Dihydrosterculic acid induces hepatic PPARα target gene expression in mice

Abstract

Abstract

Cottonseed oil (CSO) is a seed oil with a unique fatty acid composition and the ability to reduce lipid levels in humans and mice. The present study aimed to characterize the effects of dihydrosterculic acid (DHSA), a cyclopropyl fatty acid found in CSO, on lipid metabolism. First, male wild‐type mice were fed CSO‐ or isocaloric oil‐enriched diets (lacking DHSA) for 6 weeks. Tissues were analyzed via RNA‐sequencing which identified 45 differentially expressed genes within the CSO group, the majority of which are associated with lipid metabolic processes. Despite being a moderate‐fat diet, no changes in hepatic or plasma triglyceride were observed in the CSO group. Confirmational tissue analysis showed an increase in hepatic peroxisome proliferator‐activated receptor alpha (PPARα) and PPARα target gene expression in the CSO group compared to control groups, suggesting that DHSA effects may be mediated through increased PPARα transcriptional activity and fatty acid oxidation (FAO). To test this hypothesis, female PPARα knockout mice were fed a CSO‐enriched diet. In the absence of PPARα, the lipid‐lowering effect of the CSO diet was lost. Next, FAO was assessed in DHSA‐treated HepG2 cells by measuring mitochondrial respiration with long‐chain fatty acids and adenosine diphosphate substrates. Compared to the control, DHSA‐treated cells demonstrated a higher capacity to utilize FAO for energy production. Lastly, CSO‐fed mice exhibited significantly lower respiratory exchange ratio with an elevated energy expenditure (EE) compared to SFO‐fed mice. In total, these data suggest that the effects of CSO are the result of a DHSA‐dependent increase in EE via PPARα induction of FAO pathways.

Key points

• Previous studies with cottonseed oil‐ (CSO) enriched diets showed reductions in hepatic and plasma lipids; however, it is unclear whether linoleic acid or dihydrosterculic acid (DHSA), a cyclopropyl fatty acid found in CSO, is responsible for these phenotypic changes.

• This study utilized a unique diet design in which mice were fed either a CSO‐enriched diet (DHSA + linoleic acid) or an isocaloric oil‐enriched diet (containing linoleic acid but lacking DHSA).

• RNA‐sequencing analysis indicated CSO‐fed mice demonstrated increased expression of genes associated with fatty acid oxidation (FAO) in addition to increases in the transcription factor and FAO regulator peroxisome proliferator‐activated receptor alpha (PPARα) and its oxidative target genes.

• Knockout of PPARα confirmed this transcription factor is required for the lipid‐lowering phenotype seen following CSO‐enriched diets.

• CSO‐fed mice demonstrated significantly lower respiratory exchange ratio and higher energy expenditure compared to chow‐ and SFO‐fed mice, indicative of elevated FAO exclusive to the CSO group.

Abstract figure legend Proposed mechanisms for the lipid‐reducing phenotype of cottonseed oil (CSO). Increase in PPARα and PPARα target gene expression (A), decrease in SCD1 desaturase activity (B) and increase in energy expenditure in CSO‐fed mice (C). PPARα, peroxisome proliferator‐activated receptor alpha; ACOX1, acyl‐CoA oxidase 1; CPT1A, carnitine palmitoyltransferase 1 alpha; UCP2, uncoupling protein 2; SCD1, stearoyl‐CoA desaturase 1; 16:0, palmitic acid; 16:1n7, palmitoleic acid; 18:1n9, oleic acid; 18:0, stearic acid. Created using BioRender.com.

Introduction

Metabolic dysfunction‐associated steatotic liver disease (MASLD) is characterized by the accumulation of hepatic lipids. If left untreated, it can develop into metabolic dysfunction‐ associated steatohepatitis (MASH) hallmarked by chronic inflammation and scarring. Further progression of MASH can result in cirrhosis, hepatocellular carcinoma and/or liver failure (Angulo 2002; Ascha, Hanouneh et al., 2010). Although the exact aetiology of the disease has not been fully elucidated, both obesity and metabolic dysregulation are highly correlated with the onset of MASLD (Cusi 2012; Friedman, Neuschwander‐Tetri et al., 2018). Because MASLD often has no visible symptoms, diagnosis is determined by the presence of one or more metabolic risk factors (Rinella, Lazarus et al., 2023). Alternatively, transient elastography or liver biopsy, especially for lean individuals, can be used to assess fibrosis and confirm MASLD (Arora, Biswas et al., 2024).

MASLD and MASH are reversible if addressed before significant scarring has occurred. Pharmacological interventions can be prescribed to aid in weight loss, inflammation or blood glucose control as a means to indirectly treat MASLD (Cusi 2012; Guo, Shi et al., 2023). Such treatments include pioglitazone (Della Pepa, Russo et al., 2021; Wang, Du et al., 2023) and semaglutide (Bandyopadhyay, Das et al., 2023; Newsome, Buchholtz et al., 2021). More recently, resmetirom was approved by the US Food and Drug Administration for individuals with confirmed MASH diagnosis (Harrison, Bedossa et al., 2024). Other routes for reducing liver steatosis include lifestyle changes such as aerobic exercise and weight loss (Guo, Liong et al., 2015; Vilar‐Gomez, Martinez‐Perez et al., 2015). Dietary recommendations include reducing simple sugar and saturated fatty acid (SFA) intake (Rosqvist, Kullberg et al., 2019), while also incorporating mono‐ (MUFA) or polyunsaturated fatty acids (PUFA).

Recent work from our laboratory has focused on dietary compounds that can aid in reducing hepatic lipid accumulation as a means to prevent or reverse MASLD. Our premise is based on the concept that endogenous treatments that increase fatty acid oxidation (FAO) and inhibit lipogenesis would probably lead to a reduction in lipid accumulation. Previous studies in our laboratory indicate that cottonseed oil (CSO), a commonly consumed seed oil in the USA, can reduce lipid accumulation. CSO significantly lowered total cholesterol, triglyceride (TG) and non‐esterified fatty acids (NEFA), and increased high‐density lipoprotein in humans (Polley, Oswell et al., 2018; Prater, Scheurell et al., 2022), with similarly reduced biomarkers seen in animal models fed CSO‐enriched diets (Radcliffe, King et al., 2001; Son, Shockey et al., 2023). Additionally, CSO significantly reduced hepatic and plasma lipids and lipoproteins, and improved liver health in mice with fatty liver compared to an olive oil‐(OO)‐enriched diet (Son, Shockey et al., 2023). Despite this evidence, the exact mechanism for these effects has yet to be determined but is probably a result of the unique lipid composition of CSO: ∼53% linoleic acid (18:2n6), 28% palmitic acid (16:0) and 18% oleic acid (18:1n9), as well as 0.3% dihydrosterculic acid (DHSA) (Paton, Vaughan et al., 2017).

DHSA is a cyclopropyl fatty acid with a methylene ring at the 9,10‐ position that is an intermediate in the synthesis of sterculic and malvalic acids (Fig. 1) (Knothe 2006). Both sterculic and malvalic acid have been shown to irreversibly inhibit stearoyl‐CoA desaturase 1 (SCD1) (Gomez, Bauman et al., 2003; Raju & Reiser 1967; Reiser & Raju 1964), the enzyme that synthesizes MUFA from SFA in mammals, and is a target for preventing adiposity and metabolic disease. Numerous studies of global knockout (KO) (Ntambi, Miyazaki et al., 2002; Dobrzyn et al., 2004; Lee, Dobrzyn et al., 2004; Dobrzyn et al., 2005), tissue specific loss (Sampath, Flowers et al., 2009) and pharmacological inhibition of SCD1 (Jiang, Li et al., 2005) result in a decrease in lipogenesis with a concomitant increase in β‐oxidation because of the increase in SFA content of the liver. In previous studies, in vitro treatment with DHSA in the presence of SFA effectively inhibited SCD1, reflected by increased SCD1 mRNA expression and lower desaturation index (16:1n7/16:0 and 18:1n9/18:0) (Gomez, Bauman et al., 2003; Paton, Vaughan et al., 2017), confirming that DHSA is capable of targeting this pathway for the prevention of lipogenesis similar to sterculic and malvalic acid.

Figure 1. Synthesis of sterculic and dihydrosterculic acid from oleic acid.

