Network Pharmacology of Ginseng (Part III): Antitumor Potential of a Fixed Combination of Red Ginseng and Red Sage as Determined by Transcriptomics

Background: This study aimed to assess the effect of a fixed combination of Red Ginseng and Red Sage (RG–RS) on the gene expression of neuronal cells to evaluate the potential impacts on cellular functions and predict its relevance in the treatment of stress and aging-related diseases and disorders. Methods: Gene expression profiling was conducted by transcriptome-wide mRNA microarray analyses of murine HT22 hippocampal cell culture after treatment with RG–RS preparation. Ingenuity pathway analysis (IPA) was performed with datasets of significantly upregulated or downregulated genes and the expected effects on the physiological and cellular function and the diseases were identified. Results: RG–RS deregulates 1028 genes associated with cancer and 139 with metastasis, suggesting a predicted decrease in tumorigenesis, the proliferation of tumor cells, tumor growth, metastasis, and an increase in apoptosis and autophagy by their effects on the various signaling and metabolic pathways, including the inhibition of Warburg’s aerobic glycolysis, estrogen-mediated S-phase entry signaling, osteoarthritis signaling, and the super-pathway of cholesterol biosynthesis. Conclusion: The results of this study provide evidence of the potential efficacy of the fixed combination of Red Ginseng (Panax ginseng C.A. Mey.) and Red Sage/Danshen (Salvia miltiorrhiza Bunge) in cancer. Further clinical and experimental studies are required to assess the efficacy and safety of RG–RS in preventing the progression of cancer, osteoarthritis, and other aging-related diseases.


Introduction
Various stresses/stressors of exogenous origin, such as environmental carcinogens, chemicals, and radiation, and others endogenously derived from the endoplasmic reticulum, such as hypoxia and oxidative stress, disrupt homeostasis and trigger the progression of many diseases, including cancer. With age, the accidence of cancer, cardiovascular, neurodegenerative, and some other diseases dramatically increases due to the systemic progression of aseptic chronic low-grade inflammation defined as inflammageing [1][2][3]. Inflammaging is neither harmful nor beneficial by itself and may be helpful in an adaptive manner or harmful in a maladaptive way, depending on environmental stressors and genetic background [1]. The human organism has developed a highly effective defense regulatory network, the neuroendocrine-immune complex, to adapt to these stresses. Adaptation to stress/stressors could finally lead to either beneficial or unfavorable consequences where the outcome may be aging-related diseases or healthy longevity [1].
Two well-known adaptogenic plants [4][5][6][7][8][9], Red Ginseng (Panax ginseng CA Mey) [10] and Red Sage/Danshen (Salvia miltiorrhiza Bunge) [11], are used in Traditional Chinese cells, followed by network pharmacology using Ingenuity Pathway Analysis (IPA) of significantly upregulated or downregulated genes, to disclose effects on cellular functions and diseases. That is important for understanding and predicting the potential risks and benefits of the therapeutic action of RG-RS.
The choice of murine neuronal HT22 cells as an in vitro model is based on many publications where a neuroprotective effect of Red Ginseng extracts [47][48][49], Ginsenosides Rg5, Rb2, Rg1, Re [50][51][52][53], Compound K [54,55] and tanshinons I and II [56,57] in various experiments with murine hippocampal HT22 cells was demonstrated. Furthermore, Red Ginseng root preparation HRG80TM (RG, used in this study) modulates ionotropic glutamate NMDA, and kainate receptors mediated transmission in ex vivo experiments of the excitability of hippocampal pyramidal cells of rats [58]. Finally, it was effective in preventing and mitigating the stress-induced deterioration of cognitive functions in healthy subjects [29] and elderly patients with mild cognitive disorders, substantially affecting various brain regions depending on the mental load during relaxation and cognitive tasks associated with memory, attention, and mental performance, suggesting an improvement in mood and calming effects associated with GABA-ergic neurotransmission [39]. Table 1 shows the number of genes deregulated (>20-fold compared to control) by RG-SG, RG, and ginsenosides Rb1, Rg3, Rg5, and Rk1 in the hippocampal neuronal cell line HT22; for details, see Supplemental Tables S1-S3 and Table 1. n/a n/a n/a n/a TT-total tanshinones; SA-salvianolic acid B; *-gene numbers excluding those deregulated by RG; **-calculated as tanshinone II (MW 294.3).

