Salidroside orchestrates metabolic reprogramming by regulating the Hif-1α signalling pathway in acute mountain sickness

Abstract Context Rhodiola crenulata (Hook. f. et Thoms.) H. Ohba (Crassulaceae) is used to prevent and treat acute mountain sickness. However, the mechanisms underlying its effects on the central nervous system remain unclear. Objective To investigate the effect of Rhodiola crenulata on cellular metabolism in the central nervous system. Materials and methods The viability and Hif-1α levels of microglia and neurons at 5% O2 for 1, 3, 5 and 24 h were examined. We performed the binding of salidroside (Sal), rhodiosin, tyrosol and p-hydroxybenzyl alcohol to Hif-1α, Hif-1α, lactate, oxidative phosphorylation and glycolysis assays. Forty male C57BL/6J mice were divided into control and Sal (25, 50 and 100 mg/kg) groups to measure the levels of Hif-1α and lactate. Results Microglia sensed low oxygen levels earlier than neurons, accompanied by elevated expression of Hif-1α protein. Salidroside, rhodiosin, tyrosol, and p-hydroxybenzyl alcohol decreased BV-2 (IC50=1.93 ± 0.34 mM, 959.74 ± 10.24 μM, 7.47 ± 1.03 and 8.42 ± 1.63 mM) and PC-12 (IC50=6.89 ± 0.57 mM, 159.28 ± 8.89 μM, 8.65 ± 1.20 and 8.64 ± 1.42 mM) viability. They (10 μM) reduced Hif-1α degradation in BV-2 (3.7-, 2.5-, 2.9- and 2.5-fold) and PC-12 cells (2.8-, 2.8-, 2.3- and 2.0-fold) under normoxia. Salidroside increased glycolytic capacity but attenuated oxidative phosphorylation. Salidroside (50 and 100 mg/kg) treatment increased the protein expression of Hif-1α and the release of lactate in the brain tissue of mice. Conclusions These results suggest that Sal induces metabolic reprogramming by regulating the Hif-1α signalling pathway to activate compensatory responses, which may be the core mechanism underlying the effect of Rhodiola crenulata on the central nervous system.


Introduction
Rhodiola crenulata (Hook. f. et Thoms.) H. Ohba (Crassulaceae) has a long history of widespread use as a botanical medicine in Europe, Asia and the United States to prevent and treat various common conditions and complex diseases, including acute mountain sickness (AMS) (Yi et al. 2015), fatigue (Shevtsov et al. 2003) and Alzheimer's disease . Several studies have indicated that Rhodiola crenulata prevents and treats AMS by regulating the hypoxia-inducible factor-1 (Hif-1) signalling pathway (Wang et al. 2019). Salidroside (Sal), rhodiosin, tyrosol and p-hydroxybenzyl alcohol, the major functional ingredients of Rhodiola crenulata, may form a material basis for its anti-hypoxic effects (Chang et al. 2007(Chang et al. , 2018Liang et al. 2018;Yan et al. 2018). However, the active compounds and the exact role of Rhodiola crenulata in AMS remain unclear.
Salidroside possesses neuroprotective and cardioprotective effects with prominent antioxidative stress injury, anti-apoptosis and promotion of neurogenesis (Zhang et al. 2007Liu et al. 2020). In addition, Sal stimulates hypoxia-inducible factor-1 alpha (Hif-1a) and erythropoietin (EPO) production to enhance neuronal survival under hypoxia and anti-inflammatory activities after cerebral ischaemia (Villa et al. 2003;Wei et al. 2017). Studies with cell experiments have shown that Sal stimulates the accumulation of Hif-1a protein via the reduction of Hif-1a degradation to provide an anti-hypoxic effect in kidney and liver cells (Zheng et al. 2012). However, the effect of Sal on microglia and neurons for its adaptogenic function remains unknown.
If the central nervous system is damaged, the brain is sensitive to hypoxia, including brain cell injury and brain oedema (Ereci nska and Silver 2001). The adaptogenic activity of Rhodiola crenulata appears to be associated with a shift in metabolism during cold-hypoxia-restraint exposure and post-stress recovery (Gupta et al. 2008). Salidroside protects against myocardial injury from inflammation by regulating energy metabolism (Chang et al. 2016). Studies have shown that activating compensatory responses of astrocytes decreases aerobic oxidation and oxygen consumption but increases glycolysis (Marrif and Juurlink 1999;Vangeison et al. 2008). Hif-1a, the main transcriptional regulator of glycolysis, has been reported to reduce oxygen consumption, diminish the efficiency of oxidative phosphorylation (OXPHOS) and increase the dependency on glycolysis as an energy source (Iyer 1998;Gupta et al. 2009). Few studies have focussed on metabolic changes after treatment with active compounds of Rhodiola crenulata in the central nervous system.
Our study aimed to gain new insights into the active ingredients of Rhodiola crenulata against AMS. We hypothesized that active compounds of Rhodiola crenulata activate Hif-1 signalling and alter cellular metabolism to activate compensatory responses in the central nervous system. The present study used normoxia/ hypoxia cellular models with 5% O 2 or 21% O 2 in vitro. We employed mouse microglia (BV-2), pheochromocytoma (PC-12) and C57BL/6J mice to clarify the mechanism of this compensatory function stimulated by Rhodiola crenulata in advance under normoxia or mild hypoxia.

