Anti-Hypoxic Phytochemicals in Gao-Shan-Hong-Jing-Tian Oral Liquid: LC-MS Profiling, Network Pharmacology, and Carbonic Anhydrase Inhibition

Abstract

Gao-shan-hong-jing-tian (GSHJT) Oral Liquid is a phytochemical-rich preparation derived from Rhodiola, yet its anti-hypoxic active constituents and molecular mechanisms remain poorly understood. This study aimed to identify the key anti-hypoxic phytochemicals in GSHJT Oral Liquid and clarify their mechanisms of action to support its potential use in managing acute mountain sickness (AMS). We first established and validated an HPLC method for quality control, then comprehensively profiled the chemical composition using LC-MS. Network pharmacology and molecular docking were applied to predict the core anti-hypoxic components, candidate targets and signaling pathways. The primary bioactivity was further verified through an in vitro carbonic anhydrase (CA) inhibition assay. A total of 71 constituents were identified, with kaempferol and ellagic acid emerging as the primary anti-hypoxic phytochemicals. These compounds target seven core proteins (SRC, PIK3R1, ESR1, EGFR, PTK2, IGF1R, and LYN) to regulate vascular tone, inflammation, oxidative stress, blood–brain barrier integrity, and cell survival under hypoxic conditions. By modulating pathways such as HIF-1α, PI3K/AKT, FAK/PTK2, SRC, and IGF1R, these phytochemicals ultimately influence the onset and alleviation of AMS. Enzyme inhibition assays demonstrated that kaempferol and ellagic acid inhibited CA with IC₅₀ values of 34.05 μM and 119.1 μM, respectively. Molecular docking further revealed that both compounds suppressed CA activity through a combination of hydrogen bonding and hydrophobic interactions, consistent with a zinc-bound water-anchoring mechanism. This study elucidates the phytochemical basis and molecular mechanism responsible for the anti-hypoxic effects of GSHJT Oral Liquid, providing scientific support for its potential application as a natural, plant-derived intervention for preventing and alleviating acute mountain sickness, providing scientific support for its potential application and offering a reproducible paradigm for the rational development of other Rhodiola-based phytomedicines, though further in vivo validation is required to confirm the anti-hypoxic efficacy.

1. Introduction

Acute mountain sickness (AMS) is a common stress-related disorder induced by hypobaric hypoxia at high altitudes. It typically presents with headache, fatigue, dyspnea, sleep disturbance, and gastrointestinal discomfort, which severely threaten the health, safety, and performance of high-altitude workers, travelers, and military personnel [1]. Therefore, developing safe, effective, and natural anti-hypoxic and anti-AMS agents, and uncovering their molecular mechanisms, is of great clinical significance and practical value. Hypoxia refers to a physiological state in which oxygen supply is insufficient or oxygen consumption is excessive. It is generally divided into acute and chronic forms, both of which can lead to neurological dysfunction. Acute hypoxia often causes reversible cognitive impairment, whereas chronic hypoxia may result in long-term deficits, even dementia, due to distinct underlying molecular pathways [2]. Pathological studies have firmly established hypoxia as the central driver of AMS development. Accordingly, enhancing hypoxic tolerance has become a key strategy for relieving AMS symptoms [3,4,5].

Conventional drugs for AMS, such as acetazolamide and dexamethasone, are widely used but are associated with notable side effects including metabolic disorders, electrolyte imbalance, and immunosuppression [6]. This has spurred growing interest in plant-derived phytochemicals that offer favorable safety and multi-target biological activities. The genus Rhodiola is a well-recognized traditional medicinal herb widely distributed in high-altitude regions. Its main bioactive constituents include salidroside, tyrosol, gallic acid, and various flavonoids and phenolic acids [7]. Rhodiola species have been validated for preventing and alleviating AMS [8]. Modern pharmacological studies have confirmed that Rhodiola extracts exhibit a broad spectrum of activities, including anti-hypoxic, antioxidant, anti-inflammatory, immunomodulatory, antidiabetic, antihypertensive, anti-fatigue, and anticancer effects [9,10]. Gao-shan-hong-jing-tian (GSHJT) Oral Liquid utilizes Rhodiola as its sole source of pharmacologically active ingredients. The manufacturing process involves a sequential extraction of the raw botanical material using 60% ethanol followed by water, with sucrose and potassium sorbate subsequently introduced as excipients. It functions to replenish qi, nourish yin, and promote blood circulation, and is clinically used for fatigue, lassitude, insomnia, amnesia, dizziness, and vertigo caused by qi-yin deficiency and blood stasis. Although GSHJT Oral Liquid is not officially indicated for AMS, its raw material Rhodiola has been proven effective against AMS [8]. We therefore hypothesized that the anti-hypoxic activity underlying action of Rhodiola also contributes to the effects of GSHJT Oral Liquid, supporting its potential repurposing for AMS. However, this hypothesis relies on the assumption that the manufacturing process, specifically water extraction followed by ethanol precipitation, adequately preserves the key anti-AMS phytochemicals present in the raw herb. In addition, current clinical evidence regarding the anti-AMS efficacy of Rhodiola remains inconclusive. Considering this ambiguity and controversy in the clinical evidence base, further mechanistic exploration is essential to provide fundamental experimental evidence for its potential anti-hypoxic and anti-AMS activity, which represents the primary objective of this study.

Previous LC-MS-based phytochemical profiling studies have been conducted on Rhodiola raw materials and related formulations, identifying key components such as salidroside, tyrosol, and kaempferol [11,12]. Building on this, comprehensive chemical profiling using LC-MS has become an essential first step for systematically elucidating the material basis of herbal medicines [13]. Furthermore, recent studies have demonstrated the power of integrating LC-MS-based metabolomics with network pharmacology to uncover bioactive constituents and their mechanisms of action. Network pharmacology is an integrative discipline that combines computational science, bioinformatics, and systems biology to construct component–target–pathway networks. This approach enables high-throughput screening of bioactive phytochemicals, systematic dissection of multi-component and multi-target mechanisms, and mechanistic elucidation of complex herbal medicines [14,15]. Molecular docking is a structure-based computational tool widely used in drug discovery to predict binding modes, affinities, and molecular interactions between ligands and target proteins, facilitating target validation and mechanistic exploration [16]. Carbonic anhydrase (CA, EC 4.2.1.1) is a zinc-containing metalloenzyme that catalyzes the reversible hydration of carbon dioxide to bicarbonate and protons, a reaction critical for maintaining acid–base balance [17,18]. Human CA inhibitors have long been used clinically as diuretics and for treating glaucoma, epilepsy, inflammation, and altitude sickness [19,20]. CA inhibitors are classified by their binding modes: (1) zinc-binding inhibitors (e.g., acetazolamide) that directly chelate the active-site Zn²⁺; (2) zinc-bound water-anchoring inhibitors (e.g., flavonoids and polyphenols) that interact with Zn²⁺-coordinated water and Thr199 via hydrogen bonds; and (3) active-site entrance blockers (e.g., coumarins) that act as prodrugs and block substrate access [21,22,23,24]. Inhibition of CA is a well-established approach to improve hypoxic tolerance and mitigate AMS.

