Vascular transcriptional alterations produced by juvenile obesity in Ossabaw swine

Abstract

We adopted a transcriptome-wide microarray analysis approach to determine the extent to which vascular gene expression is altered as a result of juvenile obesity and identify obesity-responsive mRNAs. We examined transcriptional profiles in the left anterior descending coronary artery (LAD), perivascular fat adjacent to the LAD, and descending thoracic aorta between obese (n = 5) and lean (n = 6) juvenile Ossabaw pigs (age = 22 wk). Obesity was experimentally induced by feeding the animals a high-fat/high-fructose corn syrup/high-cholesterol diet for 16 wk. We found that expression of 189 vascular cell genes in the LAD and expression of 165 genes in the thoracic aorta were altered with juvenile obesity (false discovery rate ≤ 10%) with an overlap of only 28 genes between both arteries. Notably, a number of genes found to be markedly upregulated in the LAD of obese pigs are implicated in atherosclerosis, including ACP5, LYZ, CXCL14, APOE, PLA2G7, LGALS3, SPP1, ITGB2, CYBB, and P2RY12. Furthermore, pathway analysis revealed the induction of proinflammatory and pro-oxidant pathways with obesity primarily in the LAD. Gene expression in the LAD perivascular fat was minimally altered with juvenile obesity. Together, we provide new evidence that obesity produces artery-specific changes in pretranslational regulation with a clear upregulation of proatherogenic genes in the LAD. Our data may offer potential viable drug targets and mechanistic insights regarding the molecular precursors involved in the origins of overnutrition and obesity-associated vascular disease. In particular, our results suggest that the oxidized LDL/LOX-1/NF-κB signaling axis may be involved in the early initiation of a juvenile obesity-induced proatherogenic coronary artery phenotype.

in contrast to previous generations, physical inactivity and overabundance of food are traits of contemporary living (6–8, 12, 76). These poor lifestyle modifications have occurred too rapidly for the human genome to adapt, with the consequences manifested as an explosion of metabolic disorders (i.e., obesity) and ultimately cardiovascular complications and death (6–8, 12, 76). The epidemic of obesity is most alarming in youth as current statistics indicate that one-third of US children and adolescents are overweight or obese (58). Childhood obesity is a global problem and increasingly extends into the developing world (44). Remarkably, today's children could be the first generation in over a century to experience a decline in life expectancy due to the epidemic of childhood obesity (59). Indeed, obesity during early life is associated with a variety of adverse health outcomes in adulthood, particularly coronary artery disease (34, 42, 54). Not only are obese children presenting with risk factors for later disease, they now can exhibit early signs of atherosclerosis and are often diagnosed with what were traditionally considered to be adult diseases (e.g., Type 2 diabetes) (3, 5, 50). Although it is becoming more accepted that obesity during early life increases the susceptibility to cardiovascular diseases, the early molecular changes (i.e., alterations in mRNAs) at the artery wall caused by childhood obesity have not been characterized. To begin addressing this knowledge gap, the development of an animal model suited to the study of childhood obesity is warranted.

Rodents are commonly utilized for the study of adult obesity; however, their small size and rapid rate of maturation limit the viability of using young rodents as a human model. In addition, rodents develop complex atherosclerotic lesions only when genetically modified. Recently, the fidelity of inflammatory/immune responses in the mouse model to replicate that of humans has been challenged (71). As an alternative, the pig is emerging as an important nonprimate biomedical model for human research due to its similar genetics, metabolic features, cardiovascular system, proportional organ sizes, and pathophysiology (21, 82). In particular, the Ossabaw pig has a number of advantages among porcine models available for the study of obesity-associated metabolic and cardiovascular diseases (25, 57). These animals have a “thrifty genotype” that enabled them to survive seasonal food shortages in their native environment (75). Similar to humans in an environment of nutritional excess and physical inactivity, when mature Ossabaw pigs are fed a high-fat diet and remain inactive, they develop classic features of the metabolic syndrome, including obesity, insulin resistance, glucose intolerance, dyslipidemia, and hypertension, and exhibit complex atherosclerotic lesions (9, 25, 57, 65).

Herein, we utilized juvenile Ossabaw pigs fed a high-fat/high-fructose corn syrup/high-cholesterol diet to model what is likely occurring in current-day children and adolescents with regard to the mechanistic initiation of cardiovascular disease associated with overnutrition and obesity. We adopted a transcriptome-wide microarray analysis approach to determine the extent to which coronary artery gene expression is altered as a result of juvenile obesity and identify obesity-responsive genes, thus providing clues regarding the initial molecular mechanisms that precede the development of coronary artery disease associated with obesity. We examined transcriptional profiles in the left anterior descending coronary artery (LAD), perivascular fat adjacent to the LAD, and the descending thoracic aorta in obese and lean juvenile Ossabaw pigs. Moreover, using state-of-the-art bioinformatic techniques we examined the impact of juvenile obesity on gene networks and interactions between individual components of these networks. In general, we hypothesized that juvenile obese pigs, relative to lean pigs, would exhibit altered expression of vascular cell genes involved in inflammation and oxidative stress pathways and that such between-group differences would be most pronounced in the LAD as this artery is highly susceptible to atherosclerosis in adulthood (2, 18). Furthermore, in view of the evidence that altered expression and secretion of inflammatory cytokines from adipose tissue may contribute to the initiation and progression of atherosclerosis associated with obesity (13, 65, 66), we hypothesized that alterations in gene expression in the LAD induced by juvenile obesity would be accompanied by an increased expression of proinflammatory genes in the surrounding perivascular fat. Finally, because bioinformatic analysis indicated that the NF-κB complex is a central component of the LAD gene network influenced by obesity, we focused additional experiments to test the hypothesis that NF-κB plays a key role in regulating the induction of genes that our results indicate are highly upregulated with juvenile obesity in the LAD.

METHODS

Experimental animals.

