One hundred-gram pieces of HQ were weighed. Preparation of HQ water extract (1 g/mL) were performed as previously described (Shi et al., 2015; Lou et al., 2019). Chemical profiling and standardization of HQ water extract using FMNT and CS was performed via UHPLC-MS/MS analysis by using an Agilent 1290 infinity LC system (Supplementary materials and methods). The content of FMNT and CS were quantitated in HQ water extract at levels of 0.3338 mg/g and 0.1998 mg/g, respectively (Figures 2A, B).

Several databases, including TCMSP (http://tcmspw.com/) (Tan et al., 2022), SuperPred (http://prediction.charite.de/) (An et al., 2020), SwissTargetPrediction (http://www.swisstargetprediction.ch/) (Tan et al., 2022), PubChem (https://pubchem.ncbi.nlm.nih.gov/) (An et al., 2020) and STITCH (http://stitch.embl.de/) (An et al., 2020) databases, which limit species for “Homo sapiens”, were used to collect the potential targets of FMNT and CS. Finally, the UniProt database (https://www.uniprot.org/) was used to identify all the targets (An et al., 2020).

Keywords of “colon cancer”, “colon adenocarcinoma” and “colorectal cancer” were used to identify targets related to colon cancer from GeneCards (https://www.genecards.org/) (Tan et al., 2022), OMIM (https://www.omim.org/) (Amberger et al., 2015), TTD (http://db.idrblab.net/ttd/) (Tan et al., 2022), DisGeNET (https://www.disgenet.org/) (Tan et al., 2022) and CTD (http://ctdbase.org) (Tan et al., 2022) databases. Finally, the UniProt database (https://www.uniprot.org/) was used to identify all the targets (An et al., 2020).

Venn diagram (http://bioinformatics.psb.ugent.be/webtools/Venn/) was used to collect the common targets of FMNT and CS and colon cancer. The detectable targets in the protein-protein interaction (PPI) of FMNT and CS against colon cancer were constructed by the STRING database (https://string-db.org/) and were further visualized by using Cytoscape software (An et al., 2020; Xu et al., 2022). Crucial biotargets of FMNT and CS against colon cancer were obtained according to the degree values in the Cytoscape settings.

The Omicshare Online tools (https://www.omicshare.com/tools), a free online platform for data analysis, was used to understand the role of the potential targets in gene functions and signaling pathways, including the Gene Ontology (GO) and Kyoto Encyclopedia of Genes and Genomes (KEGG) pathway enrichment analysis (An et al., 2020). GO (https://geneontology.org/) is one of the main resources of biological information, which provides a specific definition of protein functions. GO is subdivided into three non-overlapping ontologies: biological process (BP), cellular component (CC), and molecular function (MF) terms (Ashburner et al., 2000). Those terms with a p < 0.05 were chosen for further analysis and the top 20 of GO terms enrichment were displayed. KEGG (https://www.kegg.jp/) is a database resource for the biological interpretation by genome sequencing and other high-throughput experimental technologies (Chen et al., 2015a; An et al., 2020). Those terms with an p < 0.05 were chosen for further analysis and the top 25 enriched KEGG pathways were mapped.

Astragalus membranaceus (Fisch.) Bunge was purchased from Daxiang Chinese Pharmaceutical Co., Ltd. (Guangdong, China). FMNT and CS (purity ≥98%) were purchased from Chengdu Must Biotechnology Co., Ltd. (Chengdu, China). MTT were bought from Solarbio Science and Technology Co., Ltd. (Beijing, China). BeyoClick™ EdU cell proliferation kitwas bought from Beyotime Biotechnology (Haimen, China). Annexin V-fluorescein isothiocyanate (FITC)/propidium iodide (PI) apoptosis detection kit was purchased from Absin Bioscience Inc. (Shanghai, China). DMSO and HEPES were purchased from GBCBIO Technologies (Guangzhou, China). p-Akt (Ser473); Akt, p-ERK1/2 (Thr202/Tyr204), ERK1/2, and β-actin antibodies were purchased from Cell Signaling Technology (Danvers, MA, United States). N-cadherin, E-cadherin and Vimentin antibodies were purchased from Affinity Biosciences (OH, United States). All other chemicals not mentioned here were of analytical reagent grade or better.

