Anti‐hyperlipidaemic effects of synthetic analogues of nordihydroguaiaretic acid in dyslipidaemic rats

Abstract

Background and Purpose

Previous studies have shown that Creosote bush‐derived nordihydroguaiaretic acid (NDGA) exerts beneficial actions on the key components of metabolic syndrome including dyslipidaemia, insulin resistance and hypertension in several relevant rodent models. Here, we synthesized and screened a total of 6 anti‐hyperlipidaemic analogues of NDGA and tested their efficacy against hepatic lipid metabolism in a high‐fructose diet (HFrD) fed dyslipidaemic rat model.

Experimental Approach

HFrD fed Sprague–Dawley rats treated with NDGA or one of the six analogues were used. Serum samples were analysed for blood metabolites, whereas liver samples were quantified for changes in various mRNA levels by real‐time RT‐PCR.

Key Results

Oral gavage of HFrD‐fed rats for 4 days with NDGA analogues 1 and 2 (100 mg·kg⁻¹·day⁻¹) suppressed the hepatic triglyceride content, whereas the NDGA analogues 2, 3 and 4, like NDGA, decreased the plasma triglyceride levels by 70–75%. qRT‐PCR measurements demonstrated that among NDGA analogues 1, 2, 4 and 5, analogue 4 was the most effective at inhibiting the mRNA levels of some key enzymes and transcription factors involved in lipogenesis. All four analogues almost equally inhibited the key genes involved in triglyceride synthesis and fatty acid elongation. Unlike NDGA, none of the analogues affected the genes of hepatic fatty acid oxidation or transport.

Conclusions and Implications

Our data suggest that NDGA analogues 1, 2, 4 and 5, particularly analogue 4, exert their anti‐hyperlipidaemic actions by negatively targeting genes of key enzymes and transcription factors involved in lipogenesis, triglyceride synthesis and fatty acid elongation. These analogues have therapeutic potential.

Abbreviations

Acaca

acetyl‐CoA carboxylase α

Acly

ATP citrate lyase

Acsl1

acyl‐CoA synthase long‐chain family member 1

Cd36

CD36 molecule

ChREBP

carbohydrate response element binding protein

Cpt1b

carnitine palmitoyltransferase

Dgat2

DAG‐O‐acyltransferase 2

DNL

de novo lipogenesis

Elovl1

ELOVL fatty acid elongase 1

Fads2

fatty acid desaturase 2

Fasn

fatty acid synthase

FFAs

free fatty acids

Gpam

glycerol‐3‐phosphate acyltransferase, mitochondrial

HFrD

high‐fructose diet

Mlxipl

MLXL interacting protein like

Mttp

microsomal triglyceride transfer protein

NAFLD

nonalcoholic fatty liver disease

NDGA

nordihydroguaiaretic acid

Pklr

pyruvate kinase, liver and RBC

Ppara

PPARα

Real‐time qRT‐PCR

real‐time quantitative RT‐PCR

Scd

stearoyl‐CoA desaturase (δ‐9‐desaturase)

Srebf1

sterol regulatory element binding transcription factor 1

SREBP‐1c

sterol responsive element binding protein‐1c

T2DM

type 2 diabetes mellitus

TG

triglyceride

VLDL‐TG

very LDL‐triglyceride

Introduction

Nonalcoholic fatty liver disease (NAFLD) has become the most common chronic liver disease in Western countries and other parts of the world (Loomba and Sanyal, 2013; Brunt et al., 2015; Hardy et al., 2016). It is characterized by the accumulation of triglyceride (TG) in the liver (hepatocytes) in the absence of excessive alcohol consumption (Hardy et al., 2016). NAFLD encompasses a spectrum of liver diseases from simple steatosis (triglyceride accumulation) to severe steatohepatitis (nonalcoholic steatohepatitis), cirrhosis, end‐stage liver failure and even hepatocellular carcinoma (Michelotti et al., 2013; Brunt et al., 2015; Hardy et al., 2016). Obesity and type 2 diabetes mellitus (T2DM) have led to a major increase in the prevalence of NAFLD globally, and a pathophysiological link between NAFLD with insulin resistance and metabolic syndrome has been indicated (Smith and Adams, 2011; Anstee et al., 2013; Yki‐Järvinen, 2014; Byrne and Targher, 2015; Hardy et al., 2016). NAFLD is also a risk factor for T2DM (Smith and Adams, 2011; Anstee et al., 2013) and cardiovascular disease (Bhatia et al., 2012; Anstee et al., 2013).

Hepatic steatosis, the first and the benign stage of NAFLD, results from an imbalance between lipid acquisition (i.e. fatty acid uptake and de novo lipogenesis) and removal [i.e. mitochondrial fatty acid oxidation and export as a component of very LDL‐triglyceride (VLDL‐TG) particles] (Musso et al., 2009; Kawano and Cohen, 2013; Hardy et al., 2016). Studies in both humans and rodents show that delivery of free fatty acids (FFAs) from adipose tissue is a major source of fat accumulation in NAFLD and insulin resistance (Lewis et al., 2002; Kawano and Cohen, 2013). Whole body lipolysis increases with increased total fat mass in humans (Fabbrini et al., 2008; Mittendorfer et al., 2009). Donnelly et al. (2005) demonstrated that approximately 60% of fat (TG) deposited in the liver of NAFLD patients is derived from adipose tissue‐supplied non‐esterified fatty acids in the plasma, further confirming the importance of fatty acid uptake in the pathogenesis of NAFLD. In agreement with this observation, several reports have shown that plasma FFA concentrations are elevated in NAFLD subjects due to increased release from adipose tissue depots (Holt et al., 2006; Fabbrini et al., 2008). Although under normal physiological conditions, de novo lipogenesis (DNL) contributes very little, relative to the esterification of fatty acids, to hepatic fat storage (Diraison and Beylot, 1998), its contribution to lipid accumulation increases five to sixfold (~5% vs. 25–30%) in NAFLD patients (Diraison et al., 2003; Donnelly et al., 2005; Lambert et al., 2014). Studies of rates of hepatic fatty acid oxidation in NAFLD have yielded mixed results. While some investigators reported impaired ATP production because of decreased fatty acid catabolism, others have reported increased rates of fatty acid oxidation in NAFLD (Kawano and Cohen, 2013).

No validated therapies for NAFLD currently exist (Barb et al., 2016). Previous studies from this laboratory have shown that nordihydroguaiaretic acid (NDGA, also called masoprocol), the active ingredient of the folk medicinal plant, Creosote bush (Arteaga et al., 2005), exerts profound effects on several components of the metabolic syndrome, including lowering blood glucose, FFAs, and TG levels and attenuating elevated BP in several rodent models of dyslipidaemia, insulin resistance and hypertension (Gowri et al., 1999a,b, 2000; Scribner et al., 2000; Zhang et al., 2013, 2015, 2016). Previously, we provided evidence that the beneficial actions of NDGA on dyslipidaemia and hepatic steatosis in ob/ob mice are exerted primarily through enhanced fatty acid oxidation via PPARα‐dependent pathways (Zhang et al., 2013). More recently, we reported that both acute and chronic treatment of high‐fructose diet (HFrD) fed dyslipidaemic rats ameliorates hypertriglyceridaemia and hepatic steatosis primarily by promoting increased channelling of fatty acids towards their β‐oxidation and interfering with lipogenesis (Zhang et al., 2015, 2016).

Given that hepatic steatosis in NAFLD primarily results from increased fatty acid delivery from adipose tissue and DNL and subsequent esterification into TG, we designed, synthesized and tested the molecular actions of six NDGA analogues with a high potency against hepatic DNL and fatty acid esterification into cellular TG. For these studies, we employed a HFrD fed rat model of hypertriglyceridaemia, steatosis and elevated hepatic lipogenesis (Zhang et al., 2015, 2016). The NDGA molecule contains two catechol groups. Under oxidative conditions, the catechol (benzene‐1,2‐diol) groups are known to undergo oxidation reactions that can lead to the formation of semiquinone radicals and benzoquinones causing toxicity (Schweigert et al., 2001). To prevent these oxidation reactions, we developed different NDGA analogues by modifying its catechol groups. Among these NDGA analogues, analogue 4 demonstrated excellent anti‐lipogenic properties in the liver and decreased circulating triglyceride levels when administered p.o. to HFrD rats. The other NGDA analogues also exhibited varying degrees of efficacy at inhibiting the key enzymatic steps involved in lipogenesis and TG synthesis.