Prior to initiation of the study, approval was received from the Animal Care and Use Committee at the University of Missouri. Female Ossabaw pigs were generously provided by Michael Sturek, Ph.D., in the Ossabaw Swine Resource, Comparative Medicine Program at Purdue University and Indiana University School of Medicine (IUSM). To utilize the Ossabaw pig as a model of childhood obesity, we randomly divided 5–6 wk old Ossabaw pigs (n = 11) into two experimental groups and fed either commercially available regular miniature pig chow diet (5L80, Lab Diet; 3.03 kcal/g-10.5, 71, and 18.5% by energy for fat, carbohydrate, and protein, respectively; n = 6) or high-fat/high-fructose corn syrup/high-cholesterol (5B4L, Lab Diet; 4.14 kcal/g-43, 40.8, and 16.2% by energy for fat, carbohydrate, and protein, respectively; 17.8% high fructose corn syrup; n = 5) diets for 16 wk. This age in miniature swine is equivalent to adolescent/juvenile humans. All pigs were individually housed in a core animal care facility at the University of Missouri under temperature-controlled conditions (20–23°C) with a 12 h/12 h light-dark cycle. Animals were provided food daily. Body composition was measured via dual-energy X-ray absorptiometry (Hologic QDR-1000) at 22 wk.

Tissue sampling.

Following an 18–20 h fast, 22 wk old pigs were sedated with Telazol (5 mg/kg im) and Xylazine (2.2 mg/kg im) by injection. An ear vein catheter was placed, and the animals were deeply anesthetized with Telazol (10 mg/kg iv) and Xylazine (2.2 mg/kg iv). Blood for serum and plasma analyses was collected via jugular vein access. The animals were euthanized by removal of the heart in full compliance with the American Veterinary Medical Association Guidelines on Euthanasia. The LAD and descending thoracic aorta were dissected clean of perivascular fat and excess adventitia. Both arteries and the perivascular fat adjacent to the LAD were rinsed with ice-cold Krebs saline and immediately frozen at −80°C or used for functional assessment (see below). Samples were homogenized with a mortar and pestle. Isolation of total RNA was performed with the Qiagen's RNeasy Mini Kit (Qiagen, Valencia, CA) for the arteries and the Qiagen's RNeasy Lipid Tissue Kit for the perivascular fat. The product was eluted in a 100 μl volume with a 2 μl aliquot being used for spectrophotometric quantification. We used 25 ng each sample in an Experion StdSens RNA Analysis to confirm optimal concentration and optimal quality of RNA.

Serum and plasma measures.

Fasting serum glucose, insulin, nonesterified fatty acids (NEFAs), total cholesterol, low-density lipoprotein-cholesterol (LDL-c), high-density lipoprotein-cholesterol (HDL-c), and triglycerides (TGs) were analyzed on an Olympus AU680 automated chemistry (Beckman-Coulter, Brea, CA) analyzer using commercially available assays according to manufacturer's guidelines. In addition, plasma levels of oxidized LDL (oxLDL) were assessed using a commercially available ELISA kit according to manufacturer's instructions (Mercodia, Uppsala, Sweden).

Microarray analysis.

Biotin-labeled antisense RNA (aRNA) target was produced from porcine tissue total RNA (0.5 μg) using the MessageAmp Premier RNA amplification kit (Ambion, Austin, TX) according to manufacturer's procedures and as described previously (63). Briefly, the total RNA was reverse-transcribed to first-strand cDNA with an oligo(dT) primer bearing a 5′-T7 promoter using ArrayScript reverse transcriptase. The first-strand cDNA then underwent second-strand synthesis to produce double-stranded cDNA template for in vitro transcription. The biotin-labeled aRNA was synthesized using T7 RNA transcriptase with biotin-NTP mix. After purification, the aRNA was treated with 1× fragmentation buffer at 94°C for 35 min. We hybridized 130 μl of hybridization solution containing 50 ng/μl of fragmented aRNA to the porcine genome array GeneChip (Affymetrix, Santa Clara, CA) at 45°C for 20 h. After hybridization, the chips were washed and stained with R-phycoerythrin-streptavidin on an Affymetrix Fluidics Station 450 using Fluidics protocol Midi_euk2v3-450. The microarrays were then scanned and data acquired using an Affymetrix GeneChip Scanner 3000 driven by GCOS 1.2 software (Affymetrix). All raw microarray data (32 arrays from 11 unique animals) have been deposited in the Gene Expression Omnibus (GEO) repository, which can be publicly accessed at http://www.ncbi.nlm.nih.gov/geo/. The data are MIAME compliant, and the GEO accession number is GSE41636.

Quantitative real-time PCR.

To verify the microarray results, we selected 10 LAD genes found to be altered with juvenile obesity. First-strand cDNA was synthesized from total RNA (same RNA extract used for microarray) by reverse transcription primed by a mixture of random hexamer and oligo(dT) primers (iScript cDNA synthesis kit; Bio-Rad, Hercules, CA). The reactions were incubated in a PCR Express Hybaid thermal cycler (Hybaid, Franklin, MA). Quantitative real-time PCR was performed as previously described (61) using the ABI 7900 sequence detection system (Applied Biosystems, Foster City, CA). Primer sequences (Table 1) were designed using the National Center for Biotechnology Information (NCBI) website. All primers were purchased from IDT (Coralville, IA). A 25 μl reaction mixture containing 20 μl of Power SYBR Green PCR Master Mix (Applied Biosystems) and the appropriate concentrations of gene-specific primers plus 5 μl of cDNA template was loaded in each well of a 96-well plate (duplicate samples). PCR was performed with thermal conditions as follows: 95°C for 10 min, followed by 40 cycles of 95°C for 15 s and 60°C for 45 s. A dissociation curve analysis was performed after each run to verify the specificity of the PCR products. Swine GAPDH primers (85) were used to amplify the endogenous control product. Our group has previously established that GAPDH is a suitable housekeeping gene for RT-PCR when examining porcine vascular gene expression (83). In the present study, GAPDH threshold cycles were not significantly different between lean and obese pigs (20.89 ± 0.36 vs. 20.97 ± 0.22, respectively; P = 0.87). Furthermore, microarray analysis indicated the absence of an obesity effect on GAPDH mRNA levels. mRNA expression values are presented as 2^ΔCT, whereby ΔCT = GAPDH CT − gene of interest CT (61).