In this study, we established an HPLC fingerprint for quality control and used LC-MS to comprehensively profile the chemical composition of GSHJT Oral Liquid. Using network pharmacology, we identified core anti-hypoxic phytochemicals and their related targets and pathways. Molecular docking was used for computational validation, and CA inhibition assays were performed for experimental confirmation. This work systematically clarifies the active components and molecular basis of the anti-hypoxic effects of GSHJT Oral Liquid, providing a scientific foundation for its application in preventing and treating acute mountain sickness, as well as for the rational development of other Rhodiola-based medicines.

2. Materials and Methods

2.1. Chemicals and Reagents

GSHJT Oral Liquid was obtained from Hangzhou Huawei Pharmaceutical Co., Ltd. (Hangzhou, China). Reference standards, salicylic acid, gallic acid, salidroside, tyrosol, p-coumaric acid, ellagic acid, and kaempferol, were purchased from Desite Co., Ltd. (Chengdu, China). 4-Morpholinepropanesulfonic acid (MOPS) was obtained from Beyotime Biotechnology (Shanghai, China). Acetazolamide was purchased from Aladdin Biochemical Technology Co., Ltd. (Shanghai, China). Bovine erythrocyte CA (EC 4.2.1.1) and 4-nitrophenyl acetate (PNPA) were obtained from Sigma–Aldrich (St. Louis, MO, USA). HPLC-grade methanol and acetonitrile were purchased from Merck (Darmstadt, Germany). LC-MS-grade ultrapure water was prepared using a Milli-Q water purification system. Chromatographic separation was performed on a Waters Atlantis T3 column (5 μm, 4.6 × 250 mm).

2.2. Analysis by HPLC

2.2.1. Sample Preparation for HPLC Experiments

A 1 mL aliquot of GSHJT Oral Liquid was mixed with 4 mL of methanol, centrifuged at 10,000 rpm for 10 min, and filtered through a 0.22 μm membrane filter before injection.

2.2.2. HPLC Analysis

HPLC (Agilent Technologies, Santa Clara, CA, USA)analysis was carried out using an Agilent 1260 Infinity II system with an injection volume of 10 μL at a column temperature of 30 °C. The mobile phase, consisting of 0.1% phosphoric acid in water (A) and acetonitrile (B), was delivered at a flow rate of 0.8 mL/min with UV detection set to 210 nm. Separation was achieved using a gradient elution program with the following profile: 0–10 min, 5–12% B; 10–30 min, 12–23% B; 30–33 min, 23–25% B; 33–43 min, 25–28% B; 43–45 min, 28–30% B; 45–50 min, 30–70% B. The optimization for detection wavelength, mobile phase composition, column temperature, and flow rate are illustrated in Figures S1–S4. Method validation was subsequently performed, with the detailed results summarized in Tables S1–S6. Using the same HPLC conditions, quantitative analysis to determine the concentrations of salidroside and tyrosol was performed.

2.3. Analysis by LC-MS

2.3.1. LC-MS Sample Preparation

A 1 mL volume of GSHJT Oral Liquid was centrifuged at 10,000 rpm for 10 min and filtered through a 0.22 μm membrane filter.

2.3.2. LC-MS Conditions and Data Processing

LC-MS (SCIEX, Framingham, MA, USA) analysis was performed using an ExionL AD HPLC system coupled with an AB SCIEX X500B QTOF mass spectrometer equipped with an electrospray ionization (ESI) source. For LC-MS analysis, the injection volume was set to 10 μL, and the column temperature was 30 °C. The mobile phase comprised 0.1% formic acid in water (A) and 0.1% formic acid in acetonitrile (B), with a flow rate of 1.0 mL/min and detection at 210 nm. The gradient elution conditions were programmed as follows: 0–2 min, 5–10% B; 2–8 min, 10–15% B; 8–23 min, 15–18% B; 23–28 min, 18–20% B; 28–30 min, 20–25% B; 30–43 min, 25% B; 43–52 min, 25–50% B; 52–54 min, 50–52% B; 54–60 min, 52–90% B. Mass spectra were acquired in both positive and negative ion modes. The full-scan mass range was set from m/z 100–2000, while the MS/MS scan range covered 50–2000. ESI source parameters were configured as follows: nebulizer gas (GS1) at 55 psi; heater gas (GS2) at 55 psi; curtain gas (CUR) at 35 psi; Ion source temperature (TEM) maintained at 550 °C for both positive and negative; and ion source voltage (IS) set to 5500 V for positive and −4500 V for negative modes, respectively. For full-scan acquisition, the declustering potential (DP) and focusing voltage (CE) were set to 80 V and 10 V, respectively. MS/MS data were collected using the TOF MS-Product Ion-IDA mode, with collision-induced dissociation (CID) energy of 40 ± 20 eV. Prior to sample injection, mass axis calibration was carried out using a CDS pump, ensuring a mass accuracy error below 2 ppm. All data processing was conducted with MS-DIAL 5.5.250627 software.

2.4. Network Pharmacology Analysis

2.4.1. Active Component Screening and Target Prediction

To identify potential bioactive constituents from the 71 LC-MS-detected compounds, we submitted each compound to the Swiss ADME database (http://www.swissadme.ch/). Compounds were retained as candidate active components if they met two criteria: predicted gastrointestinal (GI) absorption was rated as “High”, and (ii) at least 2 of the 5 drug likeness rules (Lipinski, Ghose, Veber, Egan, and Muegge) returned a “yes” outcome. The SMILES identifiers of these candidate compounds were then uploaded to the SwissTargetPrediction database (http://www.swisstargetprediction.ch/) to predict potential protein targets. Targets with a probability score greater than zero were retained for further analysis.

