This study was conducted in accordance with the NIH guidelines for the care of laboratory animals and the related guidelines, and it received approval by the Animal Ethics Committee of the Hebei General Hospital (No. 202173).

Male C57BL/6JC mice (age: 6 weeks) were purchased from Hebei In vivo Biotechnology Co., Ltd [License No: SYXK (Jun) 2015-0004] and housed in the Experimental Animal Centre of the Hebei General Hospital, with four mice in each cage, incubated at 22°C ± 2°C, 55% ± 10% humidity, and a light/dark cycle of 12:12 h. Food and water were made freely available throughout the experimental period.

After 1 week of acclimatization, the animals were categorized into a blank control feeding group (group C, n = 8) and a high-fat dietary feeding group (group H, n = 24) by using the random number table method. The control group was fed a diet consisting of 20% protein, 70% carbohydrate, and 10% fat, while the high-fat diet group was fed a diet consisting of 20% protein, 20% carbohydrate, and 60% fat. The mice in each group received equal numbers of calories daily and drank water ad libitum. After 12 weeks of high-fat feeding, the mice in the a high-fat dietary feeding group were screened for high-fat diet-induced obesity considering that their bodyweight was ≥20% of the mean body weight of mice in the control group. Group H was further categorized into group H (n = 8), group H + semaglutide (group S, n = 8), and group H + empagliflozin (group E, n = 8). Then, 30 nmol/kg/day bodyweight of semaglutide was administered intraperitoneally to group S and 10 mg/kg/day bodyweight of empagliflozin was administered via gavage to group E. Then, equal amounts of saline were administered intraperitoneally to the mice in groups C and H for 12 weeks in total. After completion of the drug intervention, the mice fasted for 12 h, and blood was withdrawn from the tail vein to determine the fasting glucose level.

After completing the intervention, all mice were tested by MWM (Shanghai Jiliang Software Science & Technology Co., Ltd, Shanghai, China) to assess their cognitive function status, such as learning and memory. The device consisted of a white circular pool of diameter 120 cm and a depth of 45 cm. A pool filled with water at a temperature of 23°C ± 1°C, and a platform installed 1–2 cm under the water surface. The test consisted of two phases: Four consecutive days of positioning navigation training and a fifth day of space exploration training. In the first phase, the mice were randomly placed into the water from one of the four quadrants each day to search for an underwater escape platform, and the time taken by the mice to swim to and find the platform served as the escape latency period. Then, the platform was removed for spatial exploration training. The mice were placed diagonally from the original platform into the water and they searched the platform for 60 s. Their swimming trajectory, the number of times they passed through the original platform, the percentage of time they lingered the target quadrant, the total distance and the average swimming speed were recorded and counted within 60 s.

Blood was collected from the eyes of the mice after overnight fasting and then centrifuged at 3,000 rpm for 10 min at 4°C so as to collect the serum and store it at −80°C. After blood collection, the mice were decapitated and executed, the bilateral hippocampal regions were quickly separated on an ice tray, and the surface water was blotted out and frozen at −80°C.

The levels of interleukin-6 (IL-6) and IL-1β (IL-1β) in mice were measured by antibody sandwich ELISA (Multi Sciences Biotech Co., Ltd., Hangzhou, China); the concentration of total cholesterol (TC), triglyceride (TG), fasting plasma high-density lipoprotein (HDL-C), and fasting plasma lipoprotein (HDL-C) in the mice were measured by enzymatic method using commercially available kits (Nanjing Jian Cheng Bioengineering Institute, Nanjing, China).

The hippocampal tissues from the mice were well ground with liquid nitrogen, removed into a 1.5-mL centrifuge tube, made to a final concentration of 1 mM with a protease inhibitor (Biotech, A610425-0005) and phosphatase inhibitor (Roche, 4906837001), followed by lysing by sonication (SCIENTZ-II D). The solution was then centrifuged at ×12,000 g for 10 min at room temperature to remove the cell debris, and the supernatant was removed and then centrifuged again to remove the supernatant. Next, the supernatant was transferred into a new centrifuge tube and the BCA kit was applied for protein concentration determination. Then, 10 μg of the protein was removed from each sample and segregated by 12% SDS-PAGE. The segregated gels were stained with Komas Brilliant Blue using the eStain LG Protein Stainer (Nanjing Kingsray Biotechnology Co., Ltd., Nanjing, China), and the stained gels were imaged using a fully automated Digital Gel Image Analysis System (Shanghai Tennant Technology Co., Ltd., Shanghai, China).