Dihydrosterculic acid (DHSA) is synthesized in cottonseed by methyl group addition via S‐adenosyl methionine to the double bond in oleic acid (18:1n9) from cyclopropane synthase (Step 1). DHSA is then desaturated to form sterculic acid via cyclopropane desaturase (Step 2).

However, linoleic acid is the principal component of CSO and may also be (at least partially) responsible for the anti‐lipogenic effects observed previously. It is well‐known that PUFAs can regulate expression of key enzymes and transcription factors associated with lipid metabolism, specifically SCD1 and peroxisome proliferator‐activated receptors (PPARs) (Göttlicher, Widmark et al., 1992; Landschulz, Jump et al., 1994; Paton, Vaughan et al., 2017; Zhang, Ge et al., 2001). PUFAs, including dietary linoleic acid, suppress SCD1 transcription and activity (Landschulz, Jump et al., 1994) by interfering with binding sites on the sterol‐regulated‐element‐binding protein 1‐c, the primary isoform in the liver, to prevent the activation required for transcription (Zhang, Ge et al., 2001). Conversely, PPARs have high affinity for PUFAs, or their metabolically generated derivatives, and once activated, will dimerize with retinoid X receptor to bind specific response elements located in their respective target genes, many of which are associated with FAO. Linoleic acid has been shown to activate PPARs to levels comparable to selective agonists (Göttlicher, Widmark et al., 1992), demonstrating its ability to increase FAO gene expression. This combination of downstream effects, by both DHSA and linoleic acid, could explain how CSO effectively reduces lipid accumulation, but this has not been confirmed.

It remains unclear whether linoleic acid, DHSA, or both, are the driving force for the ability of CSO to reduce hepatic lipids and increase FAO. To highlight the specific effects of linoleic acid alone or in combination with DHSA, we fed mice either a CSO‐enriched diet (rich in linoleic acid and containing 0.3% DHSA) or a modified safflower oil (SFO)‐enriched diet (mimicking the CSO composition, but lacking DHSA). Both diets were matched for overall macronutrient and calorie content, as well as total linoleic acid content. In the present study, the SFO‐enriched diet served as the linoleic acid control, whereas the CSO‐enriched diet allowed for the analysis of downstream effects resulting from the combination of linoleic acid and DHSA. RNA‐sequencing analysis identified that CSO uniquely activated FAO pathways compared to the SFO‐fed animals receiving the linoleic acid control. Additionally, the differentially expressed genes in the CSO group included several PPARα target genes, a known regulator of FAO. To further characterize this mechanism, we analyzed the metabolic effects of the same CSO and SFO diets in female wild‐type (WT) and PPARα global KO mice. These results showed that PPARα is necessary for the lipid‐reducing effects consistently seen with CSO‐enriched diets. The effects of DHSA on mitochondrial respiration in vitro suggest that it increases the capacity for mitochondrial FAO with palmitic acid substrate. This finding was supported by respiratory exchange ratio (RER) and energy expenditure (EE) data collected over a 48 h period in chow, CSO‐ and SFO‐fed mice. Altogether, these data provide evidence for the ability of DHSA to reduce hepatic lipid accumulation by activating PPARα and FAO and may be indicative of a dietary method for addressing MASLD.

Methods

Ethical approval

All animal procedures were approved by the Institutional Animal Care and Use Committee of the University of Georgia under project approval form A2023 01‐007‐Y3‐A6. The investigators understand the ethical principles under which The Journal of Physiology operates. Additionally, this work complies with the animal ethics checklist of Journal of Physiology.

Mouse diet studies

Mice were obtained from The Jackson Laboratory (Bar Harbor, ME, USA). Thirty‐two, 8‐week‐old, male, WT C57BL/6J mice (Strain #: 000664) were divided into four groups (n = 8) and fed ad libitum either chow (3.1 kcal g⁻¹), a CSO‐enriched diet (4.0 kcal g⁻¹), a modified safflower oil‐ (SFO) enriched diet (4.0 kcal g⁻¹) or an OO‐enriched diet (4.0 kcal g⁻¹) for 6 weeks (Table 1). Custom oil‐enriched mouse diets were purchased from Envigo Teklad Diets (Madison, WI, USA): 22% CSO‐enriched diet (TD.140228), SFO‐enriched diet (TD.200070) and OO‐enriched diet (TD.130379). In a separate study, female WT C57BL/6J mice (Strain #: 000664) and female PPARα KO (Strain #: 0 08154) mice (age 8 weeks) were divided into three groups and fed ad libitum either chow, CSO‐ or SFO‐enriched diets for 4 weeks (n = 8 for each genotype and diet combination except WT SFO n = 7). The SFO diet was matched to the CSO diet in lipid composition for all components except DHSA. In essence, the SFO diet was as close to a DHSA‐free CSO diet as possible. All diets were matched for total micro‐ and macronutrient composition but varied by oil lipid composition. The OO diet used in the first study served as a matched calorie and total fat % diet that lacked n6 PUFA. Bodyweight and food intake were measured weekly. At the end of the diet intervention, mice were fasted for 4 h, then tissues collected following death via isoflurane overdose in an induction chamber with a 1‐mL saturated cloth. In all studies, mice had ad libitum access to water for the entirety of the experiment.

Table 1.

Macronutrient and principle fatty acid composition of the experimental diets fed to mice.

		Test diets
	Chow	CSO	SFO	OO
Protein (g kg−1)	243	177	177	177
CHO (g kg−1)	402	330	330	330
Total fat (g kg−1)	46	222	222	222
SFA (% of total fat)	22%	28%	24%	26%
MUFA (% of total fat)	27%	18%	23%	54%
18:2n6 (% of total fat)	47%	53%	53%	19.5%
18:3n3 (% of total fat)	4%	<0.5%	<0.5%	<0.5%
kcal g−1	3.0	4.0	4.0	4.0

CSO, cottonseed oil‐enriched high‐fat diet; SFO, safflower oil‐enriched high‐fat diet; OO, olive oil‐enriched high ‐fat diet; CHO, carbohydrates; SFA, saturated fatty acids; MUFA, monounsaturated fatty acids.

Plasma and tissue lipid analysis

A portion of liver tissue was weighed and homogenized in isopropanol for a final concentration of 50 mg ml⁻¹. The homogenate was centrifuged, and the collected supernatant was analyzed for hepatic lipid content (Oakes, Thalén et al., 2001). Blood samples were collected in EDTA when mice were anaesthetized, and plasma samples were stored at −80°C until analysis. Plasma and liver homogenate samples were analyzed for TGs and NEFAs using enzymatic colorimetric assays (Wako Chemicals USA, Richmond, VA, USA).

mRNA and protein measurements

Tissue analysis for mRNA expression was performed by homogenizing samples in TRIreagent (TR118; Molecular Research Centre, Cincinnati, OH, USA) and the isolated RNA was transcribed using High‐Capacity cDNA Reverse Transcription kits (#4368814; Applied Biosystems, Waltham, MA, USA). Quantitative real‐time PCR (RT‐PCR) was conducted with SYBR Green PCR Master Mix (#4309155; Applied Biosystems) for the following genes: acyl‐CoA oxidase 1 (ACOX1), carnitine palmitoyltransferase 1 alpha (CPT1A), PPARα, uncoupling protein 2 (UCP2) and normalized to β‐actin (see Appendix, Table A1). Protein expression was performed by first homogenizing liver samples in RIPA buffer with Halt Protease Inhibitor Cocktail (100×)(#78430; Pierce Biotechnology, Rockford, IL, USA), followed by the Bradford assay to measure protein concentration (Bradford 1976). Protein expression was assessed using western blotting for ACOX1 (AB_3 075447), CPT1A (AB_2797857) and PPARα (AB_3073567), normalized to β‐actin (AB_626632).