Methods

Chemistry

Rationale for the synthesis of new NDGA analogues: Catechol groups in the molecule, by combining with heavy metal ions such as copper and iron, can cause toxicity by promoting oxidative DNA damage and DNA strand breaks. The reduction of Cu (II) by the catechol groups forms semiquinone radicals and Cu(I). These molecules reduce molecular oxygen into superoxide and hydrogen peroxide; the latter two can cause excessive oxidative stress and produce toxicity in humans. Here, we have synthesized six novel NDGA analogues with potentially low toxicity, evoked by protecting the catechol groups using the strategy presented in Figure 2.

Figure 2.

General synthetic routes to NDGA analogues.

Synthesis of NDGA analogue 1 [5,5′‐(2,3‐dimethylbutane‐1,4‐diyl)bis(benzo[d][1,3]dioxol‐2‐one)]; 2.5 g NDGA, (4,4′‐(2,3‐dimethylbutane‐1,4‐diyl)bis(benzene‐1,2‐diol)) and 6.7 g diphenyl carbonate were dissolved in 120 mL of diethyl ether and then 0.67 g (0.92 mL) triethylamine was added and the reaction mixture was stirred at room temperature for 2 days to obtain a precipitate. The precipitate was washed with diethyl ether and then dried under vacuum to obtain the NDGA analogue 1. ¹H NMR (400 MHz, CDCl3) δ (ppm): 0.85, (d, 6H, 2CH3), 1.75 (m, 2H, 2CH), 2.38 (q, 2H, 2CH (1/2CH2)), 2.84 (dd, 2H, 2CH (1/2CH2)), 6.97 (dd, 2H, 2CH (Ar)), 7.02 (d, 2H, 2CH (Ar)), 7.14(d, 2H, 2CH (Ar)). ESI‐MS m/z = 354.35; yield: 90.8%.

Synthesis of NDGA analogue 2 [N‐(2‐hydroxyethyl)‐7‐(4‐(2‐((2‐hydroxyethyl)carbamoyl)‐2,3‐dihydrobenzo[b][1,4]dioxin‐6‐yl)‐2,3‐dimethylbutyl)‐2,3‐dihydrobenzo[b][1,4]dioxine‐2‐carboxamide]; 5.4 g of analogue 5 (see below) was dissolved in 100 mL thionyl chloride and then refluxed overnight with stirring. The solvent was evaporated to obtain 6,6′‐(2,3‐dimethylbutane‐1,4‐diyl)bis(2,3‐dihydrobenzo[b][1,4]dioxine‐2‐carbonyl chloride) (5a) (yield: 100%); 5.8 g of compound 5a was dissolved in 50 mL of dichloromethane containing 20% 2‐aminoethanol, which was added slowly while stirring at 0°C; the reaction mixture was stirred for an additional 3 h at 0°C and then overnight at room temperature. The solvents were removed under vacuum in a rotary evaporator. The residue was dissolved in ethyl acetate, then washed twice each with dilute HCl (pH ≈ 4.5), and brine. The organic layer was separated, dried over sodium sulphate and then evaporated under vacuum. The crude product was purified by use of a flash purification column using 6% methanol in dichloromethane. The solvents were evaporated under vacuum to obtain analogue 2. ¹H NMR (400 MHz, CDCl3) δ (ppm): 0.82 (d, 6H, 2CH3), 1.68 (m, 2H, 2CH), 2.25 (m, 2H, 2CH (1/2CH2)), 2.66 (m, 2H, 2CH (1/2CH2)), 3.48 (m, 4H, 2CH2), 3.72 (m, 4H, 2CH2), 4.24 (m, 2H, 2CH (1/2CH2)), 4.52 (m, 2H, 2CH (1/2CH2)), 6.66 (dd, 4H, 4CH (Ar)), 6.82 (m, 2H, 2CH (Ar)), 7.04 (s, 2H, 2NH). ESI‐MS m/z = 528.64; yield: 78.0%.

Synthesis of NDGA analogue 3 [5,5′‐(2,3‐dimethylbutane‐1,4‐diyl)bis(2‐ethoxy‐2‐methylbenzo[d][1,3]dioxole)]; 2.5 g NDGA and 8.2 g 1,1,1‐triethoxyethane were dissolved in 80 mL of toluene and then 0.53 g of Amberlyst‐15 was added. The reaction mixture was stirred overnight at 100°C under reflux. The solvents were removed under a high‐vacuum rotary evaporator, and the residue was purified by use of a flash column with hexane : dichloromethane from 98/2 to 70/30 v/v. The solvents were evaporated and dried under vacuum to obtain the analogue 3. ¹H NMR (400 MHz, CDCl3) δ (ppm): 0.83, (m, 6H, 2CH3), 1.20 (t, 6H, 2CH3), 1.73 (m, 2H, 2CH), 2.26 (m, 2H, 2CH (1/2CH2)), 2.70 (m, 2H, 2CH (1/2CH2)), 3.60 (q, 4H, 2CH2), 6.60 (m, 4H, 4CH (Ar)), 6.70 (m, 2H, 2CH (Ar)). ESI‐MS m/z = 442.54; yield: 69.2%.

Synthesis of NDGA analogue 4 [5,5′‐(2,3‐dimethylbutane‐1,4‐diyl)bis(benzo[d][1,3]dioxole)]; 2.5 g NDGA and 4.42 g sodium hydroxide were dissolved in 117 mL of DMSO and 60 mL of dichloromethane. The reaction mixture was stirred overnight at 120°C with a reflux condenser and a moisture trap. The solvents were removed under high‐vacuum rotary evaporator. The residue was purified by use of a flash column with hexane : dichloromethane from 98/2 to 70/30 v/v. The solvents were evaporated and dried under vacuum to obtain the NDGA analogue 4. ¹H NMR (400 MHz, CDCl3) δ (ppm): 0.81 (d, 6H, 2CH3), 1.70 (m, 2H, 2CH), 2.24 (q, 2H, 2CH (1/2CH2)), 2.69 (dd(q), 2H, 2CH (1/2CH2)), 5.91 (s, 4H, 2CH2), 6.58 (q, 2H, 2CH (Ar)), 6.62 (d, 2H, 2CH (Ar)), 6.71 (d, 2H, 2CH (Ar)). ESI‐MS m/z = 326.39; yield: 73%.

Synthesis of NDGA analogue 5 [6,6′‐(2,3‐dimethylbutane‐1,4‐diyl)bis(2,3‐dihydrobenzo[b][1,4]dioxine‐2‐carboxylic acid)]; 14.5 g analogue 6 (see below) was dissolved in 140 mL methanol : water (120/20 v/v). To the mixture, 10 mL of 10 M NaOH was slowly added while stirring the mixture. After 18 h, the solvents were evaporated under vacuum and the residue was dissolved in ethyl acetate, washed successively with 0.5 N HCl, 0.1 N HCl and twice with brine. The organic layer was separated, dried over sodium sulphate and then evaporated under vacuum to obtain analogue 5. ¹H NMR (400 MHz, CDCl3) δ (ppm): 0.80, (d, 6H, 2CH3), 1.60 (m, 2H, 2CH), 2.25 (m, 2H, 2CH (1/2CH2)), 2.65 (m, 2H, 2CH (1/2CH2)), 4.40 (m, 4H, 2CH2), 4.85 (t, 2H, 2CH), 6.64 (m, 2H, 2CH (Ar)), 6.75 (m, 2H, 2CH (Ar)), 6.88 (m, 2H, 2CH (Ar)). ESI‐MS m/z = 442.40; yield: 98.7%.

Synthesis of NDGA analogue 6 [diethyl 6,6′‐(2,3‐dimethylbutane‐1,4‐diyl)bis(2,3‐dihydrobenzo[b][1,4]dioxine‐2‐carboxylate)]; 10 g NDGA, 25.8 g (=14.4 mL) ethyl 2,3‐dibromopropanoate and 25.6 g potassium carbonate were dissolved in 350 mL acetone. The reaction mixture was heated and refluxed with stirring for 18 h. The reaction was stopped, cooled to room temperature and the solvent mixture was evaporated under vacuum using a rotary evaporator. The residue was purified by flash column with ethyl acetate : dichloromethane : hexane with a ratio of 3/40/57/v/v/v. The solvents were evaporated and dried under vacuum to obtain an oily product, the analogue 6. ¹H NMR (400 MHz, CDCl3) δ (ppm): 0.78, (d, 6H, 2CH3), 1.22 (t, 6H, 2CH3), 1.70 (m, 2H, 2CH), 2.21 (m, 2H, 2CH (1/2CH2)), 2.67 (m, 2H, 2CH (1/2CH2)), 4.25 (q, 4H, 2CH2), 4.34 (m, 4H, 2CH2), 4.78 (m, 2H, 2CH), 6.63 (s, 2H, 2CH (Ar)), 6.76 (m, 2H, 2CH (Ar)), 6.89 (d, 2H, 2CH (Ar)). ESI‐MS m/z = 498.44; yield: 85.8%.