Table 1.

Forward and reverse primer sequences for quantitative real-time PCR

	Primer Sequence (5′→3′)
Gene	Forward	Reverse
GAPDH	GGGCATGAACCATGAGAA GT	GTCTTCTGGGTGGCAGTGAT
ACP5	GATCCCACGGTCCAATGTGT	GCCACCAGCACATAGTCCTC
LYZ	CCCGGCTTCTCAGACAACAT	CCTATAGCCGTCCATGCCAG
CXCL14	GGGGGAAACTCGCAAAGGTA	ACTGTTCAGCACGGACGAAT
APOE	GCAAGCCAGAAGATGAGGGT	CAGCGCAGGTAATCCCAGAA
PLA2G7	ACTTCACTTTTGCAACCGGC	GTTGCTAAGGCCAAGGGCTA
LGALS3	AATGGCAGACGGTTTTTCGC	AGGCCATCCTTGAGGGTTTG
SPP1	CAGACTTTCCTAGCGCCACA	CTCAGGGCTTTCGTTGGACT
ITGB2	CCAAGAAGGTTTCGAGGGCT	CAAAGTCACACTGGCACACG
CYBB	GACAGCTGGACAGGAACCTC	AGAATTGTTGACTCGGGCGT
P2RY12	AACAGACGGCCAAGTGACAA	CCTACGCCCCTGGTTCTTAC

Functional assessment of isolated LAD rings.

Distal end of the LADs were dissected from the heart. Arteries were cut into 2 mm rings and isometrically mounted in a myograph (model 610M; DMT, Aarhus, Denmark) and maintained in PSS in 95% O₂-5% CO₂ at 37°C for the remainder of the experiment. Coronary rings were preconstricted with 10 nM U-46619 to induce ∼70–80% maximal contraction (i.e., relative to maximal U-46619-induced contraction; data not shown). Concentration-response curves were obtained by cumulative addition of either bradykinin (10⁻¹²–10⁻⁷ M) or sodium nitroprusside (10⁻⁹–10⁻⁵ M). We assessed the contributions of superoxide radicals and nitric oxide in mediating bradykinin-induced relaxation by incubating the vessels with TEMPOL (a membrane-permeable superoxide dismutase mimetic, 3 mM, 60 min) and l-NG-nitro-l-arginine (l-NAME, a nitric oxide synthase inhibitor, 100 μM, 20 min), respectively. Relaxation at each concentration was measured and expressed as percent maximum relaxation, where 100% is equivalent to loss of all tension developed in response to U-46619.

Western blot analyses for lectin-like oxLDL receptor-1.

For Western blot analysis for lectin-like oxLDL receptor-1 (LOX-1, Abcam), segments of LAD arteries were homogenized and sonicated in 250 μl lysis buffer (Cellytic MT Mammalian Tissue Lysis/Extraction Reagent, Sigma) with protease inhibitor (Sigma). Protein concentration was measured with the BCA protein assay kit (Pierce). Equal amounts of protein (15 μg) were loaded and separated by 10% SDS-PAGE and transferred to nitrocellulose membranes (Bio-Rad). GAPDH (IMGENEX) and tubulin (Abcam) were used as internal controls. Signals were visualized by enhanced chemiluminescence and scanned with a Fuji LAS 3000 densitometer. The relative amounts of protein expression were quantified and normalized to corresponding lean control, which were set to a value of 1.0.

Endothelial cell culture experiments.

Porcine aortic endothelial cells harvested from a sexually matured male farm pig were maintained under standard culture conditions with Porcine Endothelial Cell Growth Medium (Cell Applications, San Diego, CA). At passage 3, after reaching >95% confluence, endothelial cells were treated for 1 h with lipopolysaccharide (LPS, 10 μM), a known proinflammatory stimulus, or vehicle control with and without wedelolactone (20 μM), an NF-κB inhibitor. Experiments were performed in a 24-well plate, and six replicates were obtained per condition. After treatments, endothelial cells were lysed with TRIzol, and cDNA was made from isolated RNA per the methods described above.

Statistical analysis.

The primary analysis of microarray gene expression data was conducted using the Linear Models for Microarray Data (limma) package(29) and the affy package (28), available through the Bioconductor project (30) for use with R statistical software, as previously described (63). Data quality was examined via an extensive battery of metrics obtained using the affyQCReport package. Within affy, quantile cubic-spline normalization (qspline) was used for between-chip normalization and RMA for background adjustment. For summarizing probe set measurements (n = 24,123 total probe sets), we used median polish and PM-only probes. After normalization, internal control probe sets (n = 188) were excluded from further statistical analyses. Finally, nonspecific filtering was carried out on probe sets that were not putatively expressed across the tissues (as defined as signal intensity >100 for at least one-third the samples); these were excluded (n = 9,608) from further statistical analyses to reduce false positives, as they contain very little information content relative to the analysis at hand.

After this preprocessing was completed, the statistical analysis was performed by an empirical Bayesian moderated t-test (72) applied to normalized intensity for each gene (n = 14,327 remaining after preprocessing), where the obese group was compared with the lean group within each of the three tissue types. Since multiple tissue types were from the same animal, the dependency between measurements was accounted for by a modified mixed linear model that treated each animal as a block. The within-block correlations were constrained to be equal between genes (73), and then information was borrowed across genes to moderate the standard deviations between genes via an empirical Bayes method. The comparisons are expressed as fold changes (obese/lean) along with adjusted P values. Adjustment to the P values was made to account for multiple testing by the false discovery rate (FDR) method of Benjamini and Hochberg (4). We chose 10% as our FDR cutoff for declaring statistical significance. Because of the sparseness of the direct annotations with the Affymetrix Porcine array, we built a custom annotation package for the array using annotations provided through ANEXdb annotations (http://www.anexdb.org/), described in detail by Couture et al. (19). To our knowledge, this is the most comprehensive source of annotations of the porcine GeneChip array. In brief, it is based around the Iowa Porcine Assembly, which is a novel assembly of ∼1.6 million unique porcine-expressed sequence reads annotated through homolog sequence alignment to NCBI RefSeq. The result is a very dense annotation of the porcine-specific probe sets (94%), of which 80% are linked to an NCBI RefSeq entry.