Preparations of reference substances and samples
Reference substances of p-hydroxybenzyl alcohol, Sal, tyrosol and rhodiosin were mixed in methanol to yield concentrations of 5, 200, 15 and 20 lg/mL. Rhodiola crenulata was pulverized and filtered through a 40-mesh sieve. Powder (0.5 g) was extracted with 10 mL of methanol. The extract was then filtered through a 0.22 lm filter and dissolved in five methanol to yield a concentration of 12.5 mg/mL.

Cell culture
The microglial cell line (BV-2) and pheochromocytoma cell line (PC-12) were purchased from the Type Culture Collection of the Chinese Academy of Sciences (Shanghai, China). The cells were cultured in Dulbecco's modified Eagle medium (DMEM)/F12 medium containing 10% foetal bovine serum, penicillin/streptomycin and high-glucose DMEM containing 10% neonatal bovine serum and penicillin/streptomycin. Before use, the cells were maintained in a CO 2 humidified incubator (37 C, 5% CO 2 ) (Sanyo, Osaka, Japan). The cell culture medium was changed every day during culture, and the cells were passaged at $80% confluence.

Western blotting
Lysates from cells were prepared in RIPA lysis buffer containing protease inhibitors and then measured using a BCA Protein Assay Kit (BestBio, Shanghai, China). Lysates (40 lg) were loaded on SDS-PAGE gels for electrophoresis and then transferred to a polyvinylidene fluoride (PVDF) membrane (Merck Millipore, Darmstadt, Germany) at 200 mA for 2 h. After that, the PVDF membranes were blocked with non-fat dry milk in tris-buffered saline for 2 h at room temperature and probed with Hif-1a or b-actin overnight at 4 C. Membranes were then incubated with anti-rabbit IgG and HRP-linked antibodies for 1 h at room temperature. Afterward, the blots on the membranes were detected with ECL reagent (Merck Millipore, Darmstadt, Germany) using a chemiluminescence imaging system (ChemiScope Mini, Shanghai, China).

Molecular docking
Two-dimensional structures of Sal, rhodiosin, tyrosol and phydroxybenzyl alcohol were obtained using ChemDraw. Ligand preparation was performed using Chem3D by energy minimization using an MMP force field. The three-dimensional structure of HIF-1a (PDBID; 4H6J), with a resolution of 1.52 Å, was obtained from the Protein Data Bank database (http://www.rcsb. org/structure/4H6J). Remove the solvent and organic matter using PyMOL. Then, hydrogen was added to the protein using AutoDock Tools. A grid box with grid points (96, 96, 108) and spacing of 0.375 was centred (12.214, À14.282, À21.338) on the given co-crystallized ligand. Molecular docking analysis was performed via AutoDock4. In the dockings, the genetic algorithm was chosen as the search parameter. Visualization was performed using the Discovery Studio program.