Animal studies

All animal experiments were performed according to the procedures approved by the VA Palo Alto Health Care System Institutional Animal Care and Use Committee. Animal studies are reported in compliance with the ARRIVE guidelines (Kilkenny et al., 2010). Male Sprague–Dawley rats (175–200 g) obtained from Harlan (Indianapolis, IN) were used in these studies. Rats were first maintained on a chow diet for approximately 1 week to allow acclimatization to a controlled new environment (25 ± 2°C, 55 ± 5% relative humidity with a 12 h‐light–dark cycle) and then switched to a HFrD diet (TD.89247; Harlan Teklad, Madison, WI) that provided 60% of total calories as fructose (HFrD rats). Four weeks after HFrD feeding, rats were fasted for 4 h and tail vein blood was collected for baseline measurement of serum TG, glucose, total cholesterol (TC), free cholesterol (FC), insulin and FFA. After 6 weeks of HFrD feeding, rats were treated with vehicle (0.5% carboxymethyl cellulose), NDGA (100 mg·kg⁻¹ BW), or one of the six NDGA analogues (100 mg·kg⁻¹ BW) (Figure 1) once a day for 4 days, delivered by oral gavage. Eight rats were studied in each treatment group. During treatment, animals were maintained on the HFrD. After 4 days of treatment, blood was collected from the tail vein 3 h after the last dose of vehicle, NDGA or NDGA analogue, and serum samples were analysed for TG, glucose, insulin, TC and FC. Subsequently, rats were killed and liver tissues collected, weighed and quickly frozen in liquid nitrogen. The samples were stored at −80°C until analysed for mRNA levels of key enzymes or proteins involved in hepatic lipid metabolism or hepatic TG content.

Figure 1.

Chemical structures of NDGA and its analogues.

Measurement of serum triglyceride, cholesterol, free fatty acids and glucose levels

Serum glucose, TG and TC levels were determined with commercial assay kits (Stanbio Laboratory, Boerne, TX). FFAs and FC were determined with commercial assay kits (Wako Diagnostics, Richmond, VA). All serum analyses were performed on eight rats in each treatment group.

Quantification of hepatic triglyceride content

Suitable aliquots of liver tissue homogenates were extracted with chloroform‐methanol by following the procedure of Folch et al. (1957), and extracted lipid samples were analysed for their TG content using the Stanbio enzymatic assay kit. All lipid analyses were performed with eight rats in each treatment group.

RNA isolation and real‐time RT‐PCR

RNA isolation from the liver tissue samples (~20 mg) was carried out using the RNeasy Plus mini kit (Qiagen, Valencia, CA) as per the manufacturer's instructions and was reverse‐transcribed (1 μg total RNA) using a High Capacity RNA‐to‐cDNA™ Kit (Life Technologies, Grand Island, NY). Real‐time PCR was performed in a final volume of 10 μL containing 50 ng of cDNA template and specific sets of forward and reverse primers using a FastStart Universal SYBR Green Master PCR Kit (Roche Applied Science, Indianapolis, IN) and an ABI Prism 7700 system (Applied Biosystems® Life Technologies) according to the manufacturer's protocols. The oligonucleotide primers and their sequences used in RT‐PCR are shown in Table 1. The amplification process times were 95°C for 2 min; 40 cycles at 95°C for 20 s, 60°C for 20 s, 72°C for 30 s and a final 5 min extension at 72°C. Results were normalized to the housekeeping gene 36B4, and the values of the control (chow‐fed) group were set to 1. Based on the SDs from our previous experiments of RT‐PCR for gene expression, we calculated that n = 4 for each group would give us a power of 90% to detect changes of 1.5‐fold with P < 0.05.

Table 1.

Primers used in quantitative real‐time RT‐PCR for detecting gene expression in rat liver tissues

Gene	Gene product (protein)	Primer sequences	GeneBank accession
Acaca	Acetyl‐CoA carboxylase α, ACC1)	Forward: 5′‐TGAGGAGGACCGCATTTATC‐3′ Reverse: 5′‐GCATGGAATGGCAGTAAGGT‐3′	NM_022193
Acsl1	Acyl‐CoA synthase long‐chain family member 1, ACS1, LACS1	Forward: 5′‐AACGATGTACGATGGCTTCC‐3′ Reverse: 5′‐CATATGGCTGGTTTGGCTTT‐3′	NM_012829
Acsl3	Acyl‐CoA synthase long‐chain family member 3, ACS3, LACS3	Forward: 5′‐GGGACTACAATACCGGCAGA‐3′ Reverse: 5′‐ATAGCCACCTTCCTCCCAGT‐3′	NM_057107
Acsl4	Acyl‐CoA synthase long‐chain family member 4, ACS4, LACS4	Forward: 5′‐AAATGCAGCCAAATGGAAAG‐3′ Reverse: 5′‐CACTCGGCAGTTCACTTCAA‐3′	NM_053623
Acsl5	Acyl‐CoA synthase long‐chain family member 5, ACS5, LACS5	Forward: 5′‐ATCTGCCTCCTGACATTTGG‐3′ Reverse: 5′‐GCTCCTCCCTCAATCCCTAC‐3′	NM_053607
Cd36	CD36 antigen, platelet glycoprotein IV, fatty acid translocase, FAT, SCARB3	Forward: 5′‐CCAGAACCCAGACAACCACT‐3′ Reverse: 5′‐CACAGGCTTTCCTTCTTTGC‐3′	NM_03156 XM_346685
Cpt1b	Carnitine palmitoyltransferase 1B (Muscle), CPT I, M‐CPT1	Forward: 5′‐GTCGCTTCTTCAAGGTTTGG‐3′ Reverse: 5′‐AAGAAAGCAGCACGTTCGAT‐3′	NM_013200
Dgat2	Diacylglycerol O‐acyltransferase 2, diglyceride acyltransferase 2, HMFN1045	Forward: 5′‐CTTCCTGGTGCTAGGAGTGG‐3′ Reverse: 5′‐GCCAGCCAGGTGAAGTAGAG‐3′	NM_001012345 XM_341887
Elovl1	ELOVL fatty acid elongase 1, Very‐long‐chain 3‐oxoacyl‐CoA synthase 1, CGI‐88	Forward: 5′‐CCCTACCTTTGGTGGAAGAA‐3′ Reverse: 5′‐TCCAGATGAGGTGGATGATG‐3′	NM_001044275 XM_001053380
Elovl2	ELOVL fatty acid elongase 2, Very‐long‐chain‐3‐oxoacyl‐CoA synthase 2, SSC2	Forward: 5′‐TTTGGCTGTCTCATCTTCCA‐3′ Reverse: 5′‐GGGAAACCGTTCTTCACTTC‐3′	NM_001109118 XM_574001
Elovl5	ELOVL fatty acid elongase 5, very‐long‐chain‐3‐oxoacyl‐CoA synthase 5, SCA38	Forward: 5′‐TACCACCATGCCACTATGCT‐3′ Reverse: 5′‐GACGTGGATGAAGCTGTTGA‐3′	NM_134382 XM_001075440
Elovl6	ELOVL fatty acid elongase 6, very‐long‐chain‐3‐oxoacyl‐CoA synthase 6, FAE	Forward: 5′‐CAACGGAACCTGTCAGCAA‐3′ Reverse: 5′‐GTGGTACCAGTGCAGGAAGA‐3′	NM_134383 XM_001075165
Fads2	Fatty acid desaturase 2, δ‐6 desaturase, FADSD6	Forward: 5′‐TGTCCACAAGTTTGTCATTGG‐3′ Reverse: 5′‐ACACGTGCAGGCTCTTTATG‐3′	NM_031344 XM_346570
Fasn	Fatty acid synthase, FAS	Forward: 5′‐GGATGTCAACAAGCCCAAGT‐3′ Reverse: 5′‐CAGAGGAGAAGGCCACAAAG‐3′	NM_017332
Gpam	Glycerol‐3‐phosphate acyltransferase, mitochondrial, GPAT, GPAT‐1	Forward: 5′‐GCCATCTTTGTCCACACCTT‐3′ Reverse: 5′‐CTCTCCGTCCTGGTGAGAAG‐3′	NM_017274
Lpin1	Lipin 1, PAP1, KIAA0188	Forward: 5′‐TCACTACCCAGTACCAGGGC‐3′ Reverse: 5′‐TGAGTCCAATCCTTTCCCAG‐3′	NM_001012111 XM_243142
Lpin2	Lipin 2, Lipin‐2	Forward: 5′‐TAGATGCAGACCCTGTTCCC‐3′ Reverse: 5′‐CTGGTGCTGGCTTCTTTTGT‐3′	NM_001108236 XM_237521
Lpin3	Lipin 3, Lipin‐3, SMP2	Forward: 5′‐AAAGACTGGACACACCAGGG‐3′ Reverse: 5′‐TGCTGGATATCACTCAGGCA‐3′	NM_001014184 XM_342566
Mlxipl	MLX interacting protein‐like, ChREBP, MONDOB	Forward: 5′‐TACTGTTCCCTGCCTGCTCT‐3′ Reverse: 5′‐GTCAGGATGCTGGTGGAAGT‐3′	NM_133552
Mttp	Microsomal triglyceride transfer protein, MTP, ABL	Forward:5′‐CCTCCATCCTGATGAAGAA‐3′ Reverse: 5′‐TCTCTGATGTCGTTGCTTGC‐3′	NM_001107727 XM_227765
Pklr	Pyruvate kinase, liver and RBC, PK1, PKL	Forward: 5′‐CTGGATGGGGCTGACTGTAT‐3′ Reverse: 5′‐GGCGTAGCTCCTCAAACAAC‐3′	NM_012624
Ppara	PPAR α, NR1C1, PPARα	Forward: 5′‐TCACACAATGCAATCCGTTT‐3′ Reverse: 5′‐GGCCTTGACCTTGTTCATGT‐3′	NM_013196
Scd	Stearoyl‐CoA desaturase, δ‐9‐desaturase, SCD1	Forward: 5′‐TGTTCGTCAGCACCTTCTTG‐3′ Reverse: 5′‐TCTTGTCGTAGGGGCGATAC‐3′	NM_AF509569
Srebf1	Sterol regulatory element binding transcription factor 1, SREBP‐1c, SREBP‐1, ADD‐1	Forward: 5′‐GGCCCTGTGTGTACTGGTCT‐3′ Reverse: 5′‐AGCATCAGAGGGAGTGAGGA‐3′	NM_001276707 XM_213329