Hypoxia-associated targets were retrieved from four public databases using “hypoxia” as the search keyword: GeneCards (https://www.genecards.org/), OMIM (https://www.omim.org/), TTD (https://ttd.idrblab.cn/), and PharmGkb (https://www.clinpgx.org/). For GeneCards-derived targets, only those with a Relevance Score greater than 1 were included. All targets retrieved from the four databases were then merged, deduplicated, and intersected using R software (version 4.5.0, https://www.r-project.org/). A Venn diagram was generated to visualize the overlap, yielding a final set of hypoxia-related targets.

2.4.2. Network Construction and Pathway Analysis

The overlap between component-related targets and hypoxia-related targets was determined using R software, and the intersecting targets were visualized with a Venn diagram. These common targets were then submitted to the STRING database (version 11.0, https://string-db.org/) to construct a protein–protein interaction (PPI) network. The analysis was restricted to Homo sapiens with a minimum interaction confidence threshold of 0.9 (high confidence). Disconnected protein nodes were excluded, while all other parameters remained at default settings. The resulting PPI network was exported in TSV format for subsequent analysis.

Topological analysis and visualization were performed using Cytoscape 3.8.0 software (https://cytoscape.org/), and the CytoNCA plugin was applied to analyze the topological parameters of targets in the network, including Betweenness centrality, degree, and Closeness centrality. The targets with values simultaneously higher than the median values of Degree centrality (DC), Closeness centrality (CC), Betweenness centrality (BC), Eigenvector centrality (EC), Local Average Connectivity (LAC), and Network centrality (NC) were selected as hub targets for subsequent enrichment analysis.

A network working file was established for the active components of GSHJT Oral Liquid and their intersecting anti-hypoxia-related targets, and imported into Cytoscape 3.8.0 software to construct an active component-intersecting target network. Meanwhile, the CytoNCA plugin was used to calculate and analyze topological parameters, and the components with the top 5 degree values were selected as the key active components.

2.5. Molecular Docking

For each candidate key component, the two-dimensional (2D) chemical structure was retrieved from the PubChem database (https://pubchem.ncbi.nlm.nih.gov/). Three-dimensional (3D) structures were then generated and energy-minimized using Chem3D software, followed by conversion to *.mol2 format. For the core targets identified through network analysis, we first obtained their UniProt IDs from the UniProt database (https://www.uniprot.org/), filtering for “Reviewed” and “Human” entries. These UniProt IDs were then used to download the corresponding 3D protein structures from the Protein Data Bank (PDB, https://www2.rcsb.org/). Protein structures were processed with PyMOL (version 3.1.6.1, https://www.pymol.org/) to remove water molecules, bound ligands, and other non-protein small molecules. The cleaned PDB files were saved for subsequent docking. The processed protein receptor files were imported into AutoDock 1.5.7 software (https://autodock.scripps.edu/) for hydrogenation, and then saved in both *PDB and *pdbqt formats. Small molecule ligand files were subsequently imported and saved in *pdbqt format. The central coordinates of the binding pocket grid box were then identified, with the Spacing value set to 1 (blind docking), and the file was saved in *gpf format. Finally, molecular docking was performed using Vina, with the free energy range set to 5 and the maximum number of docking poses set to 20. The binding affinity and activity between the targets and active compounds were evaluated based on the Docking Score, and PyMOL software was used to visualize the docking results.

2.6. CA Inhibition Assay

Bovine erythrocyte CA was used in this study because it shares high structural and functional homology with human CA isoforms. The active site region exhibits highly similar sequence identity between bovine and human CA, and bovine CA is widely accepted as a valid and economical model for initial CA inhibition screening in phytochemical research. For each test compound (kaempferol, ellagic acid, p-coumaric acid, salicylic acid, gallic acid, and salidroside) a reaction mixture containing 40 μL of the sample and 50 μL of 4-nitrophenyl acetate (PNPA, 6 mM) was prepared in a 96-well clear microplate. The enzymatic reaction was initiated by adding 10 μL of CA solution (110 nM). The plate was then incubated at 37 °C for 20 min, and the absorbance values were recorded at 405 nm using a microplate reader [25]. The inhibition rate (%) was calculated using the following Equation (1):

Inhibition rate (%) = [1 − (A1 − A2)/(A3 − A4)] × 100%

(1)

where

A₁ = absorbance value of the sample-containing group with CA;

A₂ = absorbance value of the sample-containing group without CA;

A₃ = absorbance value of the CA-containing group without samples;

A₄ = absorbance value of the blank control group without both samples and CA.

All reaction mixtures contained PNPA, and each experiment was performed in triplicate. Data are expressed as mean ± standard deviation (n = 3).

The enzyme activity inhibition curves were plotted using GraphPad Prism 10.1.2 (https://www.graphpad.com/). The Inhibitor vs. normalized response–Variable slope model in nonlinear regression (curve fitting) was selected for fitting, and the IC₅₀ values were finally obtained. Molecular docking was performed between the monomers and CA to elucidate the mechanism underlying their inhibitory effects on CA.

3. Results

3.1. HPLC Analysis Results

Following systematic optimization of the mobile phase, column temperature, detection wavelength, flow rate, and gradient elution program, the final HPLC conditions were established as follows: mobile phase consisting of 0.1% phosphoric acid aqueous solution (A) and acetonitrile (B); column temperature maintained at 30 °C; flow rate set at 0.8 mL/min; and detection wavelength fixed at 210 nm. The gradient elution program was: 0–10 min, 5–12% B; 10–30 min, 12–23% B; 30–33 min, 23–25% B; 33–43 min, 25–28% B; 43–45 min, 28–30% B; 45–50 min, 30–70% B. Peak 2 was selected as the reference peak. Method validation was performed by evaluating the relative retention times and relative peak areas of eight common chromatographic peaks (Figure 1). The developed method demonstrated satisfactory precision, repeatability, and stability. For precision, the RSD values of relative retention times were all below 0.79%, and those of relative peak areas were below 4.55%. For repeatability, the RSD values of relative retention times were below 0.05%, and those of relative peak areas were below 4.88%. For stability, the RSD values of relative retention times were below 0.83%, and those of relative peak areas were below 4.84%. In addition to qualitative fingerprinting, the contents of two major compounds, salidroside and tyrosol, were determined to be 55 μg/mL and 37 μg/mL, respectively (Figure S5 and Table S7).

