A total of 60 eight-week-old male C57BL/6 mice with specific pathogen free (SPF) were used in this study, weighing about 25 ± 2 g, which were obtained from Liaoning Changsheng Biotechnology Co., Ltd. (Benxi, Liaoning, China; Certification number: SCXK 2020-0001). The experimental animals used in this study were housed in the Center Laboratory of Chinese Medicine of Hubei University of Chinese Medicine. The standard laboratory was controlled at the temperature of 23 ± 2°C, 60% relative humidity, and 12:12 h lightdark cycle. During the experiment, the animals had free access to a standard diet and water. The approval of animal feeding and experiment protocol were obtained from the Animal Ethics Review Committee of Hubei University of Chinese Medicine (No. 8217150845) and conducted in accordance with the ethical guidelines.

Qiangji Decoction (QJD) comprised four raw herbs, including wine-steamed roots of Rehmannia glutinosa Libosch. (30 g), air-dried mature seeds of Ziziphus jujuba Mill. var. spinosa (Bunge) Hu ex H. F. Chou (30 g), air-dried tuberous roots of Ophiopogon japonicus (L. f) Ker-Gawl. (30 g), air-dried roots and rhizomes of Polygala tenuifolia Wild. or Polygala sibirica L. (6 g), which were provided by Jinpai Chizhengtang Pharmaceutical Co., Ltd. (Huangshi, Hubei, China). All raw herbs were authenticated by Prof. Qiu-yun You, a pharmacologist from Hubei University of Chinese Medicine, and identified in accordance with the Chinese Pharmacopoeia. Metformin hydrochloride extended-release tablets (0.5 g/tablet, batch number: ABU7707) were obtained from Sino-American Shanghai Squibb Pharmaceutical Co., Ltd. (Shanghai, China).

According to the methods described in our previous report with minor modifications, the stock solution of QJD was prepared (Long et al., 2021). In general, the raw herbs were crushed into pieces, and the mixtures of raw herbs were soaked in water (1:8, w/v) for 30 min. Subsequently, the mixed herbs were decocted for 30 min. After filtration, the residue was decocted for 20 min (1:6, w/v). The two filtrates were mixed and refluxed in the rotary evaporators (QYMD-60, Qi yu industry co., LTD., Shanghai, China) for 1.5 h, and finally concentrated into a stock solution. Metformin was dissolved in saline and prepared as a 0.01 g/ml stock solution.

D-galactose (D-gal), purity>99.0%, was supplied by Solarbio Science and Technology Co., Ltd. (Beijing, China; D8310). HE staining solution, Nissl staining solution, and Fluorescein (FITC) TUNEL Cell Apoptosis Detection Kit were obtained from Servicebio Technology Co., Ltd. (Wuhan, Hubei, China; G1005, G1036, and G1501, respectively). Fluoro-Jade B (FJB) staining kit was obtained from Merck-Millipore (Darmstadt, Germany; AG310). ELISA Kits for the detection of TNF-α, IL-1β, and IL-6 were supplied by ABclonal technology (Wuhan, Hubei, China; RK00027, RK00006, and RK00008, respectively). Rabbit monoclonal anti–Bcl-2, rabbit monoclonal anti-Bax, rabbit monoclonal anti–caspase-3, rabbit polyclonal anti–cleaved caspase-3, rabbit monoclonal anti–total-AMPKα, rabbit monoclonal anti–phospho-AMPKα (Thr172), rabbit monoclonal anti-SIRT1, rabbit monoclonal anti–β-actin, and HRP-linked goat anti-rabbit IgG were purchased from ABcam (Cambridge, MA, USA; ab32124, ab32503, ab32351,ab2302, ab32047, ab133448, ab110304, ab8227, and ab6721, respectively). Rabbit polyclonal anti-GFAP, rabbit monoclonal anti-IBA1, and Cy3-conjugated goat anti-rabbit IgG were purchased from ABclonal technology (Wuhan, Hubei, China; A0237, A1527, and AS007, respectively). Rabbit monoclonal anti–total-IκBα, rabbit monoclonal anti–phospho-IκBα (Ser32), rabbit monoclonal anti–total-NF-κB p65, and rabbit monoclonal anti–phospho-NF-κB p65 (Ser536) were purchased from Cell Signaling Technology (Beverly, MA, USA; #4812, #2859, #8242, and #3033, respectively).