Networks were generated through the use of Ingenuity Pathways Analysis (Ingenuity Systems, http://www.ingenuity.com), henceforth IPA, as previously described (63). The full list of all ANEXdb-annotated microarray results were uploaded into IPA with Entrez GeneID, fold change, and adjusted P value. Since the porcine array is not included in IPA as a reference set, we used the full contents of the array as a “user dataset”; therefore, it was essential to include all genes represented on the array. Therefore, the reference set consisted of 16,521 genes (out of 24,123 probe sets on the array) that mapped uniquely to its corresponding object in the Ingenuity Knowledge Base (IKB). In the case of duplicate mapped IDs, the median values (fold change and adjusted P value) were used to represent the single results for that object. Genes that were not filtered out during preprocessing (as previously described) were assigned a fold change of 1 and an adjusted P value of 1. An adjusted P value <10% was set to identify molecules whose expression was significantly differentially regulated, which IPA terms as network eligible molecules (NEMs). Our networks were built with knowledge obtained from mammalian data (or uncategorized chemicals) and experimentally observed relationships. To construct the networks, NEMs were overlaid onto a global molecular network developed from information contained in the IKB. Networks of NEMs were then algorithmically generated to maximize their specific connectivity with each other relative to all molecules they are connected to in the IKB, where the NEMs served as “seeds” for generating networks. Additional molecules were recruited to merge smaller networks into successively larger networks, by using the default IPA network maximum size of 35 molecules.

Networks were then scored based on the number of NEMs they contained, so that the higher the score, the lower the probability of finding the observed number of NEMs in a given network by random chance. Specifically, the score is the negative log₁₀ of the P value from Fisher's exact test applied to a given network, so that a score of 9 implies a one-in-a-billion chance of obtaining a network containing at least the same number of NEMs when randomly picking 35 molecules from the IKB. For a detailed description of the network generating algorithm see Ref. 11.

Graphical representations of the networks were generated with IPA's Path Designer, which illustrates the relationships between molecules. Molecules are represented as nodes, and the biological relationship between two nodes is represented as an edge (connecting line). All edges are supported by at least one reference from the literature, from a textbook, or from canonical information stored in the IKB.

In addition, between-group differences for all descriptive variables were determined by an independent t-test. Dose-response curves from vasomotor relaxation experiments were analyzed by two-way ANOVA with repeated measures on one factor (concentration). For all nonmicroarray related statistical tests, significance was set at 0.05.

RESULTS

Over the course of the study, lean pigs consumed an average of 550 g of food per day, whereas the obese pigs averaged 1,200 g of food per day. The significant increase in body weight and percent body fat induced by the high-fat/high-fructose corn syrup/high-cholesterol experimental diet is presented in Table 2. In addition, fasting serum levels of insulin, total cholesterol, LDL-c, HDL-c, LDL-c/HDL-c ratio, total cholesterol/HDL-c ratio, NEFAs, and TGs, but not fasting glucose, were significantly higher in the obese compared with the lean group (Table 2). Furthermore, we found that plasma oxLDL and LOX-1 protein expression in LAD were greater in the juvenile obese pigs compared with lean pigs (Fig. 1). Vasomotor relaxation in the LAD was not significantly altered in obese vs. lean juvenile pigs. As depicted in Fig. 2, bradykinin-induced relaxation was similar between groups under control conditions, after pretreatment with TEMPOL, and after pretreatment with l-NAME (all P > 0.05). In addition, sodium nitroprusside produced similar (P > 0.05) magnitudes of vascular relaxation between obese and lean pigs. Of note, coronary arteries from obese pigs exhibited a blunted contraction induced by 10 nM of U-46619 compared with arteries from lean pigs (17.1 ± 3.9 vs. 41.3 ± 7.3 mN, P = 0.02).

Table 2.

Body composition and fasting serum characteristics in lean and obese juvenile pigs

	Lean	Obese	P Value
Body weight, kg	25.1 ± 0.5	47.1 ± 2.4	0.0001
Percent body fat, %	20.4 ± 2.0	30.4 ± 1.4	0.004
Total cholesterol, mg/dl	91.0 ± 12.2	673.8 ± 205.4	0.012
LDL-c, mg/dl	40.5 ± 6.9	348.4 ± 82.4	0.003
HDL-c, mg/dl	45.7 ± 4.2	74.0 ± 8.4	0.011
LDL-c:HDL-c	0.9 ± 0.1	5.4 ± 1.8	0.019
Total cholesterol:HDL-c	2.0 ± 0.1	10.1 ± 3.3	0.022
NEFA, mmol/l	0.3 ± 0.1	1.6 ± 0.4	0.012
Triglycerides, mg/dl	33.7 ± 8.9	71.8 ± 27.0	0.181
Glucose, mg/dl	128.3 ± 14.3	155.6 ± 21.0	0.298
Insulin, ng/l	0.03 ± 0.01	0.08 ± 0.02	0.041

Values are means ± SE. LDL, low-density lipoprotein; HDL, high-density lipoprotein, NEFA, nonesterified fatty acids.

Fig. 1.

Plasma oxidize (ox)LDL and lectin-like oxLDL receptor-1 (LOX-1) protein expression in left anterior descending coronary arteries (LAD) in lean and obese juvenile pigs. Values are means ± SE. *Significantly different from lean. P < 0.05.

Fig. 2.

Vasomotor relaxation of LAD rings in lean and obese juvenile pigs. Values are means ± SE.

