min. The peptides were separated by an ultrahigh-performance liquid system and then injected into the NSI ion source for ionization and then analyzed by QE plus mass spectrometry. The ion source voltage was set to 2.2 kV, and the peptide precursor ions and their secondary fragments were detected and analyzed by high-resolution Orbitrap. The scanning range of the primary mass spectrum was set to 400–1,500 m/z, and the scanning resolution was set to 70,000.00; the scanning range of the secondary mass spectrum was set to a fixed starting point of 100 m/z, and the secondary scanning resolution was set to 17,500.00. The data acquisition mode used the data-dependent scanning (DDA) program; that is, the first 20.00 peptide precursor ions with the highest signal intensity were selected after the first scan to enter the HCD collision cell, and 30% of the fragmentation energy was used for fragmentation. Next, to improve the effective utilization of mass spectrometry, the automatic gain control (AGC) was set to 5E4, the signal threshold was set to 6.3E4 ions/s, the maximum injection time was set to 50 ms, and the dynamic rejection time of the tandem mass spectrometry scan was set to 30 s to avoid precursor ion repeated scans.

The MS/MS data obtained were processed through the MaxQuant search engine (v.1.5.2.8). The search parameters were set as follows. (1) The database was Mus_musculus_10090 (17032 sequences), an anti-database was added to calculate the false positive rate (FDR) caused by random matching, and a common contamination library was added to the database to eliminate the contaminating protein from the impact identification results. (2) The restriction enzyme digestion method was set to trypsin/P. (3) The number of missed cleavage sites was set to 2. (4) The minimum peptide length of seven amino acid residues was applied. (5) The maximum modification number of peptides was set to 5. (6) The mass error tolerance was set to 20.0 and 5 ppm. (7) The mass error tolerance of the second fragment ion was 20.0 ppm. (8) Cysteine alkylation carbamidomethyl (C) was set as a fixed modification, and variable modifications were set as [“acetyl (protein N-term),” “oxidation (M),” “deamidation (NQ)”]. (9) The quantitative method was set to TMT-6plex, and the FDR for protein identification and PSM identification was tuned to 1%. The data presented in the study are deposited in the ProteomeXchange Consortium via the PRIDE partner repository, accession number (PXD024543).

We analyzed the differentially expressed proteins from the aspects of Gene Ontology (GO) enrichment, the KEGG enrichment, and domain enrichment. The first was the GO annotation of the protein, which is divided into three categories: biological process, cell composition, and molecular function. Gene Ontology annotations on proteomics were downloaded from the UniProt-GOA database¹.

The Kyoto Encyclopedia of Genes and Genomes database was used for pathway enrichment analysis. In the KEGG enrichment analysis, we first used the KEGG online service tool KAAS to annotate the submitted protein, and then, we used the KEGG mapper to match the annotated protein into the corresponding pathway in the database.

In addition, the InterPro (providing resources for functional analysis of protein sequence family classification, prediction of structural domains and special sites) database was used to analyze the enrichment of functional domains of differentially expressed proteins. The above analysis used Fischer’s precise double-ended test method to test differentially expressed proteins in the background of the identified proteins. If the p-value of the domain unit enrichment test was less than 0.05, the value was considered significant.

Afterward, the differential proteins obtained by phosphorylated proteomics were used to perform motif prediction analysis. This analysis is based on MoMo software, and the motif-x algorithm was used to analyze the motif characteristics of the modified sites. Among them, all the identified phosphorylation modification sites were composed of six amino acids upstream and downstream as the analysis object; the analysis and comparison background were the peptide sequences composed of six amino acids upstream and downstream of all potential phosphorylation modification sites in the species. When the number of peptides in the form of a characteristic sequence is greater than 20 and the statistical test p-value is less than 0.000001, the characteristic sequence form is considered to be a motif of modified peptides.

Next, we utilized the STRING database² to visualize the visual network between different proteins and then used the MCODE package³ with default parameters in Cytoscape to identify highly interconnected clusters in a network.