3.2. LC-MS Analysis Results

Compositional profiling of GSHJT Oral Liquid was performed using LC-MS coupled with MS-DIAL software for compound identification. A total of 71 constituents were identified, including phenolic acids, flavonoids, glycosides, and saccharides. Detailed identification information is summarized in

Table 1. Identified chemical constituents in GSHJT Oral Liquid by LC-QTOF-MS/MS.

No.	Name	RT (min)	m/z	Adduct	ppm	Formula
1 *	Salicylic acid	2.490	137.02460	[M-H]−	1.0	C7H6O3
2	Volemitol	2.924	211.0824	[M-H]−	0.2	C7H16O7
3	Ribose	2.960	209.0670	[M+CH3COO]−	1.6	C5H10O5
4	Gluconic acid	2.983	195.0511	[M-H]−	0.4	C6H12O7
5	(2S,3R,4S)-4-(2-{[(2E)-3-(3,4-Dihydroxyphenyl)-2-propenoyl]oxy}ethyl)-2-(beta-D-glucopyranosyloxy)-3-vinyl-3,4-dihydro-2H-pyran-5-carboxylic acid	3.007	537.1662	[M-H]−	8.9	C25H30O13
6	Lactulose	3.042	341.1079	[M-H]−	3.0	C12H22O11
7	Calceolarioside A	3.099	477.1446	[M-H]−	9.2	C23H26O11
8	Betaine	3.109	118.0858	[M+H]+	5.0	C5H11NO2
9	Trehalose	3.133	365.1050	[M+Na]+	1.2	C12H22O11
10	Gentiobiose	3.144	381.0791	[M+K]+	0.7	C12H22O11
11	D-(-)-quinic acid	3.193	191.0564	[M-H]−	1.3	C7H12O6
12	Phloroglucinol	3.214	127.0382	[M+H]+	6.1	C6H6O3
13	Maltotriose	3.216	539.1375	[M+Cl]−	0.3	C18H32O16
14	Abscisic acid	3.354	287.1238	[M+Na]+	5.5	C15H20O4
15	L-Pipecolic acid	3.644	130.0861	[M+H]+	1.2	C6H11NO2
16	Malate	3.701	133.0142	[M-H]−	0.0	C4H6O5
17	Isocitric acid	3.725	191.0198	[M-H]−	0.2	C6H8O7
18	N-Fructosyl pyroglutamate	4.106	290.0883	[M-H]−	0.6	C11H17NO8
19	Adenosine	4.841	268.1039	[M+H]+	0.3	C10H13N5O4
20	Guanosine	5.171	284.099	[M+H]+	1.6	C10H13N5O5
21	Gallic acid hexoside	5.370	331.0667	[M-H]−	1.0	C13H16O10
22 *	Gallic acid	6.158	169.0145	[M-H]−	0.5	C7H6O5
23	3-Hydroxy-3-methylglutaric acid	7.459	345.0823	[2M-2H+Na]−	5.7	C6H10O5
24	3-[3-[4,5-dihydroxy-6-(hydroxymethyl)-3-[3,4,5-trihydroxy-6-(hydroxymethyl)oxan-2-yl]oxyoxan-2-yl]oxy-4,5-dihydroxy-6-(hydroxymethyl)oxan-2-yl]oxy-2-(3,4-dihydroxyphenyl)-5,7-dihydroxychromen-4-one	7.832	787.1939	[M-H]−	0.1	C33H40O22
25	Quercetin-3,4′-O-di-beta-glucoside	9.039	627.1553	[M+H]+	0.5	C27H30O17
26	Pyrogallol	9.080	127.0382	[M+H]+	6.1	C6H6O3
27	(E)-3-[2-[(2S,3R,4S,5S,6R)-3,4,5-trihydroxy-6-(hydroxymethyl)oxan-2-yl]oxyphenyl]prop-2-enoic acid	9.095	371.0979	[M+HCOO]−	1.1	C15H18O8
28	Aucubin	9.187	345.1187	[M-H]−	1.1	C15H22O9
29	Acremine I	9.198	299.1134	[M+CH3COO]−	0.7	C12H16O5
30	3,5-Dihydroxybenzoic acid	9.279	153.0190	[M-H]−	2.0	C7H6O4
31 *	Salidroside	9.135	301.1194	[M-H]−	1.0	C14H20O7
32	(2R,3S,4S,5R,6R)-2-[[(2R,3R,4R)-3,4-dihydroxy-4-(hydroxymethyl)oxolan-2-yl]oxymethyl]-6-[2-(4-hydroxyphenyl)ethoxy]oxane-3,4,5-triol	9.394	477.1602	[M+HCOO]−	2.5	C19H28O11
33	4-{[(3S,4R,5S)-3-Hydroxy-5-(4-hydroxy-3-methoxyphenyl)-4-(hydroxymethyl)tetrahydro-3-furanyl]methyl}-2-methoxyphenyl beta-D-glucopyranoside	9.731	583.2019	[M+HCOO]−	2.4	C26H34O12
34	[3,4,5-trihydroxy-6-(3,4,5-trihydroxybenzoyl)oxyoxan-2-yl]methyl 3,4,5-trihydroxybenzoate	10.357	483.0771	[M-H]−	1.9	C20H20O14
35	3,4-di-O-galloylquinic acid	10.473	495.0762	[M-H]−	3.7	C21H20O14
36	Caffeic acid hexoside	11.316	341.0873	[M-H]−	1.3	C15H18O9
37	Phenylacetaldehyde	12.107	121.0640	[M+H]+	6.5	C8H8O
38	1,3,6-tri-O-galloylglucose	12.998	635.0877	[M-H]−	2.0	C27H24O18
39 *	Tyrosol	13.189	139.0754	[M+H]+	2.6	C8H10O2
40	3-(3,4-Dihydroxyphenyl)Prop-2-Enoic Acid	15.072	179.0350	[M-H]−	0.3	C9H8O4
41	5,7-dihydroxy-2-(4-hydroxy-3-methoxyphenyl)-6-methoxy-3-[(2S,3R,4S,5S,6R)-3,4,5-trihydroxy-6-[[(2R,3R,4S,5S,6R)-3,4,5-trihydroxy-6-(hydroxymethyl)oxan-2-yl]oxymethyl]oxan-2-yl]oxychromen-4-one	16.222	671.1826	[M+H]+	1.3	C29H34O18
42	Epigallocatechin-3-Monogallate	16.288	457.0763	[M-H]−	3.0	C22H18O11
43	Sweroside	16.600	359.1337	[M+H]+	0.0	C16H22O9
44	[(2R,3R,4S,5R,6S)-2-(hydroxymethyl)-4,5,6-tris[(3,4,5-trihydroxybenzoyl)oxy]oxan-3-yl] 3,4,5-trihydroxybenzoate	17.507	787.1003	[M-H]−	0.5	C34H28O22
45	(3R,3′R,4R,6′S,7R)-5-Hydroxy-6′,7-dimethyl-6,8-dioxo-3′,4,4′,5′,6,6′,7,8-octahydrospiro[isochromene-3,2′-pyran]-3′,4,7-triyl triacetate	17.791	451.1237	[M-H]−	1.9	C21H24O11
46	(-)-Epigallocatechin gallate	18.470	457.0769	[M-H]-	1.5	C22H18O11
47	(-)-Gallocatechin gallate	19.557	459.0916	[M+H]+	1.3	C22H18O11
48	Quercetin-3,4′-O-di-beta-glucoside	19.458	625.1391	[M-H]−	3.2	C27H30O17
49 *	p-Coumaric acid	22.917	163.04230	[M-H]−	0.7	C9H8O3
50	Luteolin-4′-O-glucoside	22.981	447.0925	[M-H]−	1.7	C21H20O11
51	(2R,3S,4S,5R,6R)-2-[[(2R,3S,4S,5R)-3,4-dihydroxy-5-(hydroxymethyl)oxolan-2-yl]oxymethyl]-6-[(E)-3-phenylprop-2-enoxy]oxane-3,4,5-triol	25.707	446.2019	[M+NH4]+	0.4	C20H28O10
52 *	Ellagic acid	27.309	300.9988	[M-H]−	0.7	C14H6O8
53	1,4-Cyclohexanedione	30.860	113.0593	[M+H]+	3.5	C6H8O2
54	Spiraeoside	31.978	463.0867	[M-H]−	3.2	C21H20O12
55	Isoquercetrin	32.263	465.1028	[M+H]+	0.2	C21H20O12
56	Tricin 5-glucoside	32.759	493.1346	[M+H]+	1.0	C23H24O12
57	Syringetin-3-O-glucoside	34.147	509.1284	[M+H]+	1.1	C23H24O13
58	Luteolin-7-glucoside	34.354	449.1080	[M+H]+	0.3	C21H20O11
59	Syringetin-3-O-galactoside	37.871	507.1138	[M-H]−	1.3	C23H24O13
60	Afzelin	38.210	431.0973	[M-H]−	2.5	C21H20O10
61	Quercetin 3-O-glucuronide	40.250	477.0663	[M-H]−	2.5	C21H18O13
62	Isorhamnetin-3-glucoside-4′-glucoside	42.849	639.1556	[M-H]−	1.7	C28H32O17
63	Lonicerin	43.109	595.1651	[M+H]+	1.5	C27H30O15
64	(2R,3R,4S,5S,6R)-2-octoxy-6-[[(2S,3R,4S,5R)-3,4,5-trihydroxyoxan-2-yl]oxymethyl]oxane-3,4,5-triol	48.219	469.2278	[M+HCOO]−	2.7	C19H36O10
65	(6,6-Dimethylbicyclo [3.1.1]hept-2-yl)methyl 6-O-[(2R,3R,4R)-3,4-dihydroxy-4-(hydroxymethyl)tetrahydro-2-furanyl]-beta-D-glucopyranoside	48.371	493.2277	[M+HCOO]−	2.7	C21H36O10
66	Kaempferol-7-O-alpha-L-rhamnoside	48.532	431.0969	[M-H]−	3.4	C21H20O10
67	Atractyloside A	48.565	471.221	[M+Na]+	1.7	C21H36O10
68	Kaempferol-7-O-rhamnoside	48.753	433.1128	[M+H]+	0.3	C21H20O10
69	Luteolin	48.775	285.0400	[M-H]−	1.5	C15H10O6
70	(2Z)-4,6-dihydroxy-2-[(4-hydroxy-3,5-dimethoxyphenyl)methylidene]-1-benzofuran-3-one	52.869	329.0663	[M-H]−	1.1	C17H14O7
71 *	Kaempferol	53.089	285.0436	[M-H]−	0.7	C15H10O6