RNA‐sequencing analysis

Total RNA samples were isolated from homogenized mouse liver tissues following the 6 week study (n = 6) with TRIreagent (TR118; Molecular Research Centre) and submitted to the Georgia Genomics and Bioinformatics Core (GGBC, Athens, GA, USA) for RNA‐sequencing (RNA‐seq) using libraries generated from a QuantSeq 3′ mRNA‐Seq library kit (Lexogen, Vienna, Austria). One outlier was detected and omitted for the SFO analysis (n = 5). Sequencing was performed on an Illumina NextSeq 500 (PE75) (Illumina, San Diego, CA, USA). Bioinformatic analyses were performed by the GGBC utilizing computational resources provided by the Georgia Advanced Computing Resource Centre (GACRC, Athens, GA, USA). Read data were quality trimmed using Trimmomatic, version 0.36 (Bolger, Lohse, et al., 2014). Contaminating ribosomal RNA sequences were removed using Bowtie2, version 2.3.4.1, by mapping against a mammalian rRNA sequence dataset. The trimmed, filtered read data were aligned to the Mus musculus mm39 genome (NCBI accession GCA_000001635.9_GRC_genomic.fna) using STAR aligner, version 2.7.1a (Dobin, Davis, et al., 2013). HTSeq, version 0.9.1 (Putri, Anders, et al., 2022), was used to extract raw counts from each sample to generate a gene expression matrix used for RNA‐seq analysis input. Differential expression analyses were performed using the Bioconductor package DESeq2, version 3.6.3 (Love, Huber, et al., 2014), and final output data analyses were performed using: Webgestalt (Liao, Wang, et al., 2019), Enrichr (Chen, Tan, et al., 2013), g:Profiler (Reimand, Kull et al., 2007) and Venny 2.1 (Oliveros, 2007). Sequence data were deposited in the NCBI Gene Expression Omnibus and are accessible using GEO Series accession number GSE256392.

In vitro studies

Human hepatocyte‐like HepG2 cells were grown under standard conditions (37°C 5% CO₂) with 10% fetal bovine serum (A5670701; Life Technologies, Carlsbad, CA, USA) and 1 × antibiotic‐antimycotic (A5955; Sigma‐Aldrich, St Louis, MO, USA) in 4.5 g L⁻¹ Du;becco's modified Eagle's medium (DMEM) (10‐017‐CV; Corning, Manassas, VA, USA). At ∼50–60% confluence, the cells were rinsed with sterile 1 × phosphate‐buffered saline and switched to a low‐glucose 1 g L⁻¹ DMEM (10‐014‐CV; Corning) with 10% fetal bovine serum and 1 × penicillin/streptomycin (100×; 30‐002‐CI; Corning). After 24 h, the cells were rinsed with 1 × phosphate‐buffered saline and treated with either 25 µm DHSA in dimethyl sulphoxide (DMSO) (#24824; Cayman Chemical, Ann Arbor, MI, USA) or an equivalent volume of DMSO in low‐glucose, serum‐free DMEM with 1 × penicillin/streptomycin for 24 h.

Mitochondrial respiration

High‐resolution oxygen consumption measurements were conducted on HepG2 cells using an Oroboros Oxygraph‐2K (Oroboros Instruments, Innsbruck, Austria). HepG2 cells were collected following 24 h of treatment with DHSA, centrifuged, and resuspended in X‐buffer (50 mm K‐MES, 7.23 mm K₂EGTA, 2.77 mm CaK₂EGTA, 20 mm imidazole, 20 mm taurine, 5.7 mm ATP, 14.3 mm phosphocreatine, 0.56 mm MgCl₂‐6H₂O, pH 7.1). Cells were centrifuged at 1000 rpm (300 xg) for 7 min at room temperature (Fisher‐Wellman, Hagen et al., 2022), and the pellets were resuspended in Z‐buffer (105 mm K‐MES, 30 mm KCl, 1 mm EGTA, 10 mm K₂HPO₄, 5 mm MgCl, 0.5 mg mL⁻¹ bovine serum albumin, pH 7.1) with 5 mm creatine monohydrates. Subsequently, cells (2 × 10⁶ cells mL⁻¹) were transferred to the chamber and permeabilized with digitonin (20 µg mL⁻¹). After recording the basal rate, mitochondria were energized by adding palmitoylcarnitine (PalC) (40 µm) and carnitine (5 mm), followed by the addition of ADP (2.5 mm). Cytochrome c (10 µm) was added to assess mitochondrial membrane integrity. Experiments were carried out at 37°C in 500 µL chambers.

Diet and liver fatty acid composition analysis

Fatty acids were extracted from two portions (5 g) of the chow, CSO, SFO and OO diets as described previously (Paton, Vaughan et al., 2017). Briefly, diets were ground with a mortar and pestle and Soxhlet extractions were performed on both replicates of frozen meal for each of the four diets. Two gas chromatography‐mass spectrometry (GC‐MS) runs were performed on each pair of oil extractions for each diet. Fatty acids were extracted from liver tissue and converted to fatty acid methyl esters (FAMEs) by placement of tissue samples in 100 mm glass GC tubes containing 2 mL of freshly prepared 5% (v/v) sulfuric acid in methanol. Submerged tissue was macerated with a metal spatula followed by incubation to 80°C for 1 h. Reactions were quenched with 2 mL of saturated brine, and FAMEs extracted in 2 mL of hexane containing 0.01% butylated hydroxytoluene included as an antioxidant to maintain PUFA structural integrity. After phase separation, the hexane fraction was concentrated under a stream of nitrogen, then analyzed by gas GC‐flame ionization detection (GC‐FID) and GC‐MS. GC‐FID was carried out on an Agilent 7890B gas chromatograph (Agilent Technologies, Inc. Santa Clara, CA, USA) fitted with a Supelco SP‐2380 capillary column (Bellefonte, PA, USA) (30 m × 0.25 mm inner diameter × 0.2 µm film thickness) was used to separate the individual FAMEs as described previously (Dowd, Shockey et al., 2023).

Representative samples were also re‐analyzed by GC‐MS, which was performed on an Agilent model 8890 gas chromatograph equipped with a 5977C mass spectrometer and a 7693A autosampler. The GC was equipped with a DB‐FastFAME capillary column (60 m × 0.25 mm inner diameter × 0.25 µm cyanopropyl film thickness). One microlitre samples were injected into the GC using standard fast plunger mode with no swell time. The inlet was held at 250°C. The inlet was operated in split mode with a 50:1 ratio. Constant flow mode was used with a linear velocity of 16.1 cm s⁻¹. The oven was held for 3 min at 160°C, then ramped to 180°C at 2°C min⁻¹, to 245°C at 4°C min⁻¹, then held for 20 min. The column was interfaced to the MS through a transfer line held at 250°C. There was a 7‐min solvent delay. The MS was operated in scan mode with a 0 threshold, N = 3 samples and a scan range of 35–400 amu. The electron multiplier was operated with a gain factor of 1.0. Data were collected and analyzed using Masshunter, version 10.2 (Agilent).

Metabolic cages

Metabolic cage data was collected using the Promethion system (Sable Systems, North Las Vegas, NV, USA). Mice were fed chow, CSO‐ or SFO‐enriched diets (n = 8) and placed in the metabolic chambers during week 2 of their respective diets. Following a 2 day acclimation period in the metabolic cages, data was collected continuously for 48 h and processed over 3 min intervals using Macro 13 (Sable Systems, North Las Vegas, NV, USA). Body weight was measured immediately following data collection for normalization. Oxygen consumption, carbon dioxide production, locomotion, food and water intake, and EE were measured. The RER was calculated by taking the ratio of the volume of CO₂ produced (${\dot{V}}_{\mathrm{C}{\mathrm{O}}_2}$) to the volume of oxygen consumed (${\dot{V}}_{{\mathrm{O}}_2}$). The CalR, version 2 (Mina, LeClair et al., 2018) software tool was used to analyze the indirect calorimetry data between diet groups.

Statistical analysis

All results are presented as the mean ± SD. Significance of the difference was analyzed using either paired Student's t test or analysis of variance (ANOVA) with Fisher's least significant difference post hoc analysis for multiple comparisons via SAS, version 9.3M2 (SAS Institute Inc., Cary, NC, USA). The α‐level was set at P < 0.05 for all planned comparisons.

Results

Diet lipid composition analysis

Previous studies investigating CSO did not clearly establish whether linoleic acid, DHSA, or both are responsible for decreased plasma or hepatic lipids and increased FAO in vivo. To address this, we designed diets that served as direct comparisons for linoleic acid (SFO) or linoleic acid with DHSA (CSO). GC‐MS was used to perform lipid composition analysis and confirm this diet design (Fig. 2). The results indicate that the CSO‐ and SFO‐enriched diets were similarly matched for total linoleic acid (18:2n6) content, with only modest amounts present in the OO‐enriched diet (Fig. 2C ). It is important to note that these values represent the percent of total lipid within each diet, therefore total linoleic acid content in chow (containing 46 g of total fat kg⁻¹) and the three oil‐enriched diets (containing 222 g of total fat kg⁻¹) are not comparable (Table 1). Additionally, these results confirm the presence of DHSA (∼0.3%) exclusively in the CSO diet (Fig. 2D ).

