Bioactive compounds from Artemisia dracunculus L. activate AMPK signaling in skeletal muscle

Abstract

An extract from Artemisia dracunculus L. (termed PMI-5011) improves glucose homeostasis by enhancing insulin action and reducing ectopic lipid accumulation, while increasing fat oxidation in skeletal muscle tissue in obese insulin resistant male mice. A chalcone, DMC-2, in PMI-5011 is the major bioactive that enhances insulin signaling and activation of AKT. However, the mechanism by which PMI-5011 improves lipid metabolism is unknown. AMPK is the cellular energy and metabolic sensor and a key regulator of lipid metabolism in muscle. This study examined PMI-5011 activation of AMPK signaling using murine C2C12 muscle cell culture and skeletal muscle tissue. Findings show that PMI-5011 increases Thr172-phosphorylation of AMPK in muscle cells and skeletal muscle tissue, while hepatic AMPK activation by PMI-5011 was not observed. Increased AMPK activity by PMI-5011 affects downstream signaling of AMPK, resulting in inhibition of ACC and increased SIRT1 protein levels. Selective deletion of DMC-2 from PMI-5011 demonstrates that compounds other than DMC-2 in a “DMC-2 knock out extract” (KOE) are responsible for AMPK activation and its downstream effects. Compared to 5-aminoimidazole-4-carboxamide ribonucleotide (AICAR) and metformin, the phytochemical mixture characterizing the KOE appears to more efficiently activate AMPK in muscle cells. KOE-mediated AMPK activation was LKB-1 independent, suggesting KOE does not activate AMPK via LKB-1 stimulation. Through AMPK activation, compounds in PMI-5011 may regulate lipid metabolism in skeletal muscle. Thus, the AMPK-activating potential of the KOE adds therapeutic value to PMI-5011 and its constituents in treating insulin resistance or type 2 diabetes.

Graphical abstract

1. Introduction

The 5-AMP-activated protein kinase (AMPK) is a master regulator of metabolism that is activated in low cellular energy conditions (increased AMP/ATP ratio) to switch on ATP-generating, catabolic pathways and switch off energy-demanding, ATP-consuming processes. AMPK functions as a serine/threonine kinase composed of a catalytic α, and regulatory β and γ subunits. AMP binding to the γ subunit allows AMPK to sense shifts in AMP/ATP ratio and causes a conformational change that exposes the activating phosphorylation site at Thr172 of the catalytic α subunit [1, 2]. AMPK activation through phosphorylation at Thr172 on the α-subunit is regulated independently by two upstream kinases, liver kinase B1 (LKB1) and Ca(2+)/calmodulin-dependent protein kinase kinase-β (CaMKKβ). Increased AMP levels stimulate LKB1-mediated phosphorylation, whereas increased intracellular calcium activates CaMKKβ-mediated phosphorylation [1, 3]. Once activated by Thr172 phosphorylation, AMPK coordinates the activities of key metabolic pathways that control lipid and glucose metabolism, protein synthesis, cell growth and proliferation, autophagy, mitochondrial biogenesis and function [4]. Given the central role of AMPK in monitoring cellular energy availability, its impact on metabolism is particularly important in tissues that undergo highly dynamic changes in energy demand, such as skeletal muscle.

Skeletal muscle is unique in its ability to vary its metabolic rate in response to dynamic changes in physical activity or metabolic states. It is responsible for 70-90% insulin-stimulated glucose uptake during feeding and is a predominant site for fatty acid oxidation during fasting or exercise because of its high-energy demands and large mass [5-7]. AMPK-dependent regulation of ATP availability is critical to maintaining the metabolic flexibility necessary to switch between glucose and fatty acid oxidation in response to different physiological states such as fed and fasted or activity states ranging from resting to vigorous exercise [8]. The inability to adapt to these changes in skeletal muscle is associated with insulin resistance and metabolic disorders [9, 10]. Impaired signaling in response to insulin in skeletal muscle, coupled with accumulation of ectopic lipid metabolites, reduced fatty acid oxidation and impaired mitochondrial function plays a pivotal role in developing obesity-associated metabolic disorders such as insulin resistance and type 2 diabetes [11-13]. Thus, a strategy focused on activating AMPK signaling in skeletal muscle has therapeutic potential as a monotherapy or in combination with other approaches to treat obesity-associated metabolic dysfunction.

A long history of pharmacological use of natural products supports their anti-diabetic effects. This includes metformin, a widely used synthetic biguanide based on compounds found in Galega officinalis L., commonly known as French lilac [14]. Metformin activates AMPK in an LKB1-dependent manner that is associated with reduced hepatic glucose output and improved glucose uptake in skeletal muscle [15, 16] although neither protein is a direct target of metformin [17]. Artemisia dracunculus L., or Russian tarragon, is an herb that has been used historically in traditional medicine as an anti-inflammatory and antioxidant agent [18], but also for its antidiabetic properties [18]. An ethanolic extract of A. dracunculus, termed PMI-5011 has been investigated extensively in rodent and cell culture models because of its antidiabetic effect [19-21]. Using bioactivity-guided fractionation method, five compounds were identified in PMI-5011 as anti-diabetic marker compounds [22, 23]. A series of experiments showed that PMI-5011 improves glucose metabolism by reducing blood glucose and insulin levels and enhancing insulin signaling in skeletal muscle in obese, insulin resistant mice [24-26]. The hypoglycemic activity of PMI-5011 is primarily attributed to 2’,4’-dihydroxy-4-methoxydihydrochalcone (DMC-2, 4-O-methyldavidigenin) and, to a lesser extent, its regioisomer, 2’,4-dihydroxy-4’-methoxydihydrochalcone (DMC-1, 4’-O-methyldavidigenin) [27]. In experiments using a recently developed DESIGNER (Deplete and Enrich Select Ingredients to Generate Normalized Extract Resources) approach, DMC-2 and DMC-1 were selectively removed from PMI-5011 to definitively determine whether DMC-2 and DMC-1 account for enhanced insulin signaling in skeletal muscle [28]. PMI-5011 depleted in DMC-1/2 is herein called Knock-Out Extract and abbreviated KOE. Using this new approach combined with an animal model of obesity-induced insulin resistance, we confirmed that DMC-2 (along with DMC-1) is the major phytochemical regulating insulin signaling through AKT-activation in skeletal muscle and lowering blood glucose levels. The impact of DMC-2 on blood glucose levels was comparable to metformin [28].