* Components identified by comparison with reference standards.

.

3.3. Compound and Target Gene Collection

From the 71 identified compounds, 16 potential active components were selected based on favorable gastrointestinal absorption and drug-likeness properties via the SwissADME database (

Table 2. Potential anti-hypoxic active components of GSHJT Oral Liquid screened by SwissADME based on GI absorption and drug-likeness.

NO.	Name
CF1	1,4-Cyclohexanedione
CF2	3,5-Dihydroxybenzoic Acid
CF3	3-Hydroxy-3-methylglutaric acid
CF4	Ellagic acid
CF5	Gallic acid
CF6	Kaempferol
CF7	L-Pipecolic acid
CF8	Luteolin
CF9	Malate
CF10	p-Coumaric acid
CF11	Phenylacetaldehyde
CF12	Phloroglucinol
CF13	Salicylic acid
CF14	Salidroside
CF15	Pyrogallol
CF16	Tyrosol

). Using the SwissTargetPrediction database, 1360 putative targets were predicted, of which 390 targets with a probability score > 0 were retained as component-related targets. Hypoxia-associated targets were retrieved from four public databases: GeneCards (2834), OMIM (13), TTD (8), and PharmGkb (20). After merging and deduplication, 2839 unique hypoxia-related targets were obtained (Figure 2).

3.4. Construction of Networks and Pathway Analysis

The intersection of the components of GSHJT Oral Liquid and hypoxia-related targets was obtained using R language software, yielding 133 intersecting targets. Among these, the targets of the active component compounds of GSHJT Oral Liquid were 220 after deduplication (Figure 3A).

