Zhinao Capsule improves learning and memory in Alzheimer's mice by reducing brain inflammation through the gut-brain connection

Zhinao Capsule (ZNJN, 0.4 g per capsule) was provided by the First Affiliated Hospital of Anhui University of Traditional Chinese Medicine (Batch No.: 20221129). The formulation contains the following primary herbal ingredients: C. pilosula, A. membranaceus, P. sibiricum, C. deserticola, C. longa, A. tatarinowii, L. chuanxiong, and Lumbricus nativus. ZNJN is an in-house preparation that has been widely applied in clinical practice for over 20 years. Previous studies by our research team have identified its key bioactive constituents as β-sitosterol, quercetin, and baicalein (Ma et al., 2022). Donepezil hydrochloride tablets (5 mg/tablet; Batch No.: 0000018324) were manufactured by Zhejiang Huahai Pharmaceutical Co., Ltd. and used as a positive control. Male APP/PS1 double-transgenic mice (AD model) and age-matched wild-type C57BL/6J mice (6 months old) were purchased from Hangzhou Ziyuan Experimental Animal Technology Co., Ltd. (License No.: 20210516Abbb05000259). All animals were maintained under specific pathogen-free (SPF) conditions with ad libitum access to food and water throughout the experimental period. All experimental procedures were conducted in accordance with the institutional guidelines for animal care and approved by the Animal Ethics Committee of Anhui Agricultural University (Approval No.: AHAUB2023024).

After a 7-day acclimatization period, forty APP/PS1 transgenic mice were randomly assigned into five groups (n = 8 per group) including high-dose ZNJN (ZNJN-H), medium-dose ZNJN (ZNJN-M), low-dose ZNJN (ZNJN-L), positive control, and model group. In addition, eight age-matched C57BL/6J mice were assigned to the normal control (Ctrl) group. Mice in both the control and model groups received physiological saline at a dose of 10 mL/kg/day via oral gavage. The ZNJN-L, ZNJN-M, and ZNJN-H groups were administered ZNJN at doses of 0.234, 0.468, and 0.936 g/kg/day, respectively. Mice in the positive control group received donepezil hydrochloride at a dose of 0.65 mg/kg/day. All treatments were administered once daily by oral gavage for a duration of 28 consecutive days.

Following the final behavioral assessments, mice were fasted for 12 h. Anesthesia was induced by intraperitoneal injection of 2% sodium pentobarbital at a dose of 30 mg/kg. Mice were executed by CO₂ inhalation anesthesia in batches, and serum, tissue and fecal samples were collected in time for subsequent testing.

Serum levels of pro-inflammatory cytokines, including TNF-α, IL-1β, IL-6, and LPS, were measured using commercially available enzyme-linked immunosorbent assay (ELISA) kits (Nanjing Jiancheng Bioengineering Institute), following the manufacturers’ protocols.

For histopathological evaluation, hematoxylin, and eosin (H&E) staining was performed on the hippocampal CA1 region and colon tissues to observe neuronal morphology and intestinal structure, respectively. The experiment with three mice per group (n = 3) was the histological assessment. Immunohistochemistry (IHC) was used to detect the expression of β-amyloid peptide (Aβ_1–42) in the brain tissues of each group, as histological observations. In addition, immunofluorescence staining was conducted to assess the expression levels of GFAP and Iba-1, which are markers of astrocyte and microglial activation, respectively.

Fresh fecal samples were collected and stored at –80 °C for microbiota analysis. Genomic DNA was extracted using a commercial DNA extraction kit. The bacterial 16S rRNA gene V3–V4 hypervariable regions were amplified by PCR and sequenced using high-throughput sequencing technology. Bioinformatics analysis was performed to evaluate the gut microbial community structure. This included operational taxonomic unit (OTU) clustering, alpha and beta diversity analysis, community composition analysis (including Venn diagrams and taxonomic classification at the genus level), and intergroup difference analysis. Linear discriminant analysis effect size (LEfSe) was used to identify taxa with significant differential abundance between groups. All sequencing and data analysis procedures were carried out by Genesky Biotechnologies Inc. (Shanghai, China).

All data are presented as mean ± standard error of the mean (SEM). Statistical analyses were performed using GraphPad Prism version 10. One-way analysis of variance (ANOVA) followed by Tukey’s honestly significant difference (HSD) post-hoc test was used to assess differences among groups. For microbial community analysis, LEfSe was employed to identify differentially abundant OTUs between groups. A p-value of less than 0.05 was considered statistically significant.