Effect of RG-RS on the Gene Expression Profile in the Hippocampal Neuronal Cell Line HT22
The combination RG-RS has a more substantial impact than RG on gene expression in hippocampal neurons, deregulating almost three-fold (1151/394) more genes, Figure 1. Therefore, we aimed in this study for the first time to uncover the mechanism of action (MOA) of RG-RS by assessing its effects on gene expression in isolated neuronal cells, followed by network pharmacology using Ingenuity Pathway Analysis (IPA) of significantly upregulated or downregulated genes, to disclose effects on cellular functions and diseases. That is important for understanding and predicting the potential risks and benefits of the therapeutic action of RG-RS.
The choice of murine neuronal HT22 cells as an in vitro model is based on many publications where a neuroprotective effect of Red Ginseng extracts [47][48][49], Ginsenosides Rg5, Rb2, Rg1, Re [50][51][52][53], Compound K [54,55] and tanshinons I and II [56,57] in various experiments with murine hippocampal HT22 cells was demonstrated. Furthermore, Red Ginseng root preparation HRG80TM (RG, used in this study) modulates ionotropic glutamate NMDA, and kainate receptors mediated transmission in ex vivo experiments of the excitability of hippocampal pyramidal cells of rats [58]. Finally, it was effective in preventing and mitigating the stress-induced deterioration of cognitive functions in healthy subjects [29] and elderly patients with mild cognitive disorders, substantially affecting various brain regions depending on the mental load during relaxation and cognitive tasks associated with memory, attention, and mental performance, suggesting an improvement in mood and calming effects associated with GABA-ergic neurotransmission [39]. Table 1 shows the number of genes deregulated (>20-fold compared to control) by RG-SG, RG, and ginsenosides Rb1, Rg3, Rg5, and Rk1 in the hippocampal neuronal cell line HT22; for details, see Supplemental Tables S1-S3 and Table 1. n/a n/a n/a n/a TT-total tanshinones; SA-salvianolic acid B; *-gene numbers excluding those deregulated by RG; **-calculated as tanshinone II (MW 294.3).

Effect of RG-RS on the Gene Expression Profile in the Hippocampal Neuronal Cell Line HT22
The combination RG-RS has a more substantial impact than RG on gene expression in hippocampal neurons, deregulating almost three-fold (1151/394) more genes, Figure 1.  The profiles (signatures) of deregulated genes by RG-RS and RS are quite different; only 24% of the genes (95 of 397) deregulated by RG were deregulated by the combination RG-RS Figure 1 and Supplemental Tables S1-S3).
These observations also show that the gene expression profile of hippocampal neurons is specific for the combination RG-RS and quite different from RG extract.  Table S4 show the predicted effects (−log p-value > 1.3, z-score > 2) of RG-RS on the canonical signaling pathways, including:

Effects of RG-RS on Signaling Canonical Pathways
• Inhibition of glycolysis, estrogen-mediated S-phase entry, PCP (planar cell polarity), mevalonate, osteoarthritis pathways, cholesterol, and geranylgeranyl diphosphate biosynthesis super-pathways, and • Activation of MIF regulation of innate immunity, regulation of cellular mechanics by calpain protease, hypoxia signaling in the cardiovascular system, and xenobiotic metabolism CAR signaling pathways.  The profiles (signatures) of deregulated genes by RG-RS and RS are quite different; only 24% of the genes (95 of 397) deregulated by RG were deregulated by the combination RG-RS Figure 1 and Supplemental Tables S1-S3).
These observations also show that the gene expression profile of hippocampal neurons is specific for the combination RG-RS and quite different from RG extract.  Table S4 show the predicted effects (−log p-value > 1.3, z-score > 2) of RG-RS on the canonical signaling pathways, including:

