Western diet-induced hepatic steatosis and alterations in the liver transcriptome in adult Brown-Norway rats

Background The purpose of this study was to investigate the effects of sub-chronic high fat, high sucrose diet (also termed ‘Westernized diet’ or WD) feeding on the liver transcriptome during early nonalcoholic fatty liver disease (NAFLD) development. Methods Brown Norway male rats (9 months of age) were randomly assigned to receive ad libitum access to a control (CTL; 14 % kcal fat, 1.2 % sucrose by weight) diet or WD (42 % kcal from fat, 34 % sucrose by weight) for 6 weeks. Results Six weeks of WD feeding caused hepatic steatosis development as evidenced by the 2.25-fold increase in liver triacylglycerol content, but did not induce advanced liver disease (i.e., no overt inflammation or fibrosis) in adult Brown Norway rats. RNA deep sequencing (RNA-seq) revealed that 94 transcripts were altered in liver by WD feeding (46 up-, 48 down-regulated, FDR < 0.05). Specifically, the top differentially regulated gene network by WD feeding was ‘Lipid metabolism, small molecular biochemistry, vitamin and mineral metabolism’ (Ingenuity Pathway Analysis (IPA) score 61). The top-regulated canonical signaling pathway in WD-fed rats was the ‘Superpathway of cholesterol biosynthesis’ (10/29 genes regulated, p = 1.68E-17), which coincides with a tendency for serum cholesterol levels to increase in WD-fed rats (p = 0.09). Remarkably, liver stearoyl-CoA desaturase (Scd) mRNA expression was by far the most highly-induced transcript in WD-fed rats (approximately 30-fold, FDR = 0.01) which supports previous literature underscoring this gene as a crucial target during NAFLD development. Conclusions In summary, sub-chronic WD feeding appears to increase hepatic steatosis development over a 6-week period but only induces select inflammation-related liver transcripts, mostly acute phase response genes. These findings continue to outline the early stages of NAFLD development prior to overt liver inflammation and advanced liver disease.


Background
Many modernized cultures consume a high-fat, high sugar diet rich in calories, termed a Western diet (WD), which contributes to obesity-related maladies. Concomitant with the staggering rise in the obesity epidemic is the prevalence of nonalcoholic fatty liver disease (NAFLD) which is now estimated to affect~30 % of the adult population in the United States, and is present in 75-100 % of obese or morbidly obese individuals [1,2]. NAFLD is a progressive liver disease ranging from simple steatosis, nonalcoholic steatohepatitis (NASH), fibrosis, and cirrhosis [3]. It is characterized by elevated hepatic triglyceride (TG) storage (≥5 % by weight) in the absence of excessive alcohol consumption (<20 g/d) and is considered to be the hepatic manifestation of the metabolic syndrome (reviewed in [3]).
Western Diet (WD) feeding in rodents leads to the rapid development of NAFLD and subsequent development of NASH [4]. High-fat and/or high-sugar feeding studies examining the liver transcriptome have reported that inflammatory and lipid metabolism markers increase while oxidative stress defense markers decrease following 4-20 weeks of high fat diet feeding [5][6][7][8]. A more recent publication examining numerous transcriptomic studies identified 31 hepatic gene candidates that are induced in mice following a high fat diet [9]; specifically, mRNAs related to fatty acid β-oxidation, fatty acid synthesis, and gluconeogenesis were up-regulated; whereas, genes involved in sterol biosynthesis, insulin signaling, and oxidative stress defense were downregulated following high fat feeding. Therefore, it is apparent that high-fat/high-sugar diets exert profound and rapid alterations in the liver transcriptome which precede and/or coincide with disease progression.
We recently performed a sub-chronic (6-week) feeding study examining the effects of an antioxidant supplement on markers of liver health during concomitant WD feeding in 9-month old adult Brown-Norway rats [10]. When comparing WD ad libitum-fed rats versus rats provided control diet (CTL; low fat, low sugar) groups, we found pronounced hepatic lipid accumulation, as determined by Oil Red O and H&E staining, suggestive of a NAFLD phenotype. However, it remains to be determined if lipid accumulation with sub-chronic WD feeding altered or coincided with serum and liver markers suggestive of a pro-inflammatory response. Furthermore, there are a limited number of investigations having examined global transcriptomic changes in the liver during the early stages of WD feeding. The information will lead to better understanding of early progression of fatty liver disease, and will help to identify potential interventional targets. Therefore, the purpose of the current study was to examine how 6 weeks of WD versus CTL feedings in adult Brown Norway rats: a) affects the liver transcriptome, and b) affects select serum and liver pro-inflammatory and fibrotic markers.