Figure 2. Lipid composition analysis of oil‐enriched diets.

Repeated GC‐MS measures of chow, cottonseed oil‐ (CSO), safflower oil‐ (SFO) or olive oil‐ (OO) enriched diets (n = 2) were performed to confirm diet design. Results represent the average percentage (%) of total palmitic acid (A), oleic acid (B), linoleic acid (C), and dihydrosterculic acid (DHSA) (D) within each diet.

Body weight changes following a 6 week diet

To isolate the distinct effects of a CSO‐enriched diet in vivo, 8‐week‐old male mice (n = 8) underwent a short‐term, 6 week feeding study with ad libitum access to either chow, CSO‐, SFO‐ or OO‐enriched diets with weekly body weight measurements (Fig. 3). CSO‐ and SFO‐fed mice had a significantly higher weight gain compared to chow‐fed mice (CSO: P < 0.001; SFO: P = 0.00507) and total weight gained in OO‐fed mice was not different from the chow group (Fig. 3B ). Average caloric intake per day was not different between the CSO and SFO groups, indicating equal doses of linoleic acid between groups (Fig. 3C ).

Figure 3. Body weight changes following 6‐week diet in male wild‐type mice.

Mice were fed chow, cottonseed oil‐ (CSO), safflower oil‐ (SFO) or olive oil‐ (OO) enriched diets ad libitum and body weight data (A) and food intake (C) were collected over a 6 week period. Change in body weight (B) is expressed as difference in mouse body weight from the initiation of the feeding study in Δg. Average caloric intake between the CSO‐ and SFO‐fed mice was not different, n = 6 animals per group. *P < 0.01 vs. chow‐fed mice. Means with different letters are significantly different (P < 0.01). Values are presented as the mean ± SD.

Hepatic and plasma lipid profiles after a 6 week diet

We first wanted to assess changes in tissue lipid levels to confirm our study design. To accomplish this, plasma and liver samples were collected following the 6 week diet and analyzed for TG and NEFA using colorimetric assays. In CSO‐fed mice, TG in the liver was not changed compared to the chow‐fed mice following 6 weeks on the diets (CSO: P = 0.432) (Fig. 4A ). Both the positive control, OO diet, and SFO diet groups had significantly higher hepatic TG than the chow‐fed mice (SFO: +6.7 mg g⁻¹ protein; P = 0.00434; OO: +9.2 mg g⁻¹ protein; P = 0.0248). Despite equal linoleic acid content between the two diets, liver lysates from SFO‐fed mice exhibit greater levels of stored hepatic TG than CSO‐fed mice. This indicates that the presence of DHSA may protect against hepatic lipid accumulation, possibly as a result of reduced hepatic lipogenesis and/or lipid storage. There were no significant differences in hepatic NEFA between diet groups (Fig. 4B ).

Figure 4. Hepatic lipid levels in male wild‐type mice after 6‐week diet.

Mice were fed chow, cottonseed oil‐ (CSO), safflower oil‐ (SFO) or olive oil‐ (OO) enriched diets for 6 weeks. Liver homogenates were analyzed for total triglycerides (TG) (A) and non‐esterified fatty acid (NEFA) (B) using colorimetric assays. SFO‐ and OO‐fed mice showed increased hepatic TG compared to chow‐fed mice, whereas no difference was observed in CSO‐fed mice. NEFA was not different between diet groups, n = 6 mice per group. *P < 0.05 vs. chow‐fed mice. Values are presented as the mean ± SD.

No differences were observed in plasma TG levels between the test diets relative to chow‐fed mice (Fig 5A ). Comparisons between the oil‐enriched diets showed that plasma TG in the SFO group was significantly higher than both the CSO (+17.1 mg dL⁻¹; P = 0.0252) and OO groups (+19.3 mg dL⁻¹; P = 0.0289). No differences were observed between diets for plasma NEFA (Fig. 5B ).

Figure 5. Lipid levels in mouse plasma samples after 6‐week diet.

Mice were fed chow, cottonseed oil‐ (CSO), safflower oil‐ (SFO) or olive oil‐ (OO) enriched diets for 6 weeks and plasma was analyzed for total triglycerides (TG) (A) and non‐esterified fatty acid (NEFA) (B) using colorimetric assays, n = 6 mice per group. *P < 0.05 vs. SFO‐fed mice. Values are presented as the mean ± SD.

CSO activates lipid catabolic processes

After confirming the effectiveness of the test diets on tissue lipids, we next sought to determine genome‐wide changes in the liver using RNA‐seq analysis. The changes in gene expression between the test diets can be used to identify potential targets or pathways uniquely impacted by CSO. Liver tissue was collected at week 6 and total RNA was isolated for unbiased, genome‐wide RNA‐seq analysis to isolate changes in gene expression between diets. The MA and PCA plots in Fig. 6A and  B highlight the differential gene expression for each diet group (n = 5 for SFO, n = 6 for all other diet groups) and the results match the pattern observed in our previous metabolomics study with CSO‐fed mice (Paton, Vaughan et al., 2017). The more similar alignment between CSO‐ and chow‐fed groups in the PCA plot indicates fewer transcriptional variations for CSO‐fed mice than the other oil‐enriched diets. This may suggest that CSO‐fed mice may undergo fewer metabolic changes in response to a higher fat diet than SFO‐ and OO‐fed mice (Fig. 6B ).

Figure 6. Genome‐wide RNA sequencing analysis.

Following a 6 week diet on either chow, cottonseed oil‐ (CSO), safflower oil‐ (SFO) or olive oil‐ (OO) enriched diets, total RNA was isolated from liver samples for RNA‐sequencing analysis. MA (log ratio/mean) plot shows differentially expressed genes (DEG) compared to chow (A) which were subsequently analyzed. Principal component analysis (PCA) depicts phenotypic variation within each diet group (B), which shows that, after 6 weeks of test diets, CSO‐fed mice display fewer modifications and more closely resemble chow‐fed mice. Enrichment analysis of DEGs for each diet group (C) illustrate commonalities in up‐ and down‐regulated genes and highlight 45 DEGs specific to CSO‐fed mice. Common pathways induced between all three groups (D) and a heat map showing the diet‐specific up‐ and down‐regulated genes shared between all groups (E) indicate that the overlapping genes between the three test diets were not primarily associated with fatty acid oxidation (FAO) processes and were consistently expressed between each diet. SFO group, n = 5, n = 6 mice per group for all other groups.

Next, the differentially expressed genes (DEGs) were analyzed for overlap in the up‐ and down‐regulated genes between the three test diets (Fig. 6C ). The genes in this group represent shared gene expression responses to the test diets, a majority of which are associated with secondary alcohol and cholesterol metabolism (Fig. 6D ), whereas lipid metabolic pathways are not shared among the three diets. Among the shared genes (Fig. 6E ), there is very little variation between CSO‐, SFO‐ and OO‐fed mice, which most probably represents the consistency in macronutrient design (i.e. 50% fat diet) between groups.

For the purpose of the present study, we were interested in the expression changes following CSO consumption. The enrichment analysis identified 18 up‐ and 27 down‐regulated genes that were exclusive to the CSO‐fed group (Fig. 6C ). Validation of these DEGs identified from RNA‐seq was performed using RT‐PCR to analyze changes in mRNA expression compared to chow‐fed mice (Table 2). Analysis of these genes indicated a significant enrichment of the pathways of mitochondrial β‐oxidation, fatty acid metabolism and regulation of lipogenesis (Fig. 7A and  B ). Normalized counts were analyzed between diet groups to account for possible down‐regulation in lipid biosynthesis pathways between diets (see Appendix, Table A2) and showed relatively consistent expression patterns between diets.

Table 2.

Validation of gene expression.