During the experimental period, all groups of mice underwent Barnes Maze training from days 1 to 4. Short-term and long-term memory were assessed on days 5 and 12, respectively. As shown in Figure 1A, representative movement trajectories demonstrated that Ctrl mice exhibited directed and efficient search paths toward the target escape hole (TEH), indicating intact spatial memory. In contrast, APP/PS1 model mice displayed disorganized and erratic movement patterns, including frequent edge-circling, random hole exploration, and repeated returns to the maze center, reflecting impaired spatial memory and a reliance on non-goal-directed search strategies. Following ZNJN administration, mice exhibited more purposeful and directed trajectories, indicating a clear improvement in cognitive function compared to the untreated model group. In the short-term memory probe trial conducted on day 5 (Figure 1B), Ctrl mice located the TEH in an average of 6.48 ± 1.51 s, whereas the model group required significantly more time, averaging 43.23 ± 2.94 s (p < 0.01). ZNJN treatment resulted in a dose-dependent reduction in escape latency. Mice in the low-dose (ZNJN-L), medium-dose (ZNJN-M), and high-dose (ZNJN-H) groups required 33.98 ± 3.33, 32.10 ± 4.15, and 26.73 ± 3.45 s, respectively, to locate the TEH. Notably, the performance of the ZNJN-H group approached that of the positive control group treated with donepezil (DA), which exhibited an escape latency of 25.12 ± 3.68 s.

Similarly, in the long-term memory test conducted on day 12 (Figures 1C,D), Ctrl mice demonstrated a latency of 10.33 ± 2.31 s. The model group showed a marked impairment, taking 65.18 ± 4.52 s to locate the TEH (p < 0.01). ZNJN treatment again significantly improved performance in a dose-dependent manner, with the ZNJN-L, ZNJN-M, and ZNJN-H groups requiring 55.02 ± 4.36, 50.50 ± 3.25, and 42.47 ± 4.13 seconds, respectively. The DA group showed the shortest latency among treated groups, at 39.75 ± 2.88 s, comparable to the Ctrl group. In addition to the Barnes Maze test, spatial working memory was assessed using the Y-maze test (Figure 1E). The percentage of entries into the novel arm was 35.90 ± 3.71% in the Ctrl group, indicating normal exploratory behavior. This proportion was significantly reduced in the model group, which showed only 25.90 ± 3.78% novel arm entries (p < 0.01), reflecting impaired working memory. Following ZNJN treatment, the proportion of novel arm entries increased progressively with dosage. Specifically, the ZNJN-L group showed 28.47 ± 3.65%, the ZNJN-M group 32.09 ± 3.78%, and the ZNJN-H group 34.55 ± 2.75%. The DA group exhibited a novel arm entry proportion of 35.12 ± 5.35%, closely resembling that of the Ctrl group. The improvement observed in the ZNJN-H group was statistically significant compared to the model group (p < 0.01), indicating a substantial enhancement of spatial working memory. Thus, these findings demonstrate that ZNJN significantly ameliorated cognitive deficits in APP/PS1 mice. The effect was dose-dependent, with the high-dose group showing the greatest improvement across both behavioral paradigms. These results suggest that ZNJN has the potential to improve both spatial learning and memory retention in Alzheimer’s disease model mice.

Aβ_1–42 is a key pathological hallmark in the development of AD, with its abnormal aggregation and deposition in the brain closely associated with disease progression. As a highly amyloidogenic isoform of the Aβ peptide, Aβ_1–42 tends to form insoluble plaques that contribute to synaptic dysfunction, neuronal loss, and cognitive decline. Its elevated levels in cerebrospinal fluid or plasma have been widely recognized as a diagnostic biomarker and an indicator of disease severity (Min-Kaung-Wint-Mon et al., 2024). As shown in Figure 2A, immunohistochemical analysis revealed distinct Aβ_1–42-positive plaques in the brain tissues of APP/PS1 model mice. In particular, the model group exhibited strong immunoreactivity, with extensive localized aggregation of Aβ_1–42 throughout the hippocampal and cortical regions. In contrast, treatment with ZNJN markedly attenuated Aβ_1–42 deposition in a dose-dependent manner. ZNJN-H and DA groups showed the most pronounced reductions in plaque burden, with smaller, less dense deposits and weaker staining intensity. Quantitative analysis of average optical density (Figure 2B) confirmed a significant increase in Aβ_1–42 deposition in the model group compared to the Ctrl group (p < 0.01). Post-treatment analysis demonstrated a significant decrease in Aβ_1–42 levels in all treatment groups, with the ZNJN-H and DA groups exhibiting reductions that were statistically significant relative to the model group (p < 0.01).