Ribonucleic acid (RNA) analysis by quantitative real-time reverse transcription-polymerase chain reaction (qRT-PCR)
To investigate the gene expression of Hif-1a in BV-2 and PC-12 cells stimulated by Sal, rhodiosin, tyrosol, hydroxybenzylalcohol or DFO for 4 h, qRT-PCR was performed. BV-2 or PC-12 cells were seeded into six-well plates at a density of 6 Â 10 5 or 8 Â 10 5 cells per well, respectively. Total RNA was extracted from BV-2 and PC-12 cells using the TRIzol reagent (Thermo Scientific, Waltham, MA). Total RNA samples (1 lg) were reverse transcribed into cDNA with HiScript V R III RT SuperMix for qPCR (Vazyme Biotech Co., Ltd., Nanjing, China) and amplified using ChamQ Universal SYBR qPCR Master Mix (Vazyme Biotech Co., Ltd., Nanjing, China) with a hot denaturation for 30 s at 95 C. Then, fluorescence intensity was recorded for 40 cycles at 95 C for 10 s and annealed at 60 C for 30 s. Glyceraldehyde 3-phosphate dehydrogenase (GAPDH) was used as a housekeeping gene to normalize the expression of Hif-1a. Relative expression was calculated using the 2 -DDCT analysis.

Measurement of lactate
BV-2 or PC-12 cells were seeded into 12-well plates at a density of 2 Â 10 5 cells/mL for 24 h. The culture medium was collected at 4 h after treatment with vehicle, Sal (1, 10 and 100 lM) or DFO (100 lM) for quantification of lactate using a Lactate Colorimetric Assay kit (BioVision, Shanghai, China) following the manufacturer's instructions. The absorbance was measured using a multimode plate reader at 450 nm (PerkinElmer, Waltham, MA).
Measurement of the extracellular acidification rate (ECAR) and oxygen consumption rate (OCR) The ECAR and OCR were measured using a Seahorse XF24 analyser (Seahorse Bioscience, Boston, MA) according to the manufacturer's instructions. Plate BV-2 cells at 8 Â 10 4 cells/100 lL and PC-12 cells at 6 Â 10 4 cells/100 lL previously in the Seahorse XF Microplate, respectively, and treated with vehicle, Sal (1, 10 and 100 lM) or DFO (100 lM). The sensor cartridge was hydrated in a Seahorse XF Calibrant and incubated overnight (37 C, CO 2 -free). For ECAR measurements, the assay medium of PC-12 was prepared by XF base medium with 4 mM glutamine, and BV-2 was prepared by XF base medium with 2.5 mM glutamine. For OCR measurements, the assay medium of BV-2 (1 mM pyruvate, 2.5 mM glutamine and 17 mM glucose) and PC-12 (XF base medium containing 1 mM pyruvate, 4 mM glutamine and 25 mM glucose) were prepared immediately before assay.

Animal experiment
Forty male C57BL/6J mice were obtained from the Beijing Vital River Laboratory Animal Technology Co., Ltd. (Beijing, China), certification no. SCXK (Jing) 2016-0006 weighing 22 ± 10 g. The mice were maintained in accordance with the Guidelines for the Care and Use of Laboratory Animals formulated by the Ministry of Science and Technology of China. All procedures were approved by the Animal Ethical and Welfare Committee of Beijing University of Chinese Medicine (approval number: BUCM-4-2021082005-3050). All mice were housed five per cage at a temperature of 21 C with a 12 h light-dark cycle (lights off at 07:00 h) and at a relative humidity of 40-50% and had free access to standard mice chow and water. The mice were acclimatized to their surroundings for one week and were randomly divided into four groups (n ¼ 10/group): control group (normal saline, 0.2 mL i.p.), SAL-L group (25 mg/kg i.p.), SAL-M group (50 mg/kg i.p.) and SAL-H group (100 mg/kg i.p.). The mice were treated with normal saline, SAL-L, SAL-M or SAL-H daily for three days. After the last treatment, the brain of each mouse was isolated after the mice were intracardially perfused with 60 mL of 0.9% normal saline.

Statistical analysis
Data processing and statistical analysis were performed using GraphPad Prism 6 (La Jolla, CA) with a one-way analysis of variance (ANOVA) using Tukey's post hoc test. All data are expressed as the mean ± SEM. A p value of less than 0.05 was considered statistically significant for all tests.

Time threshold to mild hypoxia on the proliferation of BV-2 cells and PC-12 cells
For the induction of mild hypoxia, BV-2 and PC-12 cells were treated with 5% O 2 for 1, 3, 5 and 24 h, as shown schematically in Figure 1(A). To investigate the effect of mild hypoxia on proliferation, we assessed the proliferation of BV-2 and PC-12 cells after induction of hypoxia with 5% O 2 . BV-2 cells cultured with 5% O 2 for 5 h and 24 h exhibited 11% and 18% inhibition, respectively (Figure 1(B)). PC-12 cells exhibited 17% and 21% inhibition, respectively (Figure 1(C)). However, hypoxia with 5% O 2 for 1 h and 3 h had no effect on the proliferation of BV-2 and PC-12 cells.