HT29 and Caco2 human colon carcinoma cells were purchased from American Type Culture Collection (ATCC, Manassas, VA, United States) and then routinely cultured in DMEM medium (Gibco, United States) containing 10% (v/v) fetal bovine serum (FBS) (Gibco, United States) and 100 U/mL streptomycin-penicillin solution (Gibco, United States) in a humidified atmosphere containing 5% CO₂ at 37°C.

Cell viability was measured by MTT assay. HT29 and Caco2 cells were seeded in 96-well plates and cultured with the vehicle, HQ (3.125–100 mg/mL), FMNT (3.125–200 μM) or CS (3.125–200 μM) for 48 h. 5-Fluorouracil (5-FU) (25 μM) was used as a positive control. Then, the medium was removed and each well was added 200 μL MTT solution (0.5 mg/mL) and incubated at 37°C for additional 4 h. After that, supernatant was discarded and 150 μl DMSO was added to each well to dissolve formazan crystals. Optical density (OD) values were measured at 490 nm by a microplate reader (PerkinElmer, Waltham, MA, United States). Finally, cell inhibitory rates and 50% inhibitory concentration (IC₅₀) values were calculated for each group.

Cell proliferation ability was measured by EdU assay following the manufacturer’s instructions. HT29 and Caco2 cells were seeded in 96-well plates and cultured with the vehicle, HQ (25, 50, and 100 mg/mL), FMNT (25, 50, and 100 μM) or CS (25, 50, and 100 μM) for 48 h. 5-FU (25 μM) was used as a positive control. Then, cells were incubated with EdU labeling medium (50 mM) at 37°C for 2 h. Other details of the procedure are described in literature (Wu et al., 2020). Finally, the images were captured using a fluorescence microscope (Leica, Germany) and analyzed by ImageJ software.

HT29 and Caco2 cells were seeded in 6-well plates. Scratch lines were drawn on the monolith of the fused cells using 10 μL pipette tips. Then, the cells were cultured with the vehicle, HQ (25, 50, and 100 mg/mL), FMNT (25, 50, and 100 μM) or CS (25, 50, and 100 μM) for 12, 24 or 48 h, respectively. Pictures of the scratch lines were captured using a microscope (Leica, Germany) at 0, 12, 24 and 48 h, respectively. The cell migration ratio of linear wounds was calculated by using ImageJ software.

HT29 and Caco2 cells were stained with Annexin V-FITC/PI and evaluated for apoptosis by flow cytometry according to the manufacturer’s instructions. Briefly, HT29 and Caco2 cells were seeded in 6-well plates and exposed to the vehicle, HQ (25, 50, and 100 mg/mL), FMNT (25, 50, and 100 μM) or CS (25, 50, and 100 μM) for 48 h. Then, cells were collected and stained with Annexin V-FITC and PI for 20 min at room temperature in dark. Finally, cell apoptosis was analyzed by the flow cytometry (BD Biosciences, San Diego, CA, United States).

HT29 and Caco2 cells were seeded in 6-well plates and exposed to the vehicle, HQ (25, 50, and 100 mg/mL), FMNT (25, 50, and 100 μM) or CS (25, 50, and 100 μM) for 48 h. After that, the cells were washed with cold PBS and harvested. Total proteins were extracted from cells using the RIPA lysis reagent containing a protease inhibitor and then quantified using a BCA estimation kit. Western blotting was performed as previously described (An et al., 2020) using primary antibodies against p-Akt, Akt, p-ERK1/2, ERK1/2, N-cadherin, E-cadherin, Vimentin or β-actin (1:1000). Protein bands were detected by a gel imaging analysis system and were analyzed using ImageJ software.