As depicted in Fig. 3A, a total of 171 genes were upregulated and 18 genes downregulated in the LAD with juvenile obesity (FDR < 10%). In contrast, a total of 107 genes were upregulated and 58 genes downregulated in the descending thoracic aorta using the same criteria. The effects of juvenile obesity produced an overlap of 28 genes between the LAD and the thoracic aorta, and the direction of change for those genes was the same in both arteries (26 upregulated and 2 downregulated with juvenile obesity) (Fig. 3B). For brevity, Tables 3 and 4 provide the list of the top 20 significant annotated probe sets with greatest difference (in terms of fold change) in LAD and thoracic aorta gene expression between juvenile obese and lean pigs. The full list of all significant probe sets along with their fold change is available in Supplementary Data.¹ In addition, a total of five genes were upregulated and two genes were downregulated in perivascular fat of the LAD with obesity (FDR < 10%, Table 5). Interestingly, only one gene (RSAD2) was universally upregulated with obesity in all three tissues analyzed.

Fig. 3.

Number of genes upregulated and downregulated with juvenile obesity in the LAD and descending thoracic aorta is illustrated in A. Between-artery correlation in changes of gene expression induced by juvenile obesity is illustrated in B. Effects of juvenile obesity produced an overlap of 28 genes between the LAD and thoracic arteries. Each dot represents a gene. As illustrated, the direction of change for those genes was the same in both arteries (26 upregulated and 2 downregulated with juvenile obesity). Dotted line depicts perfect agreement.

Table 3.

List of top 20 annotated probe sets with greatest difference in LAD gene expression between juvenile obese and lean pigs (sorted by magnitude of fold change)

ID	Gene Symbol	Gene Title	Adjusted P Value	Fold Change (Obese/Lean)
Ssc.1126.1.A1_at	SMPDL3A*	sphingomyelin phosphodiesterase, acid-like 3A	0.01	17.81
Ssc.575.1.S1_at	ACP5	acid phosphatase 5, tartrate resistant	0.01	10.96
Ssc.31102.1.A1_at	SCIN	scinderin	0.02	9.19
Ssc.670.1.S1_at	LYZ	lysozyme	0.06	8.38
Ssc.140.1.S1_at	AMBN	ameloblastin (enamel matrix protein)	0.01	8.27
Ssc.4984.1.S1_at	CXCL14*	chemokine (C-X-C motif) ligand 14	0.01	7.37
Ssc.1342.1.S1_at	APOE	apolipoprotein E	0.02	6.66
Ssc.19691.1.S1_at	PLA2G7	phospholipase A2, group VII (platelet-activating factor acetylhydrolase, plasma)	0.03	5.86
Ssc.17815.1.S1_at	LGALS3	lectin, galactoside-binding, soluble, 3	0.03	5.51
Ssc.17203.3.S1_at	CTSS	cathepsin S	0.01	5.35
Ssc.101.1.S1_at	SPP1	secreted phosphoprotein 1	0.02	4.91
Ssc.9330.1.A1_at	LCP1	lymphocyte cytosolic protein 1 (L-plastin)	0.01	4.82
Ssc.8375.1.A1_at	RNF128	ring finger protein 128, E3 ubiquitin protein ligase	0.02	4.5
Ssc.950.1.S1_at	BASP1	brain abundant, membrane attached signal protein 1	0.06	4.27
Ssc.24721.1.A1_at	SLC27A6	solute carrier family 27 (fatty acid transporter), member 6	0.07	4.22
Ssc.508.1.S1_at	FCER1G	Fc fragment of IgE, high affinity I, receptor for; gamma polypeptide	0.02	4.05
Ssc.14561.1.S1_at	ITGB2	integrin, beta 2 (complement component 3 receptor 3 and 4 subunit)	0.02	4.01
Ssc.151.1.S1_at	CYBB	cytochrome b-245, beta polypeptide	0.02	3.98
Ssc.8449.1.A1_at	P2RY12	purinergic receptor P2Y, G protein-coupled, 12	0.03	3.97
Ssc.248.1.S1_at	NPL	N-acetylneuraminate pyruvate lyase (dihydrodipicolinate synthase)	0.02	3.93

^*Overlap with thoracic aorta. All genes in table have an false discovery rate (FDR) adjusted P value (also called a q value) of ≤10% to account for multiple testing.

Table 4.

List of top 20 annotated probe sets with greatest difference in thoracic aorta gene expression between juvenile obese and lean pigs (sorted by magnitude of fold change)

ID	Gene Symbol	Gene Title	Adjusted P Value	Fold Change (Obese/Lean)
Ssc.4984.1.S1_at	CXCL14*	chemokine (C-X-C motif) ligand 14	0.01	8.14
Ssc.1126.1.A1_at	SMPDL3A*	sphingomyelin phosphodiesterase, acid-like 3A	0.05	8.05
Ssc.90.1.S1_at	CHI3L1	chitinase 3-like 1 (cartilage glycoprotein-39)	0.06	4.69
Ssc.286.1.S1_s_at	RSAD2	radical S-adenosyl methionine domain containing 2	0.01	4.17
Ssc.12145.1.A1_at	MALL	Mal, T-cell differentiation protein-like	0.04	3.53
Ssc.8453.1.A1_at	C14orf132	chromosome 14 open reading frame 132	0.07	0.31
Ssc.5621.1.S1_at	CAP2	CAP, adenylate cyclase-associated protein, 2 (yeast)	0.05	3.05
Ssc.924.2.A1_at	THBS1	thrombospondin 1	0.01	2.84
Ssc.1031.1.S1_at	OAS1	2′-5′-oligoadenylate synthetase 1, 40/46 kDa	0.04	2.69
Ssc.11557.1.A1_at	ISG15	ISG15 ubiquitin-like modifier	0.02	2.59
Ssc.5549.2.S1_at	FABP5P7	fatty acid binding protein 5 pseudogene 7	0.07	2.59
Ssc.17238.1.A1_at	MUSTN1	musculoskeletal, embryonic nuclear protein 1	0.05	2.56
Ssc.23632.1.S1_at	LOC728532	dynein, cytoplasmic 1, intermediate chain 2 pseudogene	0.07	0.39
Ssc.9572.1.A1_at	MYO10	myosin X	0.09	2.5
Ssc.18802.1.A1_at	STX11	syntaxin 11	0.06	2.42
Ssc.31140.1.S1_at	IFIT3	interferon-induced protein with tetratricopeptide repeats 3	0.03	2.42
Ssc.6618.1.A1_at	ENDOD1	endonuclease domain containing 1	0.02	2.38
Ssc.25207.1.A1_at	ITGA6	integrin, alpha 6	0.08	2.37
Ssc.13587.1.A1_at	ANK3	ankyrin 3, node of Ranvier (ankyrin G)	0.09	2.34
Ssc.18307.1.A1_at	RBM25	RNA binding motif protein 25	0.09	0.43

^*Overlap with LAD. All genes in table have an FDR adjusted P value (also called a q value) of ≤10% to account for multiple testing.