The interaction between the active components of GSHJT Oral Liquid and the intersecting anti-hypoxia-related targets was visualized (Figure 3B). The top 5 components with the highest degree values in the network were screened as the final core active components (

Table 3. Top five key active components of GSHJT Oral Liquid ranked by degree centrality in the component–target network.

NO.	Name	Degree
CF6	Kaempferol	69
CF4	Ellagic acid	42
CF10	p-Coumaric acid	18
CF13	Salicylic acid	13
CF5	Gallic acid	12

), namely kaempferol, ellagic acid, p-coumaric acid, salicylic acid, and gallic acid. In addition, salidroside, a key component of GSHJT Oral Liquid, was also included.

After importing the intersecting targets into the STRING database, a PPI network was constructed (Figure S6), which contained 122 nodes and 165 edges. Topological analysis and visualization were performed using Cytoscape software and its CytoNCA plugin. The node color represents the degree value of the node; the darker the node color, the higher the corresponding degree value. For the first screening, a total of 27 core targets were screened under the criteria: BC > 2, CC > 0.049, DC > 2, EC > 0.014, LAC > 0.333, and NC > 0.667. For the second screening, a total of 7 core targets were obtained under the criteria: BC > 3.452, CC > 0.426, DC > 3, EC > 0.119, LAC > 2, and NC > 3. These 7 targets were identified as the final core targets between the components of GSHJT Oral Liquid and hypoxia-related targets (Figure S7), including SRC, PIK3R1, ESR1, EGFR, PTK2, IGF1R, and LYN.

According to the results of GO functional enrichment analysis, 1936 terms were annotated in biological process, 74 in cellular component, and 247 in molecular function, respectively. The top 10 terms were selected for visualization analysis. The color of the bar chart represents the enrichment degree of the corresponding function (the darker the color, the more significant the enrichment). For biological processes, the enriched items mainly included response to amyloid-beta, ephrin receptor signaling pathway, and response to oxidative stress. For cellular components, the targets were mostly located in the cell membrane. For molecular function, the main enriched term was histone kinase activity. In the KEGG pathway enrichment analysis, 137 signaling pathways were enriched with a threshold of q < 0.05, and the top 20 pathways were selected for further analysis. Among them, multiple pathways were associated with anti-hypoxic effects, including the PI3K-Akt signaling pathway, Rap1 signaling pathway, HIF-1 signaling pathway, Ras signaling pathway, relaxin signaling pathway, and p53 signaling pathway. These results indicated that the intersecting target genes may exert anti-hypoxic effects through the synergistic regulation of multiple pathways. The results of GO and KEGG enrichment analyses are presented in Figure 4.

3.5. Molecular Docking Validation

Molecular docking was performed sequentially between the screened potential key components (kaempferol, ellagic acid, p-coumaric acid, salicylic acid, gallic acid, salidroside) and the core targets (SRC, PIK3R1, ESR1, EGFR, PTK2, IGF1R, LYN) (

Table 4. Molecular docking binding energies of selected components from GSHJT Oral Liquid against seven core anti-hypoxic target proteins.

Compound	Binding Energy (Kcal/mol)
	SRC	PTK2	PIK3R1	ESR1	EGFR	IGF1R	LYN
Kaempferol	−8.0	−7.5	−6.9	−8.3	−7.7	−8.2	−8.6
Ellagic acid	−8.6	−9.2	−7.0	−8.7	−8.2	−9.4	−8.6
p-Coumaric acid	−5.9	−6.0	−5.6	−6.3	−5.4	−6.3	−5.9
Salicylic acid	−6.4	−5.8	−5.0	−6.1	−5.0	−5.9	−5.7
Gallic acid	−6.2	−5.6	−4.8	−6.6	−5.5	−5.8	−6.1
Salidroside	−7.5	−7.4	−6.5	−7.3	−6.9	−7.4	−7.7

). According to the binding energy results of molecular docking, the binding energies of all potential key components with core targets were less than −4.8 kcal/mol, suggesting potential favorable binding interactions. Among them, the binding energy between ellagic acid and IGF1R was the lowest (−9.4 kcal/mol). Of note, docking scores alone do not constitute direct evidence of biological activity; these results should be interpreted as complementary computational support for the experimental findings.

The optimal binding conformations obtained from molecular docking between key active components and core targets were visualized (Figure 5). The binding energies of p-coumaric acid with ESR1 and IGF1R were both −6.3 kcal/mol. However, no hydrogen bond was formed between p-coumaric acid and ESR1; thus, this docking result was excluded. The other potential key components formed multiple hydrogen bonds with the core targets.

3.6. CA Activity Inhibition Determination

Acetazolamide, a well-known carbonic anhydrase inhibitor clinically used for AMS prophylaxis, was used as a positive control. The IC₅₀ of acetazolamide was determined to be 18.84 nM (Figure 6A). In comparison, the inhibitory activity against CA was subsequently determined for kaempferol, ellagic acid, p-coumaric acid, salicylic acid, gallic acid, and salidroside at a single concentration of 250 μM (Figure 6B).

Kaempferol and ellagic acid were selected for full dose–response curves because they showed the highest inhibition rates at the single screening concentration (250 μM) among all tested compounds (Figure 6B). Therefore, inhibition curves were further constructed for these two compounds. The IC₅₀ value of kaempferol was 34.05 μM (Figure 6C), and that of ellagic acid was 119.1 μM (Figure 6D), which are approximately 1800-fold (34.05 μM vs. 18.84 nM) and 6300-fold (119.1 μM vs. 18.84 nM) less potent than acetazolamide under the same assay conditions.

Finally, molecular docking simulations between kaempferol, ellagic acid and CA were performed, and the results were visualized as shown in Figure 7. In the figure, blue lines represent hydrogen bonds, yellow dashed lines represent hydrophobic interactions, and the golden sphere represents the active site of Zn²⁺.