HT29 and Caco2 cells were seeded on confocal dishes and exposed to the vehicle, HQ (100 mg/mL), FMNT (100 μM) or CS (100 μM) for 48 h. At the end of the incubation, the cells were fixed in paraformaldehyde, permeabilized with TritonX-100 and blocked with bovine serum albumin. Then, the cells were incubated with a N-cadherin, an E-cadherin, or a Vimentin (1:200) antibody at 4°C overnight and then stained with a secondary fluorescent antibody (1:200; Alexa Fluor 488). Finally, the cells were incubated with DAPI for 20 min. Fluorescence signals were detected using a Leica TCS SP8 confocal fluorescence microscope (Leica, Germany). The relative fluorescence of N-cadherin, E-cadherin and Vimentin were analyzed by ImageJ software.

Molecular docking was carried out by AutoDock Vina software to measure the interactions between the two compounds and the key upstream molecules of ERK1/2 signaling pathway, including EGFR, BRAF, MEK1/2 and ERK1/2. The crystal structures of EGFR, BRAF, MEK1/2 and ERK1/2 were retrieved from the Protein Data Bank (Table 1). The mol2 file format structures of the FMNT and CS were obtained from the PubChem database. Details of the procedure are described in literature (Ruan et al., 2020). After docking, conformations with the lowest docking energy were selected for the binding conformations between the ligands (FMNT and CS) and the target proteins. Finally, the analysis and visualization of the ligand-protein complexes were performed by PyMOL software.

All assays were performed in triplicate. Results were expressed as mean with standard deviations (mean ± SD). One-way analysis of variance (ANOVA) was performed to compare the differences between different groups by SPSS 19.0. Correlation analyses were performed using Pearson product–moment correlation by SPSS 19.0. Statistical significance of differences was accepted at p < 0.05, p < 0.01 and p < 0.001.

A total of 263 potential targets for FMNT and CS were identified after screening the databases of TCMSP, SuperPred, SwissTargetPrediction, STITCH and PubChem, (Figure 2C). Through the GeneCards, OMIM, TTD, CTD and DisGeNET databases, a total of 1866 colon cancer targets were obtained (Figure 2C).

Then, compound targets and colon cancer-related targets were mapped using a Venn diagram, 147 common targets were obtained as the potential targets of FMNT and CS against colon cancer (Figure 2C). The PPI network of the 147 targets were established by Cytoscape 3.7.2 software (An et al., 2020) (Figure 2D). There were 147 nodes and 2,382 edges in the PPI network with a medium degree centrality (DC) value of 28 (Figure 2D). Afterwards, hub nodes for FMNT and CS against colon cancer were established based on DC values. Finally, a total of 19 hub nodes were screened when the thresholds were set at DC ≥ 57. These 19 crucial targets with higher DC values were identified as the candidate targets of FMNT and CS against colon cancer, including AKT1, TP53, MAPK3, EGFR, HSP90AA1, ESR1, MYC, CASP3, TNF, JUN, PTEN, MAPK1, FOS, RELA, FN1, PTGS2, HSP90AB1, MAPK8 and PPARG (Figure 2D).