In addition to Aβ pathology, neuronal morphology in the hippocampal CA1 region was evaluated using H&E staining (Figure 2C). In the Ctrl group, CA1 neurons were large, round, and evenly distributed, arranged in tightly packed layers with well-preserved cytoplasmic and nuclear structures. Conversely, neurons in the model group exhibited severe structural damage, including nuclear condensation, cytoplasmic shrinkage, irregular cell contours, and disorganized alignment, indicative of neurodegeneration. Following treatment, neuronal morphology showed marked improvement across all ZNJN and DA groups. In the ZNJN-H and DA groups, neurons appeared more uniform in shape and size, with intact nuclei and reduced intercellular spacing, closely resembling the normal architecture observed in Ctrl mice. The ZNJN-M group also demonstrated moderate improvement, with clearer stratification and reduced morphological abnormalities. However, in the ZNJN-L group, although some neuroprotective effects were evident, neurons remained loosely arranged with noticeable nuclear swelling and increased inter-neuronal gaps, suggesting only partial recovery. Collectively, these findings indicate that ZNJN exerts neuroprotective effects by reducing Aβ_1–42 burden and preserving hippocampal neuronal structure, with the high-dose treatment showing efficacy comparable to the positive control (DA). These effects are consistent with the observed improvements in cognitive function described in the behavioral assessments.

GFAP and Iba-1 are widely recognized as key markers of astrocyte and microglial activation, respectively. In the central nervous system, GFAP is predominantly expressed in astrocytes, where it contributes to the maintenance of neuronal structural integrity and glial homeostasis. In contrast, Iba-1 is specifically expressed in microglia and is markedly upregulated in response to neuroinflammatory stimuli, injury, or neurodegenerative processes. As illustrated in Figure 3A, hippocampal GFAP expression was significantly elevated in the APP/PS1 model group compared to Ctrl group, indicating robust astrocyte activation. Following ZNJN treatment, GFAP expression levels declined across all intervention groups. This reduction was most pronounced in ZNJN-H and DA groups, both of which showed significantly lower GFAP immunoreactivity relative to the model group (p < 0.01) (Figure 3C).

Similarly, immunofluorescence analysis of Iba-1 expression revealed marked microglial activation in the model group (Figure 3B). Quantitative assessment of average optical density values (Figure 3D) confirmed a significant upregulation of Iba-1 in the model group compared to the Ctrl group (p < 0.01). ZNJN administration led to a dose-dependent reduction in Iba-1 expression, with the ZNJN-H and DA groups again exhibiting the most substantial decreases (p < 0.01). Collectively, these findings suggest that ZNJN effectively attenuates glial activation in the hippocampus of APP/PS1 mice, as evidenced by the downregulation of GFAP and Iba-1 expression. This implies a potential role for ZNJN in the suppression of neuroinflammation and the preservation of neuronal homeostasis in Alzheimer’s disease pathology.

H&E staining was performed to evaluate colonic tissue morphology across the different experimental groups, as shown in Figure 4A. In Ctrl group, the colonic mucosa exhibited intact architecture, characterized by well-organized and uniformly distributed villi. No signs of edema, vascular congestion, or inflammatory cell infiltration were observed, indicating normal intestinal integrity. In contrast, colonic tissues from the APP/PS1 model group displayed pathological alterations indicative of mild inflammation. These included mucosal edema, infiltration of inflammatory cells into the epithelial and lamina propria layers, and disruption of villus architecture. The mucosal surface appeared irregular, with shortened, uneven, and partially detached villi.