Computational analysis of the interaction of NDGA with ACC and SCD1

Molecular docking of acetyl‐CoA carboxylase (ACC) and stearoyl‐CoA desaturase 1 (SCD1) proteins with NDGA and its six analogues were carried out to elucidate the interactions and preferential binding of NDGA and its analogues with the key enzymes involved in lipogenesis, triglyceride synthesis and fatty acid elongation. 3D structures of ACC and SCD1 proteins were used. The RCSB PDB (https://www.rcsbpdb.org) database was accessed for the retrieval of the 3D structure of the proteins (PDB_ID:2YL2 & 4ZYO). Before docking studies were performed, both protein and ligand molecules were prepared using protocols of AutoDock Tools 4.2. This protocol resolved the protein structures by adding missing atoms on incomplete residues, missing loop regions, alternate conformations (disorder), nonstandard atom names, incorrect protonation state of titratable residues, etc. All structures were energy minimized using MMFF94 force field, and the active site contact residues of the ACC and SCD1 proteins were recognized from earlier studies (Wang et al., 2015a,b). Molecular docking was performed using the AutoDock Vina (ADT) 1.5.6 software that uses soft‐core potentials and an optional grid representation to dock ligands into the receptor active site (Trott and Olson, 2010). Genetic algorithms form another class of well‐known stochastic methods that have been used in AutoDock (Morris et al., 1998). Random 10 orientations of NDGA and its six analogues were generated to improve the docking studies and finally refined by applying low, but most accurate, full potential as a refined pose minimization method. Analyses of the docked complexes were performed using Discovery Studio 4.5 visualizer.

Statistical analysis

All data are expressed as means ± SEM. Two‐way ANOVA was used to study differences among groups, followed by post hoc comparison using Bonferroni's multiple‐comparison test when necessary. Differences at P < 0.05 were considered statistically significant. Prism 7 software (GraphPad, La Jolla, CA) was used for all statistical calculations. A power calculation for RT‐PCR analyses showed n = 4 for an α = 0.05 and a power of 90%.

Materials

EDTA, chloroform‐d (CHCl₃ for ¹H NMR) and other laboratory chemicals were supplied by Sigma‐Aldrich (St. Louis, MO). Acetonitrile (for LC–MS) and all other reagents and solvents for the synthesis of NDGA analogues were obtained from Fisher Scientific (Pittsburgh, PA) and were used without further purification. Glucose, TG and TC measurement kits were from Stanbio Laboratory (Boerne, TX). FFA and FC measurement kits were from Wako Diagnostics (Richmond, VA).

Nomenclature of targets and ligands

Key protein targets and ligands in this article are hyperlinked to corresponding entries in http://www.guidetopharmacology.org, the common portal for data from the IUPHAR/BPS Guide to PHARMACOLOGY (Harding et al., 2018), and are permanently archived in the Concise Guide to PHARMACOLOGY 2017/18 (Alexander et al., 2017a,b).

Results

Because catechol groups in a molecule, by combining with heavy metal ions such as copper and iron, can cause toxicity by promoting oxidative DNA damage and DNA strand breaks, we sought to generate NDGA analogues with potentially low toxicity by protecting the catechol groups. We synthesized six analogues of NDGA with selective modification of aromatic groups of NDGA (Figure 1) and tested their lipid lowering efficacy in a HFrD Sprague–Dawley rat model of insulin resistance, dyslipidaemia and hypertension (Gowri et al., 1999a,b, 2000). NDGA analogues 1–6 were synthesized by the methods shown in Scheme 1 (Figure 2). NDGA analogue 1 [5,5′‐(2,3‐dimethylbutane‐1,4‐diyl)bis(benzo[d][1,3]dioxol‐2‐one)] was prepared by reacting NDGA with diphenyl carbonate in the presence of triethylamine and diethyl ether. Synthesis of NDGA analogue 2 [N‐(2‐hydroxyethyl)‐7‐(4‐(2‐((2‐hydroxyethyl)carbamoyl)‐2,3‐dihydrobenzo[b][1,4]dioxin‐6‐yl)‐2,3‐dimethylbutyl)‐2,3‐dihydrobenzo[b][1,4]dioxine‐2‐carboxamide] was accomplished by treating NDGA analogue 5 (see below) with thionyl chloride under reflux, followed by treatment with 2‐aminoethanol in dichloromethane. NDGA analogue 3 [5,5′‐(2,3‐dimethylbutane‐1,4‐diyl)bis(2‐ethoxy‐2‐methylbenzo[d][1,3]dioxole) was produced by treating NDGA with 1,1,1‐triethoxyethane and Amberlyst‐15 in toluene and refluxing the reaction mixture at 100°C with continuous stirring. NDGA analogue 4 [5,5′‐(2,3‐dimethylbutane‐1,4‐diyl)bis(benzo[d][1,3]dioxole)] was synthesized by refluxing an alkaline solution of NDGA in DMSO and dichloromethane at 120°C overnight. NDGA analogue 5 [6,6′‐(2,3‐dimethylbutane‐1,4‐diyl)bis(2,3‐dihydrobenzo[b][1,4]dioxine‐2‐carboxylic acid)] was synthesized by the hydrolysis of NDGA analogue 6 (see below) in the presence of an alkaline aqueous solution of methanol. NDGA analogue 6 [diethyl 6,6′‐(2,3‐dimethylbutane‐1,4‐diyl)bis(2,3‐dihydrobenzo[b][1,4]dioxine‐2‐carboxylate)] was synthesized by reacting NDGA with ethyl 2,3‐dibromopropanoate in the presence of potassium carbonate in acetone and refluxing the mixture for 18 h.