Animals and dietary feeding paradigm
All experimental protocols were approved by the University of Missouri's Animal Care and Use Committee. Nine-month-old male Brown Norway rats (Charles River Laboratory, O' Fallon, Missouri) were assigned to one of the following two groups for 6 weeks: 1) ad libitum WD feeding (WD; n = 8); and 2) ad libitum low-fat/low sugar diet (CTL; n = 6). WD (TD.88137; Harlan Laboratories) contained the following nutrient make-up: 4.5 kcal/g, 15.2 % protein (expressed as %kcal), 42.7 % carbohydrate (expressed as %kcal), 42.0 % fat (expressed as %kcal), and 0.2 % cholesterol (by weight). Of note, this diet has been used in prior rodent studies to elicit NASH [11,12]. CTL (LabDiet ® Certified CR 14 % protein rodent diet) contained the following nutrient make-up: 3.5 kcal/g, 16.1 % protein (expressed as %kcal), 69.3 % carbohydrate (expressed as %kcal), 14.6 % fat (expressed as %kcal), and 0.014 % cholesterol (by weight). WD also contained a high proportion of sucrose (34 % of the total diet by weight), whereas CTL did not (1.2 % of the total diet by weight).
Rats were singly housed in standard rat cages in temperature-controlled animal quarters (21°C) with a 0700-1900 light: 1900-0700 dark cycle that was maintained throughout the experimental period. Food intakes were carefully measured weekly (i.e., monitoring cage bottoms for morsels, etc.) and total caloric intakes for the duration of the study have been reported elsewhere [10].

Euthanasia procedures
Euthanasia procedures have been described in detail elsewhere [10]. Briefly, on the day of euthanasia, animal cages were removed from the animal quarters between 0800-0900 and food was removed from each cage. Euthanasia took place between 1400-1900 and rats were sacrificed under CO 2 gas in their home cages in order to minimize stress. Whole blood was subsequently removed via heart sticks using a 21-gauge needle and syringe, placed in a serum separator tube, centrifuged at 3200 rev/min for 5 min, serum was aliquoted into 1.7 ml microcentrifuge tubes, and serum was flash frozen for later serum cytokine assessment. Separate median lobe liver sections (~50-100 mg) were subsequently removed using standard dissection techniques and were flash frozen in liquid nitrogen for RNA deep-sequencing procedures, NF-κB pathway phosphoprotein analysis, and liver trichrome staining and triacylglycerol (TAG) analysis respectively.

Liver trichrome staining and TAG content
Histochemical intrahepatic TAG content was determined as previously described [13]. To examine liver morphology, formalin-fixed, paraffin-embedded livers were sectioned and stained with trichrome stain for collagen deposition/fibrosis. Liver RNA-seq methods RNA isolation and cDNA preparation Two micrograms of liver RNA from WD (n = 6) and CTL (n = 6) rats were sent to the University of Missouri's DNA Core for RNA-seq procedures. Of note, only 6 of the 8 WD rats were used for RNA-seq due to resource constraints. High RNA integrity of each sample was confirmed using the BioAnalyzer 2100 automated electrophoresis system (Bio-Rad, Hercules, CA, USA) prior to cDNA library construction. cDNA library preparation was subsequently performed using the manufacturer's protocol with reagents supplied in Illumina's TruSeq RNA sample preparation kit v2. Poly-A containing mRNA was purified from 2 μg of total RNA, RNA was fragmented, double-stranded cDNA was generated from fragmented RNA and the index containing sample identifier adapters were ligated to the ends. The final construct of each purified library was evaluated using the BioAnalyzer 2100 automated electrophoresis system, quantified with the Qubit fluorometer using the quant-iT HS dsDNA reagent kit (Invitrogen, Life Technologies, Grand Island, NY), and diluted according to Illumina's standard sequencing protocol for sequencing on the HiSeq 2000.
Illumina sequencing of NAc cDNA and statistical analyses of RNA-seq data RNA-seq procedures occurred at the University of Missouri DNA Core and are described in more detail elsewhere [14]. Briefly, following cDNA library construction, samples were loaded onto a flow cell where clusters of each oligo were replicated. Following this procedure, flow cells were placed in the sequencer and fluorescently-labeled bases were attached to the complementary bases of each sequence. The Illumina Genome Analyzer recorded 50 bp reads. Reads were trimmed to ensure adaptor sequence removal and tiled to a custom reference using NextGENe v1.92 (SoftGenetics, State College, PA).