Effects of RG-RS on Signaling Canonical Pathways
• Inhibition of glycolysis, estrogen-mediated S-phase entry, PCP (planar cell polarity), mevalonate, osteoarthritis pathways, cholesterol, and geranylgeranyl diphosphate biosynthesis super-pathways, and • Activation of MIF regulation of innate immunity, regulation of cellular mechanics by calpain protease, hypoxia signaling in the cardiovascular system, and xenobiotic metabolism CAR signaling pathways. Fisher's Exact Test), indicating the statistical significance of the overlap of analyzed dataset genes within the pathway; the intensity of color corresponds to the log p-value in a scale from 0 to 2.2 (upper panel). The brown color shows the predicted activation, and the blue color shows the predicted inhibition of signaling pathways; the intensity of the color corresponds to the activation zone on a scale from −3.153 to +2.324 (upper panel). symbol · shows that the activation z-score was < 2 and the −log p-value <1.3 = p < 0.05. An absolute z-score ≥2 is considered significant activation (+) or inhibition (-). The activation z-score predicts the activation state of the canonical pathway using the gene expression patterns of the genes within the pathway.
A large part (about 89%) of deregulated genes consist of networks significantly associated with cancer (Table 1, Figure 3 and Supplementary Tables S5 and S6). The IPA analysis shows the predicted (activation score z > [±2], −log p-value > 1.3) inhibition of diseases, Figure 4 and Supplementary Table S6.  Tailed  Fisher's Exact Test), indicating the statistical significance of the overlap of analyzed dataset genes within the pathway; the intensity of color corresponds to the log p-value in a scale from 0 to 2.2 (upper panel). The brown color shows the predicted activation, and the blue color shows the predicted inhibition of signaling pathways; the intensity of the color corresponds to the activation zone on a scale from −3.153 to +2.324 (upper panel). symbol · shows that the activation z-score was < 2 and the −log p-value <1.3 = p < 0.05. An absolute z-score ≥2 is considered significant activation (+) or inhibition (-). The activation z-score predicts the activation state of the canonical pathway using the gene expression patterns of the genes within the pathway.
A large part (about 89%) of deregulated genes consist of networks significantly associated with cancer (Table 1, Figure 3 and Supplementary Tables S5 and S6). The IPA analysis shows the predicted (activation score z > [±2], −log p-value > 1.3) inhibition of diseases, Figure 4 and Supplementary Table S6. indicating the statistical significance of the overlap of analyzed dataset genes within the diseases; the intensity of color corresponds to the log p-value in a scale from 1.7 to 16.2 (upper panel). Disease scores are shown using a gradient from dark blue to brown for predicted activation and light to dark blue for predicted inhibition of diseases on a scale from −2.817 to +2.736 (upper panel); symbol · shows that the activation z-score was < 2, and the −log p-value <1.3 = p < 0.05. An absolute z-score ≥ 2 is considered significant activation (+) or inhibition (-).