After the acclimation, the mice were randomly assigned to the following six different groups of 10 animals each, namely, negative control group (NC), D-gal group (D-gal), metformin group (Met), low dose of QJD group (L-QJD), middle dose of QJD group (M-QJD), and high dose of QJD group (H-QJD). With reference to the previous reports (Baeta-Corral et al., 2018; Ullah et al., 2020; Ahmad et al., 2021), the mice in the negative control group were administered with 0.9% saline by subcutaneous injection for 8 weeks, while the other groups were administered with D-gal (100 mg/kg/d). Based on the previous reports and clinical equivalent doses (Farr et al., 2019; Li et al., 2019), the therapeutic dose of the metformin group and QJD-treated group was confirmed accordingly. From the fifth to the eighth week, the NC group and the D-gal group received the same amount of normal saline by oral administration, the metformin group was given metformin (200 mg/kg/d), and the QJD-treated groups were treated with three different doses of QJD extract (L-QJD: 12.48 g/kg/d, M-QJD: 24.96 g/kg/d, H-QJD: 49.92 g/kg/d). As shown in Figure 1A, the study design of this research has been illustrated.

After the treatment was completed, the hippocampus-dependent learning and memory ability of mice in each group were evaluated by the Morris water maze (MWM) test. The equipment of the MWM test was given in our previous report, and it was performed with minor modifications (Long et al., 2021). Briefly, the MWM test contains two tests, namely, the navigation and spatial probe test. During the navigation test, every trained animal was given 60 s to search for the platform, and the escape latency was documented by the MWM software (version YH-MWM, Wuhan Yihong Technology Co., Ltd., China). If the experimental animals did not locate the platform within the allocated 60 s, they were manually led to the platform and allowed to remain on the platform for at least 15 s. After the completion of the navigation test, the hidden platform was retracted manually, and then the spatial probe test was executed. MWM software automatically recorded and analyzed the number of crossing the hidden platform in the spatial probe test. In addition, we also analyzed and counted the time of staying on the target quadrant and the mean swimming speed in each group.

After the completion of MWM, four animals in each group were sacrificed, and then the brain tissues were collected to fix in 4% paraformaldehyde for HE staining, FJB staining, Nissl staining, TUNEL staining, and immunofluorescence staining. The remaining mice were anesthetized in the same way, and the hippocampal tissue was separated and split into two parts on a cold plate. One portion of the hippocampal tissue was used for ELISA, and another part was used for Western blot. After separation, the hippocampal tissue was in a cryopreservation tube and stored in liquid nitrogen for 30 min, finally kept at −80°C until further processing.

The brain tissues were immersed in 4% paraformaldehyde at 4°C for 24 h and then processed into paraffin-embedded tissues. Subsequently, a microtome (CM 2016, Leica, Germany) was employed to cut the paraffin-embedded tissues into 5-μm-thick slices. The slices were dewaxed with xylene and dehydrated with gradient ethanol (100–75%). After being rinsed with double distilled water, the slices were placed in a solution of hematoxylin for 3–5 min and then stained with a solution of eosin for 5 min. Finally, neutral gum was used to mount the slices. The stained slices were photographed and analyzed with an upright optical microscope (ECLIPSE Ni-E, Nikon, Japan).

Based on the manufacturer’s protocol and a previously described report, the FJB staining was carried out (Xie et al., 2019). Briefly, the prepared sections were immersed in xylene for 15 min. Subsequently, the slices were placed in a gradient ethanol of 85 and 75% for dehydration for 5 min each. The slices were rinsed in 0.1 M PBS for 1–2 min and then incubated to a solution of final working concentration of FJB for 10 min. The cell nuclei were labeled by 4 ′6-diamidino-2-phenylindole (DAPI). After FJB staining, the slices were visualized and photographed under an upright fluorescence microscope (DP72, Olympus, Japan), and then the number of FJB-positive cells were measured by using Image-Pro Plus 7.0 software (Media Cybernetics, Inc., Rockville, MD, USA).

The paraffin-embedded slices were dewaxed with xylene for 5 min. Later, the dewaxed slices were dehydrated in a gradient ethanol series with decreasing concentration for 5 min (100–75%). After being dehydrated, the slices were treated with Nissl staining solution at 37°C for 30 min and then washed in 0.1 M PBS for 5 min. Subsequently, the paraffinized slices were incubated in 70% alcohol differentiation at 37°C for 15 s. The sections were dehydrated with gradient alcohol (70–95%) for 2 min and finally covered with neutral gum. The images were photographed and analyzed with an imaging system (BX50, Olympus, Japan). Subsequently, the number of neurons in the hippocampus was analyzed by Image-Pro Plus 7.0 software.