Up‐regulated genes in CSO‐fed mice
Gene ID	CSO/CHOW	mRNA fold change	P value
Acox1	Up	5.1 ± 1.0	<0.001
Thrsp	Up	3.6 ± 1.1	<0.001
Por	Up	3.4 ± 0.6	<0.001
Cyp1a2	Up	3.3 ± 5.8	0.283
Nrep	Up	2.7 ± 0.8	<0.001
Hmgcr	Up	2.6 ± 1.3	0.00606
Slc16a12	Up	2.3 ± 0.4	<0.001
Aldh3a2	Up	2.1 ± 0.4	<0.001
Etnk2	Up	2.0 ± 0.4	<0.001
Hadh	Up	1.9 ± 0.2	<0.001
Nampt	Up	1.8 ± 0.2	<0.001
Rbp4	Up	1.6 ± 0.4	0.0147
Gpt2	Up	1.5 ± 0.2	0.00620
Zfpm1	Up	1.3 ± 0.8	0.298
Cdk12	Up	1.2 ± 1.0	0.622
Suclg1	Up	1.2 ± 0.3	0.103
Cyp2d26	Up	1.1 ± 0.4	0.518
Psmd11	Up	1.1 ± 0.7	0.730
Acat1	Up	1.0 ± 0.2	0.915

Down‐regulated genes in CSO‐fed mice
Gene ID	CSO/CHOW	mRNA fold change	P value
Acaa2	Down	0.98 ± 0.2	0.880
Atf5	Down	0.98 ± 0.3	0.946
Camk1d	Down	0.8 ± 0.6	0.486
Ccn1	Down	0.8 ± 0.7	0.567
Nr1d1	Down	0.8 ± 0.4	0.372
Tex2	Down	0.8 ± 0.1	0.0301
Ceacam1	Down	0.7 ± 0.1	0.0381
Crot	Down	0.7 ± 0.2	0.0494
Cyp2c68	Down	0.7 ± 0.5	0.198
Ptprk	Down	0.7 ± 0.3	0.139
Tbcel	Down	0.7 ± 0.1	0.0426
Ugt2b1	Down	0.7 ± 0.1	0.0537
Abcb11	Down	0.6 ± 0.1	0.0411
Eps8l2	Down	0.6 ± 0.3	0.0404
Gldc	Down	0.6 ± 0.3	0.0139
Slc16a1	Down	0.6 ± 0.1	0.00564
Wdtc1	Down	0.6 ± 0.1	0.0182
Clpx	Down	0.5 ± 0.3	0.00443
Coq8a	Down	0.5 ± 0.1	0.00946
Cyp3a59	Down	0.5 ± 0.2	<0.001
Ddc	Down	0.5 ± 0.3	0.00374
Insig2	Down	0.5 ± 0.3	0.00849
Slc30a10	Down	0.4 ± 0.3	0.00221
Elovl3	Down	0.3 ± 0.2	<0.001

Differentially expressed genes (DEG) identified by the RNA‐sequencing analysis for CSO‐fed mice were validated using RT‐PCR. Genes are grouped based on up‐ or down‐regulation compared to chow‐fed mice. P < 0.05 vs. chow‐fed mice. Values are presented as the fold change ± SD.

Figure 7. Over‐representation analysis.

The differentially expressed genes (DEG) following 6 weeks of a cottonseed oil‐ (CSO) enriched diet were analyzed using g:Profiler for strength of association to the listed pathways. The up‐ (A) and down‐ (B) regulated DEGs within the CSO group are primarily associated with the pathways of mitochondrial β‐oxidation, fatty acid metabolism and regulation of lipogenesis. Evidence codes are represented by different colors and strength of association is depicted by the inverse log P values.

Previous studies in our lab have shown an increase in protein expression in hepatic PPARα, PPARγ coactivator 1‐alpha (PGC‐1α), CPT1A and ACOX1 in response to a CSO‐enriched diet (Son, Shockey et al., 2023). PPARs can bind fatty acids to dimerize with retinoid X receptor and bind to specific response elements located on their respective target genes associated with FAO processes. Linoleic acid, along with other fatty acids, can regulate gene expression and FAO by activating PPARs (Göttlicher, Widmark et al., 1992). Therefore, based on the combination of previous results along with the current data, it is reasonable to hypothesize from the RNA‐seq data that a CSO‐enriched diet may function through a PPARα‐dependent mechanism. First, we assessed the effects of the diets on tissue expression of PPAR genes (Fig. 8). We measured mRNA expression using RT‐PCR for the following genes: PPARα, PPARδ, PGC‐1α, CPT1A, ACOX1 and uncoupling protein 2 (UCP2). PPARα (1.6 ± 0.5 fold increase; P = 0.0284), ACOX1 (2.4 ± 1.5 fold increase; P = 0.0278) and UCP2 (2.6 ± 1.4 fold increase; P = 0.00642) expression was significantly higher in the CSO‐fed mice compared to chow‐fed mice (Fig. 8), whereas there were no differences between diets for PPARδ or PGC‐1α expression (PPARδ: P = 0.126; PGC‐1α: P = 0.226; not shown). CPT1A expression significantly increased in both the CSO‐ (2.1 ± 1.1 fold increase; P = 0.0241) and SFO‐ (1.9 ± 1.0 fold increase; P = 0.0454) fed mice compared to chow. Western blot analysis of protein expression confirms the gene expression patterns observed between diet groups (Fig. 8E ). Taken together, the expression of oxidative genes in the CSO group, including several PPARα target genes, as well as the increase in hepatic PPARα expression, indicate that the lipid components in CSO might be dependent on this transcription factor to induce the changes in lipid profiles seen here and in other studies.

Figure 8. PPARα and its targets increase after a CSO‐enriched diet.

After 6 weeks on chow, cottonseed oil‐ (CSO), safflower oil‐ (SFO) or olive oil (OO) enriched diets, liver mRNA was isolated and analyzed for changes in expression of genes associated with PPARα‐induced fatty acid oxidation including PPARα (A), CPT1A (B), UCP2 (C) and ACOX1 (D). Western blot analysis of liver lysates for ACOX1, CPT1A and PPARα normalized to β‐actin in two representative mice per diet group (E). PPARα, peroxisome proliferator‐activated receptor alpha; CPT1A, carnitine palmitoyltransferase 1 alpha; UCP2, uncoupling protein 2; ACOX1, acyl‐CoA oxidase 1. (A) to (D) n = 8 mice per group. *P < 0.05 vs. chow‐fed group. Values are presented as the mean ± SD.

Hepatic and plasma lipids in WT and PPARα KO mice

To test the hypothesis that CSO functions through PPARα to increase FAO and induce the metabolic effects seen following a CSO‐enriched diet, we fed 8‐week‐old female WT and PPARα‐ KO mice (n = 8) chow, CSO‐ or SFO‐enriched diets ad libitum. Hepatic TG and NEFA were analyzed for both genotypes to determine whether the lipid‐lowering phenotype was still present in the CSO group in the absence of PPARα. Consistent with the first 6 week study in males, the female WT chow‐ and CSO‐fed mice had no difference in hepatic TG (P = 0.906) or NEFA (P = 0.0657) levels (Fig. 9A and  B ). SFO‐fed mice displayed significantly higher values for both measures compared to chow (TG: 25.3 ± 2.5 vs. 16.4 ± 1.2 mg g⁻¹ protein, P = 0.00344; NEFA: 0.5 ± 0.1 vs. 0.2 ± 0.02 µm mg⁻¹ protein, P = 0.0101) (Fig. 9A and  B ). These results show that the effect of CSO in females is consistent with the metabolic differences observed with males, specifically that dietary CSO can prevent hepatic lipid accumulation.

Figure 9. Liver TG and NEFA in female wild‐type and PPARα knockout mice.

Female wild‐type (WT) and PPARα knockout (KO) mice were fed chow, cottonseed oil‐ (CSO) or safflower oil‐ (SFO) enriched diets for 4 weeks. Liver homogenates were analyzed for TG (A) and (C) and NEFA (B) and (D) for both genotypes. In the absence of PPARα, a CSO‐enriched diet did not elicit the lipid‐lowering phenotype seen in the WT mice. NEFA, non‐esterified fatty acids; TG, triglycerides, WT SFO group, n = 7, n = 8 mice for all other groups. *P < 0.05 vs. chow‐fed mice. Values are presented as the mean ± SD.

In the absence of PPARα, this phenotype was no longer observed. KO mice fed a CSO‐enriched diet had two‐fold higher hepatic TG (32.5 ± 1.6 protein vs. 16.3 ± 0.4 mg g⁻¹ protein; P < 0.001) and six‐fold higher NEFA (1.8 ± 0.1  protein vs. 0.3 ± 0.03 µm mg⁻¹ protein; P < 0.001) than CSO‐fed WT mice (Fig. 9C and  D ), indicating that PPARα is required for the reduction in hepatic lipids by CSO. Analysis of plasma lipids shows no difference in TG or NEFA between diets for the WT females, with the exception of SFO‐fed WT NEFA levels, which were significantly lower than chow‐fed females (P = 0.0280) (Fig. 10). Similarly, there was no difference in plasma NEFA between diets in the KO mice. Circulating TG was significantly lower in the CSO (P < 0.001) and SFO (P < 0.001) groups compared to chow in the KO females.