Each sample was diluted to the same concentration and volume with a lysis solution. Dithiothreitol (DTT) was added to the abovementioned protein solution to a final concentration of 5 mM and then reduced at 56°C for 30 min. Next, a corresponding volume of iodoacetamide was added to a final concentration of 10 mM and left for 15 min at room temperature away from light. To this solution, 6 times the volume of acetone was added to precipitate the protein and then left overnight at −20°C. The precipitate was collected via centrifugation at ×8,000 g for 10 min at room temperature, and acetone was evaporated for 2–3 min, followed by re-solubilization of the precipitate with ammonium bicarbonate (50 mM), the addition of trypsin in a 50:1 mass ratio (protein: trypsin), enzymatic digestion overnight at 37°C, and collection by vacuum centrifugation, to be followed by desalting and lyophilization.

The column was placed in a 2-mL microcentrifuge tube and centrifuged at ×1,000 g for 30 s to remove the storage buffer. Then, dissolve the peptide dry powder in 200 µL of binding/wash buffer and centrifuged at the same rate, and repeated once again. For phosphorylated peptide enrichment, 200 µL of the suspended peptide sample was first added to an equilibrated centrifuge column and the resin was mixed with the sample by tapping the bottom stopper and incubated for 30 min. The column was then placed in a microcentrifuge tube and centrifuged at ×1,000 g for 30 s before discarding the flow-through solution. The column was then washed with 200 µL of the binding/wash buffer, centrifuged at ×1,000 g for 30 s (repeated twice), and the flow-through solution was finally discarded. Then, 200 µL of the LC-MS-grade water was added to wash the column and centrifuged at ×1,000 g for 30 s. Finally, the column was placed in a new microcentrifuge tube, and 100 µL of the elution buffer was added to the column, centrifuged at ×1,000 g for 30 s (repeated once again). The eluate was then immediately dried in a high-speed vacuum concentrator to remove the elution buffer.

The peptides were dissolved in buffer A (0.1% formic acid [FA]), separated using an EASY-nLC 1200 UPLC system (ThermoFisher), and then sampled onto a 25 cm C18 (RP-C18, ionopticks) analytical column at a flow rate of 300 nL/min for gradient elution. B buffer (80% ACN plus 0.1% FA) was set as follows: 3%–27% for 0–66 min; 27%–46% for 66–73 min; 46%–100% for 73–84 min; and 100% for 84–90 min. The peptides were separated by the UHPLC system and injected into an NSI ion source for ionization and then analyzed by the timsTOF Pro mass Spectrometer with three biological replicates.

Raw MS data were disposed employing the MaxQuant version (v1.5.2.8) with FDR <0.01 at the protein, peptide, and modification levels. Enzyme specificity was set to trypsin and searches included cysteine carbamidomethylation as a fixed modification, protein N-acetylation, oxidation of methionine, and/or phosphorylation of Ser, Thr, Tyr residues as variable modifications. A maximum of two missed cleavages were allowed during protease digestion, and the peptide was required to be fully trypsinized.

For the identified phosphorylation site corresponding proteins, each annotation information was extracted based on the Uniprot, KEGG, GO, KOG/COG, and other databases to mine the protein functions. GO/KEGG enrichment analysis of the proteins corresponding to the differentially phosphorylated sites was performed to characterise their function. p < 0.05 was considered to indicate statistical significance.

The STRING (v.10.5, https://string-db.org/) database is a database of predicted functional correlations among the proteins. A protein–protein interaction (PPI) network of differentially phosphorylated proteins was constructed by setting the criterion “confidence level to ≥0.7".

Figures were expressed by using the indicator of mean ± standard deviation, which also analyzed using the SPSS.25 software (IBM Corp.). Assessment of escape latency in MWM by repeated-measures analysis of variance (ANOVA). Between-group comparisons of MWM spatial probe tests, bodyweight, and serological indicators were conducted by one-way ANOVA and Fisher’s least significant difference test. Data that did not satisfy the normal distribution were compared using non-parametric tests, expressed as median and interquartile spacing [M (P25%, P75%)], and differences between groups were compared using Bonferroni. p < 0.05 was considered to indicate statistical significance.