4. Discussion

The optimized and validated HPLC method provided reliable qualitative and selected quantitative evaluation of major compounds of GSHJT Oral Liquid. Comparison with published Rhodiola phytochemical profiles indicates that GSHJT Oral Liquid retains most key constituents, including salidroside, tyrosol, and kaempferol, despite processing losses [11,12]. While LC-MS profiling identified 71 constituents and HPLC provided quantitative data for two major compounds (salidroside and tyrosol), absolute quantification of kaempferol and ellagic acid, the two key anti-hypoxic phytochemicals identified in this study, was not performed. Therefore, future studies should develop and validate a sensitive LC-MS/MS method for the absolute quantification of kaempferol and ellagic acid in GSHJT Oral Liquid. LC-MS profiling revealed that the oral liquid is rich in flavonoids, phenolic acids, and their glycoside derivatives, which are widely recognized as bioactive phytochemicals with anti-hypoxic, antioxidant, and anti-inflammatory properties. Flavonoids are widely distributed in vegetables, fruits, cereals, and tea, and exhibit diverse biological activities including antioxidant, antitumor, and anti-inflammatory effects [26]. Kaempferol displays anticancer, antioxidant, and anti-inflammatory activities [27,28,29] and exerts anti-hypoxic effects by suppressing the activation of the PI3K and AKT1 signaling pathways [30]. Ellagic acid, a dimeric derivative of gallic acid, possesses anti-inflammatory, antibacterial, and antitumor properties [31,32,33] and attenuates hypoxia-induced inflammation and reactive oxygen species (ROS) accumulation by inhibiting the phosphorylation of JAK1, JAK2, and STAT1 and downregulating NOX4 expression [34]. Hypoxia triggers SRC phosphorylation, activating EGFR, PIK3R1, ESR1, and IGF1R pathways to cause oxidative stress, inflammation, and vascular barrier disruption during AMS. Kaempferol and ellagic acid act on these predicted targets to reverse these pathological changes. They inhibit SRC and EGFR phosphorylation to block upstream signals, suppress the hyperactivated PIK3R1/PI3K/AKT/HIF-1α pathway, and regulate ESR1-mediated antioxidant functions and IGF1R-driven energy metabolism. Together, these signaling modifications mitigate hypoxia-induced tissue damage and vascular leakage, thereby improving hypoxia tolerance in AMS [35,36,37,38,39].

Integrating network pharmacology and LC-MS profiling, we systematically characterized the active components, potential targets, and related signaling pathways of GSHJT Oral Liquid. Database mining yielded 16 putative active components and 220 corresponding unique hypoxia-associated targets, along with 2839 hypoxia-related genes. Intersection analysis identified 133 overlapping targets, and the resulting component–target network revealed that the 16 bioactive ingredients may collectively modulate these 133 anti-hypoxic targets in a synergistic manner. The loose probability threshold (>0) and broad keyword “hypoxia” represent inherent limitations of this network pharmacology approach, as they may introduce false-positive targets and artificially inflate network complexity. However, this permissive threshold was chosen intentionally to maximize sensitivity for hypothesis generation, ensuring that weakly bound functional proteins with potential relevance to hypoxia were not prematurely excluded. Potential false positives were subsequently minimized through multi-step topological screening and rigorous experimental validation. Notably, the experimentally validated target, CA, did not rank among the top predicted core targets. This discrepancy arises because current target databases have limited coverage of metabolic enzymes like CA, which are underrepresented in cheminformatic prediction algorithms. Therefore, CA was selected for experimental validation based on its well-established physiological role in hypoxia, specifically in acid-base balance, CO₂ transport, and oxygen delivery, and its clinical track record as a druggable target for AMS, rather than prediction scores alone. Future studies should adopt stricter filtering criteria (e.g., probability > 0.1) combined with pathway-restricted target selection, treating network pharmacology as a hypothesis-generating tool that must be complemented by biochemical validation. From the 133 overlapping targets, two-stage topological screening based on betweenness centrality, closeness centrality, degree centrality, eigenvector centrality, local average connectivity, and network centrality identified seven core hub targets: SRC, PIK3R1, ESR1, EGFR, PTK2, IGF1R, and LYN. Under hypoxic conditions, SRC phosphorylation is initiated, sequentially activating the EGFR, PIK3R1, ESR1, and IGF1R signaling pathways to drive oxidative stress, inflammation, and vascular barrier damage in AMS [35,36]. Within the erythrocyte membrane, the abnormally elevated activity of SRC and its family kinase LYN jointly disrupts membrane structure, amplifies oxidative stress, and impairs cellular function, thereby aggravating AMS pathology [35]. Kaempferol and ellagic acid can effectively counteract these pathological changes. Specifically, they inhibit SRC and EGFR phosphorylation, restrain the overactive PIK3R1-mediated PI3K/AKT/HIF-1α axis, and modulate both the antioxidative functions of ESR1 and the energy metabolism governed by IGF1R, ultimately mitigating tissue injury and vascular leakage to enhance hypoxia tolerance [35,36,37,38,39]. Parallel to these mechanisms, several key molecules regulate vascular tone and pulmonary adaptation at high altitudes. ESR1 activates eNOS via the PIK3R1/PI3K axis to stimulate NO synthesis, improving vasodilation during high-altitude climbing [40]. Simultaneously, hypoxia upregulates HIF-1α, EGF, and EGFR expression to drive alveolar epithelial proliferation, supporting lung tissue adaptation and injury repair [41]. During hypoxia-induced vascular remodeling, HIF-1α also modulates FAK/PTK2 activity to inhibit smooth muscle cell adhesion and migration [42]. Conversely, in severe AMS and high-altitude pulmonary edema (HAPE), hypobaric hypoxia triggers excessive M1 macrophage polarization and pro-inflammatory cytokine release, disrupting the endothelial barrier and inducing pulmonary edema. The macrophage IGF-1/IGF-1R signaling pathway helps resolve this hypoxia-induced lung injury by promoting M2 polarization, suppressing NF-κB-mediated inflammation, enhancing efferocytosis, and restoring endothelial tight junctions, which minimizes vascular leakage and accelerates edema resolution [43].