Interestingly, phosphorylation of AS160 occurred in the complete extract and the KOE containing all components of PMI-5011 other than DMC-2 and DMC-1, although AKT phosphorylation did not occur in the KOE [28]. Since AS160 is also activated by AMPK, we hypothesize that PMI-5011 and the KOE activate the AMPK-signaling pathway in skeletal muscle. In support of this possibility, recent studies demonstrate that PMI-5011 supplementation results in favorable changes in lipid metabolism by increasing mitochondrial fatty acid oxidation, improving metabolic flexibility in skeletal muscle, and decreasing triglyceride content in the skeletal muscle and liver of obese, insulin resistant mice without changes in body weight or adiposity [26, 29, 30]. Moreover, earlier studies showed that PMI-5011 increased insulin secretion from β cells through AMPK activation [31]. The current study used a high fat diet-fed mouse model of obesity-related insulin resistance and murine C2C12 skeletal muscle cells to test the effect of PMI-5011 and the KOE on AMPK activation in skeletal muscle.

2. Materials and Methods

2.1. Sourcing and characterization of PMI-5011 extract

Artemisia dracunculus was grown in a Rutgers University (New Brunswick, NJ) greenhouse under uniform and strictly controlled conditions using commercially available seeds (Sheffield’s Seed Company, Locke, NY). Preparation of the ethanolic extract from A. dracunculus (PMI-5011), detailed information about quality control, and biochemical characterization were previously reported [22, 23, 28, 32-35]. Bioactivity guided fractionation using in vitro bioassays followed by confirmation in vivo, identified five bioactive compounds [22]. The knock-out extract (KOE) lacking DMC-1 and DMC-2 and the enriched DMC-2 fraction were generated by the Center of Natural Products Technologies (CENAPT) (University of Illinois at Chicago) using DESIGNER technology [28]. DMC-2 was synthesized as described [33]. In Table 1, the concentrations of DMC-1 and DMC-2 in PMI-5011 and the KOE were determined as previously described [28]. Due to lack of available standards, concentrations of the compounds in the KOE were approximated based on structural similarity to DMC-2 and Sakuranetin. DMC-2 and sakuranetin were quantified by LC-MS using standard curves of pure compounds.

Table 1.

Bioactive Compounds in Artemisia dracunculus L.

Bioactive compounds in Artemisia dracunculus L.	PMI50 11	DMC enrich ed fraction	Knockout fraction “KOE”
2',4'-Dihydroxy-4-methoxydihydrochalcone (DMC-2)	1.76	68.73	Below LOD
2',4-Dihydroxy-4'-methoxydihydrochalcone (DMC-1)	0.81	21.86	Below LOD
Davidigenin	0.74*		0.74*
6-Demethoxycapillarisin	1.73*		1.32*
Sakuranetin	3.4*		3.13*

Compounds detected in each fraction (% [w/w]); LOD, limit of detection

^*Approximations based on DMC-2 and Sakuranetin equivalents

2.2. Experimental animals

Male C57BL/6J mice (N=50) were obtained from Jackson Laboratories (Bar Harbor, ME, USA). All animal experiments were approved by the Rutgers University Animal Care and Use Committee (Protocol #04-023) and were compliant with the NIH Guide for the Care and Use of Laboratory. The animals were housed singly with a 12-hour light–dark cycle at 24 °C. At 4 weeks of age, the mice were placed on a very high fat diet (60 kcal% fat, D12492, Research Diets) for 16 weeks. Body weight was measured weekly. At 18 weeks of age, the mice were acclimated to gavage over a 2-week period prior to administering a single, one time dose of water as a control (N=7, data not shown), the vehicle control (N=5, Labrasol^®), PMI-5011 complete extract (N=7), KOE (N=7), enriched DMC-2 fraction (N=5), synthetic DMC-2 at 100 mg/kg (N=7), synthetic DMC-2 at 300 mg/kg (N=7), or metformin at 300 mg/kg (N=5) via gavage at 20 weeks of age. Blood glucose levels were assayed at baseline and each hour thereafter up to 6 hours post gavage. The mice were euthanized and tissues collected for analysis at 6 hours after gavage. Body weight, fasting glucose, insulin and C-peptide levels at 6 hours post gavage were previously reported [28].

2.3. Skeletal muscle cell culture

Murine C2C12 myoblasts cells were obtained from the American Type Culture Collection (#CRL-1458) and cultured in Dulbecco’s modified Eagle’s medium (DMEM), high glucose (25 mM) with 10% fetal bovine serum, 2 mM glutamine, and antibiotics (100 units/mL penicillin G and 100 μg/mL streptomycin), in a humidified chamber at 37 °C and 5% CO₂. To obtain fully differentiated myotubes, the medium was exchanged for DMEM, high glucose with 2% horse serum, glutamine, and antibiotics, when the myoblasts reached confluence. The medium was replaced every 48 hours, and the cells were maintained in this medium until fully differentiated, when the medium was exchanged for DMEM, low glucose (5 mM) with 2% horse serum. The myotubes were fully formed by the fourth day post-induction. The experiments analyzing the impact of each experimental material on insulin signaling in the presence of palmitate-induced insulin resistance were performed as described previously [26]. After serum starvation, cells were treated overnight with PMI-5011, KOE, enriched DMC-2/DMC-1 fraction, or synthetic DMC-2, all at 10 μg/ml, dissolved in DMSO as indicated in the text or figures. The myotubes were also treated FK-506 (10 μM, Cayman), STO-609 (10 μg/ml, Fisher Scientific), 5-Aminoimidazole-4-carboxamide ribonucleotide (AICAR, 0.5 mM, Sigma), metformin (2 mM, Sigma), MG132 (10 μM, Boston Biochem), or radicicol (1 μM and 5 μM) as indicated. Time course experiments in the myotubes were carried out under basal conditions by exchanging the media at day 4 post-induction to DMEM, low glucose with 0.3% fatty acid free BSA at time zero (6 AM) along with addition of PMI-5011, DMC-2 or KOE and each treatment as indicated in the figures.

