Exploring the Effect and Mechanism of Liraglutide in Treating Depression Based on Network Pharmacology and Experimental Analysis

Abstract

Depression is a disorder caused by various reasons, with low mood as the main symptom, and it has a serious impact on mental health. Liraglutide (Lir) has been confirmed to alleviate neuroinflammation and depression-like behaviours induced by chronic stress, but its underlying mechanisms remain unclear. This study investigated the regulation of Lir for microglia-associated inflammation in depression through network pharmacology. In vivo experiments demonstrate that Lir reduces depressive-like behaviours by activating Nrf2 and subsequently downregulating HMGB1 expression, while also reducing the generation of pro-inflammatory mediators and oxidative stress damage. In vitro studies confirmed that the downregulation of HMGB1 depends on Nrf2 activation, and Lir activates Nrf2 via the PI3K/AKT pathway. Additionally, indirect co-culture of BV2 and HT22 cells demonstrated Lir's neuroprotective effects against neuronal apoptosis, consistent with findings from in vivo experiments. The study results first demonstrate that Lir exerts antidepressant effects through the PI3K/Nrf2/HMGB1 pathway, which reveals a novel mechanism of action for the antidepressant effects of Lir.

Keywords: CUMS; depression; liraglutide; microglia; network pharmacology; neuroinflammation.

FIGURE 1

The network pharmacology analysis of Lir in the treatment of depression. (A) Venn diagram showing the overlapping targets between Lir‐related targets and depression‐related targets. (B) Drug–target–disease interaction network constructed using Cytoscape, illustrating the potential targets through which Lir may exert antidepressant effects. (C) Number of targets. (D) The PPI network of genes for Lir treatment of depression by the STRING database. (E) The core targets in the PPI network obtained using the CytoHubba plugin. (F) The top 20 biological processes (BPs). (G) The top 20 cellular components (CCs) (H) The top 20 molecular functions (MFs). (I) The top 20 KEGG signalling pathways.

FIGURE 2

Effect of Lir on depression‐like behaviours and serum CORT levels in mice. (A) Representative OFT movement trajectory maps showing the exploratory behaviour and locomotor activity of mice (n = 7). (B) Heat map of movements in the OFT (n = 7). (C–E) The total distance travelled, movement speed and time spent in the centre area within 5 min for each group of mice (n = 7). (F) Immobility time in the TST, indicative of behavioural despair (n = 10). (G) Immobility time in the FST, another classical index of depression‐like behaviour (n = 10). (H) Serum CORT levels as a physiological marker of stress response (n = 3). Normally distributed data were analysed by one‐way ANOVA with Bonferroni or Tamhane's T2 post hoc tests, whereas non‐normally distributed data were assessed using the Kruskal‐Wallis test. Data are presented as mean ± SEM. *p < 0.05, **p < 0.01, ***p < 0.001.

FIGURE 3

The neuroprotective effects of Lir on brain neurons in mice. Representative histopathological images of the HIP (A) and PFC (B) were obtained using H&E staining. Neuronal damage was assessed in specific hippocampal regions: CA1 (C), CA3 (D) and DG (E), as well as in the PFC (F), across different experimental groups. Nissl staining of HIP (G) and PFC (H) tissues further corroborated these findings. Quantitative analysis revealed neuronal integrity in HIP CA1 (I), CA3 (J), DG (K) regions and in PFC (L) across experimental groups. Nissl body counts in HIP CA1 (M), CA3 (N), DG (O) and PFC (P) regions were also quantified (n = 3). Scale bar = 20 μm. The arrows indicate damaged neurons, and the zoomed views highlight the cells containing Nissl bodies. Normally distributed data were analysed by one‐way ANOVA with Bonferroni or Tamhane's T2 post hoc tests, whereas non‐normally distributed data were assessed using the Kruskal‐Wallis test. Data are presented as mean ± SEM. *p < 0.05, **p < 0.01, ***p < 0.001.