GO and KEGG enrichment analyses were performed on the 133 hypoxia-associated targets. GO terms revealed significant involvement in biological processes including response to amyloid-beta, epidermal growth factor receptor signaling, and response to oxidative stress, implying that these processes contribute to the anti-hypoxic action of GSHJT Oral Liquid. Cellular component annotation indicated predominant localization in membrane rafts, membrane microdomains, and protein kinase complexes, consistent with the subcellular sites of drug action. Molecular function enrichment highlighted histone kinase activity (including histone H3 kinase and histone H2AX kinase), supporting multi-target regulatory effects. KEGG pathway analysis indicated that the anti-hypoxic mechanisms may involve the PI3K-Akt, Rap1, and HIF-1 signaling pathways. The PI3K-Akt pathway regulates mTOR activity, which in turn controls the expression of hypoxia-inducible factor (HIF) [44]. HIF-1 functions as a key oxygen-sensitive transcription factor that orchestrates adaptive cellular responses to hypoxia [45]. Rap1, a telomere-associated protein, also serves as a novel regulator of hypoxia-induced apoptosis [46]. Furthermore, vitexin has been reported to mitigate hypoxia/reoxygenation injury in H9c2 cardiomyocytes via the Epac1/Rap1 pathway and by reducing ERK phosphorylation [47].

Molecular docking confirmed that key components of GSHJT Oral Liquid, including kaempferol, ellagic acid, and salidroside, display strong binding affinity toward SRC, PTK2, IGF1R, ESR1, and LYN, as evidenced by low binding energies. These results support stable and specific interactions between the active constituents and core anti-hypoxic targets.

Specifically, kaempferol formed two hydrogen bonds with LYN through residues including MET-322; ellagic acid formed five hydrogen bonds with IGF1R via VAL-1102, PHE-1124, GLY-1122, and GLU-1020; p-coumaric acid formed three hydrogen bonds with IGF1R through GLU-1020 and PHE-1124; salicylic acid formed six hydrogen bonds with SRC via GLU-181, THR-182, ARG-158, and ARG-178; gallic acid formed three hydrogen bonds with ESR1 through ILE-386, GLU-323, and GLU-353; and salidroside formed five hydrogen bonds with LYN via LYS-275, GLU-290, ASP-385, THR-319, and MET-322.

In the CA inhibition assay, kaempferol and ellagic acid exhibited moderate inhibitory activity. Supported by molecular docking simulations, both compounds were classified as zinc-bound water-anchoring inhibitors. They inhibited CA activity through a combined mechanism of hydrogen bonding and hydrophobic interactions, without direct coordination to the catalytic zinc ion and independent of prodrug hydrolysis. Mechanistically, kaempferol contains multiple phenolic hydroxyl groups, particularly ortho-hydroxyl groups on the B-ring, which form stable hydrogen-bond networks with Thr 199 and zinc-bound water at the CA active site, thereby interfering with zinc-mediated substrate activation. Its planar aromatic skeleton readily inserts into the hydrophobic pocket of the CA active site (e.g., with Val 121 and Val 143), strengthening hydrophobic interactions and stabilizing the inhibitor-enzyme complex. Similarly, ellagic acid inserts into the CA active site via hydrophobic interactions; its bulky molecular structure also generates steric hindrance that blocks substrate access to the catalytic pocket [24]. Beyond the individual activities of kaempferol and ellagic acid, GSHJT Oral Liquid is a multicomponent phytochemical preparation containing multiple flavonoids, phenolic acids, and glycosides. It is therefore plausible that the overall anti-hypoxic effect of the oral liquid may arise from synergistic interactions among these constituents rather than simply the additive effects of individual compounds. For instance, salidroside and gallic acid, though exhibiting only weak CA inhibition in the present single-concentration (250 μM) screening, may contribute to the overall bioactivity by modulating other hypoxia-related pathways or by enhancing the solubility, stability, or cellular uptake of kaempferol and ellagic acid. Conversely, the possibility of antagonistic matrix effects, where abundant excipients or co-eluting phytochemicals interfere with target binding, cannot be completely excluded based on the current experimental design. The present study employed a single screening concentration for the less active compounds, which provides an initial ranking of inhibitory potential but does not capture full concentration-response relationships or assess potential interactions between components. Therefore, future investigations should systematically evaluate the combinatorial effects of kaempferol, ellagic acid, salidroside, and gallic acid using established synergy quantification methods.

The present study has several limitations that should be acknowledged. First, while our network pharmacology and molecular docking analyses identified multiple potential targets, the experimental validation was confined to CA inhibition. The roles of these computationally predicted targets in the anti-hypoxic effects of GSHJT Oral Liquid remain to be experimentally verified. Second, the lack of cellular and animal validation means that the anti-hypoxic and anti-AMS effects remain speculative at this stage. Third, the absolute concentrations of kaempferol and ellagic acid within the oral liquid formulation were not quantified. Furthermore, we did not verify whether oral administration achieves therapeutic plasma concentrations in vivo. Therefore, future investigations should prioritize: (i) validation of CA inhibition in human cell-based hypoxia models; (ii) Quantification of kaempferol and ellagic acid concentrations within the oral liquid formulation; (iii) evaluation of kaempferol and ellagic acid in established animal models of acute mountain sickness; and (iv) systematic assessment of synergistic effects among the key phytochemicals using combination index assays.

5. Conclusions

This study systematically characterized the phytochemical profile of GSHJT Oral Liquid using HPLC and LC-MS, and explored its anti-hypoxic mechanisms using an integrated strategy of network pharmacology, molecular docking, and enzymatic validation. A total of 71 components were identified, with kaempferol, ellagic acid, p-coumaric acid, gallic acid, salicylic acid, and salidroside defined as key anti-hypoxic constituents. These components are predicted to target seven core hub targets (SRC, PIK3R1, ESR1, EGFR, PTK2, IGF1R, LYN) and synergistically regulate PI3K-Akt, HIF-1, Rap1, and p53 signaling pathways to enhance hypoxic tolerance. Kaempferol (IC₅₀ = 34.05 μM) and ellagic acid (IC₅₀ = 119.1 μM) exert moderate CA inhibitory effects via a water-anchoring binding mode, providing direct experimental evidence for the anti-hypoxic action. This work explores the active components and molecular mechanisms responsible for the anti-hypoxic effects of GSHJT Oral Liquid, establishing a scientific foundation for future preclinical and clinical investigations into its potential application in the prevention and adjuvant treatment of acute mountain sickness. It also provides a reproducible analytical and mechanistic paradigm for the modern research and development of Rhodiola-based phytomedicines. However, the findings are primarily based on in vitro and computational data; therefore, further validation using cellular and animal models is essential before any clinical recommendations can be made.