For subcellular fractionation, C2C12 cell pellets were harvested in hypotonic buffer (20 mM Tris-Cl, pH 7.4 with 10 mM NaCl, 3 mM MgCl2) and incubated on ice for 15 minutes prior to adding 10% Igepal to a final concentration of 0.5% Igepal. The lysate was briefly vortexed followed by centrifuging for 10 minutes at 800 x g at 4 °C. The resulting supernatant was saved as the cytoplasmic fraction and the nuclear pellet was resuspended in 20 mM Tris-Cl, pH 7.4, 420 mM NaCl, 1.5 mM MgCl₂, 0.2 mM EDTA and 25% glycerol.

For siRNA transfection, C2C12 cell culture was carried out in a 24-well plate format. At 100% confluence, the growth medium was replaced by differentiation medium with 2% horse serum. After 3 days, fully differentiated myotubes were transfected with AMPKα1 siRNA (sc-29674, Santa Cruz) or AMPKα2 siRNA (sc-38924, Santa Cruz) using Dharmafect Duo Transfection Reagent (Thermo Scientific, Cat# T-2010-02) according to the manufacturer’s instruction. Myotubes were harvested at 24 or 48 hours after the transfection. Sixteen hours prior to harvesting the cells, the vehicle (DMSO) or PMI-5011 (10 μg/ml) was added to the differentiation medium.

2.4. Protein Expression Analysis

Skeletal muscle and liver tissue lysates were prepared from powdered tissue by homogenization in 25 mM HEPES, pH 7.4, 1% Igepal CA630, 137 mM NaCl, with 10 mM Na₄P₂O₇,100 mM NaF, 2 mM Na₃VO₄1 mM PMSF, 10 μg/mL aprotinin, 1 μg/mL pepstatin, and 5 μg/mL leupeptin. C2C12 myotube lysates were prepared by sonication in 20 mM Tris pH 7.4, 1% Igepal CA630, 5mM EDTA with 10 mM Na₄P₂O₇,100 mM NaF, 2 mM Na₃VO₄, 1 mM PMSF, 10 μg/ml aprotinin, 1 μg/ml pepstatin, and 5 μg/ml leupeptin. The tissue and cell lysates were centrifuged at 14,000 x g for 10 minutes at 4 °C. Protein concentrations were determined using BCA assays (Thermo Fisher Scientific, Rockford, IL, USA) and the resulting supernatants (50 μg) were resolved by 10% SDS-PAGE and subjected to immunoblotting (Supplemental Table S1) (https://figshare.com/s/b723288ab24c26e1af1a)). Nitrocellulose membranes were incubated with antibodies for 1–2 hours at RT or overnight at 4 °C. Image J software was used for quantification of the western blot band intensities in Figure 2.

Fig. 2.

PMI-5011 is more efficient in activating AMPK pathway in muscle cells compared to AICAR and metformin. A: Phosphorylation of AMPK at Thr172 and protein expression of AMPKα1 and AMPKα2 in C2C12 muscle cells treated with PMI-5011 (10 μg/ml), AICAR (0.5 mM), metformin (2mM) and DMSO as vehicle for up to 16 hours. Tubulin is included as a loading control. B: Comparison of the kinetics of phosphorylation of AMPK at Thr172 in C2C12 muscle cells treated with PMI-5011, AICAR and metformin. Image J software was used for densitometry quantification of the immunoblots in (A). Phosphorylation levels of AMPK at Thr172 were evaluated as ratios of phospho-AMPK to total AMPK, where AMPKα1 and AMPKα2 was combined to calculate total AMPK. The values at 0-time point for each treatment were set to 1 and fold change for each time points was calculated as fold over 0-time for each treatment. C: Phosphorylation levels of AMPK at Thr172 in C2C12 muscle cells treated with PMI-5011, AICAR and metformin at the 16-hr time point. The phosphorylation levels were determined as described in (B) using immunoblot images in (A). Results shown are representative of three independent experiments. Data are means and Std dev; *p < 0.05 significance for treatments with AICAR or metformin vs PMI-5011.

2.5. Statistical analysis

Data are presented as the mean and standard deviation of results from at least three independent experiments. The Student's t-test was used to determine the significance of differences between two groups. Differences with a p value < 0.05 were regarded as statistically significant and denoted as indicated. Statistical analysis was conducted and all graphs were generated using GraphPad Prism 8.4.

3. Results

3.1. PMI-5011 is a complex mixture of bioactive phytochemicals.

DMC-2 (and likely DMC-1) is the major phytochemical constituent responsible for the enhanced insulin signaling activity in skeletal muscle and the glucose lowering effect of the PMI-5011 extract. However, DMC-2 does not account for all the detectable biological activities present in the extract. This is typical of botanical extracts containing multiple phytochemicals that have the potential to act on multiple pharmacological targets, where each compound could potentially affect different cellular pathways independently, or in combination, to impact cellular function [36]. Moreover, several of the identified bioactive compounds in PMI-5011 are present in skeletal muscle cell lysates along with DMC-2 [28].

A DMC-2 depleted “knock-out extract” (KOE) was generated from PMI-5011 by selectively removing DMC-2 and DMC-1, which only comprise approximately 1.8% and 0.8% of PMI-5011 by weight, respectively. Elemicin and iso-elemicin, volatile methoxyallylbenzenes of Russian tarragon and PMI-5011 are significantly decreased, but not eliminated, from the KOE due to solvent evaporation during processing [28]. Otherwise, the KOE contained all of the other components of PMI-5011, including davidigenin, 6-demethoxycapillarisin and sakuranetin at essentially the same concentrations as detected in PMI-5011 (Table 1).