FIGURE 4

Lir reduced microglial activation in the HIP and PFC regions of CUMS‐induced depressed mice. Representative immunofluorescence images showing GLP‐1R (red) and microglial (IBa‐1, green) co‐staining in the HIP (A, D, G) and PFC (J) regions. Quantitative analysis of GLP‐1R expression levels in the HIP (B, E, H) and PFC (K) was performed for each experimental group. The number and total area of microglia were assessed in the HIP (C1‐C2, F1‐F2, I1‐I2) and PFC (L1‐L2) regions. Microglial morphological complexity was evaluated by analysing the branches in the HIP (C3‐C4, F3‐F4, I3‐I4) and PFC (L3‐L4), and the end‐point voxels was quantified in the HIP (C5, F5, I5) and PFC (L5). (n = 3). Scale bar = 20 μm. Normally distributed data were analysed by one‐way ANOVA with Bonferroni or Tamhane's T2 post hoc tests, whereas non‐normally distributed data were assessed using the Kruskal‐Wallis test. Data are presented as mean ± SEM. *p < 0.05, **p < 0.01, ***p < 0.001.

FIGURE 5

Effects of Lir on Nrf2 and HMGB1 protein expression levels, oxidative stress and inflammatory factors in the HIP and PFC tissues. (A) Representative WB images showing the expression levels of Nrf2 and HMGB1 proteins. Quantitative analysis of Nrf2 (B) and HMGB1 (C) protein levels in HIP tissues across all experimental groups. Quantitative analysis of Nrf2 (D) and HMGB1 (E) protein levels in PFC tissues. Oxidative stress factors in HIP tissues, including MDA (F), GSH (G) and SOD (H) levels. Oxidative stress factors in PFC tissues, including MDA (I), GSH (J) and SOD (K) levels. Levels of inflammatory factors including IL‐1β (L), IL‐6 (M) and TNF‐α (N) in HIP tissues, and IL‐1β (O), IL‐6 (P) and TNF‐α (Q) in PFC tissues. (n = 3–6). Normally distributed data were analysed by one‐way ANOVA with Bonferroni or Tamhane's T2 post hoc tests, whereas non‐normally distributed data were assessed using the Kruskal–Wallis test. Data are presented as mean ± SEM. *p < 0.05, **p < 0.01, ***p < 0.001.

FIGURE 6

Effects of Lir on Nrf2 expression and oxidative stress following CUMS exposure. (A) Representative movement trajectory and activity heat maps of mice in the OFT. (B) Distance travelled (n = 7). (C) Movement speed (n = 7). (D) Time spent in the central area (n = 7). (E) Immobility time in TST (n = 13). (F) Immobility time in FST (n = 13). (G) Representative WB images of total protein expression in HIP and PFC tissues, showing (H) Nrf2, (I) Keap1, (J) HO‐1 and (K) NQO1 expression in HIP and (L) Nrf2, (M) Keap1, (N) HO‐1 and (O) NQO1 expression in PFC. (P) Representative WB images for nuclear and cytoplasmic fractions of Nrf2 in HIP and PFC tissues. Quantification of Nrf2 levels in the nuclear (Q, S) and cytoplasmic (R, T) compartments of HIP and PFC tissues. Levels of oxidative stress factors including MDA (U), GSH (V), SOD (W) in HIP tissue and MDA (X), GSH (Y) and SOD (Z) in PFC tissue (n = 3–5). Normally distributed data were analysed by one‐way ANOVA with Bonferroni or Tamhane's T2 post hoc tests, whereas non‐normally distributed data were assessed using the Kruskal–Wallis test. Data are presented as mean ± SEM. *p < 0.05, **p < 0.01, ***p < 0.001.

FIGURE 7

The inhibitory effect of Lir on HMGB1 expression and related behavioural improvements in CUMS‐exposed mice. (A) Representative movement trajectory and activity heat maps of mice in the OFT. (B) Movement distance (n = 7). (C) Movement speed (n = 7). (D) Time spent in the central area (n = 7). (E) Immobility time in the TST (n = 10). (F) Immobility time in the FST (n = 10). (G) Representative WB images showing protein expression. Quantification of total HMGB1 (H) and TLR4 (I) protein levels in HIP tissue and total HMGB1 (J) and TLR4 (K) levels in PFC tissue. (L) Representative WB images. Quantification of HMGB1 protein levels in the cytoplasmic and nuclear fractions of the HIP (M, N) and PFC (O, P) tissue (n = 3–5). Levels of inflammatory cytokines IL‐1β (Q), IL‐6(R), TNF‐α (S) in HIP tissue and IL‐1β (T), IL‐6(U), TNF‐α (V) in PFC tissue (n = 3). Normally distributed data were analysed by one‐way ANOVA with Bonferroni or Tamhane's T2 post hoc tests, while non‐normally distributed data were assessed using the Kruskal–Wallis test. Data are presented as mean ± SEM. *p < 0.05, **p < 0.01, ***p < 0.001.