Table 5.

List of annotated probe sets with significant difference in LAD perivascular fat gene expression between juvenile obese and lean pigs (sorted by magnitude of fold change)

ID	Gene Symbol	Gene Title	Adjusted P Value	Fold Change (Obese/Lean)
Ssc.18175.1.A1_at	FASN	fatty acid synthase	0.03	0.25
Ssc.16336.1.S1_at	ME1	malic enzyme 1, NADP+-dependent, cytosolic	0.03	0.27
Ssc.286.1.S1_s_at	RSAD2	radical S-adenosyl methionine domain containing 2	0.09	3.57
Ssc.27622.1.S1_at	KLF11	Kruppel-like factor 11	0.03	2.19
Ssc.25076.1.A1_at	FASN	fatty acid synthase	0.03	0.48
Ssc.25076.2.S1_at	FASN	fatty acid synthase	0.09	0.62
Ssc.2327.1.S1_at	VSTM2L	V-set and transmembrane domain containing 2 like	0.09	0.76

All genes in table have an FDR adjusted P value (also called a q value) of ≤10% to account for multiple testing.

Quantitative real-time PCR was carried out on 10 genes whose expression were altered in the LAD with juvenile obesity and that represent a selection from the list of top 20 genes provided in Table 3. Importantly, current literature indicates that all 10 genes are implicated in atherosclerosis based on published literature (1, 10, 22–24, 27, 32, 33, 45, 48, 51, 56, 64, 74, 77, 84). As illustrated in Fig. 4, real-time PCR and microarray analysis produced similar results for all 10 genes, providing confirmation of the findings obtained by the microarray approach.

Fig. 4.

Verification of microarray results by quantitative real-time PCR in a subset of relevant genes found to be upregulated in the LAD. Values are means ± SE. *Significantly different from lean, P < 0.05. For abbreviations see Table 3.

Figures 5 and 6 illustrate the top-scoring and highly significant gene networks influenced by juvenile obesity in the LAD and descending thoracic aorta, respectively. The scores for these gene networks were 39 in the LAD and 38 in the thoracic aorta. See Statistical analysis for a description of the calculation of the scores.

Fig. 5.

Top-scoring gene network influenced by juvenile obesity in the LAD (score = 39). Nodes represent genes/molecules. The number below each node is the fold change and the shading is in proportion to the size of the fold change (red, upregulation; green, downregulation). White nodes denote network members that were not represented on the array. Gray nodes denote network members that did not reach the false discovery rate (FDR) < 10%. Solid and dotted lines denote direct and indirect relationships, respectively.

Fig. 6.

Top-scoring gene network influenced by juvenile obesity in the descending thoracic aorta (score = 38). Nodes represent genes/molecules. The number below each node is the fold change and the shading is in proportion to the size of the fold change (red, upregulation; green, downregulation). White nodes denote network members that were not represented on the array. Gray nodes denote network members that did not reach the FDR < 10%. Solid and dotted lines denote direct and indirect relationships, respectively.

Given that our pathway analysis revealed the NF-κB complex was a central component of the LAD gene network influenced by juvenile obesity (Fig. 5), we designed a follow-up experiment in cell culture to determine the extent to which activation of NF-κB plays a role in regulating expression of obesity-responsive genes in the LAD. We found that LPS-induced upregulation of CYBB and LGALS3 in porcine cultured endothelial cells was NF-κB dependent (P < 0.05) (Fig. 7). Endothelial expression of remaining genes evaluated (i.e., list in Fig. 4) was not influenced by LPS and/or NF-κB inhibition under our cell culture conditions (data not shown, P > 0.05).

Fig. 7.

LPS-induced upregulation of CYBB and LGALS3 in porcine cultured endothelial cells is NF-κB dependent; n = 6 per condition. *Significantly different from vehicle (Ve); P < 0.05. #Significantly different from LPS without NF-κB inhibition; P < 0.05.

DISCUSSION

This is the first study to perform a comprehensive analysis of vascular transcriptional profiles in a porcine model of childhood overnutrition and obesity. We found that the LAD, a classically “atheroprone” coronary artery in adulthood (2, 18), is more susceptible to transcriptional alterations as a result of juvenile obesity, compared with the more “atheroresistant” descending thoracic aorta (17). Notably, we identified a number of genes in the LAD (e.g., ACP5, LYZ, CXCL14, APOE, PLA2G7, LGALS3, SPP1, ITGB2, CYBB, P2RY12) that are implicated in atherosclerosis based on published literature (1, 10, 22–24, 27, 32, 33, 45, 48, 51, 56, 64, 74, 77, 84) and whose expression was markedly upregulated with juvenile obesity. Because the alterations in expression of these genes occurred in the absence of coronary artery vascular dysfunction, this study provides information regarding the initial molecular events potentially involved in creating a permissive state for the development of vascular dysfunction and disease, which is indeed manifested when obesity in the Ossabaw pig model persists into adulthood (9, 25, 57, 65, 80).