Figure 3.
Predicted effects of RG-RS (SAL) at concentrations of 20 μg/mL, RG at concentrations of 10 μg/mL (HRG1), white ginseng (WG), and ginsenosides Rb1, Rg3, Rg5, Rk1 at a concentration of 0.1 μM. The violet color shows the p-value (calculated using a Right-Tailed Fisher's Exact Test), indicating the statistical significance of the overlap of analyzed dataset genes within the diseases; the intensity of color corresponds to the log p-value in a scale from 1.7 to 16.2 (upper panel). Disease scores are shown using a gradient from dark blue to brown for predicted activation and light to dark blue for predicted inhibition of diseases on a scale from −2.817 to +2.736 (upper panel); symbol · shows that the activation z-score was < 2, and the −log p-value <1.3 = p < 0.05. An absolute z-score ≥ 2 is considered significant activation (+) or inhibition (-). The largest part of deregulated genes (1011 genes, about 88%) consist of networks significantly associated with decreased tumorigenesis (Figure 4 and Supplementary Tables S7 and S8), 139 genes (12%) with reduced metastasis, 81 (7%) with increased autophagy, and 312 (27%) with increased apoptosis (Figure 5 and Supplementary Tables S7 and  S9).  The function scores are shown using a gradient from dark blue to brown for predicted activation and light to dark blue for predicted inhibition of diseases; the intensity of color corresponds to the activation zone on a scale from −2.817 to +2.736 (upper panel). symbol · shows that the activation z-score was < 2 and the −log p-value <1.3 = p < 0.05. An absolute z-score ≥2 is considered significant activation (+) or inhibition (-). The activation z-score predicts the activation state of the function using the gene expression patterns. The largest part of deregulated genes (1011 genes, about 88%) consist of networks significantly associated with decreased tumorigenesis (Figure 4 and Supplementary Tables S7 and S8), 139 genes (12%) with reduced metastasis, 81 (7%) with increased autophagy, and 312 (27%) with increased apoptosis ( Figure 5 and Supplementary Tables S7 and S9). In this study, we, for the first time, demonstrate the inhibitory effect of RG-RS on gene expression regulating glycolysis metabolic enzymes ( Figure 6 and Supplementary  Figures S2 and S3), which is crucial in cancer.