3.2. PMI-5011 upregulates AMPK-signaling in skeletal muscle, but not in the liver of obese mice.

Previous studies revealed that PMI-5011 supplementation increased mitochondrial fatty acid oxidation and reduced intracellular triglyceride levels in skeletal muscle of diet-induced obese male and female mice [26, 30]. Moreover, experiments in vitro using skeletal muscle homogenates from high fat fed male and female mice indicated PMI-5011 supplementation enhanced the ability of skeletal muscle to switch between fat and carbohydrate oxidation, an indicator of metabolic flexibility [26, 30]. Based on these results, and given that AMPK is a key metabolic sensor, we tested if PMI-5011 could activate AMPK-signaling in skeletal muscle.

Activation of AMPK was evaluated by phosphorylation at Thr172 on the α-catalytic subunit of AMPK, and PMI-5011 supplementation enhanced AMPK phosphorylation in skeletal muscle in high fat-fed obese mice compared to vehicle-supplemented control mice (Fig. 1A). Notably, the increase in AMPK phosphorylation in skeletal muscle tissue from mice supplemented with KOE only, a PMI-5011 extract lacking DMC-1 and DMC-2, was comparable to the level in mice with PMI-5011 supplementation. However, AMPK phosphorylation in skeletal muscle was not changed from vehicle (Labrasol^®) gavage when the mice were supplemented with enriched DMC-2, a fraction of PMI-5011 containing only DMC-1 and DMC-2, or synthetic DMC-2 (at 100 μg/kg or 300 μg/kg body weight). Interestingly, the increase in AMPK phosphorylation induced by PMI-5011, and the effect invoked by the KOE was similar to the phosphorylation level of AMPK by metformin, a well-known and clinically important AMPK-activator.

Fig. 1.

PMI-5011 activates AMPK-signaling pathway in C2C12 muscle cells and skeletal muscle tissue, but not in liver of mice fed HFD. A-B: Phosphorylation of AMPK at Thr172 and protein expression of AMPKα1 and AMPKα2 in skeletal muscle (A) and liver (B) of mice administered PMI-5011 (500 mg/kg), knockout extract, KOE (500 mg/kg), enriched DMC-2 (100 mg/kg), synthetic DMC-2 at 100 or 300 mg/kg and metformin (300 mg/kg) via gavage (n = 5-7 animals per group). C: Phosphorylation of AMPK at Thr172 and protein expression of AMPKα1 and AMPKα2 in C2C12 myotubes treated with PMI-5011 (10 μg/ml) and DMSO as vehicle for up to 4 hours. D: Phosphorylation of AMPK at Thr172 and AKT at Ser473, and protein expression of AMPKα1, AMPKα2 and total AKT in C2C12 myotubes treated with PMI-5011 (10 μg/ml), synthetic DMC-2 (10 μg/ml), KOE (10 μg/ml) and DMSO as vehicle for up to 16 hours under basal conditions. E: Phosphorylation of AMPK at Thr172 and AKT at Ser473, and protein expression of AMPKα1, AMPKα2 and total AKT in C2C12 myotubes. The myotubes were treated with PMI-5011 (10 μg/ml), KOE (10 μg/ml), enriched DMC-2 (10 μg/ml), synthetic DMC-2 (10 μg/ml), and DMSO as vehicle in the presence of palmitate (200 μM) for 16 hours followed by a 10 minutes insulin (200 nM) treatment prior to harvesting cells. Tubulin or β-actin are included as a loading control. Immunoblots shown are representative of three or more independent experiments.

Previous reports showed that PMI-5011 supplementation significantly reduced liver triglyceride content in high fat-fed male mice, but not in high fat-fed female mice [26, 30]. For that reason, this study also examined the effect of PMI-5011 supplementation on hepatic AMPK activation. Notably, phosphorylation of AMPK at Thr172 in liver was not changed by treatment with PMI-5011, the KOE, the enriched DMC-2 fraction or the synthetic DMC-2 (Fig. 1B).

Taken together, the data demonstrate that PMI-5011 stimulates phosphorylation of AMPK at Thr172 in skeletal muscle, but not in the liver of high fat-fed obese mice. Also, the KOE appears to mediate PMI-5011 action on activating AMPK in skeletal muscle, while enriched DMC-2 fraction has no effect on phosphorylation of AMPK, at least under basal condition.

3.3. PMI-5011 increases AMPK-signaling in muscle cells.

To understand the mechanism of PMI-5011 action on AMPK signaling pathway in skeletal muscle, we used the murine C2C12 skeletal muscle cell culture model. First, we determined whether PMI-5011 treatment enhances activity of AMPK pathway in C2C12 myotubes. Notably, phosphorylation of AMPK at Thr172 was increased in myotubes treated with PMI-5011 for up to 4 hours (Fig. 1C). The AMPK activating effect of PMI-5011 and KOE occurred within 1 hour and was sustained in myotubes exposed to PMI-5011 or KOE for up to 16 hours (Fig. 1D). This was consistent with skeletal muscle from the obese mice (Fig. 1A). In agreement with the previous data [28], phosphorylation of AKT was preserved in myotubes exposed to enriched DMC-2, however, it was not observed in KOE-treated myotubes (Fig. 1D). Interestingly, increased phosphorylation of AMPK was observed in DMC-2-treated myotubes after 16 hours exposure to DMC-2, a time point when DMC-2-mediated AKT phosphorylation is diminished under basal conditions. Given the rhythmic response of cells in culture to serum shock, it is possible that variations in AMPK and AKT phosphorylation over time in the myotubes also reflect an impact of each treatment on circadian patterns of protein modifications (also see Supplementary Fig. 1) [37].

To mimic lipid-induced insulin resistant condition in vitro, C2C12 myotubes were treated with palmitate for 16 hours, followed by insulin for 10 minutes prior to cell collection. Consistent with the in vivo data, phosphorylation of AMPK was increased in myotubes treated with PMI-5011 and KOE in the presence of insulin and palmitate (Fig. 1E). A slight increase in phosphorylation of AMPK was observed in myotubes treated with enriched DMC-2 and synthetic DMC-2 after 16 hours, corresponding to the results in Figure 1D. In agreement with earlier findings [28], PMI-5011 increased insulin stimulated AKT phosphorylation and this effect was mediated by enriched DMC-2 and synthetic DMC-2, whereas KOE alone did not change insulin-stimulated AKT phosphorylation in myotubes (Fig. 1E).

