Effects of semaglutide on gut microbiota, cognitive function and inflammation in obese mice

Animals and experimental design

A total of 24 male specific-pathogen-free grade C57BL/6J mice (6-week-old) were obtained from Hebei In vivo Biotechnology Co., Ltd (Hebei, China) and acclimated for 1 week. Animal care was carried out according to established guidelines. The mice were housed in ventilated cages (485 ×200 ×200 mm; 4 mice per cage; temperature: 20−24 °C; humidity: 55% ±10%). Thereafter, they were randomly divided into a normal-chow diet group (NCD, n = 8) or high-fat diet group (HFD, n = 16). The HFD group was administered a HFD (20% carbohydrate, 20% protein, and 60% fat) for 12 weeks to induce obesity in mice. The HFD group was further assigned into two groups according to a randomized block design: the HFD group (n = 8) and the Sema group (HFD + semaglutide, n = 8). The Sema group received a daily subcutaneous injection of semaglutide (Bagsværd, Novo Nordisk, Denmark, 30 nmol/kg/d), while the NCD and HFD groups were injected with equal volumes of saline for a further 12 weeks. The weight of all mice was measured after the intervention. The Morris water maze test assessed cognitive performance in each group (n = 8). Fecal and serum samples were collected for 16S rRNA sequencing (n = 6) and cytokines measurement (n = 6). Feces were collected in sterile cryopreservation tubes and then labeled and stored in a −80 °C refrigerator. At the end of the study, all mice were anesthetized with 1% sodium pentobarbital solution (50 mg/kg, intraperitoneal injection). This study was approved by the Animal Ethics Committee of Hebei General Hospital (202332), and every effort was made to minimize animal suffering.

Measurement of plasma cytokines

Serum samples were collected from the retro-orbital sinus after the mice received an intraperitoneal injection of pentobarbital sodium. Blood was centrifuged at 3000 × g for 10 min at 4 °C to obtain plasma for measurement of cytokines such as tumor necrosis factor (TNF) α, IL-6, and IL-1β. Cytokine concentrations are expressed in pg/mL.

Morris water maze test

The Morris water maze (MWM) test assessed spatial memory in all mice. The MWM test was conducted in a circular pool (diameter 120 cm, height 45 cm) filled with water at 24 to 26 °C. A visible escape platform (diameter 11 cm) was submerged approximately 1.5 cm under the water surface. In each acquisition trial, the mice started from one of the four quadrants by facing the wall of the tank. Each mouse was trained to reach the platform within 60 s for five consecutive days. If a mouse did not find the platform, it was gently guided to the platform and allowed to stay for 10 s. On the sixth day, the platform was removed and all mice were subjected to probe trials. Mice were allowed to swim in the pool for 60 s. Data were collected using a computerized animal tracking system (Shanghai Jiliang Software Science & Technology Co., Ltd, Shanghai, China), which recorded the path length, swimming speed, the time and latency to reach the platform and the number of platform crossings.

Gut microbiota analysis

Fecal DNA was extracted using the OMEGA Soil DNA Kit (D5625-01) (Omega Bio-Tek, Norcross, GA, USA) and subjected to PCR amplification using specific primers targeting the V3 and V4 hypervariable regions of the 16S rRNA gene. The specific forward primer was 341F 5′-CCTAYGGGRBGCASCAG-3′, and the reverse primer was 806R 5′-GGACTACNNGGGTATCTAAT-3′. Thermal cycling consisted of initial denaturation at 98 °C for 1 min, 30 cycles of denaturation at 98 °C for 10 s, annealing at 50 °C for 30 s, and elongation at 72 °C for 30 s. Finally, 72 °C for 5 min.

The amplified PCR products were extracted by 2% agarose gel electrophoresis and purified by Quant-iT PicoGreen dsDNA Assay Kit. The library was constructed using Illumina’s TruSeq Nano DNA LT Library Prep Kit. Agilent Bioanalyzer 2100 and Promega QuantiFluor were utilized to assess library quality. Raw sequencing data were in FASTQ format. 250 bp paired-end reads were generated and preprocessed by cutadapt software. Clean sequence reads were imported into QIIME2 (Bolyen et al., 2019), and variant calling was carried out using DADA2 (Callahan et al., 2016). The amplicon sequence variant (ASV) was clustered based on 100% similarity. Species annotation was performed using a pre-trained Naive Bayes classifier aligned with the SILVA 138 reference database.