Statistics and bioinformatics for liver RNA-seq data
Differential gene expression patterns were analyzed for annotated genes between the WD and CTL groups using reads per kilobase per million mapped reads (RPKM) values. Differentially expressed liver mRNAs were considered to be significant when WD/CTL fold-change false discovery rate (FDR) p-values were < 0.05. Liver transcripts that were found to be altered with WD feeding were subsequently entered into Ingenuity Pathway Analysis (Qiagen) to elucidate gene networks and/or biological functions within the liver that were altered.

Statistics for non-RNA-seq data
Unless otherwise stated, dependent variables are presented as mean ± standard error and between-group comparisons were performed using independent samples t-tests. Significance was set at an alpha level of 0.05.

Results
Phenotypic changes following 6 weeks of WD versus CTL feeding Phenotypic changes in WD versus CTL rats are presented in Table 1 below. As reported previously [10], WD feeding significantly increased body mass at sacrifice by 6 % (p = 0.0019), omental fat pad mass by 52 % (p < 0.001), perirenal fat pad mass by 74 % (p < 0.001), and liver fat deposition by 4.2-fold (p = 0.004). We also previously reported that WD rats consumed 15 % more calories compared to CTL rats (p < 0.001), with a diet higher in total fat, total sucrose, and total cholesterol. We also expand upon our previous findings of increases in neutral lipid staining by Oil-Red O in WD rats by reporting that liver TAG content was 2.2 fold higher in WD versus CTL rats (p < 0.001; Table 1). Moreover, trichrome staining revealed no difference in liver fibrosis in WD versus CTL rats (data not shown).

WD feeding disrupts the hepatic expression of genes involved with cholesterol and fatty acid synthesis
Using a statistical cut-off of FDR < 0.05, we demonstrate that 6 weeks of WD feeding up-regulated 46 liver transcripts and down-regulated 48 transcripts. IPA readouts demonstrate that the top regulated gene network in WD versus CTL rats included 'Lipid metabolism, small molecular biochemistry, vitamin and mineral metabolism' (IPA score 61; Fig. 1). Serum cholesterol levels tended to be higher in WD-fed rats as previously reported [10] (Fig. 2a), which coincides with the aforementioned RNA-seq readout suggesting a dysregulation in cholesterol biosynthesis-related transcripts in WD versus CTL rats (Fig. 2b). Interestingly, the top regulated canonical signaling pathway in WD versus CTL rats was the 'Superpathway of cholesterol biosynthesis' (10/29 genes down-regulated, p = 1.68E-17; Fig. 2c).
Of the transcripts meeting the statistical cut-off of FDR < 0.05, only 4 mRNAs were up-regulated > 2.0-fold between WD versus CTL rats, whereas 17 mRNAs were down-regulated > −2.0-fold ( Table 2) WD feeding does not affect systemic or hepatic proinflammatory markers, but modestly affects markers of the acute phase protein response Surprisingly, six weeks of WD feeding did not significantly affect serum cytokines (Fig. 3a-e) and/or liver phospho-IkBα and/or phospho-NF-κB p65 patterns ( Fig. 3f/g). Likewise, RNA-seq readouts for genes related to TNF-α signaling, IL-6 signaling, and IL-1β signaling suggest that 6 weeks of WD feeding only elicits modest alterations in pro-inflammatory cytokine pathway signaling genes (Table 3). Specifically, IL-1β mRNA and Jak3 mRNA (IL-6 downstream effector) were up-regulated in WD versus CTL rats (nominal p-value < 0.05), albeit these increases did not meet the FDR cut-off values. Interestingly, WD feeding caused a differential regulation of liver mRNAs related to the acute phase response (Table 3). Specifically, WD feeding caused an upregulation in platelet factor 4 (Pf4) mRNA (Table 3) which is a putative chemoattractant protein that recruits neutrophils and monocytes. When examining a Values are means ± SE. Body mass, total kcal consumed, and Oil-Red O data previously reported [10]; symbols: g, grams; AUs, arbitrary units spectrum of acute phase response genes, WD feeding also significantly increased serum amyloid P component (Apcs) mRNA, haptoglobin (Hb) mRNA and superoxide dismutase 2 (Sod2) mRNA while significantly decreasing apolipoprotein H (Apoh) mRNA and lipopolysaccharide binding protein (Lpb) mRNA. Finally, WD feeding down-regulated the mRNA expression of the growth hormone receptor (Ghr) (WD/CTL: −1.33, FDR = 0.04). This is of particular interest given that: a) the liver is a very responsive tissue to growth hormone, and b) the macrophage-specific knockout of Ghr causes an enhanced inflammatory response in peripheral tissues when animals were challenged with a high-fat diet [15]. Collectively, these modest transcript alterations may be initial changes that occur with the development of WD-induced liver inflammation that has been shown to occur with longer feeding schedules [16,17].