To evaluate the pharmacological effects of FMNT and CS against colon cancer. The potential 147 targets were further utilized for the GO analysis and KEGG pathway enrichment analysis. GO analysis showed that the majority of these 147 targets were enriched in cells with molecular function of protein binding. Specifically, these targets were abundantly enriched in cellular process, metabolic process, biological regulation and response to stimuli (Figure 3A). These results implicated that various cellular functions could be involved in the diverse and synergistic effects of FMNT and CS against colon cancer. The top 20 of GO terms enrichment were mapped (Figures 3B, C). The results showed that targets were closely related to multiple biological processes (BPs), including cellular response to chemical stimuli, response to chemical, response to oxygen-containing compounds, response to organic substances, regulation of programmed cell death, regulation of cell death, regulation of apoptotic process, and apoptotic process, etc (Figure 3B). In molecular functions (MFs), the significant terms were mainly enriched in enzyme binding, kinase binding, protein kinase binding, ubiquitin-like protein ligase binding, protein domain specific binding, identical protein binding, ubiquitin protein ligase binding, and organic cyclic compound binding, etc (Figure 3C). Moreover, cellular components (CCs) analysis showed that the significant terms mainly included cytoplasmic part, membrane-bounded organelle, mitochondrion, intracellular organelle lumen, organelle lumen, membrane-enclosed lumen, cytoplasm and nucleoplasm, etc (Figure 3D).

To further elucidate the potential mechanisms of FMNT and CS against colon cancer, KEGG pathway enrichment analysis of those 147 key targets was conducted. KEGG pathway annotation showed that 91 and 89 targets were involved in the signal transduction and cancer, respectively (Figure 4A). The top 20 KEGG pathway enrichments with a p-value <0.05 were displayed in Figure 4B. The significant pathways mainly included hepatitis C, pathways in cancer, hepatitis B, MAPK signaling pathway, prostate cancer, PI3K-Akt signaling pathway, toxoplasmosis and apoptosis, etc. It further showed that 42 and 46 targets were involved in the MAPK (ko04010) and PI3K-Akt (ko04151) signaling pathways, respectively (Figure 4C). The MAPK signaling pathway is essential in regulating many cellular processes including cell differentiation, cell stress response, cell division, metabolism, cell proliferation, motility and apoptosis. The role of the MAPK pathway in various diseases, including cancer, has been widely studied and well recognized (Haigis et al., 2008; Ahearn et al., 2011). As shown in Supplementary Figure 1, FMNT and CS could modulate multiple signaling molecules and pathways to suppress cancer proliferation, promote apoptosis, and inhibit angiogenesis. The key targets of the MAPK signaling pathway mainly include MAPK1, MAPK3, EGFR, AKT1, TP53, MYC, FOS, and MAPK8 (Supplementary Figure S1A). The PI3K-Akt signaling pathway is another crucial intracellular signal transduction pathway that closely regulate cell growth, migration, metabolism, proliferation and angiogenesis [28]. The PI3K-Akt is also a crucial signaling pathway contributing to initiation and development of various cancers (Noorolyai et al., 2019; Pompura and Dominguez-Villar, 2018). The key targets of the PI3K-Akt signaling pathway mainly include AKT1, HSP90AA1, HSP90AB1, TP53, PTEN, MYC, EGFR, and RELA (Supplementary Figure S1B). Hence, the potential mechanisms associated with the MAPK and PI3K-Akt signaling pathways were further verified using experimental validation coupled with molecular docking.

MTT assay was performed to detect the effects of HQ, FMNT and CS on the viability of HT29 and Caco2 cells. It was found that treatment with HQ at 3.125–100 mg/mL for 48 h significantly suppressed the viability of HT29 and Caco2 cells in a dose-dependent manner (Figures 5A, B, p < 0.001). The IC50 value of HQ in HT29 and Caco2 cells after treatment was 26.67 and 25.31 mg/mL, respectively. After treating cells with FMNT and CS at 3.125–200 μM for 48 h, cell viabilities were also significantly suppressed in a dose-dependent manner (Figures 5C–F, p < 0.05 or p < 0.001). 5-FU, which was used a positive control, also produced significant inhibition toward the viability of HT29 and Caco2 cells as expected (p < 0.001). Accordingly, three concentrations of 25, 50, and 100 mg/mL for HQ, and three concentrations of 25, 50, and 100 μM for FMNT or CS, and a duration of 48 h treatment were selected to conduct the following experiments.