FIGURE 8

Statistical analysis of BV2 cell viability assessed by CCK‐8 assay. (A) Effect of different concentrations of LPS on the viability of BV2 microglial cells. (B) Effect of varying concentrations of Lir on LPS‐stimulated BV2 cells, showing the potential protective effect of Lir against LPS‐induced reduction in cell viability (n = 4). Normally distributed data were analysed by one‐way ANOVA with Bonferroni or Tamhane's T2 post hoc tests, whereas non‐normally distributed data were assessed using the Kruskal–Wallis test. Data are presented as mean ± SEM. *p < 0.05, **p < 0.01, ***p < 0.001.

FIGURE 9

Representative WB images showing the expression of proteins involved in the Nrf2 and HMGB1 signalling pathways. (A) Representative WB images showing the expression of related proteins of Nrf2 and HMGB1 signalling pathways. Quantitative analysis of protein expression levels for Nrf2 (B), keap1 (D), HO‐1 (E), NQO1 (F), HMGB1 (G) and TLR4 (C). (H) Representative WB images. Expression of Nuclear Nrf2 (I), Nuclear HMGB1 (J), Cytoplasmic Nrf2 (K) and Cytoplasmic HMGB1 (L). (M) Detection of ROS by immunofluorescence. (N) Statistical analysis of ROS fluorescence intensity, scale bar = 100 μm. (O) Representative immunofluorescence images showing colocalisation of Nrf2 and HMGB1 in BV2 cells, scale bar = 20 μm. Quantification of fluorescence intensity for Nrf2 (P) and HMGB1 (Q). Levels of oxidative stress markers MDA (R), GSH (S) and SOD (T), as well as inflammatory cytokines IL‐1β (U), IL‐6 (V) and TNF‐α (W) (n = 3–5). Normally distributed data were analysed by one‐way ANOVA with Bonferroni or Tamhane's T2 post hoc tests, while non‐normally distributed data were assessed using the Kruskal‐Wallis test. Data are presented as mean ± SEM. *p < 0.05, **p < 0.01, ***p < 0.001.

FIGURE 10

Investigation of the mechanistic pathway of Lir in the treatment of depression. (A) Representative WB images showing the expression of pathway‐related proteins in the in vitro experiment. Quantitative analysis of protein expression levels, including p‐PI3K/PI3K (B), p‐AKT/AKT (C), Nrf2 (D) and HMGB1 (E). (F, K) Representative WB images of pathway proteins in the HIP and PFC tissues of each group in the in vivo experiment. Quantitative analysis of protein expression levels in the HIP, including p‐PI3K/PI3K (G), p‐AKT/AKT (H), Nrf2 (I) and HMGB1 (J), and in the PFC, including p‐PI3K/PI3K (L), p‐AKT/AKT (M), Nrf2 (N) and HMGB1 (O) (n = 3–5). Normally distributed data were analysed by one‐way ANOVA with Bonferroni or Tamhane's T2 post hoc tests, whereas non‐normally distributed data were assessed using the Kruskal–Wallis test. Data are presented as mean ± SEM. *p < 0.05, **p < 0.01, ***p < 0.001.

FIGURE 11

Neuroprotective effects of Liraglutide in the treatment of depression. (A) Representative WB images from in vitro experiments. Quantitative analysis of Bcl‐2 (B), Bax (C) and Cleaved Caspase‐3 (D) protein levels in BV2 cells. (E, F) Representative WB images from in vivo experiments. Quantitative analysis of Bcl‐2 (G, J), Bax (H, K) and Cleaved Caspase‐3 (I, L) protein levels in the HIP and PFC tissues of each group of mice (n = 3–5). Normally distributed data were analysed by one‐way ANOVA with Bonferroni or Tamhane's T2 post hoc tests, whereas non‐normally distributed data were assessed using the Kruskal–Wallis test. Data are presented as mean ± SEM. *p < 0.05, **p < 0.01, ***p < 0.001.