Discussion
These RNA-seq data-mining efforts provide a unique snapshot of liver transcriptomic alterations that occur with sub-chronic WD feeding in adult Brown Norway rats. While it has demonstrated that longer high-fat/ high-sugar feeding schedules cause increases in liver inflammation [17], oxidative stress [18], fibrosis [17,18] and mitochondrial dysfunction [19,20], we here provide evidence that sub-chronic WD feeding causes a rapid dysregulation in select hepatic lipid and cholesterol metabolism genes in rats; an effect which: a) may contribute to the doubling in hepatic TG content presented herein; and b) may be responsible for serum cholesterol levels tending to increase after 6 weeks of WD feeding as previously reported [10]. Furthermore, the RNA-seq analyses reveal that while a large majority of pro-inflammatory mediators remained unaltered with western diet feeding, select liver transcripts related to acutephase inflammation signaling were altered with subchronic WD feeding in the setting of hepatic steatosis.

Six weeks of WD feeding primarily affect hepatic mRNA expression patterns related to lipid metabolism and cholesterol biosynthesis
The primary finding from the current study is that subchronic WD feeding causes rapid alterations in select lipid and cholesterol metabolism genes which, in turn, likely contributes to the observed increases in liver fat accumulation and dysregulation of serum cholesterol levels. The striking 28.3-fold increase in liver Scd1 mRNA with 6 weeks of WD feeding emphasizes the potential importance of this gene in the development of NAFLD. The Scd1 enzyme catalyzes monounsaturated long-chain fatty acid synthesis from saturated fatty acyl-CoAs [21], and Scd1-deficient mice have been  shown to possess less liver fat accumulation [22][23][24]. Indeed, pharmacologically [25] and genetically [26] knocking down Scd1 has been shown to favorably alter fatty acid composition in mouse liver tissue. Thus, our data is in agreement with the aforementioned literature suggesting that a dietary-induced overexpression of liver Scd1 may be crucial in the early development of NAFLD. Liver cytochrome P450 polypeptide 18 (Cyp2c18) was also up-regulated 2.7fold in WD versus CTL rats. In adipocytes, Cyp2c18 is thought to work in concert with Scd1 to oxidize fatty acids once they are desaturated [27]. Cyp2c18 is also thought to be involved in hepatic arachidonic acid and linoleic acid metabolism [28] and has been shown to be up-regulated in other models of NAFLD [29]. Contrary to a select number of hepatic genes being >2fold up-regulated in WD versus CTL rats, more hepatic genes were >2-fold down-regulated in WD versus CTL rats with several of these genes being involved in cholesterol biosynthesis which is likely a partial reflection in the increased cholesterol content in the WD. Interestingly, Values are means ± SE. WD feeding tended to elevate serum leptin compared to CTL rats (p = 0.07), while not affecting other pro-inflammatory serum markers. Moreover, WD feeding did not affect liver markers of NF-κB signaling hepatic Tm7sf2, which encodes the endoplasmic reticulum enzyme 3β-hydroxysterol Δ14-reductase that is involved with cholesterol biosynthesis [30], was robustly downregulated with WD feeding. Recent evidence suggests that Tm7sf2 −/− mice present increases in NF-κB activation; this being a hallmark feature of inflammation signaling. Thus, while markers of NF-κB signaling were not altered with WD feeding in the current study, this may be an early transcriptomic event that precedes NF-κB signaling with NAFLD development. Other notable hepatic genes downregulated in WD included: a) Cyp51a1 and Fdps which have been genotype-associated with higher HDL-C levels [31] and NAFLD [32] in humans, respectively; b) Hmgcs1 which has been shown to be positively associated with HDL-C levels [33]; and c) Insig1 which is a gene that, when overexpressed in livers of Zucker diabetic fatty rats, attenuates hepatic steatosis [34]. Interestingly, hepatic Pcsk9, which encodes for a serine protease that reduces LDL receptor content [35], was paradoxically down-regulated in WD versus CTL rats. The WD-induced down-regulation in hepatic Acly mRNA is also paradoxical given that its inhibition in db/db mice has been shown to markedly protect against de novo lipogenesis and hepatic steatosis [36]. Notwithstanding, we contend that the marked dysregulation in hepatic lipid metabolism and cholesterol biosynthesis genes after 6 weeks of WD feeding is reflective of the phenotypic changes we report at the tissue level. Moreover, these genetic alterations likely precede hepatic inflammation and fibrosis development.
Six weeks of WD feeding affect select hepatic acute phase protein and fibrosis-related mRNA expression patterns, but not local or systemic pro-inflammatory biomarkers Various markers of liver inflammation were generally not elevated with 6 weeks of WD feeding in adult Brown Norway rats despite significant development of hepatic steatosis. These results extend mouse data to rats. Stanton et al. [16] previously demonstrated that C57BL/6 mice fed a high fat/high cholesterol diet for 6 weeks did not present increases in several pro-inflammatory liver chemokine mRNAs, but instead presented significant increases in these markers 16 and 26 weeks after feeding.
Notwithstanding, we demonstrated that the mRNA expression of numerous acute phase proteins was altered (some increasing while others decreasing) following 6 weeks of WD feeding in adult rats. Acute phase proteins are proteins produced by the liver that are thought to increase or decrease in the plasma by 25 % in response to an immunological or inflammatory stimulus [37]. The presence of an acute phase response to NAFLD and NASH also extends to mouse data [38,39]. Radonjic et al. [40] demonstrated that liver mRNAs for several acute phase reactants within days of high-fat diet feeding in ApoE3L mice. Hence, our current data suggests that liver acute phase response genes can be induced early during WD-induced NAFLD; a finding which may demarcate an early pro-inflammatory stimulus. Interestingly, the 2-fold WD-induced increase in hepatic platelet factor 4 (Pf4 or Cxcl4) mRNA may be an early indicator or even initiator of fibrosis that follows the development of NAFLD. To our knowledge, no other studies using models of NAFLD have shown this gene to change, although Pf4 has been shown to induce fibrosis [41]. Indeed, Pf4 mRNA has been shown to be up-regulated in alcoholic hepatitis patients [42]. Recombinant murine Cxcl4 has also been shown to stimulate the proliferation, chemotaxis, and chemokine expression of hepatic stellate cells [41]; a phenomena which suggests that there may be a potential role for its induction in the development of nonalcoholic steatohepatitis.
Of the serum pro-inflammatory cytokines measured in the current study, only serum leptin tended to be 2.4 times greater in WD versus CTL rats. Indeed longer high-fat and/or high-sugar feeding schedules have been shown to increase serum levels of select proinflammatory cytokines [43]. While leptin is mainly regarded as an adipokine involved with appetite regulation, some evidence supports its role in promoting an inflammation response [44,45]. For instance, leptindeficient ob/ob mice present with impairments in macrophage function [46]. Other studies have also shown that leptin is required for hepatic fibrosis development in leptin-deficient ob/ob mice model [47,48]. Thus, our data supports the hypothesis that WDinduced increases in leptin levels may precede and even be permissive for subsequent liver fibrosis to occur with a longer term WD challenge. Finally, the lack of increased NF-κB signaling in WD-fed rats may be due to the short-term feeding schedule employed in the current study. Indeed, genetically-altered murine models (Ldlr −/− , and Apoe2-knock-in) have been shown to present robust hepatic inflammation (via NF-κB signaling) and steatosis within 21 days of high fat/high cholesterol feeding, although wild-type mice in this same study developed only ectopic fat deposition without overt inflammation [49]. Notwithstanding, the 6-week WD feeding in the current study caused hepatic steatosis but appeared to be insufficient in duration to develop histological NASH with inflammation and fibrosis.

