Alzheimer’s disease (AD) is an irreversible, chronic, and relentless neurodegenerative disease, characterized by memory loss, psychoses, affective disorders, and behavioral alterations (Gong et al., 2015; Fang et al., 2017; Cai F. F. et al., 2018; Pang et al., 2018; Sun et al., 2018). Numerous details of AD-related neuropathological changes have been described, including the formation of amyloid plaques, neurofibrillary tangles, and loss of neurons and synapses. Nonetheless, key events leading to the pathogenesis and progression of the disease remain poorly understood (Kumar et al., 2015).

Considering the multifactorial causation of AD, pharmacotherapy approaches targeting a single molecule may not optimal. And As an example, although the anti-amyloid antibody, aducanumab, developed by Biogen, has been shown to lower the number of amyloid plaques in the brain, there is no evidence that it improves cognitive outcomes. Thus, the accelerated approval of aducanumab by the FDA remains controversial (Mullard, 2021). Given the worldwide impact of AZ effective novel approaches to the management of the disease are urgently needed (Luo et al., 2020). TCM has been in clinical use for thousands of years, particularly in the management of chronic diseases. Numerous Chinese herbal formulations show efficacy in the treatment of AD. These are gaining gradual worldwide acceptance due their holistic approach acting simultaneously of multiple symptoms, while exhibiting limited side effects (Wang and Zhang, 2017).

Yuan-Zhi Decoction (YZD) is a natural plant-derived TCM formulation. It can be traced back to the book “Qi xiao liang fang” written by Fang Xian during the Ming Dynasty. YZD is a mixture of eight Chinese herbs, namely: Polygala tenuifolia Willd. (yuan-zhi in Chinese), Panax ginseng C. A. Mey. (ren-shen in Chinese), Acorus tatarinowii Schott (shi-chang-pu in Chinese), Notopterygium incisum Ting ex H. T. Chang (qiang-huo in Chinese), Asrum heterotropoides Fr. Schmidt var. mandshuricum (Maxim.) Kitag. (xi-xin in Chinese), Ephedra sinica Stapf (ma-huang in Chinese), Paeonia lactiflora Pall. (chi-shao in Chinese), Atractylodes macrocephala Koidz. (bai-zhu in Chinese). This formulation has been in use to treat a variety of medical issues. Recent pharmacological studies have demonstrated that many ingredients of YZD have beneficial effects in AD. Yuanzhi (Polygala tenuifolia Willd.) is an ingredient of multiple formulas use to managed age related memory impairment (May et al., 2016; Howes et al., 2017; Li B. et al., 2021). Shichangpu (Acori Tatarinowii Rhizoma), has pronounced neuro-pharmacological effects (Oh et al., 2004; Chang and Teng, 2018; Song et al., 2018; Wang L. et al., 2019; Deng et al., 2020). Xixin decoction was shown to reduce phosphorylation toxicity at specific sites of the tau protein in the brains of rats (Diwu et al., 2013). However, due to the complex composition of YZD, a comprehensive elucidation of the molecular mechanisms involved in tits beneficial effects in AD remains incomplete.

Network pharmacology, first proposed by Andrew Hopkins in 2007 is a relatively novel approach in elucidating the mechanism of action of multicomponent drugs (Luyen et al., 2015; Lee et al., 2018; Wang N. et al., 2019). Its core concept is to optimize treatment strategies based on the biological network of disease features, bioactive agents, and drug targets that are connected to each other (Hopkins, 2007). With rapid progress in bioinformatics and systems biology, network pharmacology is becoming an increasingly viable approach in identifying active compounds and elucidating the underlying pharmacological mechanisms involved in their beneficial effects (Li et al., 2011; Li and Zhang, 2013; Cai H. et al., 2018; Zhou et al., 2020). Through the integration of chemical component identification, target prediction, and network construction, this approach can effectively predict the molecular mechanisms affected by the active compounds in TCM formulations (Luo Z et al., 2020). As such network pharmacology represents a promising tool in the study of the mechanism of action of complex formulations containing a multitude of active ingredients (Li et al., 2020).