ZNJN treatment resulted in a dose-dependent improvement in colonic morphology. In the ZNJN-L, ZNJN-M, and ZNJN-H groups, reductions in mucosal edema and inflammatory infiltration were observed, along with partial restoration of villus structure. Although mild epithelial shedding persisted in some cases, the overall mucosal architecture was markedly more preserved compared to the model group. Among the treatment groups, the ZNJN-H group exhibited the most substantial histological improvement, with relatively intact and orderly villi, suggesting a strong protective effect on the intestinal barrier. Interestingly, DA group showed minimal morphological improvement relative to the model group, indicating that ZNJN may have unique protective effects on intestinal structure that are not shared by standard anti-AD pharmacotherapy. These findings suggest that ZNJN not only exerts neuroprotective effects but may also contribute to the maintenance of intestinal mucosal integrity, potentially through modulation of systemic inflammation or the gut–brain axis.

As shown in Figures 4B–E, modeled mice demonstrated substantial increase TNF-α, IL-6, IL-1β, and LPS (p < 0.01) levels in serum compared with Ctrl. After ZNJN intervention, TNF-α, IL-6, IL-1β, and LPS (p < 0.01) levels in serum were markedly lowered in all treatment groups in comparison with the model (p < 0.01). Among the treatments, the ZNJN-H and DA groups showed the most pronounced effects, which was statistically significant (p < 0.01). The results indicate that ZNJN, particularly at higher doses, can effectively reduce inflammation markers in AD mice, highlighting their potential anti-inflammatory effects in AD.

As shown in Figures 4B–E, serum levels of pro-inflammatory mediators, including TNF-α, IL-6, IL-1β, and LPS, were significantly elevated in the APP/PS1 model group compared to Ctrl group (p < 0.01), indicating a pronounced systemic inflammatory response associated with Alzheimer’s disease pathology. Following ZNJN treatment, all dosage groups exhibited marked reductions in serum concentrations of TNF-α, IL-6, IL-1β, and LPS when compared to the model group (p < 0.01). Notably, ZNJN-H and DA groups demonstrated the most substantial decreases in these inflammatory markers, with differences reaching statistical significance relative to the model group (p < 0.01). These findings indicate that ZNJN possesses significant anti-inflammatory properties, particularly at higher doses. By effectively suppressing systemic inflammatory mediators, ZNJN may contribute to the mitigation of neuroinflammation and associated pathological processes in Alzheimer’s disease.

RT-qPCR analysis of inflammatory cytokine mRNA expression levels in mice hippocampal tissue (as shown in Figures 4F,G) revealed significantly elevated IL-1β and TNF-α mRNA expression in the Model group compared to the Control group (p < 0.05). Following drug treatment, mRNA expression levels of IL-1β and TNF-α in hippocampal tissue were significantly reduced in treated mice, with more pronounced effects observed in the ZNJN-H and DA groups (p < 0.05).

To investigate the impact of ZNJN on gut microbiota, a comprehensive analysis of microbial diversity and taxa composition was conducted. As shown in Figure 5A, a Venn diagram illustrating OTUs revealed that all six experimental groups shared 104 OTUs. However, Ctrl group exhibited a total of 731 OTUs, whereas the APP/PS1 model group displayed a marked reduction to 326 OTUs, indicating a significant loss of microbial diversity associated with Alzheimer’s disease pathology. PCoA was subsequently performed to assess β-diversity among groups (Figure 5B). The first two principal components (PC1 and PC2) accounted for 23.1 and 13.05% of total variance, respectively. The model group was clearly separated from the Ctrl group, reflecting substantial microbial dysbiosis. Notably, the microbiota profiles of ZNJN-treated mice, particularly ZNJN-H group, shifted closer to the Ctrl group, suggesting partial restoration of microbial community structure following treatment. To evaluate α-diversity, Shannon and Simpson indices were calculated (Figures 5C,D). The model group exhibited a significant reduction in the Shannon index, indicative of decreased microbial diversity. ZNJN administration led to an upward trend in Shannon index values across all treated groups, although the differences were not statistically significant. The Simpson index demonstrated a corresponding decrease following ZNJN intervention, consistent with improved community evenness; however, this change also lacked statistical significance. Overall, these data suggest that ZNJN treatment contributed to microbial diversity in APP/PS1 mice.