Figure 10. Plasma TG and NEFA in female wild‐type and PPARα knockout mice.

Female wild‐type (WT) and PPARα knockout (KO) mice were fed chow, cottonseed oil‐ (CSO) or safflower oil‐ (SFO) enriched diets for 4 weeks. Plasma was analyzed for TG (A) and (C) and NEFA (B) and (D) for both genotypes. There was no difference in TG between diet groups for the WT mice, whereas NEFA was lower in the SFO‐fed mice only. TG was significantly lower for both oil‐enriched diets compared to chow‐fed mice in the KO females, whereas NEFA showed no difference. NEFA, non‐esterified fatty acids; TG, triglycerides; WT SFO mice, n = 7, n = 8 for all other groups. *P < 0.05 vs. chow‐fed mice. Values are presented as the mean ± SD.

Hepatic lipid composition

Previous studies indicated that CSO‐fed mice were protected from hepatic TG accumulation, possibly through one of two pathways: increased flux through FAO or suppression of lipogenesis. As the rate‐limiting step of lipogenesis, we aimed to assess SCD1 activity between diet groups. To indirectly analyze differences in SCD1 activity, lipids were extracted from liver samples and analyzed via GC‐MS to determine the total percent composition of hepatic lipid species following 4 weeks on the respective diets. Palmitic acid (16:0) and stearic acid (18:0), SCD1 substrates, were analyzed relative to the SCD1 products palmitoleic acid (16:1n7) and oleic acid (18:1n9) (Table 3). In the WT mice, 16:1n7/16:0 desaturation index was significantly lower in the CSO group (P < 0.001) but not the SFO group (P = 0.257) relative to chow‐fed mice (Table 3). Additionally, the 16:1n7/16:0 desaturation index was significantly lower in the CSO‐fed mice compared to the SFO group (P = 0.0205). 18:1n9/18:0 desaturation index in the CSO‐fed WT group was significantly reduced compared to SFO‐fed WT mice (P = 0.00321) and SFO showed no difference compared to chow (P = 0.120) (Table 3). The effects of genotype on desaturation index were not significant. The lower indices observed, or rather the accumulation of SFA, in the CSO‐fed mice relative to SFO and chow groups is indicative of reduced SCD1 activity that is dependent on the presence of DHSA and not linoleic acid alone.

Table 3.

Diet effects on desaturation indices.

16:1n7/16:0 desaturation index
Diet	Wild‐type	PPARα knockout
Chow	0.05 ± 0.02	0.06 ± 0.02
CSO	0.01 ± 0.008*	0.02 ± 0.007**
SFO	0.03 ± 0.02 ‡	0.02 ± 0.006**

18:1n9/18:0 desaturation index
Diet	Wild‐type	PPARα knockout
Chow	1.4 ± 0.3	1.3 ± 0.3
CSO	1.0 ± 0.2*	1.6 ± 0.4
SFO	1.6 ± 0.4 ‡	1.8 ± 0.3*

Liver samples from female WT and PPARα KO mice were analyzed by gas chromatography (GC) for total percent lipid species following a 4 week diet with chow, cottonseed oil‐ (CSO) or safflower oil‐ (SFO) enriched diets. Indices were determined for palmitoleic acid (16:1n7) to palmitic acid (16:0) ratio and oleic acid (18:1n9) to stearic acid (18:0) ratio, which are reflective of SCD1 activity. These results, specifically for 16:1n7/16:0, indicate that SCD1 activity is significantly impaired following a CSO‐enriched diet. SCD1, stearoyl‐CoA desaturase 1. Values are presented as the mean ± SD of percent total neutral lipid. WT SFO, n = 7, n = 8 mice for all other groups. *P < 0.05 vs. chow group within genotype, **P < 0.001 and ^‡ P < 0.05 vs. CSO group within genotype.

Effects of DHSA on mitochondrial function

Based on the increase in PPARα and its mitochondrial FAO target genes in the CSO diet group, we assessed the effects of DHSA on mitochondrial respiration in vitro (Fig. 11). HepG2 cells were treated with DHSA and oxygen consumption rate (OCR) was measured after the serial addition of PalC/carnitine, ADP and cytochrome c. 16‐carbon PalC/carnitine substrate was used to assess FAO‐dependent mitochondrial respiration, whereas glycolytic intermediates were omitted. Solely using long‐chain fatty acids, such as PalC, without malate or carnitine can exert a rate‐limiting effect because the accumulation of acetyl‐CoA from β‐oxidation can inhibit key enzymes in the tricarboxylic acid cycle, thereby decreasing the generation of reducing equivalents (i.e. NADH and FADH₂). However, carnitine facilitates the export of excess acetyl‐CoA to the cytosol as acetyl‐carnitine, alleviating acetyl‐CoA‐mediated feedback inhibition and enhancing mitochondrial respiratory efficiency. OCR was not significantly different between groups in the PalC/carnitine phase. Following the addition of ADP to induce maximal respiration, there was a significant increase in OCR for the DHSA group (P = 0.0384) compared to the control group. These data suggest that DHSA‐treated cells have greater capacity to perform FAO using palmitic acid as substrate.

Figure 11. OCR following DHSA‐treatment.

HepG2 cells were treated with 25 µM unconjugated DHSA for ∼24 h in low‐glucose, serum free media and analyzed for differences in oxygen consumption rate (OCR) compared to the DMSO vehicle. There was no difference in OCR between groups following PalC/carnitine. DHSA‐treated cells had significantly higher maximal OCR following the addition of ADP. Cytochrome c was used to assess the integrity of mitochondrial membranes, n = 6 replicates per treatment. *P < 0.05 vs. control. Values are presented as the mean ± SD.

A CSO‐enriched diet increases energy expenditure

To better characterize the effects of these diets on metabolic rate and whole‐body EE, mice were placed in metabolic cages and data collected over a 48 h period. Food intake over the 48 h test period was not different between the three diets (Fig. 12B ). Average hourly CO₂ production and O₂ consumption (Fig. 12C ) were used to calculate the respiratory exchange ratio (${\dot{V}}_{\mathrm{C}{\mathrm{O}}_2}$/ ${\dot{V}}_{{\mathrm{O}}_2}$) (Fig. 12D ). The dark cycle RER values for chow‐fed mice are indicative of primarily carbohydrate oxidation (0.93 ± 0.01). RER values for both oil‐enriched diet groups, which were significantly lower compared to chow during both light (CSO: P < 0.0001; SFO: P = 0.000519) and dark (P < 0.0001 CSO and SFO) cycles, indicate the expected increased rates of fat oxidation. However, despite the expected decrease in RER from the introduction of fat enriched diets, we saw a further significant decrease in CSO‐ compared to SFO‐fed mice (P = 0.0135). EE was analyzed for each diet group across both light and dark periods. CSO‐fed mice had significantly higher relative whole‐body EE compared to both chow‐ (P < 0.0001) and SFO‐fed (P < 0.0001) mice (Fig. 12E ) in both the light and dark cycles. It is important to note that EE for the SFO group was not different from chow‐fed mice in the light (P = 0.0850) or dark cycle (P = 0.134). Analysis of total locomotion determined that activity during the light or dark cycles was not different between CSO‐ and SFO‐fed mice (P = 0.0769) (see Appendix, Fig. A1). These data suggest that the increased EE observed exclusively in the CSO‐fed group is probably attributed to elevated metabolic activity that is dependent on the presence of DHSA.

Figure 12. Metabolic cage data.

Mice were fed chow, cottonseed oil‐ (CSO) or safflower oil‐ (SFO) enriched diets and transferred to metabolic cages at week 2 of feeding. Following the acclimation period, carbon dioxide production, oxygen consumption, and energy expenditure were collected continuously for 48 h. Mice were weighed immediately following data collection (A) and food intake (kcal g⁻¹) was analyzed between each diet group (B). Average hourly carbon dioxide production and oxygen consumption (C) was used to calculate respiratory exchange ratio (RER) for each diet group (D). Energy expenditure was normalized by body weight (kcal kg⁻¹ h⁻¹) and was significantly higher in the CSO group compared to chow and SFO groups (E). Data was analyzed using CalR (version 2) software, n = 8 mice per group. *P < 0.05 vs. chow‐fed mice; # P < 0.05 vs. CSO‐fed mice. Values are presented as the mean ± SD.