Molecular docking studies use computer simulations to study the interactions between receptors and small molecules potentially binding to them. By predicting certain features, such as the location and affinity of these interaction it is an invaluable tool in drug design and screening of chemical compounds (Chen et al., 2014; Lionta et al., 2014; Saikia and Bordoloi, 2019; Reddyrajula and Dalimba, 2020). Recently the combination of network pharmacology and molecular docking analysis was successfully applied to study the active component of TCM formulations (Wang Y. et al., 2019).

In the study presented here we systematically identified chemical constituents of YZD using liquid chromatography/mass spectrometry and explored possible targets and related pathways with potential relevance to the treatment of AD. The validity of this approach was confirmed by behavioral evaluation, microscopic analysis of the brains, and by Western blotting.

AD is a complex disease affecting multiple biochemical pathways. Given its unclear etiology and pathogenesis, there is an increasing interest in the use of traditional medical formulations in the treatment of the condition (Poornima et al., 2016). TCM generally uses natural products, with a good safety record and low toxicity demonstrated during extensive periods of clinical use. TCM also relies on multi-compound formulations affecting several targets and multi-component pathways. Ultimately, this approach may have significant advantages over single compound medications in the treatment of complex, multifactorial diseases (Poornima et al., 2016).

To the best of our knowledge, this is the first study to systematically investigate the mechanism of action of YZD in combatting AD using network pharmacology analysis. The main chemical constituents of YZD were identified using liquid chromatography coupled to mass spectroscopy. This was followed by network pharmacology studies to assess potential interactions between the detected chemical components and putative AD-related protein targets identified in database searches. With this approach, both the protein targets and candidate active constituents of YZD were predicted in silico. By comparing the list of all potential target proteins affected by the constituents of YZD with a comprehensive list of proteins relevant to the development of AD we identified those proteins YZD is likely to act to deliver its beneficial effects in AD. Next, pathway enrichment analysis was performed to investigate the molecular pathways affected by the 27 components of YZD that were predicted act in AD. Following this computational phase of the work, the APP1/PS1 transgenic mouse model of AD was used to assess the in vivo efficacy of YZD in improving learning and memory deficits. These experiments demonstrated that YZD administration had a positive impact on spatial learning and memory, evidenced by reducing the time needed to reach a hidden platform during MWM trials. Finally, some of the network pharmacology predictions were validated by Western blot analysis and immunohistochemistry, confirming that several AD-related pathways predicted in silico indeed played a critical role in a live animal model.

In this study, data from UHPLC-Q Exactive HF mass spectrometry was complemented with TCM database searches to maximize the number of identified chemical constituents of YZD (Chu et al., 2016). In total 211 compounds were identified experimentally, including high levels of Prenol lipids, Phenol ethers, Benzopyrans, Benzene and substituted derivatives. Searches conducted in TCM-relevant databases found 1026 chemical compounds (Supplementary Table S2). Out of these two lists, a total of 27 compounds were identified that were uniquely present in YZD. Four of these were derived from the herb YuanZhi, 5 from RenShen, 2 from ShiChangPu, 6 from QiangHuo, 4 from MaHuang, 3 from XiXin, 4 from ChiShaoYao, and a single compound from BaiZhu, with 3 ingredients being present in more than one herb.

Oligosaccharide esters are a unique class of sugar esters in Polygonum consisting of a sucrose core and connect monosaccharides, mainly glucose and rhamnose, linked with different glycosidic bonds (Zhao et al., 2020). This raises the possibility that oligosaccharide esters may be biologically active substances in Polygonum japonica that could be utilized in the development of new anti-aging and neuroprotective drugs. While oligosaccharide esters exist in a variety of plants, a sugar core containing three or more monosaccharides can only be found in Polygala. It has been suggested that these unique oligosaccharide esters may have unusual pharmacologic effects (Li, 2008). Studies have shown that tenuifoliside A, tenuifoliside C (E), tenuifoliose B (D), tenuifoliside K, and tenuifoliside I (J) are reasonably well absorbed and enter the systemic circulation, exerting biological effects (Wang et al., 2017).