EdU assay was used to further evaluate the effects of HQ, FMNT and CS on the proliferation of HT29 and Caco2 cells. As shown in Figures 6A, B, treatment with HQ at 25, 50, and 100 mg/mL for 48 h significantly decreased EdU positive cells in a dose-dependent manner over the control (p < 0.05 or p < 0.001), indicating that HQ could significantly suppressed the proliferation of HT29 and Caco2 cells. The number of EdU positive cells in the FMNT or CS treated groups were also significantly reduced in a dose-dependent manner (Figures 6C–F, p < 0.05, p < 0.01 or p < 0.001), indicating that both FMNT and CS significantly suppressed the proliferation of HT29 and Caco2 cells. 5-FU, as a positive control, markedly inhibited the proliferation of HT29 and Caco2 cells as expected (p < 0.001).

An Annexin V-FITC/PI staining assay was performed to evaluate whether HQ, FMNT and CS could induce apoptosis of HT29 and Caco2 cells. Treatment with HQ at 25, 50, and 100 mg/mL for 48 h significantly induced the apoptosis rate of HT29 and Caco2 cells in a dose-dependent manner (Figure 7A, p < 0.001). Apoptosis rate was also dose-dependently increased in HT29 and Caco2 cells exposed to FMNT (Figure 7B, p < 0.01 or p < 0.001) or CS (Figure 7C, p < 0.001) at 10, 20, and 40 μM for 48 h.

Wound healing assay was used to evaluate the effects of HQ, FMNT and CS on the migration of HT29 and Caco2 cells. As shown in Figures 8A, B, treatment with HQ at 25, 50, and 100 mg/mL for 12–48 h significantly suppressed the migration ratio of HT29 and Caco2 cells dose-dependently over the control (p < 0.05, p < 0.01 or p < 0.001). Incubation of FMNT and CS at 10, 20, and 40 μM also exhibited significant inhibitory effects on the migration of HT29 and Caco2 cells (Figures 8C–F, p < 0.05, p < 0.01 or p < 0.001).

E-cadherin, N-cadherin, and Vimentin are critical epithelial–mesenchymal transformation (EMT)-related proteins, which that modulate cell migration and tumor invasiveness. E-cadherin is notably downregulated while N-cadherin and Vimentin are upregulated during EMT [30]. Hence, the intracellular protein levels of E-cadherin, N-cadherin, and Vimentin were further determined. As shown in Figure 9A, exposure to HQ at 25, 50, and 100 mg/mL for 48 h significantly upregulated the E-cadherin protein levels, while dose-dependently downregulated the protein levels of N-cadherin and Vimentin over the control (p < 0.05, p < 0.01, or p < 0.001). It was observed that incubation of FMNT and CS at 10, 20, and 40 μM for 48 h also strikingly increased the E-cadherin protein levels, while decreased the protein levels of N-cadherin and Vimentin in a dose-dependent manner (Figures 9B, C, p < 0.05, p < 0.01, or p < 0.001). Moreover, the results of immunofluorescence assay further revealed the expression of intracellular E-cadherin, N-cadherin, and Vimentin with the same treatment. Of note, the fluorescence intensity of E-cadherin was markedly enhanced, while the fluorescence intensity of N-cadherin and Vimentin was notably decreased in the cells exposed to HQ, FMNT and CS over the control cells (Figure 10).

Western blot analysis was further performed to understand the roles of ERK1/2 and PI3K-Akt signaling in mediating the inhibitory effects of HQ, FMNT and CS against colon cancer. Compared with the control cells, exposure to HQ at 25, 50, and 100 mg/mL for 48 h strikingly downregulated the p-ERK1/2 expression in a dose-dependent manner (Figure 11A, p < 0.01, or p < 0.001), while did not notably alter the total ERK1/2 expression. Thus, p-ERK1/2/ERK1/2 ratio was significantly decreased in the cells exposed to HQ than that in the control cells (Figure 11A, p < 0.01, or p < 0.001). The treatment of HQ at high dose also slightly downregulated the p-AKT expression without marked change in total Akt expression. Similarly, Western blot analysis demonstrated a decreased p-ERK1/2 expression in the cells exposed to FMNT or CS (Figures 11B, C, p < 0.05 or p < 0.001). Thus, FMNT and CS treatment significantly decreased the p-ERK1/2/ERK1/2 ratio compared with the control cells (Figures 11B, C, p < 0.01, or p < 0.001). The same treatment of FMNT and CS at their high doses also slightly downregulated the p-Akt protein levels, while did not notably alter the total Akt expression (Figures 11B, C).