Discussion

The present study aimed to assess the distinct effects of linoleic acid vs. the cyclopropyl fatty acid, DHSA, on lipid metabolism. To accomplish this, we first fed male WT mice one of four diets for 6 weeks: one rich in linoleic acid (SFO), one rich in linoleic acid and containing DHSA (CSO) and one lipogenic control (OO) compared to chow‐fed mice. This diet design for CSO and SFO was based on the idea that if linoleic acid was necessary and sufficient to produce the effects of CSO, then the SFO diet would produce similar effects to CSO. If DHSA was necessary, then neither the SFO nor OO diet would match the CSO results. Unfortunately, we were not able to test the sufficiency of DHSA alone because of its scarcity, although it would have been ideal to have been able to feed mice a DHSA‐enriched diet in the absence of linoleic acid.

Previous feeding studies with CSO have demonstrated its ability to reduce both circulating lipids and hepatic steatosis. Following a short‐term diet in mice, we observed elevated hepatic TG in SFO mice compared to chow‐fed mice, with no difference observed in CSO‐fed mice. Similarly, plasma TG was higher in SFO‐fed mice after 6 weeks compared to CSO‐fed mice. These measures indicate that the two components of CSO, linoleic acid and DHSA, are required for the observed changes in lipids after the CSO diet.

RNA‐seq analysis of animals fed these diets confirmed that the composition of CSO probably reduces lipids through the activation of lipid oxidation pathways, which were not induced by the matched high linoleic acid diet. CSO‐fed animals displayed increased expression of genes involved in mitochondrial and peroxisomal β‐oxidation, lipid metabolism and regulation of lipogenic pathways compared to the other two diet groups. Moreover, the expression of genes regulating lipid biosynthesis pathways was consistent between diets, further supporting that CSO specifically up‐regulates oxidative processes compared to the other oil‐enriched diets. The unique DEGs identified in CSO‐fed mice confirmed that changes in expression are dependent on the specific lipid composition of CSO; specifically, that linoleic acid alone is not sufficient to change expression of these oxidative genes but requires the presence of DHSA. We validated the changes in these genes, which showed similar increases in expression for FAO genes seen in previous studies with CSO‐enriched diets (Son, Shockey et al., 2023). These data indicate that the consumption of CSO is capable of improving lipid levels by increasing FAO and the linoleic acid composition of the oil is not the sole cause of this result. Rather it is most probably the result of an additional aspect of CSO, such as DHSA. Whether or not DHSA alone is sufficient remains to be determined, but the results at hand indicate that linoleic acid + DHSA is required for the lipid‐lowering effects of CSO.

Further analysis of the DEGs obtained from RNA‐Seq highlighted multiple key regulators of lipid oxidation pathways, several of which are under transcriptional regulation of PPARα. Although no difference in PPARα protein expression was observed between diet groups, the consistent increases in PPARα target gene and protein levels suggest a selective increase in transcriptional activity in the CSO‐fed mice compared to other test diets. To determine the necessity of PPARα for the CSO phenotype, we fed female WT and PPARα KO mice chow, CSO and SFO diets for 4 weeks. Similar to the first study, we found that liver TG and NEFA levels were maintained with a CSO‐enriched diet, but SFO was not sufficient to prevent lipid increases in the liver. Conversely, in PPARα KO mice, we found that the lipid‐lowering effects of CSO were lost. Hepatic TG and NEFA were higher in both CSO and SFO KO mice compared to the chow and the WT counterparts. Sex differences were not accounted for in the PPARα KO model, therefore conclusions cannot be made on the effects of CSO in male KO mice and future studies must be done. Overall, these data indicate that PPARα is necessary for the phenotypic responses seen with CSO‐enriched diets.

Consistent increases in mitochondrial FAO genes led us to investigate changes in mitochondrial OCR following DHSA treatment in vitro. Maximal OCR was significantly higher compared to untreated cells after the ADP stimulus, suggesting an increased capacity to oxidize palmitic acid substrate. The increase in OCR observed in DHSA‐treated cells could be the result of increased mitochondrial content or increased enzyme activity. To test this hypothesis, we repeated the OCR experiment using liver tissues from mice fed chow, CSO‐, SFO‐ and OO‐enriched diets but found no difference in mitochondrial OCR between diet groups (data not shown). Given that there was no difference in mitochondrial OCR between CSO and SFO liver tissues, it is possible that the effects on hepatic FAO are not exclusive to mitochondria. Following CSO diet interventions, we have observed consistent increases in ACOX1 gene and protein expression in vivo, suggesting an increase in peroxisomal content may also be responsible for preventing hepatic lipid accumulation. ACOX1 is the first and rate‐limiting step of peroxisomal β‐oxidation. This enzyme catalyses the desaturation of acyl‐CoA to 2‐trans‐enoyl‐CoA and generates hydrogen peroxide from molecular oxygen as a byproduct of the reaction. The preferred substrate of ACOX1 is primarily straight chain, saturated or unsaturated, long‐chain or very‐long chain fatty acids. Recent studies suggest that ACOX1 mRNA expression is inducible by high‐fat diets and is positively correlated with the obesity status (Lu, He et al., 2024). ACOX1 liver‐specific knockout yielded mice with higher EE and that were resistant to high‐fat diet‐induced obesity, insulin resistance and inflammation (Lu, He et al., 2024). Similarly, administration of ACOX1 inhibitor 10,12‐tricosadiynoic acid increased PPARα target gene expression and mitochondrial FAO, and decreased body weight, liver TG and hydrogen peroxide formation (Zeng, Deng et al., 2017). Similar to the feedback loop observed with SCD1, it is possible that a lipid component in the CSO diet could inhibit ACOX1 and result in elevated mRNA expression, but further studies are needed to confirm this. In theory, increased ACOX1 activity would correspond to increased OCR; however, the PalC substrate used with the liver tissues in the present study is not readily transported into peroxisomes for degradation and might explain why no difference was observed. Further investigation is needed to clarify the role of ACOX1 in hepatic lipid maintenance.

We then collected metabolic cage data to analyze RER and whole‐body EE between chow, CSO and SFO diet groups. As expected, RER was lower in the oil‐enriched diet groups vs. chow‐fed mice, indicating increased reliance on fatty acid substrates for fuel. Between the CSO and SFO groups, the RER was significantly lower in CSO‐fed mice. The low RER was accompanied by a significantly higher EE in the CSO group compared to chow and SFO across both photoperiods. Analysis of locomotor activity suggests that an increase in spontaneous activity is probably not the cause of the elevated EE in the CSO group but may be the result of increased metabolic activity across multiple tissues. These data suggest that linoleic acid alone is insufficient to increase EE in SFO mice. Moreover, the DHSA‐dependent increase in FAO in CSO mice is supported by the identified DEGs as well as the enhanced mitochondrial OCR response to DHSA treatment in vitro. It remains unclear whether peroxisomal activity is the driving force of the elevated EE. One diet study utilizing the peroxisomal substrate dicarboxylic acids (DCA) and a fat/calorie matched high‐fat diet design, found that DCA addition increased EE, prevented hepatic lipid accumulation and up‐regulated ACOX1 mRNA, in addition to other metabolic markers (Goetzman, Zhang et al., 2024). Future studies are needed to clarify the role of peroxisomal degradation and whole‐body EE.

It is still unknown how DHSA can induce PPARα expression or activity. We and others have observed that inhibition of SCD1 causes an accumulation of saturated fatty acid (SFA), which in turn can activate endoplasmic reticulum (ER) stress, increase cellular efflux of toxic unesterified lipids and promote mitochondrial lipid oxidation via PGC‐1α. Although these observations are consistent and clear, the mechanistic link(s) between SFA accumulation and PPARα activity needs to be elucidated. It is interesting to note that ER stress was shown to suppress FAO in one study (DeZwaan‐McCabe, Sheldon et al., 2017), but another found that loss of PPARα induced both mitochondrial and ER stress (Su, Baker et al., 2014). In this first case, PPARα and its target genes were significantly down‐regulated following robust ER stress, resulting in reductions in mitochondrial oxidation efficiency and subsequent steatosis (DeZwaan‐McCabe, Sheldon et al., 2017). In the second scenario, unrestricted lipogenesis induced by a high‐fructose diet suppressed PPARα activity and simultaneously increased cell stress (Su, Baker et al., 2014). Although these studies represent cell signalling under extreme circumstances, they suggest that ER stress may determine PPARα expression in order for the cell stressor (i.e. SFA) to be removed through PPARα‐dependent pathways. Thus, the role of SFA‐induced ER‐stress in mediating PPARα activity needs to be characterized.