BV-2 were sensitive to mild hypoxia
We chose DFO as a positive control, which mimics hypoxic conditions because of inducing Hif-1a. We investigated the protein expression of Hif-1a. Western blotting results showed that induction with 5% O 2 for 3 h increased Hif-1a protein expression in BV-2 cells (Figure 2(A)) and that induction with 5% O 2 for 5 h increased Hif-1a protein expression in PC-12 cells (Figure 2(B)). These data suggest that BV-2 cells were more sensitive to mild hypoxia than PC-12 cells after induction with 5% O 2 . Therefore, we used 3 h for mild hypoxia in the subsequent experiments.

Salidroside induced metabolic reprogramming from oxidative phosphorylation to glycolysis
Glycolysis enables cell proliferation, migration, cytokine secretion and phagocytosis because the glucose metabolism rate in glycolysis is 10-100 times faster than that of oxidative phosphorylation (Baik et al. 2019). Cellular energy metabolic pathways in the central nervous system may be closely linked to their effector function. To determine whether Sal is involved in metabolic reprogramming, we assessed the status of glycolysis and mitochondrial respiration in BV-2 and PC-12 cells after treatment with Sal. Cell proliferation analyses showed that Sal did not show any toxic effects in BV-2 and PC-12 cells at a concentration of 200 lM ( Figure 5(A,B)). After Sal treatment for 4 h, the lactic acid in the supernatant was increased compared with the control, which indicated Sal induced glycolysis in BV-2 and PC-12 cells ( Figure 5(C,D)).
The ECAR was measured as a measure of glycolysis after treatment with glucose (a saturating concentration of glucose), oligomycin (an ATP synthase inhibitor) and 2-deoxy-D-glucose (2-DG) (a glucose analogue). The results revealed that the glycolytic capacity level was increased by 91.5%, 104.0% and 102.4% after exposure to Sal at doses of 1, 10 and 100 lM for 4 h in BV-2 cells compared with the control (Figure 6(A)). The glycolytic capacity was increased by 7.6%, 7.1% and 15.9% at 1, 10 and 100 lM of Sal for 4 h in PC-12 cells (Figure 6(B)). These data indicate a shift towards glycolytic metabolism in BV-2 and PC-12 cells.
As we observed that Sal regulated glycolysis levels, we hypothesized that Sal also regulated oxidative phosphorylation efficiency. To confirm this, the changes in the OCR were measured, an indicator of mitochondrial respiration after treatment with oligomycin (an ATP synthase inhibitor), cyanide p-trifluoromethoxyphenyl-hydrazone (FCCP) (H þ ionophore), and a mixture of rotenone and antimycin A (electron-transport chain inhibitor), to assess mitochondrial function. The results implied that Sal treatment at doses of 10 and 100 lM significantly diminished the maximal respiratory capacity of mitochondria in BV-2 and PC-12 cells compared with the control (Figure 6(C,D)). These data demonstrate that Sal increased glycolysis while attenuating oxidative phosphorylation, suggesting that metabolic reprogramming towards glycolysis is critical for the prevention of AMS by Sal.

Salidroside induced glycolytic metabolism in mice
To better verify the mechanisms of the central nervous system functions induced by Sal against AMS, we performed animal experiments to detect the protein expression level of Hif-1a and the secretion of lactate in the brain of mice. After administering Sal at 50 and 100 mg/kg i.p. for three days, Sal led to a significant upregulation of Hif-1a protein expression compared with the control (Figure 7(A,B)). In addition, the lactic acid in the brain tissue increased after Sal treatment, suggesting that Sal induced glycolytic metabolism in vivo.