Person correlation analysis showed that the inactivation of HQ, FMNT and CS on ERK1/2 signaling pathway was closely related to their inhibitory effects against colon cancer. It was observed that HQ-induced decrease in ratio of p-ERK1/2/ERK1/2 was positively related to inhibition of cell viability, proliferation, and migration, while were negatively related to promotion of the cell apoptosis (Supplementary Figure S2, p < 0.001). Similarity, a strong positive correlation was found between the decrease in p-ERK1/2/ERK1/2 ratio and inhibition of cell viability, proliferation, and migration by FMNT and CS. Conversely, a significant negative correlation was observed between the decreased p-ERK1/2/ERK1/2 ratio and promotion of the cell apoptosis by FMNT and CS (Supplementary Figures S3, S4, p < 0.001).

Overactivation of the RAS/RAF/MEK/ERK signaling pathway is associated with the carcinogenesis and progression of numerous cancers (Zhang et al., 2019; Zhu et al., 2020). Epidermal growth factor receptor (EGFR), a typical receptor tyrosine kinase, is activated by binding with ligands and takes part in cell differentiation, proliferation, migration, and apoptosis via activating the subsequent intracellular signaling pathways (Zhang et al., 2019). RAS/RAF/MEK/ERK is one of the most key EGFR downstream signaling pathways (Zhang et al., 2019). Thus, EGFR/RAS/RAF/MEK/ERK pathway is an essential target in clinical cancer therapy. Development of inhibitors of this pathway is considered as promising strategy for the prevention and treatment of numerous cancers (Supplementary Figure S5). Hence, molecular docking study was further conducted to explore the binding affinity of FMNT and CS with EGFR/RAS/RAF/MEK/ERK pathway. The pattern diagram of molecular docking was shown in Figure 12. Protein residues Glu738 and Asp831 in EGFR formed hydrogen bonds with FMNT and CS, respectively (Figures 12A, B). FMNT could bind to the protein residue Lys483, forming one single hydrogen bond (Figure 12C), while CS could bind to the protein residues Cys532 and Lys483 in BRAF through hydrogen bonds (Figure 12D). Protein residues Met146 and Gly144 in MEK1 formed hydrogen bonds with FMNT (Figure 12E), while only residue Met146 in MEK1 and CS bonded to each other through one single hydrogen bond (Figure 12F). FMNT could interact with the residue Met150 (Figure 12G), while CS could interact with the residues Met150 and Asn199 in MEK2 through hydrogen bonds (Figure 12H). Both FMNT and CS could interact with the residue Met125 in ERK1through hydrogen bonds (Figures 12I, J). CS could also interact with the residue Asp184 in ERK1 (Figure 12J). Protein residues Lys52 and Glu107 in ERK2 formed one single hydrogen bond with FMNT and CS, respectively (Figures 12K, L). The docking scores for FMNT and CS with the six crystal structures were showed in Table 1. The bond strength of hydrogen bonds represents the binding affinity between ligand compounds and receptor molecules, and when the docking score is lower, the binding affinity is stronger. An affinity <−7 indicates strong binding activity (Trott and Olson, 2010). Thus, the results revealed that FMNT and CS have a strong binding affinity for the key targets of EGFR/RAS/RAF/MEK/ERK pathway, and the binding affinity of CS with the targets was stronger than that of FMNT to the same targets.