Furthermore, a group-based prediction system (GPS5.0)⁴ was adopted to predict the phosphokinase upstream of the phosphorylation site to predict the kinase family that may regulate the specific phosphorylation site. Then, the corresponding kinase protein in the kinase family was obtained by comparison with the kinase sequence in integrated annotations for eukaryotic protein kinases, protein phosphatases, and phosphoprotein-binding domains, ver2.0 (iEKPD2.0), and finally, the protein–protein interaction information (PPI) was used to filter out potential false positive interactions. In addition, we used the iKAP (Mischnik et al., 2016) algorithm to predict kinase activity based on the expression of phosphorylation sites.

The schematic flow is shown in Figure 1A. We applied TMT techniques for labeling, high pH reverse HPLC for fractionation, and TiO₂ chromatography for phosphopeptide enrichment and utilized liquid chromatography tandem mass spectrometry analysis (Figure 1A). To ensure that the mass spectrometry data is credible, the protein expression levels of GSK-3β, p-GSK-3β, and GSK-3α in GSk-3β knockdown cells than in control cells were determined by Western blot (Supplementary Figure 1). To verify the repeatability of the sample, we performed principal component analysis (PCA) (Supplementary Figure 2).

We recognized that the ratio between the two groups was greater than 1.3 times that of the differentially expressed proteins, so 149 proteins were identified as differentially expressed proteins. A total of 34 proteins with higher levels and 115 proteins with lower levels were found in the lentivirus-mediated GSK-3β knockdown group than in the control group (Figure 1B). A total of 581 phosphosites with higher phosphorylation levels and 288 phosphosites with lower phosphorylation levels were identified in the lentivirus-mediated GSK-3β knockdown group than in the control group (Figure 1D). The analysis of QDEPs and the differentially expressed proteins of the phosphorylated proteome (PDEPs) is helpful to identify proteins that are more closely related to GSK-3β.

In this study, we found that the three proteins (Supplementary Figure 3) Trhde (GSK3B_KD/control ratio = 2.454), Spp1 (GSK3B_KD/control ratio = 1.311), and Mical3 (GSK3B_KD/control ratio = 1.314) had reduced phosphorylation levels but increased total levels. In addition, quantitative proteome and phosphoproteome of differentially expressed protein subcellular location maps (Figures 1C,E) showed that the protein is mostly distributed in the nucleus and cytoplasm. Differential protein localization shows that GSK-3β also plays a role in cell structures such as the extracellular space, plasma membrane, and mitochondria.

A total of 5,661 quantitative proteins were identified in the proteome analysis (Supplementary Figure 4A). We defined proteins that were significantly different (Student’s t-test, p < 0.05) and used the criterion of 1.3-fold or greater change as the criteria to screen candidate proteins and identified 34 proteins with higher levels and 115 proteins with lower levels in the lentivirus-mediated GSK-3β knockdown group than in the control group (Figure 2A). Heatmaps were applied to indicate the expression levels of the differentially expressed proteins screened by the volcano map in three replicate samples of the experimental group and the control group (Figure 2B).

The GO enrichment of differentially expressed proteins obtained is shown in Figure 2C. GO enrichment is mainly divided into three parts: Biological process, cell composition, and molecular function. From the perspective of biological process, differentially expressed proteins are mainly enriched in negative regulation of cell growth, negative regulation of cell migration, axon extension involved in axon guidance, negative regulation of cell motility, ISG15–protein conjugation, etc. From the perspective of cell composition, differentially expressed proteins were mainly enriched in intrinsic components of the plasma membrane, intrinsic components of the membrane, integral components of the plasma membrane, integral components of the membrane, and myofilaments. In the molecular function classification, cell adhesion molecule binding, toxic substance binding, receptor binding, integrin binding, oxidoreductase activity, and oxidizing metal ions were significantly enriched.