In addition to increasing oxidation, we believe CSO elicits further changes in lipid profiles through the suppression of lipogenesis. Percent total lipid composition was measured in the livers of the female WT and KO mice following 4 weeks on the CSO and SFO diets. Desaturation indices indicated a significant increase in 16:0 and 18:0 relative to 16:1n7 and 18:1n9 in the CSO group compared to both SFO and chow groups, suggesting that SCD1 activity was reduced in these animals. PUFAs are known suppressors of SCD1 by interfering with binding sites located in the gene promoter region to prevent transcription (Zhang, Ge et al., 2001). Dietary linoleic acid has been shown to suppress SCD1 mRNA expression (Landschulz, Jump et al., 1994) and is abundant in both the CSO and SFO diets. DHSA is an intermediate in sterculic acid synthesis, which is a known inhibitor of SCD1 (Gomez, Bauman et al., 2003) and is capable of reducing plasma and liver TG content following chronic dietary consumption (Ortinau, Nickelson et al., 2013). Here, we hypothesize that there is a potentially additive effect in vivo through co‐ordinated suppression of lipogenesis in the liver by both linoleic acid and DHSA components in CSO. The expression of SCD1 is probably reduced by linoleic acid, with the subsequent translated product being inhibited by DHSA.

The present study has demonstrated that the lipid‐lowering effects seen following a CSO‐enriched diet are accomplished through linoleic acid/DHSA‐dependent activation of FAO pathways that require the presence of the transcription factor PPARα. The mechanism by which DHSA activates the FAO pathways remains to be determined, but it is dependent on PPARα activity. Future studies should aim to determine whether the demonstrated metabolic changes are the result of direct interaction between DHSA and PPARα to influence transcriptional activity, or whether other upstream activators are involved. Specifically, studies utilizing DHSA‐enriched diets will better determine its effects on metabolism and whether the presence of linoleic acid is a necessary co‐activator for metabolic changes. These findings indicate that the addition of CSO to the diet may be beneficial in the prevention or treatment of diseases that involve excessive hepatic lipid accumulation.

Competing interests

The authors declare that they have no competing interests.

Author contributions

C.M.P was responsible for the conception and design of the experiments. L.S.H, Y.S, J.H, J.A.C, J.S, W.W.L and C.M.P were responsible for the acquisition, analysis or interpretation of data for the work. All authors contributed to writing and critically reviewing the manuscript. All authors approved the final version of the manuscript submitted for publication and agree to be accountable for all aspects of the work in ensuring that questions related to the accuracy or integrity of any part of the work are appropriately investigated and resolved. All persons designated as authors qualify for authorship, and all those who qualify for authorship are listed.

Funding

Funding was provided by the US Department of Agriculture and Cotton Incorporated.

Supporting information

Peer Review History

Biography

Dr Chad M. Paton is conducting research focusing on the bioactive properties of two specific lipids: dihydrosterculic acid (DHSA) and linoleic acid. DHSA is a cyclopropene fatty acid that is found naturally in cottonseed oil and it blocks endogenous lipid synthesis and cholesterol biogenesis. He is actively pursuing methods to assess its ability to treat hyperlipidaemia and hypercholesterolemia in animal and human models. His work on the bioactive properties of linoleic acid has centred on its role in modulating skeletal muscle myogenesis and mitochondrial biogenesis. As a long‐chain omega‐6 polyunsaturated fatty acid, it promotes transcriptional activity via nuclear receptor‐DNA interactions and more recently Chad has found that its ability to induce angiopoietin‐like protein 4 production in muscle significantly impairs the capacity for muscle differentiation and metabolic function. The purpose of this research is to help restore normal metabolic function in disease states using molecular biology and biochemistry to understand how macronutrient metabolism is regulated in cell and animal models.

Table A1.

Mouse primers for tissue analysis.

Mouse primers	Forward	Reverse
ACOX1	5′‐ GGCTTGGAAACCACTGCCACATA ‐3′	5′‐ TCCCAATCTCACGGATAGGGACAAC ‐3′
CPT1A	5′‐ TTGATCAAGAAGTGCCGGACGAGT ‐3′	5′‐ GTCCATCATGGCCAGCACAAAGTT ‐3′
PPARα	5′‐ GTGCCCTGAACATCGAGTGTCGAATA ‐3′	5′‐ TCGTACACCAGCTTCAGCCGAATAG ‐3′
UCP2	5′‐ AGCCCAGCCTACAGATGTGGTAAA ‐3′	5′‐ GCATTACGGGCAACATTGGGAGAA ‐3′
β‐actin	5′‐ TTGCTGACAGGATGCAGAAGGAGA ‐3′	5′‐ ACTCCTGCTTGCTGATCCACATCT ‐3′

ACOX1, acyl‐CoA oxidase 1; CPT1A, carnitine palmitoyltransferase 1 alpha; PPARα, peroxisome proliferator‐activated receptor alpha; UCP2, uncoupling protein 2.

Table A2.

Gene expression associated with lipid synthesis.

Lipid synthesis
Gene ID	CHOW	CSO	OO	SFO
Fasn	4800.6	6336.4	12 376.1	6331.2
Srebf1	1487.2***	2827.8	3838.9	2244.6
Srebf2	617.1	812.8	842.7	777.1
Acaca	626.0	630.8	875.7	620.7
Acly	2768.8	3364.8	4734.2	3238.9
Lpin1	2679.3	1964.0	853.2	2133.7
Lpin2	4246.3	4572.9	3005.6*	3690.7
Lpin3	85.0	94.8	36.0	60.0
Plin2	5680.7	5804.2	6281.7	6416.5
Acss2	1537.8	2083.9	2318.4	1862.5

Lipid uptake
Gene ID	CHOW	CSO	OO	SFO
Scarb1	1757.9	2146.9	2759.8	2208.1
CD36	805.5	899.4	629.7	366.8**
Vldlr	43.3	21.5	18.3	3.1
Ldlr	1780.5	2203.7	2515.2	2059.5
Fatp	110.5	170.1	56.5*	49.0*
Fabp1	16 914.2	24 511.5	14 100.9*	15 525.5
Fabp4	93.0	46.8	6.8***	38.1

Lipolysis and lipid export
Gene ID	CHOW	CSO	OO	SFO
Pnpla2	920.6	1011.2	1007.7	954.1
Abhd5	311.3	254.2	346.0	316.1
Lipe	98.7***	233.7	296.7	154.4
Mgll	4337.0	4523.0	3833.3	3186.5**
Apoe	196 110.4	215 604.9	231 820.9	185 046.8
Apob	74 798.7	79 045.3	94 991.0*	80 517.0
Mttp	4206.9	4542.2	5069.7	3663.0**

Following a 6 week diet on either chow, cottonseed oil‐ (CSO), safflower oil‐ (SFO) or olive oil‐ (OO) enriched diets, total RNA was isolated from liver samples for RNA‐sequencing analysis. Differentially expressed genes were identified using DESeq2 software, and the normalized expression values are presented as the median of ratios for genes involved in lipid synthesis, uptake and export. Values are presented as the mean for each diet group, n = 6 for all groups, except SFO, n = 5. *P < 0.05, **P < 0.01 and ***P < 0.001 vs. CSO‐fed mice.

Figure A1. Locomotor activity.

Mice were fed chow, cottonseed oil‐ (CSO) or safflower oil‐ (SFO) enriched diets and transferred to metabolic cages at the start of week 2. Following an acclimation period, sensors recorded beam breaks continuously for 48 h. Locomotion was not different between diet groups during the light cycle. Locomotion was greater in CSO and SFO groups relative to chow‐fed mice during the dark cycle, with no difference observed between CSO and SFO groups. Data was analyzed using CalR, version 2. Values represent group means overall, in the dark cycle and in the light cycle respectively, n = 8 mice per group. *P < 0.05 vs. chow‐fed mice.

Data availability statement

This article contains supplemental data. All data supporting the findings presented here are included in this paper or in the supplemental data.