To quantify the changes in gut microbiota after the intervention, we compared the ASV profiles of the three groups. The QIIME2 software was used to analyze alpha and beta diversity (https://docs.qiime2.org). Alpha-diversity indexes (ACE, Chao1, Shannon, and Simpson) were estimated. Principal coordinates analysis (PCoA) based on unweighted UniFrac distances was then used to assess beta diversity (Lozupone et al., 2007), showing the similarity of the microbial community. Differential abundance analysis of gut microbiota was performed through the Kruskal-Wallis test. To identify the representative taxa of each group, linear discriminant analysis effect size (LEfSe) was used to detect different features among the three groups (Segata et al., 2011). LDA > 4 and p < 0.05 were set as cutoff values to define significantly different genera.

Statistical analysis

Baseline differences were evaluated using one-way ANOVA. Tukey post hoc test was used for comparisons between more than two groups. Data that did not satisfy the normal distribution were compared using non-parametric tests, and differences between groups were compared using Kruskal-Wallis. Repeated-measures analysis of variance (ANOVA) was used to analyze spatial MWM data. Spearman’s correlation analysis was performed to explore the relationship between inflammatory markers or cognitive function and gut microbiota composition. The SPSS software package 26.0 (IBM Corporation, Armonk, NY, USA) and R, version 4.1.0 (R Core Team, 2021) were used for the above analyses. LEfSe analysis used the non-parametric factorial Kruskal-Wallis rank-sum test (Kruskal & Wallis, 1952) and the (unpaired) Wilcoxon rank-sum test (Wilcoxon, 1945; Mann & Whitney, 1947) in combination with liner discriminant analysis (LDA) (Fisher, 1936) effect sizes to find robust differential species between groups. Figures were drawn using Graph Pad Prism 8.0 software. A value of p < 0.05 was regarded as a significant difference.

Semaglutide decreased body weight gain and ameliorated inflammatory markers

Inflammation is a prevalent process in obesity. At the end of the experiment, HFD induced an increase in body weight (Fig. 1A). Levels of inflammatory markers levels, including TNF α, IL-6, and IL-1β were elevated in the HFD group compared to the NCD group (Figs. 1B–1D). However, semaglutide treatment significantly reduced body weight and levels of inflammatory markers (all p < 0.05).

Semaglutide improved cognitive function

To explore the effects of semaglutide treatment on cognitive function, the MWM test was used to examine learning and memory function. During the 5-day learning phase, escape latencies were significantly shorter in the Sema group than in the HFD group (Fig. 2A). On the sixth day of the probe test period, the HFD group showed decreased time spent in the target quadrant (TSTQ) and the number of times crossing the platform area (NTCPA) compared with the NCD group (all p < 0.05). In contrast, TSTQ and NTCPA were greater in the Sema group than in the HFD group (all p < 0.05). There was no statistical difference between the three groups in total swimming distance (TSD) and average swimming speed (ASP) within 60 s (Figs. 2B–2E).

Alterations of gut microbiota diversity associated with semaglutide treatment

The microbiota was analyzed by 16S rRNA gene sequencing. The amplicon sequence variant (ASV) rank abundance curve was based on ASV serial number and relative abundance as axes to draw a hierarchical clustering curve. The curve was wide and smooth in the horizontal direction, indicating that the evenness and richness of the three groups were good (Fig. 3A). As shown in the Venn diagram, the total number of ASV was 2632, and the number of shared ASV was 211. In addition, 922, 544, and 654 ASV were unique to the NCD, HFD and Sema groups, respectively (Fig. 3B). For the diversity index analysis, abundance-based coverage estimators (ACE) was significantly higher in the Sema group than in the HFD group (p < 0.05). Compared with the NCD group, the Chao1, Shannon, and Simpson indices were also reduced in the HFD group (p < 0.05), while there was no significant difference in the values between the Sema group and HFD groups (Figs. 4A–4D). The Principal Coordinate Analysis (PCoA) based on unweighted UniFrac distances (Fig. 4E) showed that the bacterial composition was significantly different among the three groups (ANOSIM, unweighted UniFrac R = 0.816, P = 0.001). The PCoA plots also showed that the HFD and Sema groups were distinctly separated, indicating that the semaglutide treatment significantly affects intestinal flora diversity in obese mice.

Alterations in gut microbiota composition of mice treated with Semaglutide

To determine the effects of semaglutide on the microbial community, we compared the microbial taxa at the phylum, order, family, and genus level among three groups (Figs. 5A–5D).