At first, there was no significant difference in bodyweight between the four groups of mice. After 12 weeks of high-fat feeding, the bodyweight of mice in group H increased significantly compared to group C, increasing more than 20%, and there was a significant difference between the two groups (27.93 ± 2.16 vs. 40.96 ± 2.49, p < 0.001), as shown in Figure 1A, indicating that the mouse obesity model had been successfully established. The bodyweight of mice decreased significantly with the combination of semaglutide and empagliflozin, while the bodyweight of group S decreased more significantly (43.31 ± 5.03 vs. 30.06 ± 3.59, p < 0.001; 43.31 ± 5.03 vs. 30.68 ± 0.98, p < 0.001). When compared to group C, fasting blood glucose, TC, TG, LDL-C, and HDL-C levels were all increased in group H [3.32 (2.98, 3.72) vs. 7.77 (6.51, 9.07), p < 0.001; 4.36 ± 0.59 vs. 7.86 ± 1.65, p < 0.001; 0.64 ± 0.10 vs. 0.92 ± 0.15, p < 0.001; 2.55 ± 0.78 vs. 6.44 ± 1.58, p < 0.001; 2.71 ± 0.93 vs. 6.30 ± 0.90, p < 0.001]. However, after intervention with semaglutide and empagliflozin, TG, LDL-C, and HDL-C were significantly lower in groups S compared to group H (0.92 ± 0.15 vs. 0.74 ± 0.09, p = 0.009; 6.44 ± 1.58 vs. 4.99 ± 0.786, p = 0.026; 6.30 ± 0.90 vs. 4.90 ± 0.83, p = 0.017), and fasting blood glucose and TC decreased compared to group H, but there was no statistical difference [7.86 ± 1.65 vs. 6.74 ± 1.52, p = 0.089; 7.77 (6.51, 9.07) vs. 5.89 (5.7, 6.24), p = 0.106]; LDL-C and HDL-C were significantly lower in group E compared to group H (6.44 ± 1.58 vs. 3.53 ± 1.53, p < 0.001; 6.30 ± 0.90 vs. 4.25 ± 1.59, p = 0.001), and fasting glucose, TC and TG decreased compared to group H, but there was no statistical difference [7.86 ± 1.65 vs. 6.59 ± 1.08, p = 0.056; 7.77 (6.51, 9.07) vs. 6.26 (4.55, 6.90), p = 0.445; 0.92 ± 0.15 vs. 0.86 ± 0.18, p = 0.366].

We examined the changes in serum levels of IL-1 and IL-6. As shown in Figure 2, serum levels of IL-1 and IL-6 were significantly higher in group H compared to group C (3.55 ± 0.79 vs. 7.51 ± 1.51, p < 0.001; 15.80 ± 2.38 vs. 29.27 ± 9.47, p = 0.002); however, intervention with semaglutide and empagliflozin significantly reduced serum IL-1 and IL-6 levels in groups S and E, there was a statistically significant difference (7.51 ± 1.51 vs. 5.22 ± 1.27, p < 0.05; 29.27 ± 9.47 vs. 20.51 ± 1.47, p < 0.05; 7.51 ± 1.51 vs. 5.12 ± 1.86, p < 0.05; 29.27 ± 9.47 vs. 20.93 ± 6.03, p < 0.05).

As shown in Figure 3, the escape latency of mice in group H was longer than that of mice in group C on days 2–4 of the training phase (day 2: p < 0.05; day 3: p < 0.01; day 4: p < 0.01), indicating that high-fat diet-induced obesity leads to reduced spatial learning ability in mice. In contrast, the latency of mice in groups S and E was shorter than that of mice in group H (S vs. H: day 3: p < 0.05, day 4: p < 0.01; E vs. H: day 3: p < 0.05, day 4: p < 0.01), suggesting that semaglutide or empagliflozin therapy alleviated learning deficits. On the fifth day of the spatial probe test, HFD mice showed poor mobility, leading to a significant decrease in the total swimming distance and average swimming speed during the test period compared to group C mice (all p < 0.05), and a significant decrease in the time spent in the target quadrant and the number of platform crossings in group H (all p < 0.01). In contrast, groups S and E stayed in the target quadrant longer and had a higher frequency of platform crossing (all p < 0.05), but there was no statistical difference in total swimming distance and average swimming speed compared to group H mice (all p > 0.05). This is further supported by the swimming track (Figure 3F). In conclusion, both semaglutide and empagliflozin partially restored the deficit in spatial learning and memory.

We applied the phosphorylation-based 4D-LFQ technique to identify and quantify phosphopeptides and phosphoproteins, identifying a total of 21,239 phosphorylation sites on 20,493 peptides and 4,290 phosphorylated proteins (Figure 4A). Of the 20,493 detected phosphopeptides, 17,003 were uniquely phosphorylated, 2,824 were doubly phosphorylated, 571 were triple phosphorylated, and 95 had more than triple phosphorylated peptides (Figure 4B). In contrast, 17,577 of the 21,239 detected phosphorylation sites were located on serine residues, 3,424 on threonine residues, and 238 on tyrosine residues, accounting for 82.76, 16.12%, and 1.12%, respectively (Figure 4C).