Some of the identified YZD components were previously shown to be beneficial in AD. Tenuifolin attenuates Amyloid-β₄₂-induced neuroinflammation (Wei, 2016; Wang L. et al., 2019; Chen and Jia, 2020). This mechanism was protective against neuronal apoptosis and improved learning and memory deficits in a mouse AD model (Wang N. et al., 2019). Ginsenoside Rg1 significantly improved cognitive behavioral impairments in several animal models of AD (Nie et al., 2017; Liang et al., 2021), while ginsenoside Rg2 had a protective effect against memory impairments and neuronal death seen in Aβ25-35 transgenic rats (Cui J. et al., 2017; Cui et al., 2021). Ginsenoside Rg3 prevented cognitive impairment in a rat AD model (Zhang et al., 2019) and Notopterol effectively ameliorated cognitive deficits in APP/PS1 mice by attenuating pathology caused by the accumulation of amyloid-β and tau (Jiang X. et al., 2020; Jiang X. W. et al., 2020). Other studies have shown that Paeoniflorin exerted neuroprotective effects in a transgenic mouse model of AD (Zhang et al., 2015; Gu et al., 2016; Ma et al., 2018; Huang et al., 2020; Kong et al., 2020). Albiflorin ameliorated memory deficits in APP/PS1 transgenic mice (Xu et al., 2019) and Atractylodes macrocephala Koidz. was also used to treat Alzheimer’s disease (Zhu et al., 2018). β-Asarone, a crucial compound of YZD, was reported to prevent amyloid β-induced apoptosis by regulating the PI3K/AKT pathway in PC-12 cells (Li Z. et al., 2021; Meng et al., 2021). β-Asarone could also inhibit Amyloid deposition and reverse cognitive deficits in experimental animals (Wang et al., 2021). Nodakenin exhibited potent inhibitory activities against the β-site amyloid precursor protein cleaving enzyme 1 (BACE1) and is regarded as a promising therapeutic option in the treatment of AD (Yousof Ali et al., 2015).

Network pharmacology analysis is able to identify interactions between chemical constituents and their targets/pathways, and establish “compound-target-pathway” maps identifying chemical constituents and the mechanisms associated with their pharmacological activity (Ren et al., 2018). Our work is the first to explore the possible actions of YZD in AD using network pharmacology, laying the foundation for the use of this approach in further research. The potential pharmacological targets of the 27 YZD-derived compounds were identified with the help of the SEA, STITCH, Swiss Target Prediction database searches and PharmMapper. At the same time, protein molecules associated with the pathogenesis of AD were identified from the Drugbank, OMIM, GeneCards, and DisGeNET databases and by the analysis of differentially expressed genes in GEO datasets. This work identified 34 target molecules in AD that could be affected by ingredients in YZD.

After network topology analysis 17 of these genes, including IL6, APP, VEGFA, MPO, ITGB1, IL1B, ACE, MME, MAPT, IL4, CXCR2, GRIN2B, VCP, GSK3B, LTB4R, CHRM1, and BACE1 were considered to be the key molecules targeted by the chemical constituents of YZD. These proteins show complex interactions with each other, forming a PPI network. In such networks a change affecting a single molecule can result in a series of functional abnormalities, potentially leading to the emergence of pathologic abnormalities (Gu et al., 2020). Our results showed that APP, IL6, VEGFA, GSK3B and BACE1 could extensively interact with other members of the network. For example, APP interacts with BACE1, GSK3B, IL6, and PSEN2. APP is a widely expressed transmembrane protein playing a key role in the formation of Aβ in the pathogenesis of AD (Zhu et al., 2017). BACE1 is the main rate-limiting enzyme in the production of Aβ. Previous studies experimentally validated BACE1 as a YZD target in the treatment of AD (Ahmad et al., 2017; Prati et al., 2017). Furthermore, there is emerging evidence that BACE1 inhibitors may be potentially interesting tools in the management of AD (Rueeger et al., 2021), while the inhibition of GSK3B can reduce the phosphorylation of Tau and the conversion of APP to Aβ. These observations provide strong evidence that several components of YZD have significant association with various AD-related signaling pathways. Our molecular docking models investigating the interactions between validated targets (i.e., APP, BACE1 and GSK3B) and the 27 identified key chemical compounds found in YZD provides additional support for a role of YZD in the treatment of AD.