Discussion
Taking Rhodiola crenulata in advance of exposure to high altitudes can prevent and treat AMS. Our study revealed that cellular metabolism was reprogrammed to activate the compensatory response of the central nervous system after the administration of Sal under normoxia. Our results demonstrated the following: (1) microglia were more sensitive to mild hypoxia than neurons; (2) Sal dampened Hif-1a degradation under normoxia; and (3) Sal-triggered metabolic reprogramming from oxidative phosphorylation to glycolysis. Our results provide strategies for the protection of AMS from the perspective of cellular metabolism.
Hif-1a serves as the core initiator molecule for the cells to initiate the hypoxic protection reaction, which mediates the basic oxygen sensing mechanism under hypoxic conditions (Semenza 2004). During transient mild hypoxia, intracellular accumulation of Hif-1 strongly protects against hypoxic injury. Hypoxic preconditioning can increase cell anoxic tolerance and survival (Wang et al. 2008). In this study, we observed that treatment with 5% O 2 increased Hif-1a expression in BV-2 cells for 3 h while PC-12 cells for 5 h, indicating that BV-2 cells were more sensitive to low oxygen levels than PC-12 cells. Our results confirmed that microglia responded to mild hypoxia earlier than the neurons.
Microglia and neurons are highly susceptible to hypoxia, which leads to brain cell injury and neuronal death. Many studies suggest that Sal, rhodiosin, tyrosol and p-hydroxybenzyl alcohol exert anti-hypoxic pharmacological effects, including neuroprotection and the activation of microglial cells. Our results clearly showed that Sal had a stronger binding ability to Hif-1a with hydrogen-bonding interactions and van der Waals and Alktl forces in different directions. Consistent with these studies, western blotting demonstrated that the protein level of Hif-1a was higher after treatment with Sal than with rhodiosin, tyrosol and p-hydroxybenzyl alcohol treatments in BV-2 and PC-12 cells under normoxia or hypoxia. Previous studies have shown that Sal modulates microglial polarization to protect neurons after cerebral ischaemia ). However, Sal did not affect the mRNA levels of Hif-1a under normoxia. Taken together, these data suggest that Sal induces an increase in Hif-1a protein in microglia triggered neuroprotection under normoxia, which might be associated with the prevention of AMS in Rhodiola crenulata.
Hif-1 mediates hypoxia and regulates more than 1000 direct transcriptional targets, including metabolic adaptation, angiogenesis, cell cycle and apoptosis (Schodel and Ratcliffe 2019). Moreover, Hif-1a upregulates the transcription of glycolytic genes, resulting in a metabolic switch from oxidative phosphorylation to glycolysis under hypoxia to supply cellular energy demands (Semenza 2011(Semenza , 2012. Recent studies have revealed that immune cell metabolism reflects their status. In the present study, Sal changed brain energy metabolism in mice and induced microglial and neural metabolic reprogramming. Sal enhanced lactic acid production in BV-2 cells. Furthermore, ECAR increased after exposure to Sal and reached a plateau at a dose of 10 lM, while the maximal respiratory capacity of mitochondria was diminished. These findings indicate that Sal induced metabolic reprogramming towards glycolysis in immune cells, activating inflammation. In addition, we observed metabolic reprogramming towards glycolysis in neurons. Neurons increase glycolysis in the face of acute energy demand, which provides a faster response to energy (Yellen 2018). Glycolytic enzymes located at the plasma membrane and synaptic vesicles are conducive to supply acute ATP to neurons (Ikemoto et al. 2003;Hinckelmann et al. 2016). This might explain why Sal increases glycolysis as the main energy source for neurons in response to stimulation. The mammalian target of the rapamycin (mTOR) pathway senses cellular energy status by regulating glucose metabolism, which regulates the expression of Hif-1a, which serves as the master transcriptional regulator of glycolysis (Cheng et al. 2014). These findings collectively suggest that Sal induced metabolic reprogramming depends on the Hif-1a pathway in the central nervous system, which might be enhanced by the phosphorylation of mTOR. Nevertheless, this speculation requires further investigation. We will consider the cell metabolism of astrocytes in future experiments to better explain the mechanisms of the functions of the central nervous system against AMS.

Conclusions
In conclusion (Figure 8), we have demonstrated that microglia respond to mild hypoxia earlier than neurons in the central nervous system. Furthermore, we have confirmed that Sal activates compensatory responses in the mouse brain, activates microglia and improves neural function by dampening Hif-1a degradation under normoxia. The improved endurance performance mechanism of Sal is potentially associated with the orchestration of metabolic reprogramming of neurons and microglia from oxidative phosphorylation to glycolysis.

Author contributions
Xiaoning Yan conducted the experiments and wrote the manuscript. Meixia Zhu, Yinhuan Zhang and Zhixin Jia analysed the data. Lirong Liu and Menghan Feng contributed reagents, materials and analysis tools. Jie Liu and Yijun Chen revised the manuscript. Hongbin Xiao designed the study as the corresponding author.

Disclosure statement
No potential conflict of interest was reported by the author(s).