Furthermore, sites with expression values ≥ 50% in any group of samples were retained based on the table of qualitative phosphorylation sites obtained from the screening. Loci with missing values ≤ 50% were filled with the mean values of the same group of samples, screened using Localization prob≥0.75 and Delta score≥8, and a total of 11,681 phosphorylation loci with high confidence were found using Median Normalization and log2 log transformation.

Based on confidence loci, two criteria were selected to compute the difference between groups. The Fold change (FC) method was used to assess the fold change in expression levels of a locus across groups; the p-value derived by the t-test demonstrated the significance of the difference between groups. Figure 5A demonstrates a comparison of the expression of phosphorylated proteins in group S/H, group E/H, group H/C, group E/C and group S/E. 844 differentially significant phosphorylated loci were identified in the H/C group. There were 1,084 differentially significant phosphorylation sites identified in the S/H group, with 686 being upregulated and 398 being downregulated. In the E/H group, 1,552 differentially significant phosphorylation sites were identified, with 955 being upregulated and 597 being downregulated. We created volcano diagrams to depict the differentially expressed proteins more graphically, as shown in Figures 5B–D.

To gain more insight into the biological significance of differentially phosphorylated proteins, GO and KEGG enrichment analyses were performed, and the GO and KEGG enrichment results were clustered. As shown in Figure 6, the cellular components of differentially phosphorylated proteins in the H/C, S/H, and E/H groups were essentially the same in the CC analysis, with neuronal cell bodies being the most involved, followed by postsynaptic density, glutamatergic synapses, synaptic cell components, axons, and cytoskeleton. The three groups of differentially phosphorylated proteins in MF analysis mostly had molecular functions such as protein kinase binding, actin-binding, calmodulin-binding, GTPase activator activity, and SH3 structural domain binding. Further analysis showed that the three differentially phosphorylated proteins were primarily involved in neural projection development, synaptic plasticity control, axonogenesis, and microtubule cytoskeleton organization.

The differentially expressed phosphorylated proteins were further localized and classified subcellularly, and the results revealed that the differentially phosphorylated proteins in the three comparison groups of H/C, S/H, and E/H were mainly distributed in subcellular structures such as cytoskeletal, mitochondrial, cytoplasmic, and peroxisomal (Figure 7).

KEGG is the most widely used public database for studying protein metabolic pathways in cells. Using the KEGG database, the proteins that were differentially phosphorylated in the H/C group were found to be involved in dopaminergic synapses, axon guidance, myocardial contraction, central carbon metabolism in cancer, neurotrophic factor signaling pathways, and pathways of neurodegeneration—Metabolic pathways in multiple diseases (Figure 8A). The S/H group, on the other hand, showed differential phosphorylated proteins mainly focused on dopaminergic synapses, MAPK signaling pathway, cholinergic synapses, oxytocin signaling pathway, regulation of the actin cytoskeleton, and axon guidance (Figure 8B). In contrast, the differentially phosphorylated proteins in the E/H group were mainly focused on dopaminergic synapses, neurodegenerative pathways-multiple diseases, regulation of the actin cytoskeleton, axon guidance, myocardial contraction, ErbB signaling pathway, and insulin secretory metabolic pathway (Figure 8C). Furthermore, we discovered that the differentially phosphorylated proteins in the three comparison groups act together in dopaminergic synapse and axon guidance signaling pathway processes by combining the aforementioned findings. This is consistent with previous GO analyses, which found that proteins were concentrated in neuronal projection development, regulation of synaptic plasticity, and axonogenesis findings. Figure 9 depicts the dopaminergic synapse pathway.

When the results of differentially phosphorylated proteins from groups H and C, S and H, and E and H were combined, 44 proteins were found to be upregulated (or downregulated) at phosphorylation sites in group H and downregulated (or upregulated) at phosphorylation sites in group S and involved in the first 20 KEGG pathways (Table 1), and 60 proteins were found to be upregulated (or downregulated) at phosphorylation sites in group H and downregulated (or upregulated) at phosphorylation sites in group E and involved in the first 20 KEGG pathways (Table 2). The interactions of the proteins corresponding to the differential sites were obtained using the STRING database, it was discovered that voltage-dependent L-type calcium channel subunit alpha-1D (CACNA1D), voltage-dependent P/Q-type calcium channel subunit alpha-1A (CACNA1A), and voltage-dependent N-type calcium channel subunit alpha-1B (CACNA1B) in the different comparison groups were highly functionally related and operated together in the dopaminergic synapse pathway (Figures 9, 10).