The TUNEL staining was employed to assess the level of hippocampal neuron apoptosis in each group, and it performed as described previously (Li et al., 2021). After being dewaxed and rehydrated, the paraffin-embedded slices were incubated with proteinase K at room temperature for 25 min. After incubation, the sections were washed in 0.1 M PBS for 5 min and then incubated in permeabilize working solution at room temperature for 20 min. The slices are slightly dried and then incubated with TUNEL mixture at room temperature for 2 h. Subsequently, the cell nuclei were counterstained by using 4′ 6-diamidino-2-phenylindole (DAPI), and the slices were finally mounted with anti-fade mounting medium. The slices were observed under an upright fluorescence microscope (DP72, Olympus, Japan). In this study, we used Image-Pro Plus 7.0 software to measure the number of TUNEL-positive cells, so as to assess whether QJD could improve the hippocampal neuronal apoptosis.

Based on the previous report, the immunofluorescence staining was conducted with minor modifications (Rong et al., 2019). In brief, the slices were dewaxed with xylene and dehydrated with gradient ethanol (100–75%), and then washed in 0.1 M PBS for 5 min. Next, the dehydrated slices were placed in EDTA antigen retrieval buffer for 8 min at a sub-boiling temperature. Subsequently, the slices were blocked in 3% bovine serum albumin (BAS). After washing three times using 0.1 M PBS 3 min, the primary antibody (IBA1, 1:100; GFAP, 1:200) was selected to incubate the slices overnight at 4°C. The following day, the Cy3-conjugated secondary antibody (1:200) was added to incubate at 37°C for 50 min. DAPI was employed to label the cell nuclei, and the slices were finally mounted with anti-fade mounting medium. In this study, we used Image-Pro Plus 7.0 software to calculate the mean fluorescence intensity of images.

To evaluate whether QJD could inhibit the neuroinflammation, we employed enzyme-linked immunosorbent assay (ELISA) kits to estimate the contents of pro-inflammatory factors in hippocampal tissue, such as TNF-α, IL-1β, and IL-6. In brief, the separated hippocampus was immersed in PBS (1:8, w/v), homogenized in a cold tissue homogenizer (JXFSTPRP-CL-24, Tuohe Electromechanical Technology Co., Ltd., Shanghai, China), followed by centrifugation (12,000 rpm) at 4°C for 10 min. The supernatant was then collected. Subsequently, the level of protein was quantified using the BCA Protein Assay kit. Finally, the contents of TNF-α, IL-1β, and IL-6 in hippocampus were detected using ELISA kits, following the manufacturer’s protocol.

In brief, the separated hippocampus was lysed in RIPA lysis buffer (1:8, w/v) containing phenylmethylsulfonyl fluoride (PMSF) and phosphatase inhibitors to extract the total proteins of hippocampus. After homogenizing, the hippocampal homogenates were subjected to centrifugation (12,000 rpm) at 4°C for 30 min, and the supernatants were collected. Subsequently, BCA Protein Assay kit was used to calculate the protein concentration of each sample. After that, the 10–12% separation gel buffer and 5% stacking gel buffer were prepared, and the protein of each sample (30 μg) was separated with SDS-PAGE. The protein was then transferred to the PVDF membranes (Millipore, Darmstadt, Germany). After blocked in 5% nonfat milk at room temperature for 30 min, the membranes were incubated with primary antibodies including anti–Bcl-2 (1:1,000), anti-Bax (1:1,000), anti–cleaved caspase-3 (1:1,000), anti-AMPKα (1:1,000), anti–p-AMPKα-Thr172 (1:1,000), anti-SIRT1 (1:1,000), anti-IκBα (1:1,000), anti–p-IκBα-Ser32 (1:1,000), anti–NF-κB p65 (1:1,000), anti–p-NF-κB p65-Ser536 (1:1,000), and anti–β-actin (1:1,000) at 4°C overnight. The next day, the membranes were incubated with HRP-linked secondary antibody (1:10,000) at room temperature for 30 min. After washing three times with TBST, ECL kit was used to capture the blots, and the band intensity of each sample was quantified by using the Image J software (Bethesda, United States).