The prevalence and severity of obesity among children (age 6–11 yr) and adolescents (age 12–19 yr) in modern societies have increased dramatically over the past 30 yr (54, 58). This is a product of the combinatorial effect of overnutrition and lack of physical activity, a set of behaviors facilitated by our society (20, 38, 47, 79). Importantly, obesity during childhood has subsequently been linked with a variety of adverse health outcomes in adulthood, particularly coronary artery disease (34, 42, 54). While it has been proposed that coronary artery disease has its origins in childhood and adolescence (34), the molecular precursors triggered by obesity during early life have not been delineated. In the present study, we employed a porcine model of childhood obesity and performed whole genome transcriptome analyses to identify genes in the arterial wall that are responsive to juvenile obesity. As illustrated in Fig. 3, expression of 189 vascular cell genes in the LAD and expression of 165 genes in the thoracic aorta were altered with juvenile obesity. Remarkably, obesity produced an overlap of only 28 genes (e.g., SMPDL3A, CXCL14) between both arteries. The observation that the effects of juvenile obesity on vascular pretranslational regulation were different between arteries was also confirmed by IPA. Figures 5 and 6 illustrate the top vascular gene networks influenced by juvenile obesity in the LAD and descending thoracic aorta, respectively. As noted, there was no overlap between the top two networks. More importantly, the top gene network impacted by obesity in the LAD reflects the upregulation of proinflammatory and pro-oxidant pathways with central players being the NF-κB complex and interferons (i.e., cytokines). This differential pretranslational regulation between arteries indicates that vascular cells respond differently to a systemic insult such as high-fat feeding and obesity. It is thus likely that systemic factors interact with local factors in the regulation of vascular cell phenotype (40, 62). For example, it is possible that differences in vascular wall shear stress profiles between arteries alters the susceptibility of the vascular cells to the proatherogenic effect of circulating factors associated with obesity including oxLDL.

Because it is well established that coronary arteries, particularly the LAD, are highly susceptible to atherosclerosis (2, 18) relative to other vascular beds such as the descending thoracic aorta (17), the changes in vascular gene expression produced by juvenile obesity that are exclusive to the LAD are of particular interest. In this regard, a number of genes found to be markedly upregulated in the LAD of obese pigs are implicated in the progression of atherosclerosis, including ACP5, LYZ, CXCL14, APOE, PLA2G7, LGALS3, SPP1, ITGB2, CYBB, and P2RY12 (Table 3) (1, 10, 22–24, 27, 32, 33, 45, 48, 51, 56, 64, 74, 77, 84). We verified the upregulation of these 10 transcripts by real-time PCR (Fig. 4); owing to space limitation, the role of only a subset of these genes is briefly discussed below.

PLA2G7 gene encodes the phospholipase A2 enzyme (Lp-PLA₂), which is a lipoprotein-bound enzyme that catalyzes the hydrolysis of oxidized phospholipids, generating bioactive oxidized free fatty acids with potent proinflammatory and proatherogenic actions (84). Indeed, increased circulating levels of Lp-PLA₂ are associated with greater incidence of coronary artery disease. In addition, there is evidence that Lp-PLA₂ expression is increased within the necrotic core and surrounding macrophages of vulnerable plaques, relative to less advanced lesions (45). Thus, our observation that juvenile obesity increases PLA2G7 mRNA levels in the LAD is significant, considering the growing evidence that PLA2G7 is involved in the modulation of atherosclerosis.

ITGB2 gene, also known as CD18, encodes the integrin beta chain beta 2. Integrins are integral cell-surface proteins involved in cell adhesion. Cellular adhesion molecules mediate the recruitment of circulating leukocytes to the arterial wall and their subsequent migration into the subendothelial space, thus playing a key role in all stages of atherosclerosis (51). In a study using C57BL/6 mice fed a high-fat diet, the authors found a ∼50% reduction in atherosclerotic fatty streaks in mice deficient in ITGB2 (56). This finding, together with our data indicating that juvenile pigs fed a high-fat diet increased expression of ITGB2 in the LAD, suggests the implication of ITGB2 in controlling the progression of vascular disease in the context of high-fat feeding and obesity.

LGALS3 gene encodes Galectin-3, a member of the lectin family. Galectin-3 participates in a number of biological processes including cell adhesion, cell activation chemoattraction, and apoptosis (24). Current research indicates that LGALS3 is upregulated at the mRNA and protein level in unstable plaque regions of carotid endarterectomy specimens, compared with stable regions from the same patients (64). Furthermore, in a recent microarray study, the LGALS3 transcript was found to be upregulated in endothelial cells isolated from obese mice fed a high-fat diet (22). Similarly, LGALS3 is highly expressed in aortic samples from ApoE−/− mice on a high-fat diet and colocalized with macrophages (64). In addition, it has been proposed that LGALS3 may be regarded as a novel target for antiatherosclerotic therapies (64). This proposition may be of particular interest in view of our finding that juvenile obesity increases mRNA levels of LGALS3 in the LAD.

CYBB gene encodes NOX2, also known as gp91^phox, the catalytic subunit of vascular NADPH oxidase. NADPH oxidase represents the largest source of reactive oxygen species in vascular cells (41). In fact, recent evidence indicates that endothelial NOX2 overexpression is sufficient to increase vascular superoxide production and induce macrophage recruitment via activation of endothelial cells (23). Our finding that CYBB mRNA levels were increased in the LAD of juvenile obese pigs is highly relevant since NOX2 is markedly upregulated in human coronary artery disease (32) and it is also associated with severity of atherosclerosis (74).

Taken together, these results clearly indicate that obesity in young pigs produced a proatherogenic shift in the pretranslational regulation of the coronary arteries. Notably, these pretranscriptional alterations seen in juvenile obesity occurred prior to measurable vascular dysfunction (Fig. 2) and before visible signs of atherosclerotic lesions. The lack of significant differences in coronary vasomotor relaxation between juvenile lean and obese Ossabaw pigs is consistent with findings by others using this swine model (46). However, our vasomotor relaxation data are in contrast with findings from studies in children and adolescents demonstrating that increased adiposity is associated with impaired brachial artery flow-mediated dilation (37, 53, 60). Discrepancies between swine and human studies may be attributed to the interrogation of different arteries (coronary arteries vs. brachial arteries) and/or the use of different stimuli to signal endothelium-dependent relaxation (bradykinin vs. shear stress). Our finding that vasomotor contraction to U-46619 (a thromboxane A2 agonist) was blunted in coronary arteries from obese pigs compared with lean pigs warrants further investigation. The molecular changes reported herein may be considered as very early steps that set the stage for the instigation of the disease processes. Indeed, there is compelling evidence that when the Ossabaw pig remains obese during adulthood, the coronary arteries exhibit severe vascular dysfunction and complex atherosclerotic disease (9, 25, 57, 65, 80).