At the phylum level, compared with the NCD group, the relative abundance of Firmicutes and Desulfobacterota was higher, while the relative abundance of Verrucomicrobiota and Proteobacteria was lower in the HFD group. Compared with the HFD group, the relative abundance of Firmicutes and Desulfobacterota was decreased, while the relative abundance of Verrucomicrobiota and Proteobacteria was increased in the Sema group (Fig. 5A).

At the class level, the relative abundance of Bacilli, Clostridia, and Desulfovibrionia was higher in the HFD group compared to the NCD group, while the relative abundance of Verrucomicrobiae and Gammaproteobacteria was lower in the HFD group. Compared with the HFD group, the relative abundance of Bacilli, Clostridia, and Desulfovibrionia was decreased while the relative abundance of Verrucomicrobiae and Gammaproteobacteria was increased in the Sema group (Fig. 5B).

At the family level, compared with the NCD group, the relative abundance of Erysipelotrichaceae, Lachnospiraceae, Desulfovibrionaceae and Peptostreptococcaceae was higher, while the relative abundance of Lactobacillaceae, Muribaculaceae and Akkermansiaceae was lower in the HFD group. Compared with the HFD group, the relative abundance of Erysipelotrichaceae, Lachnospiraceae, Desulfovibrionaceae and Peptostreptococcaceae were decreased while the relative abundance of Lactobacillaceae, Muribaculaceae and Akkermansiaceae was increased in the Sema group (Fig. 5C).

At the genus level, compared with the NCD group, the relative abundance of Dubosiella, Romboutsia and Odoribacter was higher, while the relative abundance of Muribaculaceae, Lactobacillus, Akkermansia was lower in the HFD group. Compared with the HFD group, the relative abundance of Dubosiella, Romboutsia, and Odoribacter was decreased, while the relative abundance of Muribaculaceae, Lactobacillus, and Akkermansia was increased in the Sema group (Fig. 5D).

Additionally, the linear discriminant analysis effect size (LEfSe) analysis (α = 0.05, linear discriminant analysis (LDA) score >4.0) was applied to display the species composition and differences of the samples visually (Fig. 5E). The HFD group displayed a significant increase in the abundance of Firmicutes at the phylum level, Bacilli at the class level, Erysipelotrichales and Peptostreptococcales_Tissierellales at the order level, Erysipelotrichaceae, Peptostreptococcaceae and Eggerthellaceae at the family level, Romboutsia, Dubosiella and Enterorhabdus at the genus level. The Sema group was characterized by Actinobacteriota and Verrucomicrobiota phylum. Additionally, the Sema group was enriched in Coriobacteriia and Verrucomicrobiae at the class level, Coriobacteriales and Verrucomicrobiales at the order level, Atopobiaceae and Akkermansiaceae at the family level, and Akkermansia and Coriobacteriaceae _UCG_002, Clostridia _UCG_014 at the genus level.

Relationship between gut microbiota composition and inflammatory markers

As shown in Fig. 6, Enterorhabdus (r =0.549), Romboutsia (r =0.482) and Dubosiella (r =0.501) were positively correlated with TNF α, while Clostridia _UCG_014 (r =  − 0.552) and Muribaculaceae (r =  − 0.495) were negatively correlated with TNF α (all p < 0.05). Romboutsia (r =0.568) was positively correlated with IL-6, whereas Clostridia _UCG_014 (r =  − 0.575) and Muribaculaceae (r =  − 0.485) were negatively correlated with IL-6 (all p < 0.05). Enterorhabdus (r =0.656, p < 0.01) and Romboutsia (r =0.550, p < 0.05) were positively correlated with IL-1β, whereas Clostridia _UCG_014 (r =  − 0.735, p < 0.01) was negatively correlated with IL-1β. Interestingly, these negatively related genera were deficient in the intestine of mice in the HFD group but enriched in the Sema group.

Relationship between gut microbiota composition and cognitive function

As shown in Fig. 7, TSTQ and NTCPA were positively correlated with Muribaculaceae (r = 0.537, 0.497, respectively; all p < 0.05) and Clostridia _UCG_014 (r = 0.655, 0.595, respectively; all p < 0.01). TSTQ and NTCPA were negatively correlated with Romboutsia (r =  − 0.566, −0.560, respectively; all p < 0.05) and Dubosiella (r =  − 0.548,−0.671, respectively; all p < 0.05). TSD was negatively correlated with Enterorhabdus (r =  − 0.492; p < 0.05). Interestingly, these positively related strains were enriched in the intestine of mice in the Sema group, but deficient in the HFD group.