Discussion
This study's primary aim was to employ transcriptomics of neuronal cells to uncover the mechanisms of action (MOA) and potential pharmacological activities of a fixed combination RG-RS, consisting of two adaptogenic plants traditionally used in aging-related diseases and disorders. The MOA of both plants and many of their active constituents have been described in numerous publications. However, they might be quite different if both plants are combined in one formulation due to various synergistic and antagonistic interactions within the organism. Consequently, pharmacological and toxicological profiles might be considerably diverse in purified constituents. The present study confirms this common rule, as evident from Figure 1 and Table 1; e.g., RG-RS has to impact the expression of 302 genes specifically deregulated by RG. Furthermore, the gene expression profiles activated or inhibited the signaling pathways and predicted that the pharmacological effects are different (see Figures 2-4 and Supplementary Tables S1-S3).
To our best knowledge, only two clinical studies have been conducted with the RG-RS combination [40,41], suggesting its efficacy in the recovery of muscle soreness and preventing an increase in proinflammatory protein in exercising healthy subjects. Several in vitro and in vivo [42][43][44][45] studies conducted with a combination of purified active constituents of the plant extracts, i.e., salvianolic acids, ginsenosides, and ginseng polysaccharides, suggest the potential efficacy in preventing cancer progression and tumor metastasis [42][43][44][45], which are presumably associated with an inhibition of the PI3K/AKT signaling pathway [43]. From the chemical point of view, the substance RG-RS studied in our experiments differs from the substance used in any other study of RG, RS, and their constituents, except for two clinical studies in athletes. RG consists of nearly 300 compounds (more than 200 have been isolated from Red Ginseng and more than 100 from Red Sage), while OCF [42][43][44][45] has only 3 constituents-salvianolic acids, ginsenosides, and ginseng polysaccharides. Consequently, it is not surprising that the pharmacological profile of the combination RG-RS is unique and different from other substances observed in other studies. It is noteworthy that, in our study, we have observed an effect on the transcriptomic level of regulation of cellular response in vitro model; our experiments did not cover many other The results of this study, for the first time, describe the effect of RG-RS on the gene expression profile of isolated neurons, the potential impacts of RG-RS on diseases and disorders, biological processes, cellular and physiological function, and implications on intracellular signaling and metabolic pathways ( Figure 5).
In this study, we, for the first time, demonstrate the inhibitory effect of RG-RS on gene expression regulating glycolysis metabolic enzymes ( Figure 6 and Supplementary  Figures S2 and S3), which is crucial in cancer.
Most cancer cells are characterized by a high rate of aerobic glycolysis for the energy supply of proliferation, tumor growth, and survival, even with sufficient oxygen supply [59][60][61][62]. Highly proliferating cancer cells undergo the Warburg effect, i.e., a metabolic switch from oxidative phosphorylation providing 36 molecules of energy molecules (ATP) to an abnormal and ineffective energy supply metabolic pathway, aerobic glycolysis, providing only 4 ATP molecules and mostly lactate instead of pyruvate ( Figure 6). The glucose metabolic pathway is activated by several tumor-promoting (P3IK/AKT, HIF-1α, mTOR, K-Ras, c-Myc, miRNAs,) and inhibited by tumor-suppressing (AMPK, p53) signaling pathways in highly proliferative cancer cells [60,62].
Several transcription factors and signaling molecules, such as HIF-1α, c-Myc, Akt, and mTOR, and other upstream regulators, including oncogene K-Ras, tumor suppressor p53, energy sensor adenosine monophosphate-activated protein kinase (AMPK), noncoding RNAs, and sirtuin family proteins and deacetylation, are also involved in this process [59][60][61][62]. Decreased glycolytic flux in response to AMPK or p53 may be an adaptive response to shut off proliferative metabolism during periods of oxidative stress or low energy availability [59]. Figure 6 and Supplementary Table S10 show that RG-RS strongly downregulates the expression of five genes upregulated in cancer cells: 2562-fold downregulated neuronspecific enolase (NSE) ENO1 gene, 1426-fold downregulated phosphoglycerate kinase 1 PGK1 gene, 1094-fold downregulated pyruvate kinase PKM gene, 855-fold downregulated aldolase ALDOA gene, and 810-fold downregulated triose phosphate isomerase TPI gene. Moreover, RG-RS significantly inhibits the HIF-1α signaling pathway, a master regulator of glycolysis, and activates antitumor monophosphate-activated protein kinase (AMPK), suggesting the substantial inhibition of the conversion of glucose into lactate (Warburg effect) in cancer, which is crucial for the proliferation and survival of tumor cells. Figure 6 shows that RG-RS targets 10 (!) disease-specific molecules (5 directly and 5 indirectly via upstream regulators HIF-1α and AMPK) in the tumor's energy supply source.
Overall, RG-RS targets 1028 targets associated with cancer and 139 with metastasis, Table 1 and Supplementary Tables S8 and S9. In silico IPA analysis of the effects of RG-RS on gene expression in isolated neurons predicts the antitumor activity of RG-RS, which is in line with other studies [42][43][44][45]. Figures 2-4 show the potential effects of RG-RS on the canonical signaling pathways, cellular and physiological functions, and aging-related diseases, particularly the inhibition of the osteoarthritis signaling pathway where the probability of prediction is significant (−log p-value 1.7, z-score -3.2; Supplementary Table S6 and Supplementary Figure S1). That is in line with other studies of RG and RS where the potential health effect of Reg Ginseng [63][64][65][66][67] and Red Sage [68][69][70][71][72] in osteoarthritis was demonstrated.
A limitation in this study is related to the lack of scientific literature about the direction (negative or positive) of correlations between gene expression and physiological function or disease for predicting the effects of some experimental data used in silico analysis.  Table S6) the oncogenic Hypoxia-inducible factor-1 (HIF-1α) and activates (p > 1.3; z-1.7, Supplement Table S6) tumor suppressor adenosine monophosphate-activated protein kinase (AMPK) signaling pathways.
Most cancer cells are characterized by a high rate of aerobic glycolysis for the energy supply of proliferation, tumor growth, and survival, even with sufficient oxygen supply  Table S6) the oncogenic Hypoxia-inducible factor-1 (HIF-1α) and activates (p > 1.3; z-1.7, Supplement Table S6) tumor suppressor adenosine monophosphate-activated protein kinase (AMPK) signaling pathways.