All experimental data were dealt by using Statistical Analysis System software (SAS, version 9.4, SAS Institute Inc., Cary, NC, USA), and the results were expressed as mean ± standard error of the mean (SEM). The two-way analysis of variance (ANOVA) with repeated measures was used to analyze the escape latency, and the remaining data were analyzed by one-way ANOVA. If the variance was homogeneous, the Bonferroni post hoc test was adopted. If not, the Tamhane T2 test was used. When the p-value was less than 0.05, values were considered statistically significant.

MWM is one of the most widely used behavioral testing methods, which has been commonly used to evaluate the learning and memory ability of brain aging–related neurodegenerative diseases (Garthe and Kempermann, 2013; Wahl et al., 2017). Accordingly, we assessed the mice’s spatial learning ability in each group by using the navigation test. During the navigation test, there was no obvious alteration in the escape latency among all groups during 3 days of consecutive training (p > 0.05, Figure 1B). Starting from the fourth day, the D-gal treatment significantly increased the escape latency, whereas this phenomenon was significantly reversed by the metformin, M-QJD, and H-QJD (p < 0.05 or 0.01; Figures 1B, C). The representative path tracings of each group in the navigation test were presented in Figure 1D, which indicated the M-QJD significantly shortened the escape latency in D-gal-induced mice. Besides, we examined the spatial memory ability of mice by using the spatial probe test. According to the spatial probe test, we found the M-QJD group and H-QJD group had a higher number of crossing platform when compared to the D-gal group (p < 0.05 or 0.01; Figure 1E). Meanwhile, the M-QJD group and H-QJD group significantly increased the time that mice spent in the target quadrant (p < 0.05 or 0.01; Figure 1F). Nevertheless, the L-QJD group did not significantly change the abovementioned results (p > 0.05; Figures 1E,F). In mean swimming speed, no statistical difference was found among all groups (p > 0.05; Figure 1G), which implies that mice’s swimming ability had no difference. These results indicated that chronic D-gal administration could induce the cognitive impairment, and the QJD, especially the M-QJD, reversed this phenomenon, whereas the L-QJD did not mitigate the cognitive impairment in D-gal–induced mice.

The MWM test results show that the QJD could alleviate the cognitive deficits induced by D-gal, and thereby, the HE staining and FJB staining were utilized to evaluate whether QJD could rescue the pathological alterations of the hippocampus induced by D-gal. In HE staining, we noticed that the D-gal group’s neurons in the hippocampal CA1 and CA3 regions were degenerated with dark staining, deformed, and denatured nuclei in comparison with the NC group (Figure 2A). However, the neuronal damages in these regions of the M-QJD group and H-QJD group were decreased in comparison to the D-gal group, and the stained neurons and deformed nuclei were also alleviated. In addition, neurodegenerative changes in the hippocampus were detected by using FJB staining. As showed in FJB staining, compared with the NC group, there were remarkably increased FJB-positive cells in the hippocampal CA1 and CA3 regions of D-gal group (p < 0.01, Figures 2B–D). However, after treatment with M-QJD and H-QJD, the number of FJB-positive cells in these regions were significantly diminished (p < 0.05 or 0.01; Figures 2B–D). The above findings manifested that QJD could relieve the pathological alterations of hippocampus induced by D-gal.

We also employed Nissl staining and TUNEL staining to investigate whether QJD could improve the hippocampal neuron loss and apoptosis in D-gal–induced mice. In Nissl staining, we found that there was a decreased number of neurons in the CA1 and CA3 regions, when compared to the NC group (p < 0.01; Figures 3A,C,D). Whereas compared to the D-gal group, the decreased neurons were rescued by the M-QJD and the H-QJD (p < 0.05 or 0.01; Figures 3A,C,D). We also used the TUNEL staining to evaluate whether QJD could restrain neuronal apoptosis and quantified the number of TUNEL-positive cells by using Image-Pro Plus 7.0 software. The TUNEL staining results showed that the decreased TUNEL-positive neurons in the CA1 and CA3 regions were inhibited by the M-QJD and the H-QJD (p < 0.05 or 0.01; Figures 3B,E,F). In addition, the apoptotic markers including Bcl-2 (anti-apoptotic protein), Bax (apoptosis regulator), and cleaved caspase-3 (pro-apoptotic protein) were detected by Western blot (Ullah et al., 2020). Our finding verified that the D-gal treatment could upregulate the protein levels of Bax and cleaved caspase-3 in the hippocampus compared to the NC group (p < 0.01; Figures 3G,I,J), while the protein levels of Bcl-2 were significantly downregulated (p < 0.01; Figures 3G,H). After 4 weeks of M-QJD and H-QJD treatment, the above phenomenon was reversed (p < 0.05 or 0.01; Figures 3G–J). These results implied that the QJD could alleviate the hippocampal neuron loss and apoptosis in D-gal–induced mice.