Table 2 compares the effects of oral gavage of HFrD fed rats with vehicle, NDGA or the NDGA analogues on serum TG, TC, FC, FFAs and glucose concentrations. Treatment of HFrD rats with NDGA, and NDGA analogue 2, 3 or 4 significantly decreased serum TG levels. In contrast, treatment with only NDGA analogue 4 reduced serum TC and FC levels, whereas neither NDGA nor any of the six NDGA analogues had any effect on serum levels of FFAs or glucose. NDGA analogues 1 and 2 also modestly decreased hepatic TG content (Table 3). Other NDGA analogues had no effect on hepatic TG content.

Table 2.

Plasma triglyceride, total cholesterol, free cholesterol, free fatty acid and glucose levels in Sprague Dawley rats fed a HFrD or a HFrD with NDGA, or one of the NDGA analogues, or vehicle p.o.

NDGA analogues	Triglyceride	Total cholesterol	Free cholesterol	Free fatty acids	Glucose
	(mM)	(mM)	(mM)	(mEq.L‐1)	(mM)
Control (CMC)	2.54 ± 0.58	1.95 ± 0.12	0.64 ± 0.07	1.55 ± 0.62	11.59 ± 0.93
NDGA	0.71 ± 0.11*	1.59 ± 0.07	0.47 ± 0.02	0.76 ± 0.28	10.56 ± 0.27
NDGA analogue 1	2.09 ± 0.37	1.5 ± 0.11	0.53 ± 0.04	0.68 ± 0.26	10.19 ± 1.38
NDGA analogue 2	1.15 ± 0.11*	1.64 ± 0.12	0.54 ± 0.04	1.16 ± 0.41	11.14 ± 0.85
NDGA analogue 3	1.09 ± 0.29*	1.72 ± 0.19	0.55 ± 0.06	1.05 ± 0.39	12.02 ± 1.33
NDGA analogue 4	0.61 ± 0.06*	1.44 ± 0.15*	0.43 ± 0.06*	0.90 ± 0.38	10.64 ± 1.46
NDGA analogue 5	1.75 ± 0.33	1.74 ± 0.17	0.57 ± 0.04	1.05 ± 0.44	10.24 ± 1.16
NDGA analogue 6	1.49 ± 0.25	1.97 ± 0.24	0.6 ± 0.06	0.97 ± 0.43	12.14 ± 0.5

Rats were maintained on a HFrD for 6 weeks and, subsequently, rats were divided into groups with 8 rats in each group: orally gavaged with NDGA or one of the six analogues at a dose of 100 mg·kg^‐1 BW a day for 4 days, or vehicle (control). Values are mean ± SEM.

^* P < 0.05 vs control (CMC)

Table 3.

Effect of NDGA administered p.o. on liver TG content in HFrD hypertriglyceridaemic rats

NDGA analogue	Liver TG content (μg·mg−1 liver tissue ± SEM)
Control (CMC)	4.78 ± 0.54
NDGA	5.82 ± 1.20
NDGA analogue 1	2.36 ± 0.12*
NDGA analogue 2	2.53 ± 0.35*
NDGA analogue 3	3.55 ± 0.24
NDGA analogue 4	4.62 ± 0.43
NDGA analogue 5	3.62 ± 0.53
NDGA analogue 6	5.69 ± 0.39

Rats were maintained on a HFrD for 6 weeks, and subsequently, rats were divided into groups with eight rats in each group: administered NDGA p.o. or one of the six analogues at a dose of 100 mg·kg⁻¹ BW a day for 4 days or vehicle (control). Liver samples were subsequently quantified for TG content. Values are mean ± SEM.

^* P < 0.05.

As noted above, both acute and chronic treatment of HFrD fed rats with NDGA has been shown to ameliorate hypertriglyceridemia and hepatic steatosis primarily by promoting increased channelling of fatty acids towards their β‐oxidation and interfering with lipogenesis (Zhang et al., 2015, 2016). Thus, we evaluated the relative effects of the six NDGA analogues and NDGA on mRNA levels of key enzymes/proteins and transcription factors involved in hepatic lipid metabolism, including fatty acid activation (Cooper et al., 2015), fatty acid uptake (Pepino et al., 2014), lipogenesis (Guillou et al., 2010; Strable and Ntambi, 2010; Wang et al., 2015a,b), TG synthesis (Takeuchi and Reue, 2009; Coleman and Mashek, 2011), lipogenic transcription factors, SREBP and carbohydrate response element binding protein (ChREBP) (Horton et al., 2002; Jump, 2011; Xu et al., 2013; Wang et al., 2015a,b), VLDL assembly (Cohen and Fisher, 2013) and fatty acid oxidation (Kersten, 2014), by real‐time qRT‐PCR; and results are presented in Figures 1, 2, 3, 4, 5, 6. Initially, we measured mRNA levels of acyl‐CoA synthase long‐chain family member 1, 3, 4 and 5 (Acsl1, Acsl3, Acsl4, Acsl5) enzymes, which catalyse the formation of a thioester between a CoA and a long‐chain fatty acid, resulting in the formation of long‐chain fatty acyl CoA (Cooper et al., 2015). Long‐chain acyl CoAs serve as metabolic intermediates in β‐oxidation, elongation of fatty acids and synthesis of phospholipids, as acyl donors in protein acylation and activators of enzymes and intracellular protein transport (Tomoda et al., 1991; Black and DiRusso, 2003; Cooper et al., 2015). In addition, they are also involved in the regulation of cell signalling, transcriptional control and many other metabolic pathways (Tomoda et al., 1991; Black and DiRusso, 2003; Cooper et al., 2015). Results presented in Figure 3 demonstrate that none of the six analogues had any significant effect on Acsl1, 3, 4, or 5 mRNA levels, whereas NDGA treatment only affected (stimulated) Acsl1 mRNA levels. Next, we evaluated mRNA levels of lipid transcription factors, SREBP‐1c (Srebf1) and ChREBP (Mlxipl), which regulate genes related to fatty acid and TG biosynthesis (lipogenesis). In addition, we measured mRNA levels of key genes whose enzyme protein products catalyse the formation of fatty acid and TG. SREBP‐1c is a bHLH‐LZ transcription factor that binds as dimers to the sterol regulatory element located within the promoter of some of the enzymes involved in fatty acid and TG synthesis, such as Srebf1, Fasn, Acaca, Acly, Scd, Gpam, Fads1, Fads2, Elovl2, Elovl5, Elovl6, Pklr and glucokinase (Horton et al., 2002; Ma et al., 2006; Jump, 2011; Wang et al., 2015a,b). ChREBP is another lipogenic/glucose metabolism transcription factor primarily responsive to glucose. It also contains a bHLH‐LZ domain and forms a heterodimeric complex with another bHLH‐LZ protein, Max‐like protein X (MLX), and binds to the carbohydrate response element located within the promoter of a number of genes encoding enzymes/proteins involved in glucose metabolism, such as Pklr, glycerol‐3‐phosphate dehydrogenase 1 (Gdp1), glucose transporter 4 (GLUT4), G6pc and glucokinase (hexokinase 4) regulator, and lipogenic enzymes, such as Fasn, Acaca, Acly, Scd and Elovl6 (Ma et al., 2006; Jump, 2011; Wang et al., 2015a,b). Figure 4A shows that treatment of HFrD fed rats with analogues 2, 4 or 5 significantly decreased mRNA levels of Srebf1 (the gene for SREBP‐1c), with analogue 4 being most effective. Analogue 6 or NDGA had no discernible effect on Srebf1 mRNA levels. Interestingly, none of the NDGA analogues or NDGA itself had any significant effect on mRNA levels of Mlxipl (the gene for ChREBP) (Figure 4B). Administration of analogues 1, 2, 4 or 5 caused significant reduction in Acaca mRNA levels, the gene that encodes cytosolic acetyl‐CoA carboxylase α (ACC1) (Figure 4C). Again, analogue 4 was most effective in lowering Acaca mRNA levels. ACC1 is a biotin containing multienzyme system, which catalyses the carboxylation of acetyl‐CoA to malonyl‐CoA, the rate limiting step in fatty acid biosynthesis (Strable and Ntambi, 2010; Wang et al., 2015a,b). In contrast, only analogue 4 caused a robust inhibition of Fasn (the gene for fatty acid synthase or FAS) mRNA content (Figure 4D). FAS catalyses the formation of long‐chain fatty acids from acetyl‐CoA, malonyl‐CoA and NADPH. This multifunctional protein consists of seven catalytic activities and acyl carrier protein.