Next, the effect of PMI-5011 on AMPK activation in myotubes was compared to the other well-known AMPK activators such as AICAR and metformin (Fig. 2A). Activation of AMPK was evaluated as a ratio of phosphorylated portion of AMPK at Thr172 on α subunits to total AMPK α subunits, where protein expression of α1 and α2 subunits were combined and presented as total AMPK (Fig. 2B). The effect of PMI-5011 on AMPK activation appeared very early and the 6-fold increase in phosphorylation of AMPK was observed in myotubes as early as 1 hour of exposure to PMI-5011 and continued to increase (Fig. 2A and 2B). However, increased phosphorylation of AMPK by AICAR and metformin was delayed compared to PMI-5011, and the increase was only 2.4- and 1.4-fold at 1 hour of the treatment, respectively. In myotubes exposed to AMPK-activators for 16 hours, the level of AMPK-phosphorylation by PMI-5011 was significantly higher compared to the levels achieved by AICAR and metformin (Fig. 2C).

Thus, the present outcomes demonstrate that the positive effects of PMI-5011 on activation of AMPK observed in skeletal muscle tissue are reproducible in the muscle cell model. Specifically, compounds remaining in the KOE mediate PMI-5011’s effect on activating AMPK signaling, while the major marker compound, DMC-2, mediates PMI-5011’s effect on insulin-stimulated activation of AKT in skeletal muscle cells as previously shown [28]. Moreover, in skeletal muscle cells, PMI-5011 appears to be a more efficient AMPK-activator than AICAR or metformin. The ability to distinguish mechanistic effects arising from a major phytoconstituent from those caused by other (minor) components of a botanical extract significantly advanced the understanding of botanical pharmacology.

3.4. LKB1 kinase mediates PMI-5011 effect on activation of AMPK-pathway.

Two upstream kinases, liver kinase B1 (LKB1) and Calcium/calmodulin-dependent kinase kinase beta (CaMKKβ) phosphorylate Thr172 of the α subunits of AMPK [1]. To determine whether these upstream kinases are involved in action of PMI-5011 on enhancing the phosphorylation of AMPK, we used C2C12 myotubes. LKB1 activity was evaluated by its phosphorylation at Ser428, which is required for nuclear export of LKB to the cytoplasm [38, 39] where it interacts with AMPK. PMI-5011 treatment increased the phosphorylation of LKB1 in myotubes and the increase was apparent after 1-hour exposure to PMI-5011 (Fig. 3A, Supplemental Fig. 2), which was consistent with the phosphorylation profile of AMPK in myotubes treated with PMI-5011 (Fig. 2B). Phosphorylation of LKB1 in the myotubes was increased with prolonged exposure to enriched DMC-2 (Fig. 3B). This data suggests that other phytochemicals present in PMI-5011 contribute to phosphorylation of LKB1 in myotubes observed at the earlier time points with PMI-5011.

Fig. 3.

AMPK activation by PMI-5011 involves LKB1, but not CaMKKβ-dependent regulation. A-B: Phosphorylation of LKB1 at Ser428 and protein expression of LKB1 in C2C12 myotubes treated with PMI-5011 (10 μg/ml) (A) and synthetic DMC-2 (10 μg/ml) (B) or DMSO as the vehicle control for up to 16 hours. C: Phosphorylation of LB1 at Ser428 and protein expression of LKB1 in skeletal muscle of mice administered PMI-5011 (500 mg/kg), knockout extract - KOE (500 mg/kg), enriched DMC-2 (100 mg/kg), synthetic DMC-2 at 100 or 300 mg/kg, and metformin (300 mg/kg body weight) via gavage (n = 5-7 animals per group). Tubulin or β-actin are included as a loading control. D: Phosphorylation of AMPK at Thr172 and protein expression of AMPKα1 and AMPKα2 in C2C12 myotubes treated with PMI-5011 (10 μg/ml), synthetic DMC-2 (10 μg/ml) and KOE (10 μg/ml) in the presence of FK-506 (10 μM) or STO-609 (10 μg/ml) for 16 hours. E. Phosphorylation and total protein levels of AMPKα and LKB in myotubes treated with PMI-5011 (10 μg/ml), synthetic DMC-2 (10 μg/ml) and KOE (10 μg/ml) for 16 hours in the absence or presence of a 24 hour pretreatment with radicicol (1μM and 5 μM). Results shown are representative of three or more independent experiments.

To determine if KOE activates LKB1, phosphorylation of LKB1 at Ser428 was examined in skeletal muscle tissue. In agreement with the myotubes, LKB1-phosphorylation increased in skeletal muscle when mice were administered PMI-5011 total extract compared to vehicle gavaged mice (Fig. 3C). Phosphorylation of LKB1 Ser428 was markedly increased in skeletal muscle of KOE-treated mice and this level was comparable to the level achieved by administering metformin. Although a modest increase in phosphorylation of LKB1 was detected in skeletal muscle tissue when mice were treated with the enriched DMC-2 fraction, the levels of LKB1-phosphorylation observed with synthetic DMC-2 treatment were not different from the level detected in the skeletal muscle tissue of vehicle-treated mice. As a separate pathway to activate AMPK, CaMKKβ binds to AMPK through a direct interaction of their kinase domains [40]. FK-506 and STO-609 are small molecules that selectively inhibit CaMKKβ by blocking that interaction [41, 42]. Notably, the presence or absence of either CaMKKβ inhibitor did not change the phosphorylation profiles of AMPK in myotubes treated with PMI-5011 or its KOE and enriched DMC-2 fractions (Fig. 3D). This result indicates CaMKKβ kinase is not involved in AMPK activation by PMI-5011, specifically by KOE. LKB-1 phosphorylation at serine 248 is robustly stimulated by KOE in the myotubes (Fig. 3E, no radicicol), but inhibition of LKB-1 phosphorylation by radicicol leads to a dose dependent increase in KOE-mediated phosphorylation of AMPKα (Fig. 3E).