KEGG pathway enrichment analysis showed that the identified molecules preferentially affected the transcription of RNA polymerase II promoters, the process of apoptosis, cell proliferation, signal transduction, synaptic transmission, neuroactive ligand-receptor interactions, and the PI3K-Akt, HIF-1, and calcium signaling pathways. The PI3K/AKT signaling pathway is involved in regulating cell proliferation, differentiation, metabolism, and apoptosis. It also exerts anti-oxidative, anti-inflammatory, and neuroprotective effects. Altered regulation of this pathway has been found to play a central role in the development of AD (Cui W. et al., 2017; Razani et al., 2021). PI3K activation can act as a second messenger to transmit signals into cells, activating AKT phosphorylation. In turn, AKT could exert its anti-apoptotic effects by promoting the phosphorylation of downstream substrates, such as GSK-3β and mTOR. However, it was found that the activation of GSK-3β could lead to an abnormal increase in the level of phosphorylation of tau protein (Jimenez et al., 2011). Gene ontology analysis has also demonstrated that the molecular targets of YZD show a varied distribution within cells.

We also conducted behavioral tests to assess the efficacy of YZD in treating learning and memory deficits in aged APP/PS1 transgenic mice. These mice exhibit impairments in spatial learning and memory, as evidenced by their poor performance in MWM tests. Our results showed that YZD treatment restored learning and memory deficits in these animals, demonstrating a protective effect of this formulation in APP/PS1 transgenic AD mice.

The PI3K/AKT/GSK-3β signaling pathway plays a crucial role in cell growth, survival, apoptosis, and proliferation, and is also one of the important pathways that lead to the phosphorylation of tau protein and then the development of AD. (Kitagishi et al., 2014) Activation of PI3K will promote the phosphorylation of Thr308 within AKT, activating the protein. Thus, the activity of PI3K is reflected by the phosphorylation status of AKT. Activated AKT inhibits the action of GSK-3β downstream in this signaling pathway. GSK-3β is normally in the activated state in the cytoplasm of neurons. The activation of AKT results in the phosphorylation of the Ser9 residue of GSK-3β, inhibiting the activity of GSK-3β. GSK-3β is an important threonine/serine protein kinase, which can promote the phosphorylation of tau protein at multiple sites. (Kitagishi et al., 2014).

Our Western blot experiments indicated that treatment with YZD promoted the expression of PI3K and p-Akt, and reduced the levels of p-GSK3β, proving that YZD can regulate the activated PI3K/AKT/GSK3β pathway. The expression of BACE1 and β-amyloid were also down-regulated significantly, providing experimental confirmation of predictions reached by using network pharmacology and molecular docking analyses. Based on these findings, we propose that YZD may protect against the progression of AD by inhibiting the expression of BACE1, thereby reducing the production of beta-amyloid. The other main action of the formulation is via the regulation of the PI3K/AKT/GSK-3β pathway.

Although this work shed light on some of the therapeutic mechanisms of YZD in AD, the results need to be interpreted with caution. Additional pharmacologic experiments will be necessary to validate further mechanisms of YZD predicted to occur in our network analysis. Second, network pharmacology fails to identify dose-effect relationships of multi-component formulations. These issues will be addressed in subsequent studies. Finally, further work is needed to study global metabolic changes in other species to fully elucidate the mechanisms by which YZD improves AD. Enhanced metabolomics technologies, such as targeted metabolomics, are required to detect metabolite biomarkers (Wei et al., 2020).

In summary, multiple compounds in YZD show promise in the potential treatment of AD (Figure 10). Our experiments showed that YZD could reduce the expression of BACE1 and the production of beta-amyloid and regulate the PI3K/AKT/GSK-3β pathway. These findings are consistent with the results of our network pharmacology and molecular docking-based predictions. Thus, we propose that YZD is likely to affect a multitude of molecular mechanisms in the treatment of AD.

Herbal medicines have provided valuable tools in preventing and treating AD for centuries. In this study, we have identified 27 chemical constituents of YZD. Using a network pharmacology approach combined with molecular docking analyses identified some of the mechanisms involved in the beneficial effects of YZD treatment in AD. These involve effects on neuroactive ligand-receptor interaction pathways, the PI3K/AKT signaling pathway, as well as HIF-1 and calcium signaling. The results showed that YZD has potential benefits on learning and memory deficits in APP/PS1 transgenic AD mice, and regulated the apoptotic process and signal transduction. The key proteins of related pathways were quantitatively analyzed using Western blots indicating that the extract of YZD could inhibit or activate key protein pathways. The insights provided by this preliminary work will pave the way for further studies into the potential mechanisms by which YZD acts as a therapeutic agent in AD.