Figure 3.

(A–D) Effect of acute NDGA or NDGA analogue treatment on the expression of hepatic Acsl1, Acsl3, Acsl4 and Acsl5 genes in HFrD fed hypertriglyceridaemic rats. Quantitative RT‐PCR of RNA from rats fed a HFrD for 6 weeks and subsequently treated with vehicle, NDGA or one of the six NDGA analogues at a dose of 100 mg·kg⁻¹ BW once a day for 4 days by oral gavage. (A) Acsl1 (acyl‐CoA synthase long‐chain family member 1, ACS1, LACS1), (B) Acsl3 (acyl‐CoA synthase long‐chain family member 3, ACS3, LACS3), (C) Acsl4 (acyl‐CoA synthase long‐chain family member 4, ACS4, LACS4) and (D) Acsl5 (acyl‐CoA synthase long‐chain family member 5, ACS5, LACS5). Results are mean ± SEM of three independent RT‐PCR experiments for each gene with four rats within each group. Data are presented relative to Rplp0 (36B4) in the same samples, and relative mRNA levels were determined by qRT‐PCR using the comparative C_T method. *P < 0.05.

Figure 4.

(A–D) Effect of acute NDGA or NDGA analogue treatment on the expression of enzyme genes and transcription factors involved in hepatic lipogenesis in HFrD fed hypertriglyceridaemic rats. Experimental conditions were the same as described under Figure 1. (A) Srebf1 (sterol regulatory element binding transcription factor 1, SREBP‐1, ADD‐1), (B) Mlxipl (MLX interacting protein like, ChREBP, MONDOB), (C) Acaca (acetyl‐CoA carboxylase α, ACC1) and (D) Fasn. Results are mean ± SEM of three independent RT‐PCR experiments for each gene with four rats within each group. Data are presented relative to Rplp0 (36B4) in the same samples, and relative mRNA levels were determined by qRT‐PCR using the comparative C_T method. *P < 0.05.

Figure 5.

(A–D) Effect of acute NDGA or NDGA analogue treatment on the expression of key enzyme genes involved in hepatic triglyceride synthesis in HFrD fed hypertriglyceridaemic rats. Experimental conditions were the same as described under Figure 1. (A) Gpam (glycerol‐3‐phosphate acyltransferase, mitochondrial GPAT, GPAT‐1), (B) Lpin1 (Lipin1, PAP1, KIAA0188), (C) Lpin2 (Lipin2, Lipin‐2, KIAA0249) and (D) Lpin3 (Lipin3, Lipin‐3, SMP2). Results are mean ± SEM of three independent RT‐PCR experiments for each gene with four rats within each group. Data are presented relative to Rplp0 (36B4) in the same samples, and relative mRNA levels were determined by qRT‐PCR using the comparative C_T method. *P < 0.05.

Figure 6.

(A–D) Effect of acute NDGA or NDGA analogue treatment on the expression of key enzyme genes involved in hepatic triglyceride synthesis and VLDL assembly in HFrD fed hypertriglyceridaemic rats. Experimental conditions were the same as described under Figure 1. (A) Dgat2 (DAG‐O‐acyltransferase 2, diglyceride acyltransaferase 2, HMFN1045), (B) Fads2 (fatty acid desaturase 2, δ‐6 desaturase, FADSD6), (C) Scd (stearoyl‐CoA desaturase, δ‐a‐desaturase, SCD1) and (D) Mttp (microsomal triglyceride transfer protein, MTP, ABL). Results are mean ± SEM of three independent RT‐PCR experiments for each gene with four rats within each group. Data are presented relative to Rplp0 (36B4) in the same samples, and relative mRNA levels were determined by qRT‐PCR using the comparative C_T method. *P < 0.05.

We next evaluated the effects on the hepatic mRNA expression of key enzymes involved in TG synthesis (Gpam/glycerol‐3‐phosphate acyltransferase, mitochondrial, Lpin1/Lipin1, Lpin2/Lipin2, Lpin3/Lipin3, Dgat2/DAG‐O‐acyltransferase 2), fatty acid desaturase (Fads2/fatty acid desaturase 2, Scd/stearoyl‐CoA desaturase), fatty acid elongase [Elovl1/fatty acid elongase 1 (Elovl1), Elovl2/fatty acid elongase 2 (Elovl2), Elovl5/fatty acid elongase 5 (Elovl5), Elovl6/fatty acid elongase 6 (Elovl 6)], and a protein involved in VLDL‐TG assembly (Mttp/microsomal triglyceride transfer protein) (Takeuchi and Reue, 2009; Strable and Ntambi, 2010; Coleman and Mashek, 2011; Jump, 2011; Cohen and Fisher, 2013; Pepino et al., 2014). Treatment of HFrD fed rats with NDGA analogues 1, 2, 4 and 5 decreased Gpam expressions, and analogue 4 was most effective in down‐regulating Gpam mRNA expression (>70%) (Figure 5A). None of the analogues or NDGA had any significant effect on Lpin1, Lpin2 or Lpin3 mRNA expression with the exception of analogue 4, which reduced Lpin2 expression by approximately 60% (Figure 5B–D). Levels of Dgat2 mRNA, whose enzyme protein product (DAG‐O‐acyltransferase 2) catalyses the final and only committed step in TG synthesis, were decreased between 30% and 60% in response to treatment with NDGA analogues 1, 2, 4 and 5 as well as NDGA (Figure 6A); again, analogue 4 was most effective. In contrast, Fads2 mRNA expression was not affected by any of the NDGA analogues or NDGA itself (Figure 6B). The enzyme product of Fads2 gene, fatty acid desaturase 2/∆⁶‐desaturase, catalyses the first and rate‐limiting step in the biosynthetic pathway of polyunsaturated fatty acids, which is the desaturation of linoleic acid (18:2n‐6) and α‐linolenic acid (18:3n‐3) into γ‐linolenic acid (18:3n‐6) and stearidonic acid (18:4n‐3), respectively, and other desaturation steps, leading to formation of long‐chain polyunsaturated fatty acids such as arachidonic acid (20:4n‐6), eicosapentaenoic acid (20:5n‐3) and docosahexaenoic acid (20:6n‐3). Results presented in Figure 6C show that NDGA and NDGA analogues 1, 2, 3, 4 and 5 inhibited Scd mRNA levels to a varying extent, with analogues 2 and 4 producing maximal effect. The Scd gene encodes stearoyl‐CoA desaturase, which catalyses a rate limiting step in the synthesis of unsaturated fatty acids, primarily the synthesis of oleic acid. Only analogue 2 significantly decreased hepatic mRNA expression of microsomal TG transfer protein (Mttp), a protein that facilitates the assembly of VLDL‐TG particles (Figure 6D).

Figure 7A–D shows effects on mRNA expression of some of the hepatic fatty acid elongases. The family of elongase enzymes consists of at least six members of condensing enzymes that, together with 3‐keto acyl‐CoA reductase, a dehydratase and trans‐2,3‐enoyl‐CoA reductase, carry out substrate‐specific elongation of fatty acids with different chain lengths and degrees of unsaturation (Jakobsson et al., 2006; Guillou et al., 2010). Administration of analogues 1, 2 and 4 significantly decreased mRNA content of Elovl2 (ELOVL fatty acid elongase 2) (Figure 7B). Likewise, administration of NDGA analogues 1, 2, 3, 4 or 5 or NDGA significantly reduced mRNA levels of Elovl5 (ELOVL fatty acid elongase 5) (Figure 7C). In contrast, none of the treatments produced any significant effect on mRNA levels of Elovl1 (ELOVL fatty acid elongase 1) or Elovl6 (ELOVL fatty acid elongase 6) (Figure 7A, D). We also examined the effects on mRNA levels of PPARα (Ppara), carnitine palmitoyltransferase 1b (Cpt1b) and CD36 (Cd36), the key molecules involved in fatty acid oxidation, as well as the glycolytic enzyme pyruvate kinase, liver and RBC (Pklr). Among these, PPARα is the major regulator of hepatic fatty acid oxidation and is also the target of fibrate drugs commonly used in the clinical management of dyslipidaemia (Kersten, 2014). Carnitine palmitoyltransferase 1b, a member of the carnitine/ChAT family, is the rate‐limiting enzyme of the long‐chain fatty acid β‐oxidation pathway in mitochondria. Cd36 encodes CD36 protein or fatty acid transporter, whose major function in liver is enhancing flux of fatty acids for their oxidation and TG accumulation (Kersten, 2014; Pepino et al., 2014). Pklr encodes pyruvate kinase, liver and RBC type, a glycolytic enzyme that catalyses the transphosphorylation from phosphoenolpyruvate to ADP, yielding pyruvate and ATP. Pyruvate enters the citric acid cycle to produce energy (ATP) and provide substrates (acetyl‐CoA/citrate) for lipogenesis. Results presented in Figure 8A demonstrate that NDGA analogues and NDGA did not significantly affect Ppara mRNA abundance, whereas only NDGA treatment increased mRNA expression of Cpt1b and Cd36 (Figure 8B,C). On the other hand, only analogue 4 and NDGA significantly lowered Pklr mRNA levels (Figure 8D).