Conclusions
The current study provides a unique perspective in adult rats as to how a Westernized diet affects liver transcriptomic changes over an initial 6-week feeding period.
The transcriptome-wide interrogation employed herein gives insight into the rapid liver gene expression changes that occur with early WDinduced hepatic steatosis development prior to the development of secondary outcomes (i.e., inflammation and fibrosis) and support other transcriptomic work in this sub-chronic timeframe [50,51]. This study points to the potential sensitivity of the rodent liver to dietary fat, sucrose, and cholesterol and helps add to the framework for other researchers interested in examining gene targets or metabolic pathways that occur with NASH development. Our data suggest that the expression of numerous lipid metabolism and cholesterol biosynthesis genes are rapidly altered with sub-chronic WD feeding; an effect which was manifested with significant hepatic steatosis. Furthermore, the 28-fold increase in liver Scd1 mRNA with subchronic WD feeding continues to underscore the potential importance of this gene in the development of NAFLD. While WD feeding caused alterations in select acute phase response mRNAs, other proinflammatory signaling mediators and mRNAs appear unaltered with 6 weeks of WD feeding. Hence, this may suggest that diet-induced liver fibrosis and inflammation may either: a) occur much later with more chronic WD rat feeding paradigms; and/or b) may be initiated by other factors that precede eventual perturbations in local and systemic proinflammatory signalers with chronic WD feeding.