Taken together, our data indicates LKB1, but not CaMKKβ is the upstream kinase that regulates phosphorylation of AMPK in myotubes and skeletal muscle in mice treated with PMI 5011 or KOE. Consistent with the AMPK phosphorylation data, PMI-5011 phytochemicals remaining in KOE are responsible for PMI-5011-mediated LKB1 activation. However, LKB-1 activation in response to KOE reduces the impact of KOE on AMPK phosphorylation, suggesting LKB-1 phosphorylation at serine 428 acts to inhibit KOE-mediated upregulation of AMPK activity.

3.5. PMI-5011 and KOE modulate downstream events of AMPK activation to link PMI-5011 and KOE to cellular metabolism in skeletal muscle.

AMPK activation increases catabolic activities to generate energy and decreases anabolic activities that consume energy, thus, AMPK integrates energy demands with cellular metabolic pathways. AMPK regulates fatty acid metabolism through acetyl-CoA carboxylase (ACC) that has two isoforms ACC1 and ACC2 [39]. ACC catalyzes the carboxylation of acetyl-CoA to produce malonyl-CoA, leading to increased fatty acid synthesis and inhibition of fatty acid oxidation [43]. AMPK inhibits ACC by directly phosphorylating both isoforms of ACC, thus, activation of AMPK reduces lipid synthesis and enhances fatty acid oxidation. To determine whether PMI-5011 modulates activity of ACC through AMPK-activation, we examined phosphorylation of ACC at Ser79, the AMPK target site. Notably, phosphorylation of ACC was enhanced in myotubes treated with PMI-5011 and as well as KOE in the presence of insulin and palmitate compared to the level of the phosphorylation in the control myotubes in the presence of insulin and palmitate (Fig. 4A). However, ACC phosphorylation was unaffected in myotubes treated with enriched DMC-2 and synthetic DMC-2. This pattern was repeated when ACC phosphorylation was examined in skeletal muscle (Fig. 4B). These data indicate that activation of AMPK by PMI-5011 and specifically by KOE inhibits ACC activity.

Fig. 4.

Downstream effectors of AMPK signaling are regulated by PMI-5011. A: Phosphorylation of ACC at Ser79 and protein expression of ACC in C2C12 myotubes treated with PMI-5011 (10 μg/ml), KOE (10 μg/ml), enriched DMC-2 (10 μg/ml), synthetic DMC-2 (10 μg/ml) in the presence of palmitate (200 μM) for 16 hours, followed by insulin (200 nM) for 10 minutes. B: Protein expression of p-ACC, ACC and SIRT1 in skeletal muscle of mice administered PMI-5011 (500 mg/kg), KOE (500 mg/kg), enriched DMC-2 (100 mg/kg), synthetic DMC-2 at 100 or 300 mg/kg and metformin (300 mg/kg) via gavage (n = 5-7 animals per group). Tubulin or β-actin is included as a loading control.

AMPK also enhances the activity of the NAD-dependent deacetylase Sirtuin 1 (SIRT1) by increasing cellular NAD⁺ levels or the NAD⁺/NADH ratio [44]. We previously demonstrated that PMI-5011 increases protein expression of SIRT1 in skeletal muscle but not in the liver of obese mice [26]. The current study found that protein expression of SIRT1 increased noticeably in skeletal muscle of mice supplemented with KOE, whereas SIRT1 expression in skeletal muscle of mice treated with enriched DMC-2 or synthetic DMC-2, was comparable to the level of SIRT1 in muscle tissue of vehicle-treated mice (Fig. 4B). Although, recent reports raised the possibility of metformin as a direct SIRT1-activating compound using computational modeling and cell-free assays [45], no increase in SIRT1 protein expression was observed in skeletal muscle of the metformin treated mice (Fig. 4B).

3.6. Both isoforms of alpha-subunit of AMPK are involved in PMI-5011 action in myotubes.

The AMPK α catalytic subunit is encoded by two genes (α1 and α2) that have tissue-specific expression patterns. The α1 isoform is ubiquitously expressed while the α2 subunit is highly expressed in skeletal and cardiac muscle, and at low levels in other tissues [46]. A previous report indicated that PMI-5011 supplementation did not affect gene expression of AMPKα1 and AMPKα2 in skeletal muscle of mice [26]. However, functional differences between two isoforms of catalytic subunits of AMPK have been reported. Particularly in skeletal muscle, AMPKα1 has been shown to control mTORC1 signaling, and thus controls muscle cell size, while AMPKα2 mediates muscle metabolic adaptation [47]. To determine whether PMI-5011 acts on AMPK-activation in an isoform-specific manner, we used siRNA to specifically deplete AMPKα1 or α2 in C2C12 myotubes. Protein expression of AMPKα1 and AMPKα2 in siRNA-treated myotubes was depleted after 24 hours and remained markedly reduced after 48 hours (Fig. 5A). Basal phosphorylation of AMPK with the nonspecific siRNA was observed at 24 hours, however, this level was lower at 48 hours. On the other hand, the PMI-5011 induced phosphorylation of AMPK was dramatically enhanced compared to the control even after 48 hours (Fig. 5B). In addition, PMI-5011 induced phosphorylation of LKB1 Ser428 was still observed at 48 hours. Interestingly, individually depleting α1 or α2 did not substantially affect AMPK phosphorylation, suggesting that both isoforms contribute to PMI-5011-mediated AMPK activation.

Fig. 5.

PMI-5011 acts on both AMPKα1 and AMPKα2 by enhancing their phosphorylation at Thr172 in the cytoplasm and nucleus. A: Phosphorylation of AMPK at Thr172 and protein expression of AMPKα1 and AMPKα2 in C2C12 myotubes in the presence of siRNA for AMPKα1 or AMPKα2 and non-targeting (NT) control. B: Phosphorylation of AMPK at Thr172 and protein expression of AMPKα1 and AMPKα2 in C2C12 myotubes treated with PMI-5011 (10 μg/ml) and DMSO as vehicle in the presence of siRNA for AMPKα1 or AMPKα2 and non-targeting (NT) control for 48 hours. Tubulin is included as a loading control. C. Cytoplasmic and nuclear AMPK phosphorylation at Thr172 is increased by PMI-5011. Tubulin is a marker of total lysate and the cytoplasmic fraction; histones mark the nuclear fraction. Results shown are representative of three independent experiments.