Figure 7.

(A–D) Effect of acute NDGA or NDGA analogue treatment on the expression of elongases in HFrD fed hypertriglyceridaemic rats. Experimental conditions were same as described under Figure 1. (A) Elovl1 (ELOVL fatty acid elongase 1, very‐long‐chain‐3‐oxoacyl‐CoA synthase 1, CGI‐88), (B) Elovl2 (ELOVL fatty acid elongase 2, very‐long‐chain‐3‐oxoacyl‐CoA synthase 2, SSC2), (C) Elovl5 (ELOVL fatty acid elongase 5, very‐long‐chain‐3‐oxoacyl‐CoA synthase 5, SCA38) and (D) Elovl6 (ELOVL fatty acid elongase 6, very‐long‐chain3‐oxoacyl‐CoA synthase 6, FAE). Results are mean ± SE of three independent RT‐PCR experiments for each gene with four rats within each group. Data are presented relative to Rplp0 (36B4) in the same samples, and relative mRNA levels were determined by qRT‐PCR using the comparative C_T method. *P < 0.05.

Figure 8.

(A–D) Effect of acute NDGA or NDGA analogue treatment on the expression of enzyme genes and transcription factors involved in hepatic fatty acid oxidation in HFrD fed hypertriglyceridaemic rats. Experimental conditions were the same as described under Figure 1. (A) Ppara (PPARα, NR1C1), (B) Cpt1b (carnitine palmitoyltransferase 1b, CPTI, M‐CPT1), (C) Cd36 (CD36 molecule, CD36 antigen, fatty acid translocase, FAT, SCARB3) and (D) Pklr (pyruvate kinase, liver and RBC, PK1, PKL). Results are mean ± SEM of three independent RT‐PCR experiments for each gene with four rats within each group. Data are presented relative to Rplp0 (36B4) in the same samples, and relative mRNA levels were determined by qRT‐PCR using the comparative C_T method. *P < 0.05.

To check for molecular interactions of ACC and SCD1 with NDGA, analogue 1, 2, 3, 4, 5 and 6, we performed blind docking with NDGA and its analogues. After molecular docking, a total of 10 poses were generated with the best pose identified using highest binding energy. Calculated binding energies of these molecular complexes are shown in Table 4. Details of all molecular docking interactions of NDGA and its analogues are shown in Supporting Information Tables S1–S14. NDGA and its analogues showed binding energies ranging between −7.6 to −8.9 Kcal·mol⁻¹ with ACC and −7.4 to −10.2 Kcal·mol⁻¹ for SCD1. These interactions are governed by hydrogen and hydrophobic bonds, specifically conventional H‐bond, carbon H‐bond, Pi‐donor hydrogen bond, Pi‐alkyl, Pi‐Pi T‐shaped and amide‐Pi stacked. The analysis of the docked NDGA analogues with the ACC and SCD1 proteins also brought into focus some important interactions operating at the molecular level. The results of NDGA, NDGA1, NDGA4 and NDGA5 docking showed that the binding pocket involves amino acid residues LEU78, MET79, LEU82, THR100, TRP101, VAL105, TYR108, PHE146, GLN147, TRP153, ASN163, GLY164, TRP184, LEU185, TRP246, VAL248, LEU252, ARG253, VAL256, LEU258, THR261, ASN265, GLY355, ILE358, HIS394, SER409, HIS421, LYS423, PRO429, ALA430, GLU442, VAL446, LYS447, LYS450, GLU461, LEU463, LEU473, GLU474, ASN476, ARG478, GLN480, VAL481, PHE529, ALA533, PRO,536, PHE594 and HIS596. The important hydrogen bond forming amino acid residues were THR100, TRP101, GLN147, ASN163, GLY164, VAL248, THR261, ASN265, GLY355, SER409, LYS423, ALA430, GLU442, GLU461, ASN476, ARG478 and GLN480. To analyse the potential compound with desirable pharmacokinetic (absorption, distribution and excretion) and pharmacodynamic properties, Lipinski's Rule‐of‐Five (Ro5) (Lipinski et al., 1997) was applied to analyse NDGA and the analogues. Table 5 shows the MW, hydrogen bond acceptor, hydrogen bond donor and log P from the analysis.

Table 4.

Computed binding affinities of NDGA and its analogues with the enzymes (ACC and SCD1)

S. No.	Ligand	Binding affinity (kcal·mol−1) 2YL2 (ACC)	Binding affinity (kcal·mol−1) 4ZYO (SCD1)
1.	NDGA	−7.6	−7.7
2.	NDGA1	−8.2	−8.8
3.	NDGA2	−8.8	−9.8
4.	NDGA3	−7.6	−7.4
5.	NDGA4	−8.0	−9.3
6.	NDGA5	−8.9	−9.9
7.	NDGA6	−8.7	−10.2

Table 5.

Predicted drug‐likeness properties of potential compounds

Chemical_name	MW	HBA	HBD	Log_P
NDGA	302.26	4	4	4.71
NDGA analogue 1	354.38	6	0	5.22
NDGA analogue 2	528.60	10	4	2.56
NDGA analogue 3	442.54	6	0	8.06
NDGA analogue 4	326.39	4	0	5.21
NDGA analogue 5	442.40	8	2	4.48
NDGA analogue 6	458.44	8	0	5.63

HBA: hydrogen bond acceptor; HBD: hydrogen bond donor; Log P: Lipophilicity Lipinski's Rule‐of‐Five (Ro5) proposes that molecules with poor permeation and oral absorption have MW > 500, ClogP > 5, more than five hydrogen‐bond donor and more than 10 acceptor groups.

The observation of drug‐likeness is an attempt to optimize the selection and to choose the best hits to become leads using Lipinski's Rule‐of‐Five (Lipinski et al., 1997). This concept includes molecular features well‐matched with biological activity and desirable pharmacokinetic (absorption, distribution and excretion) and pharmacodynamic properties. These physicochemical properties complement pharmacophoric sites that do not affect its chemical functions and give them the potential for adequate pharmacology (Vistoli et al., 2008). Compounds with large numbers of hydrogen bond donors and acceptors are often rather insoluble in many solvents due to the formation of intermolecular hydrogen bonds in the crystal lattice. Data shown in Table 5 suggest that NDGA analogues possess drug like properties. It can be predicted that NDGA, analogue 1, 4 and 5 are smaller molecules and have good permeability and solubility across the cell membrane. Also, these synthesized compounds are likely to be orally active and did not violate more than one of the Lipinski Ro5.

Discussion

NDGA, a potent lipoxygenase and COX inhibitor and antioxidant, is a plant lignan commercially prepared from the leaves and twigs of shrub creosote bush, Larrea tridentate (Arteaga et al., 2005; Lü et al., 2010). Previous studies from this laboratory have shown that NDGA exerts profound effects on a number metabolic abnormalities including obesity, dyslipidaemia, insulin resistance, hypertension, hepatic steatosis and altered glucose metabolism in several relevant rodent models. NDGA has also been tested as a therapeutic agent for the treatment of breast, prostate, lung, oesophageal and skin cancers (Lü et al., 2010). It also functions as neuroprotective agent and possesses antiviral properties (Lü et al., 2010). Furthermore, it has been demonstrated that NDGA increases the lifespan of genetically heterogeneous male mice (Strong et al., 2016). Considering that NDGA exerts its modulatory effect on a large number of metabolic processes raises the possibility of selectively modifying the NDGA molecule to generate an NDGA derivative with a desired selectivity, specificity and efficacy.