Neuroinflammation acts as the contributor in accelerating the pathological progression of aging-related neurodegenerative diseases, and the increasing researches have demonstrated that chronic D-gal administration can stimulate the microglia and astrocytes activation in central nervous system, thereby inducing chronic neuroinflammation (Lu et al., 2010; Cao et al., 2019; Yahfoufi et al., 2020). Therefore, in this study, we measured the contents of TNF-α, IL-1β, and IL-6 in the hippocampus through ELISA. When compared to the D-gal group, the NC group had a lower level of TNF-α, IL-1β, and IL-6 (p < 0.01; Figures 4A–C). However, after treatment with M-QJD and H-QJD, these pro-inflammatory cytokines were suppressed (p < 0.05 or 0.01; Figures 4A–C). The microglia and astrocytes activation is the critical factor for the release of pro-inflammatory cytokines. Growing researches have confirmed that chronic D-gal administration can induce the microglia and astrocytes activation, thereby accelerating the release of pro-inflammatory cytokines (Lu et al., 2010; Jeong et al., 2021; Oskouei et al., 2021). So, we used the immunofluorescence staining to detect the levels of IBA1 and GFAP. Our findings showed that the D-gal group had higher fluorescence intensity of IBA1-positive microglia and GFAP-positive astrocytes (p < 0.01; Figures 4D–I), when compared to the NC group. After treatment with M-QJD and H-QJD, the fluorescence intensity of IBA1-positive microglia and GFAP-positive astrocytes was significantly decreased (p < 0.05 or 0.01; Figures 4D–I). Collectively, these results manifested that QJD could alleviate chronic neuroinflammation through suppressing the activation of microglial and astrocytes in D-gal–induced mice.

In the aging-related neurodegenerative diseases, the AMPK/SIRT1 signaling pathway acts as a crucial role in regulating energy metabolism, apoptosis as well as neuroinflammation (Domise and Vingtdeux, 2016; Chen et al., 2020). Previous research has confirmed that the inactivation of this signal pathway can accelerate neuroinflammation, while the activation of this signal pathway can naturally inhibit neuroinflammation (Peixoto et al., 2017). In this study, we examined the levels of AMPK and SIRT1 and the phosphorylation of APMK in the hippocampus by using Western blot. Compared with NC group, the levels of SIRT1 and the phosphorylation of APMK in the D-gal group were significantly decreased (p < 0.01; Figures 5A–D), which indicated that the AMPK/SIRT1 signaling pathway was inhibited by D-gal. After treatment with QJD, the M-QJD and H-QJD had a higher levels of SIRT1 and the phosphorylation of APMK, when compared to the D-gal group (p < 0.05 or 0.01; Figures 5A–D). Furthermore, we noticed that the H-QJD had the similar effect with M-QJD. Those results showed that QJD could activate the AMPK/SIRT1 signaling pathway in the hippocampus of D-gal–induced mice.

AMPK/SITR1 signaling pathway is an important regulator of NF-κB, and the activity of NF-κB can be restrained by the AMPK/SITR1 signaling pathway. Growing researches have suggested that the activation of the AMPK/SITR1 signaling pathway can inhibit the activation of NF-κB, so we further investigate whether QJD alleviates the neuroinflammation through regulating the AMPK/SITR1–mediated NF-κB signaling pathway. In this study, we first used Western blot to examine the IκBα and NF-κB p65 in the hippocampus. Subsequently, the phosphorylated IκBα (Ser32) and NF-κB p65 (Ser536) were also detected by Western blot. As revealed by Figures 6A–E, the D-gal treatment significantly increased the levels of phosphorylated IκBα and phosphorylated NF-κB p65 in the hippocampus of the D-gal group, when compared to the NC group (p < 0.01). However, the H-QJD, especially the M-QJD, suppressed the levels of phosphorylated IκBα and phosphorylated NF-κB p65 (p < 0.05 or 0.01; Figures 6A–E). These data indicated that QJD could inhibit the NF-κB signaling pathway activation in the hippocampus of D-gal–induced mice.