In mammalian cells, AMPK signaling is compartmentalized with activation of different pools of AMPK depending on various metabolic and stress conditions [48, 49]. The two α subunits of AMPK have a differential subcellular localization pattern. AMPKα1 is detected in the cytoplasm and nucleus, but AMPKα2 is localized in the nucleus [8, 50]. In the PMI-5011 treated C2C12 myotubes, phosphorylation of AMPK was observed in both cytoplasmic and nuclear fractions as well as total cell lysates (Fig. 5C). PMI-5011 did not enhance cytoplasmic AMPK phosphorylation in the presence of AMPKα1 alone, but substantially increased nuclear AMPK phosphorylation, indicating AMPKα2 is the predominant form of AMPK regulated by PMI-5011.

Reduced levels of AMPKα1 and α2 proteins, but not LKB1 in the myotubes treated with PMI-5011 (Fig. 5B), as well as evidence that AMPK protein stability is regulated by the ubiquitin-proteasome system [51, 52] prompted examination of whether degradation of the AMPKα subunits is proteasome dependent. The presence of the proteasome inhibitor MG132 partially preserved α1 and α2 proteins in myotubes treated with PMI-5011 or KOE (Fig. 6A and 6B). Interestingly, proteasome inhibition reduced AMPK phosphorylation at Thr172, indicating degradation of an unidentified inhibitor of AMPK activation is essential for AMPK activity. The impact of the inhibitor was overridden by PMI-5011 extract or the KOE. Contrary to the KOE-treated myotubes, neither AMPK-phosphorylation nor AMPKα1 and AMPKα2 protein expressions were changed in myotubes treated with enriched DMC-2 in the presence or absence of MG132 (Fig. 6C). In addition, changes in protein expression with PMI-5011 or KOE treatments specifically affected the α-subunits of AMPK but not LKB1 protein expression (Fig. 6A-C). Nevertheless, protein abundance of AMPKα1 and AMPKα2 in skeletal muscle of mice appears slightly lower with KOE-supplementation, and unchanged with enriched DMC-2 supplementation, consistent with PMI-5011-mediated regulation of AMPK activity independent of DMC-2.

Fig. 6.

PMI-5011 and KOE enhance phosphorylation of AMPK despite proteasome-dependent reduction of protein expression of AMPKα1 or AMPKα2 in muscle cells. A-C: Phosphorylation of AMPK at Thr172 and protein expression of AMPKα1, AMPKα2 and LKB1 in C2C12 muscle cells treated with PMI-5011 (10 μg/ml) (A), KOE (10 μg/ml) (B) and synthetic DMC-2 (10 μg/ml) (C) in the absence or presence of MG132 (10 μM) for up to 4 hours. Tubulin is included as a loading control. Results shown are representative of three independent experiments.

Collectively, these data demonstrate that compounds (other than DMC-2) in the KOE from PMI-5011 activate AMPK through phosphorylation of both α subunits in skeletal muscle. PMI-5011 and the KOE reduced protein abundance of AMPKα1 and AMPKα2 subunits while preserving AMPK phosphorylation. Although protein levels of both isoforms of AMPKα subunit are reduced by PMI-5011 and KOE in the C2C12 myotubes, AMPKα2 subunit protein levels are predominantly affected in the skeletal muscle tissue (Fig. 1A).

4. Discussion

Our previous studies demonstrated the anti-diabetic effect of the ethanol-based extract from Artemisia dracunculus L., termed PMI-5011 corresponded to enhanced insulin-AKT signaling in the obese insulin resistant state, resulting in improved glucose homeostasis [19, 21, 25, 29]. When 2’, 4’-dihydroxy-4-methoxydihydrochalcone (DMC-2) is selectively depleted from the PMI-5011 extract, the resulting “knock-out extract” (KOE) no longer promotes AKT activation in skeletal muscle, identifying DMC-2 as the compound responsible for AKT activation in skeletal muscle [28]. However, AS160 phosphorylation is enhanced in the KOE even though AKT phosphorylation is reduced, pointing to possible regulation of AMPK signaling by the KOE, given that AS160 is also a downstream target of AMPK [53]. This possibility is consistent with our earlier findings that PMI-5011 promotes fatty acid oxidation and improves metabolic flexibility in skeletal muscle of high fat-fed male and female mice [26, 30]. The current study advances knowledge by showing that compounds in the KOE robustly stimulate AMPK phosphorylation in skeletal muscle independent of AKT activation, leading to inactivation of ACC, a critical regulator of fatty acid oxidation. Thus, the complex mixture of phytochemicals in PMI-5011 affects both insulin signaling and energy sensing in skeletal muscle, integrating the reciprocal relationship between AKT and AMPK signaling in skeletal muscle (Supplemental Fig. S1 https://figshare.com/s/b723288ab24c26e1af1a) ) [54]. Moreover, KOE administration had no effect on hepatic AMPK activity, suggesting the effect is specific to skeletal muscle.

Examining two major upstream kinases that directly phosphorylate the AMPK alpha subunit at Thr172 led to the conclusion that the energy sensing AMP-dependent liver kinase B1 (LKB1) mediates the effect of PMI-5011 and KOE on AMPK activation. LKB1 links fatty acid uptake by CD36 to AMPK activation and fatty acid oxidation [55]. Previously results showed that expression of CD36 was increased in skeletal muscle tissue of female mice fed a diet with PMI-5011 [30]. Coupled with KOE-stimulated ACC phosphorylation and increased protein levels of the energy-sensing NAD⁺-dependent SIRT1 deacetylase, both substrates of AMPK that promote fatty acid oxidation when modified by AMPK [1, 44, 56], our results place KOE as regulating early signaling events that coordinate fatty acid uptake and oxidation (Fig. 7A).