The current study was initiated with the objective of synthesizing and characterizing NDGA analogues with high potency against hepatic DNL and fatty acid esterification into cellular TG. We identified four NDGA analogues, 1, 2, 4 and 5, which exhibited varying potencies against genes encoding lipogenic transcription factors and enzymes involved in lipogenesis and TG synthesis in a unique nutritionally induced model of dyslipidaemia and insulin resistance (high‐fructose fed rats). We found that NDGA analogue 4 in general was more effective than three other analogues (1, 2 and 5). Oral administration of NDGA analogue 4 not only decreased plasma TG levels (like NDGA) but also caused a significant reduction in plasma FC and TC levels. However, the observed decline in plasma TG levels in response to NDGA analogue 4 treatment was not accompanied by any significant reduction in hepatic TG content. In contrast, NDGA analogues 1, 2 and 3 differentially affected plasma and liver TG levels, whereas NDGA analogues 5 and 6 showed no effect on either of these parameters. These findings suggest that plasma and hepatic TG levels may be independently regulated.

To evaluate the effects of NDGA analogues on the key genes involved in hepatic lipogenesis and TG synthesis, we carried out qRT‐PCR measurements to assess the levels of two major lipogenic transcription factors, insulin‐regulated SREBP‐1c (Srebf1), and glucose‐sensitive ChREBP (Mlxipl). NDGA analogues 2, 4 and 5 selectively down‐regulated Srebf1 mRNA levels without impacting Mlxipl mRNA. Interestingly, NDGA analogue 4 produced the most pronounced inhibitory effect on Srebf1 mRNA expression. We also examined the effects of the six analogues on the mRNA expression of Acaca and Fasn, the downstream target genes of both SREBP‐1c and ChREBP (Wang et al., 2015a,b), which participate in DNL. Although NDGA analogues 1, 2, 4 and 5 significantly down‐regulated Acaca mRNA expression, only NDGA analogue 4 caused a robust reduction in Fasn mRNA levels. These data suggest one potential mechanism by which NDGA analogues, particularly NDGA analogue 4, inhibit lipogenesis is via inhibition of Srebf1. We also evaluated the effects of NDGA analogues on key genes (Gpam, Lpin1, Lpin2, Lipin3, Dgat2, Fads2 and Scd) involved in hepatic TG synthesis. Treatment of HFrD animals with NDGA analogues 1, 2, 4 or 5 specifically reduced the expression of Gpam, Dgat2 and Scd. We further observed that expression of Scd was also inhibited by NDGA analogue 3 and NDGA itself and that NDGA analogues 2 and 4 caused near complete inhibition of Scd mRNA expression. The inhibition of Dgat2 and Scd expression by NDGA analogues is noteworthy, given that enzyme/protein products of these two genes (DGAT2 and SCD) are critically involved in TG synthesis. It is well known that DGAT2 is an essential acyltransferase that catalyses the terminal and only committed step in TG synthesis in which DAG is covalently linked to long‐chain fatty‐acyl‐CoAs (Choi et al., 2007; Yen et al., 2008; Liu et al., 2012). It is abundantly expressed in liver, acts upstream of DGAT1 and utilizes nascent DAG and de novo synthesized fatty acids as substrates to synthesize triglyceride destined for VLDL‐TG assembly or intracellular storage, especially in the liver (steatosis) (Choi et al., 2007; Zammit, 2013; Futatsugi et al., 2015; Li et al., 2015). Likewise, SCD1 is a crucial enzyme for hepatic TG synthesis that catalyses the formation of major monounsaturated fatty acids, such as oleate and palmitoleate, required for the synthesis of TG (Ntambi et al., 2002). Interestingly, previous studies have provided evidence that DGAT2 (but not DGAT1), SCD1 and FATP1 (with acyl‐CoA synthase activity) exist in close proximity to each other in mammalian cells, presumably to facilitate efficient channelling of substrates within the endoplasmic reticulum for TG synthesis (Man et al., 2006; Choi et al., 2007). Taken together, our data indicate that NDGA analogues attenuate hypertriglyceridemia by (i) limiting the availability of de novo synthesized FFAs for TG synthesis through decreased SREBP1c‐mediated Acaca, Fasn, and Scd mRNA expression and (ii) suppression of DGAT2 gene.

In summary, we have designed and synthesized six anti‐hyperlipidaemic NDGA analogues (analogues 1–6) based on the selective modification of the aromatic groups of NDGA. We tested their efficacy against hepatic lipid metabolism in an HFrD‐fed rat model of dyslipidaemia, insulin resistance and hypertension. Oral gavage with NDGA analogues 1, 2, 4 and 5 differentially and to various extents attenuated both hepatic TG content as well as plasma TG levels. qRT‐PCR measurements demonstrated that among the NDGA analogues 1, 2, 4 and 5, analogue 4 was most effective in inhibiting the mRNA levels of some key enzymes and transcription factors enzymes involved in TG synthesis, all four analogues almost equally inhibited the genes such as Scd, Gpam and Dgat2 involved in TG synthesis and fatty acid elongation (Elovl 2 and Elovl 5). The mechanism of action is illustrated in Figure 9. Unlike NDGA, none of the NDGA analogues affected the genes of hepatic fatty acid oxidation or transport. More detailed pharmacological, molecular and biochemical characterization of NDGA analogue 4 using several dyslipidaemic rodent models is underway and will be described in future publications.

Figure 9.

Mechanism of action of NDGA analogue 4. NDGA analogue 4 exerts its anti‐hyperlipidaemic actions by negatively targeting genes of key enzymes and transcription factors involved in lipogenesis, triglyceride synthesis and fatty acid elongation. ACC (acetyl‐CoA carboxylase α), FAS, SCD, SREBP‐1c (sterol regulatory element binding transcription factor 1c), ChREBP (carbohydrate‐responsive element‐binding protein, MLX interacting protein like), TG.

Author contributions

M.S., S.B., Y.L., W.‐J.S., F.B.K. and S.A. participated in research design. M.S., S.B., Y.L., A.B., L.H. and Y.C. conducted experiments. M.I. and J.R. contributed new reagents and analytic tools. M.S., S.B., A.B., Z.A. and R.P. performed the data analysis. M.S., S.B., A.B., W.‐J.S., M.I., J.R., M.R.N., F.B.K. and S.A. wrote or contributed to the writing of the manuscript.

Conflict of interest

The authors declare no conflicts of interest.

Declaration of transparency and scientific rigour

This Declaration acknowledges that this paper adheres to the principles for transparent reporting and scientific rigour of preclinical research recommended by funding agencies, publishers and other organisations engaged with supporting research.

Supporting information

Table S1 Docking Analysis between Best Pose of NDGA with ACC Protein.

Table S2 Docking Analysis between Best Pose of NDGA1 with ACC Protein.

Table S3 Docking Analysis between Best Pose of NDGA2 with ACC Protein.

Table S4 Docking Analysis between Best Pose of NDGA3 with ACC Protein.

Table S5 Docking Analysis between Best Pose of NDGA4 with ACC Protein.

Table S6 Docking Analysis between Best Pose of NDGA5 with ACC Protein.

Table S7 Docking Analysis between Best Pose of NDGA6 with ACC Protein.

Table S8 Docking Analysis between Best Pose of NDGA with SCD1 Protein.

Table S9 Docking Analysis between Best Pose of NDGA1 with SCD1 Protein.

Table S10 Docking Analysis between Best Pose of NDGA2 with SCD1 Protein.

Table S11 Docking Analysis between Best Pose of NDGA3 with SCD1 Protein.

Table S12 Docking Analysis between Best Pose of NDGA4 with SCD1 Protein.

Table S13 Docking Analysis between Best Pose of NDGA5 with SCD1 Protein.

Table S14 Docking Analysis between Best Pose of NDGA6 with SCD1 Protein.

Singh, M. , Bittner, S. , Li, Y. , Bittner, A. , Han, L. , Cortez, Y. , Inayathullah, M. , Arif, Z. , Parthasarathi, R. , Rajadas, J. , Shen, W.‐J. , Nicolls, M. R. , Kraemer, F. B. , and Azhar, S. (2019) Anti‐hyperlipidaemic effects of synthetic analogues of nordihydroguaiaretic acid in dyslipidaemic rats. British Journal of Pharmacology, 176: 369–385. 10.1111/bph.14528.