Materials and Methods
All the materials and methods used in this study have been recently described in detail in our recent publication [24], which was conducted by the same experimental protocol [24]. They include mRNA microarray hybridization conducted by the same experimental protocol, Ingenuity Pathway Analysis (IPA), and Statistical Analysis [24].  Belgium). Korean Ginseng was steamed to Red Ginseng, which was dried, powdered, and standardized for the content of ginsenosides Rk1 and Rg5 to obtain Red Ginseng HRG80TM preparation containing 7.6% of total ginsenosides. HRG80TM preparation was exhaustively extracted by 70% methanol and evaporated to dryness to obtain HRG extract (DER 4: 1) containing 1.9% Rg5 and 1.0% Rk1; the content of total ginsenosides in the HRG extract was 30.32%.
HPLC conditions. Analytical samples were analyzed by Waters LC system using reverse phase HPLC column ACE C18 5 µm (250 × 4.6 mm) and the solvent system containing gradually increasing concentration (from 21 to 90% in eight steps) of acetonitrile in water solution of 0.1% ortho-phosphoric acid at 30 • C, flow rate-1.2 mL/min.; diode array detection at 203 nm. The content of analytical markers in the analytes was quantified by reference standards used to generate calibration curves. The analytical method was validated for selectivity, precision, and accuracy (RSD < 5%).

mRNA Microarray Hybridization
Murine hippocampal neuronal HT22 cells were seeded and attached for 24 h before drug treatment. Cells were then treated for 24 h at final concentrations of 10 µg/mL of RG and 10 µg/mL RS in RG-RS, 10 µg/mL of RG, 0.1 µM of ginsenosides Rb1, Rg3, Rg5, and Rk1 (Table 1) or DMSO as solvent control (0.5%). The total RNA Nanochip assay was applied on an Agilent 2100 bioanalyzer (Agilent Technologies GmbH) for quality control of RNA. An RNA index value > 8.5 served as a threshold for further processing of RNA samples. Every two independent experiments have been performed. The mRNA microarray hybridization has been performed at the Genomics and Proteomics Core Facility, German Cancer Research Center (Heidelberg, Germany), using Affymetrix GeneChips ® with mouse Clariom S assays. A quantile normalization algorithm has normalized the measurements without background subtraction. The standard deviation differences were calculated in one-by-one comparisons to identify differentially regulated genes. The Chipster software (The Finnish IT Center for Science CSC) was used for further evaluation of the results.

Ingenuity Pathway Analysis (IPA)
The interpretation of microarray data and gene expression changes was performed using Ingenuity Pathways Analysis (IPA) software, summer release 2021 (QIAGEN Bioinformatics, Aarhus C, Denmark), which relies on the Ingenuity Knowledge Base, a continuously updated database gathering observations with more than 8.1 million findings manually curated from the biomedical literature or integrated from 45 third-party databases. The IPA network contains 40,000 nodes representing mammalian genes, molecules, and biological functions, linked by 1,480,000 edges representing experimentally observed cause-effect relationships (either activating or inhibiting) related to gene expression, transcription, activation, molecular metabolism, and binding. Network edges are also associated with a direction (either activating or inhibiting) of the causal effect [73].
Using the IPA Core Analysis tool for all tested transcriptomic datasets, we performed predictive analyses of the impact of test samples on canonical signaling and metabolic pathways, which displayed the molecules of interest within well-established pathways; and diseases, disorders, molecular and cellular functions that are activated or inhibited downstream and upstream of the genes, whose expression has been altered.

Statistical Analysis
Two statistical methods of analysis of gene expression datasets were used in the IPA: (i) the gene-set-enrichment method, where differentially expressed genes are intersected with sets of genes that are associated with a particular biological function or pathway providing an "enrichment" score (Fisher's exact test p-value) that measures the overlap of the observed and predicted regulated gene sets [74,75], (ii) the method based on cause-effect relationships related to the direction of effects reported in the literature [76,77] that provides the so-called z-score measuring the match of observed and predicted up/down-regulation [73]. The predicted effects are based on gene expression changes in the experimental samples relative to the control; z-score > 2, −log p-value > 1.3.

Conclusions
The results of this study present evidence of the potential efficacy of the fixed combination of Red Ginseng (Panax ginseng C.A. Mey.) and Red Sage/Danshen (Salvia miltiorrhiza Bunge) in cancer. Further preclinical and clinical studies are warranted to assess the efficacy and safety of RG-RS in preventing the progression of cancer and other agingrelated diseases.