Fig. 7.

Schematic presentation of PMI-5011 or KOE action on AMPK-signaling pathway in muscle cells. A: Main outcomes of AMPK activation by PMI-5011/KOE at the cellular level. Dashed line indicates proposed, but not shown in this study. B: Model of proposed mechanisms of activation of AMPK by PMI-5011/KOE; allosterically activating by direct binding to γ subunit of AMPK (mechanism 1), promoting Thr172-phosphorylation of α subunit of AMPK and activation of feedback by LKB1 (mechanism 2), or inhibiting Thr172-dephosphorylation of α subunit of AMPK possibly by blocking protein phosphatases (PP) access to Thr172 site (mechanism 3).

LKB1 also links energy sensing to protein synthesis via AMPK [57]. In our earlier studies, proteomic analysis in primary human muscle cells derived from skeletal muscle of obese, insulin resistant individuals revealed the mTOR pathway, and its downstream targets p70 S6K and eIF4 were differentially regulated by PMI-5011 [58]. Coupled with recent studies demonstrating inhibition of mTOR phosphorylation at serine 2448 by both PMI-5011 and KOE [28], the present data suggest PMI-5011 or KOE may reduce mTORC1 activity in myotubes (Fig. 7A).

The rapid and sustained activation of AMPK by PMI-5011 may be explained by several potential mechanisms underlying PMI-5011 KOE-mediated AMPK activation. LKB1 dependent phosphorylation of AMPK at Thr172 is stimulated by AMP binding to the AMPK γ-subunit [1], an event that is reproduced by AICAR as an AMP analog. The PMI-5011 KOE modulates AMPK Thr172 phosphorylation and AMPKα1 protein level in a pattern similar to that of AICAR, raising the possibility that one or more compounds in KOE directly bind the AMPKγ subunit (Fig. 7B, mechanism 1).

Alternatively, the strong stimulation of LKB1 phosphorylation at Ser428 by KOE supports a mechanism by which KOE mediates LKB phosphorylation and nuclear export either directly or indirectly to affect AMPK Thr172 phosphorylation. LKB-1 phosphorylation in response to KOE inhibits KOE-mediated AMPK activation, consistent with an indirect effect of KOE on LKB activation and a possible modulatory role for LKB-1 phosphorylation on AMPK activity. Although PMI-5011 regulates AMPK phosphorylation of the cytoplasmic and nuclear AMPK alpha subunits, our studies do not address whether PMI-5011 or KOE affect LKB1/AMPK interaction in either cellular compartment.

However, an interesting mechanism is suggested by metformin-mediated LKB1 phosphorylation at Ser428 via an atypical protein kinase C zeta (aPKCζ) [59] in the skeletal muscle of diabetic subjects [60]. The aPKCζ-dependent LKB1 Ser428 (399 short form) phosphorylation promotes LKB nuclear export [38] and both KOE and metformin stimulate LKB1 Ser428 phosphorylation in skeletal muscle, raising the possibility that KOE acts via aPKCζ to indirectly affect AMPK activity. Although KOE mirrors the effect of metformin on AMPK Thr172 phosphorylation, metformin does not alter AMPKα1 protein level, suggesting KOE has properties related to AMPK activation that are distinct from metformin (Fig. 7B, mechanism 2).

A third possible mechanism is related to the sustained effect of PMI-5011 and KOE on AMPK Thr172 phosphorylation even though both PMI-5011 and KOE stimulate proteasome-dependent degradation of the AMPK alpha subunits. The present findings revealed that AMPK Thr172 phosphorylation is inhibited when proteasome activity is blocked thus indicating that a factor responsible for AMPK Thr172 dephosphorylation is degraded by the proteasome. This effect is overridden by either PMI-5011 or KOE, consistent with our previous report showing a significant decrease in phosphatase activity in skeletal muscle of genetically obese mouse model supplemented with PMI-5011 [25] (Fig. 7B, mechanism 3).

The multiple possible mechanisms of action of both PMI-5011 and its KOE on AMPK activation can be attributed to the phytochemical composition characterizing PMI-5011 that is preserved in the KOE. Of the four phytochemicals present in the KOE that have known antidiabetic activity [22], sakuranetin and 6-demethoxycapillarisin are particularly interesting as a potential regulator of AMPK activity. Sakuranetin is a flavonoid derived from naringenin, a compound found to improve insulin sensitivity, skeletal muscle glucose uptake and mitochondrial function through AMPK phosphorylation [61-63] that may potentially bind AMPK [64]. 6-demethoxycapillarisin is related to coumarins, a class of phytochemicals that activate AMPK signaling via increased phosphorylation of Thr172 in skeletal muscle and adipocytes [65, 66]. Thus, both sakuranetin and 6-demethoxycapillarisin are candidate phytochemicals in both PMI-5011 and its KOE possibly involved in regulating activation of AMPK (Supplemental Table 2).

In summary, while DMC-2 in PMI-5011 improves glucose metabolism by enhancing insulin signaling and AKT activation in skeletal muscle, the compounds present in the KOE of PMI-5011 improve lipid metabolism in skeletal muscle by activating AMPK, a central regulator of energy homeostasis. The multi-target, “polypharmacological” paradigm of the complex mixture of phytochemicals found in PMI-5011 offers an attractive strategy for treating the combined dysregulation of glucose and lipid metabolism characteristic of obesity-related metabolic diseases.

Highlights:

• Phytochemicals present in Artemisia dracunculus regulate AMPK activity in skeletal muscle.

• A subset of phytochemicals are identified as responsible for activating AMPK in skeletal muscle.

• The mechanism of action of the phytochemicals involves regulation of LKB-1 activity.

Abbreviations:

DMC-1

2’,4-dihydroxy-4’-methoxydihydrochalcone

DMC-2

2’,4-dihydroxy-4-methoxydihydrochalcone

DESIGNER

Deplete and Enrich Select Ingredients to Generate Normalized Extract Resources

KOE

DMC-1 and DMC-2-depleted Knock-Out Extract

PMI-5011

ethanolic extract from Russian tarragon