We performed domain enrichment and KEGG enrichment on the differentially expressed proteins to obtain a deeper understanding of the role and function of GSK-3β in podocytes. For domain enrichment of differentially expressed proteins (Figure 3A), we identified eight domains: core histone H2A/H2B/H3/H4, metallothionein, integrin beta cytoplasmic domain, integrin beta tail domain, integrin, beta chain, complement Clr-like EGF-like, GRAM domain, and calponin homology (CH) domain. Next, we performed KEGG enrichment for differentially expressed proteins and identified four pathways (Figure 3B) including axon guidance, focal adhesion, mineral absorption, and regulation of actin cytoskeleton. In addition, QDEPs were analyzed for GO, KEGG, and domain enrichment based on the difference expression multiplied QDEPs [Q1 (<0.667), Q2 (0.667–0.769), Q3 (1.3–1.5), Q4 (>1.5)] (Supplementary Figure 5).

Finally, we used STRING and Cytoscape to visualize the interaction between differentially expressed proteins (Figure 3C) and used the MCODE plug-in in Cytoscape software to further analyze the PPI network to obtain four more critical and core protein groups that were more closely connected in the network (Figure 3D). MCODE 1 (MCODE score = 5) consisted 5 nodes and 10 edges, MCODE 2 (MCODE score = 5) consisted 5 nodes and 10 edges, MCODE 3 (MCODE score = 4.5) comprised 5 nodes and 9 edges, and MCODE 4 (MCODE score = 3) comprised 3 nodes and 3 edges. We chose the top 20 hub proteins in the PPI network based on betweenness (Supplementary Table 1), which usually plays a paramount role in network stability due to its high degree of connection/interaction.

A total of 2,094 quantitative proteins were identified in the experimental group and the control group (Supplementary Figure 4B). Similarly, we identified 581 proteins with higher phosphorylation levels and 288 proteins with lower phosphorylation levels in the lentivirus-mediated GSK-3β knockdown group than in the control group (Figure 4A). Besides, a total of 1,175 phosphosites were identified, of which phosphosites on the serine, threonine, and tyrosine accounted for 90.89% (1,068/1,175), 8.85% (104/1,175), and 0.26% (3/1,175), and we distributed phosphoproteins based on the number of phosphorylation sites per protein (Supplementary Figure 6). The phosphorylation site of Dpysl2/CRMP2 that become less phosphorylated was the most obvious. Research has shown that it plays a role in repairing the integrity of microtubules (Xu et al., 2015). The R package “pheatmap” was used to draw a heatmap (Figure 4B), which shows the expression levels of the differentially expressed proteins at the phosphorylation site screened by the volcano map. Afterward, we performed GO enrichment for phosphorylated differentially expressed proteins (Figure 4C). From the dimension of cell composition, PDEPs are mainly enriched in actin cytoskeleton, cell cortex region, anchoring junction, basal cortex, adherens junction, lamellipodium, cell cortex, and cortical cytoskeleton. Molecular functions were enriched in actin binding, structural molecule activity, microtubule plus-end binding, protein C-terminus binding, mRNA 5′-UTR binding, actin filament binding, structural constituent of cytoskeleton, and ankyrin binding. In the biological process classification, mRNA processing, RNA splicing, nucleocytoplasmic transport, regulation of mRNA metabolic process, nuclear export, regulation of mRNA processing, protein export from nucleus, RNA export from nucleus, and regulation of RNA splicing were significantly enriched.