At the phylum level (Figures 6A,C), the relative abundance of Bacteroidetes was significantly reduced in the model group compared to the Ctrl (p < 0.01), while Firmicutes and Desulfobacterota were significantly increased, resulting in a notably elevated Firmicutes/Bacteroidetes (F/B) ratio. ZNJN-H treatment partially reversed these changes, increasing the relative abundance of Bacteroidetes and decreasing that of Firmicutes and Desulfobacterota, although the differences were not statistically significant. At the genus level (Figures 6B,D), the gut microbiota was predominantly composed of Allobaculum, Ligilactobacillus, Akkermansia, Lachnoclostridium, Dubosiella, and Bacteroides. Relative to the Ctrl group, the model group showed significant reductions in Bacteroides and Akkermansia (p < 0.01), while Allobaculum, Desulfobacterota, and Turicibacter were markedly elevated (p < 0.01). ZNJN-H treatment resulted in a significant increase in the relative abundance of Lachnoclostridium (p < 0.01), and partial recovery of Bacteroides and Akkermansia, though these changes were not statistically significant. Additionally, a significant decrease in Turicibacter was observed in the ZNJN-H group compared to the model group (p < 0.01), indicating a potential reversal of pro-inflammatory microbial shifts.

To identify taxa most influenced by ZNJN treatment, LEfSe was performed on OTUs with relative abundance > 0.5% (Figures 7A,B). In the model group, Allobaculum, Desulfovibrio, and Turicibacter were dominant, whereas the Ctrl group was characterized by Ruminococcus, UCG_010_genus, and Adlercreutzia. The ZNJN-H group showed enrichment in several potentially beneficial genera, including Dubosiella, Ileibacterium, Lachnoclostridium, Parasutterella, Coriobacteriaceae_UCG_002, and Intestinimonas, as well as two specific species, Anaerofustis stercorihominis and Ileibacterium valens. In contrast, the DA group was dominated by Helicobacter and Lachnospiraceae_NK4A136_group.

To evaluate the relationship between systemic inflammation and gut microbiota, Spearman correlation analysis was conducted, aiming to identify microbial taxa significantly associated with inflammatory mediators. The results are visually presented at both the phylum level (Figure 7C) and genus level (Figure 7D). At the phylum level, Firmicutes, Desulfobacterota, Campylobacterota, and Patescibacteria exhibited significant positive correlations (p < 0.01) with pro-inflammatory cytokines TNF-α, IL-1β, and IL-6. Additionally, LPS levels were positively correlated (p < 0.01) with Firmicutes and Campylobacterota, but not with Desulfobacterota or Patescibacteria. In contrast, Verrucomicrobiota, Bacteroidota, and Cyanobacteria showed significant negative correlations (p < 0.01) with TNF-α, IL-1β, IL-6, and LPS, suggesting a potential anti-inflammatory role for these phyla.

At the genus level, 14 genera demonstrated significant positive correlations with TNF-α, IL-1β, and IL-6 (p < 0.01), including Facklamia, Helicobacter, Turicibacter, Atopostipes, Ileibacterium, Lachnoclostridium, Lactococcus, Candidatus_Soleaferrea, Gordoni- bacter, Allobaculum, Odoribacter, Desulfovibrio, Rikenella, and Coriobacteriaceae_UCG-002. Conversely, multiple genera were negatively correlated (p < 0.01) with all four inflammatory mediators including TNF-α, IL-1β, IL-6, and LPS. These included Intestinimonas, [Eubacterium]_ventriosum_group, Butyrici- monas, Marvinbryantia, Erysipelatoclostridiaceae_genus, Bacter- oides, Akkermansia, Tyzzerella, Staphylococcus, Erysipelo- trichaceae_genus, Adlercreutzia, and Clostridia_vadinBB60_ group_genus. These taxa may exert anti-inflammatory effects through gut–immune interactions.

Among the identified genera, a substantial number exhibited statistically significant correlations with the measured inflammatory mediators. Specifically, IL-1β was significantly associated with 34 genera, of which 15 were positively and 19 negatively correlated. Similarly, TNF-α showed significant correlations with 38 genera (16 positive, 22 negative), and IL-6 was linked to 43 genera, including 19 with positive and 24 with negative associations. In the case of LPS, significant correlations were observed with 39 genera, comprising 14 positive and 25 negative relationships.

These results highlight a consistent pattern wherein a greater proportion of microbial genera demonstrated negative correlations with systemic inflammatory markers, suggesting that many of these microbes may play a protective or anti-inflammatory role in the host. Conversely, a subset of genera exhibited positive associations, implying potential involvement in pro-inflammatory pathways. This duality underscores the complexity of hostd positive associations, implying potential involvement disease and supports the hypothesis that modulation of specific microbial taxa may influence systemic inflammation and disease progression.