Domain enrichment analysis (Figure 5A) and KEGG enrichment analysis were performed to obtain a more comprehensive understanding of GSK-3β in podocytes (Figure 5B) on phosphorylated differentially expressed proteins. Seven domains were enriched in the domain enrichment analysis, namely, SPOC domain, bromodomain, P21-Rho-binding domain, immunoglobulin I-set domain, variant SH3 domain, gelsolin repeat, and kinase-associated domain 1. Subsequently, 10 pathways were enriched, namely, RNA transport, spliceosome, adherens junction, tight junction, ribosome biogenesis in eukaryotes, thyroid hormone signaling pathway, proteoglycans in cancer, focal adhesion, mRNA surveillance pathway, and Hippo signaling pathway. Furthermore, PDEPs were analyzed for GO, KEGG, and domain enrichment based on the difference expression multiplied PDEPs [Q1 (<0.667), Q2 (0.667–0.769), Q3 (1.3–1.5), Q4 (>1.5)] (Supplementary Figure 7).

Afterward, we used STRING and Cytoscape to visualize the interaction between PDEPs (Figure 5C) and used the MCODE plug-in in Cytoscape software to further analyze the PPI network and obtained 22 key and core proteins that were tighter in the network connection group (Figure 5D). We displayed the first four most compact MOCDEs (Figure 5D): MCODE 1 (MCODE score = 28.714) consisted of 29 nodes and 402 edges, MCODE 2 (MCODE score = 19.714) consisted of 50 nodes and 483 edges, MCODE 3 (MCODE score = 11.053) was composed of 20 nodes and 105 edges, and MCODE 4 (MCODE score = 10.727) was composed of 12 nodes and 59 edges. In addition, we chose the top 20 hub proteins in the PPI network based on the betweenness of the phosphorylated proteome (Supplementary Table 2), which usually plays a paramount role in network stability due to its high degree of connection/interaction.

Protein phosphorylation modification is regulated by protein kinases, and different protein kinases prefer specific substrates with conserved motifs. We applied a large number of phosphorylation sites identified in this study for bioinformatics analysis, and we used the motif-x program to perform sequence analysis of the six amino acids upstream and six amino acids downstream of the identified phosphate sites of serine and threonine residues.

Moreover, we performed motif enrichment of phosphoserine (Figure 6A) and phosphothreonine (Figure 6B) in the form of a heatmap. The numbers on the abscissa represent the upstream and downstream phosphorylation modification sites. We identified 89 conserved motifs based on 1,068 phosphoserine (pS) phosphate sites and 15 conserved motifs and phosphoserine based on 104 phosphothreonine (pT) phosphate sites (Supplementary Table 1). We chose the top four hub motifs based on motif score, respectively (Figures 6C,D). The amino acids aspartate (D), glutamate (E), glycine (G), lysine (K), proline (P), and arginine (R) had a tendency to be present in the proximity of serine phosphosites. The amino acids glutamic acid (E), lysine (K), proline (P), arginine (R), and serine (S) had a tendency to be present in the proximity of threonine phosphosites.

Apparently, the associated upstream kinase is unknown for the majority of the GSK-3β-regulated phosphosites. We therefore predicted the upstream kinase for each of these sites using the group-based prediction system (GPS) and iEKPD2.0. In accordance with the p-value of the identified protein ratio being under 0.05, 156,515 regulatory relationships between 299 protein kinases and 3,460 significantly changed phosphorylation modification sites on 1,574 proteins were identified (Supplementary Table 2). According to the prediction results, 1,299 regulatory sites of GSK-3β were obtained (Supplementary Table 3). In line with the changes in the expression level of the regulatory sites, the sites with a change of more than 1.8 times are displayed as shown in Figure 7A. The iKAP algorithm was used to obtain nine kinases with high levels (Figure 7B) and 18 kinases with low levels (Figure 7C). The activities of NIMA-related kinase 6 (NEK6) and casein kinase 2 alpha 2 (CSNK2A2) are apparently increased. The activity of SRSF Protein Kinase 1 (SRPK1) and MAPK Interacting Serine/Threonine Kinase 1 (MKNK1) are obviously decreased. Next, Cytoscape was used to demonstrate the activity of kinase regulatory networks (Figure 7D). Protein phosphorylation sites with a GSK3B-KD/control ratio of 1.8 times or more are displayed in the network.