To gain insight into possible factors that may contribute to the modulation of coronary artery health in our porcine model of childhood obesity, we evaluated circulating levels of total cholesterol, LDL-c, HDL-c, NEFA, oxLDL, TGs, glucose, and insulin (Table 2). Plasma levels of oxLDL were found to be fourfold higher in the juvenile obese pigs. Notably, it is now well recognized that oxLDL is implicated in the pathogenesis of atherosclerosis (31, 52, 68), including both the formation of foam cells and the induction of endothelial dysfunction. Human data indicate that elevated circulating levels of oxLDL are associated with cardiovascular risk (35) as well as coronary artery disease (26, 36, 78). Furthermore, there is evidence from studies in children and adolescents that circulating levels of oxLDL are associated with obesity and impaired brachial artery flow-mediated dilation (49, 81). Along these lines, it is thought that vascular dysfunction induced by oxLDL precedes tissue morphological changes and is believed to be one of the initiators of atherosclerosis (39).

The effects of oxLDL in the vascular wall are mediated by several scavenger receptors including LOX-1. LOX-1 expression in vascular cells is relatively low under normal conditions, but it is highly inducible as a result of inactivity, obesity, hypertension, diabetes, and hyperlipidemia (14, 16, 55, 61). For example, we recently found that LOX-1 expression was increased in the iliac artery of juvenile rats after exposure to 7 days of inactivity (61). Interestingly, LOX-1 expression is highest in atherosclerotic lesions (15, 43) but also is enhanced in the endothelium of prelesion areas (14), suggesting that LOX-1 expression may lead to vascular pathology. Indeed, LOX-1 overexpression in vivo causes vascular inflammation, oxidative stress, and atherosclerosis (39). Our findings that juvenile obese pigs display greater circulating levels of oxLDL and greater expression of LOX-1 in the LAD (Fig. 1) could lead to the speculation that obesity-associated changes in coronary artery gene expression reported herein may be downstream from oxLDL-LOX-1 signaling.

There is evidence that oxLDL/LOX-1 signaling induces the activation of NF-κB (67, 70), a key transcriptional regulator of inflammatory gene expression in vascular cells. Importantly, pathway analysis indicated that the NF-κB complex was a central component of the LAD gene network influenced by juvenile obesity (Fig. 5). This observation stimulated the hypothesis that NF-κB may be playing a role in regulating expression of many of the determined proatherogenic genes in the LAD (list in Fig. 4). Our cell culture experiments revealed that CYBB (i.e., gp91phox) and LGALS3 mRNAs were increased with LPS and pharmacological inhibition of NF-κB with wedelolactone prevented this effect (Fig. 7). To our knowledge this is the first study to provide evidence that endothelial expression of CYBB and LGALS3 is in part NF-κB dependent. Based on our in vivo and in vitro data we speculate that oxLDL/LOX-1/NF-κB signaling axis may be involved in the early initiation of a juvenile obesity-induced proatherogenic coronary artery phenotype.

An unexpected finding of the present study was the relatively small number of transcripts altered in perivascular fat in response to juvenile obesity (Table 5). This was surprising considering the emerging literature indicating that perivascular adipose tissue contributes to obesity-related coronary artery disease through an increased expression and secretion of inflammatory cytokines (13, 65, 66). Future research should evaluate whether this lack of responsiveness to juvenile obesity is specific to perivascular fat or also extendable to other fat depots. In view of the present data, we speculate that the changes in LAD transcriptional profiles promoted by juvenile obesity are not attributable to alterations in the phenotype of the adjacent perivascular fat and consequent derived signals.

Limitations of the present investigation should be considered. First, because we studied mRNA levels from whole artery homogenates, it is unknown whether differences in gene expression reported in this study are originating from the endothelium, smooth muscle, or adventitia. Examination of the impact of juvenile obesity on endothelial vs. smooth muscle cells in isolation will indeed be a priority for future studies. Nevertheless, our current nonreductionist approach is reasonable considering that atherosclerosis is a disease of the entire artery wall, not confined to a specific cell type (69). Second, the present study identified artery-specific changes in vascular mRNAs that likely reflect upstream events underlying a shift in vascular phenotype that occurs with persistent obesity (9, 25, 57, 65); thus, future time-course studies are required to establish when changes in mRNA levels triggered during juvenile obesity result in vascular dysfunction. Third, due to the lack of information in the specific research question addressed here, a descriptive study as first step is novel by nature. Identification of genes altered with juvenile obesity is necessary before future research can examine the extent to which overexpression/knockdown of such genes plays a significant role in atherogenesis in vivo. In addition, whether vascular changes prompted by childhood obesity predispose the individual to disease development later in life even without a persistent obesity insult is an important issue that needs to be addressed in future investigations.

In summary, we provide evidence for the first time in pigs that juvenile obesity produces marked alterations in coronary artery gene expression, which is mainly characterized by the induction of proinflammatory and pro-oxidant pathways. In addition, our data indicate that the influence of juvenile obesity is not uniform throughout the vascular tree in that we observed little overlap in the number and function of altered genes between the LAD and the descending thoracic aorta. Since modulation in expression of these genes occurred prior to overt signs of coronary artery vascular dysfunction, our data may offer mechanistic insights regarding the molecular precursors involved in the origins of overnutrition and obesity-associated vascular phenotypic changes. In particular, our results suggest that the increases in circulating oxLDL and LOX-1 protein expression in coronary arteries as a result of juvenile obesity may contribute to the expression of proinflammatory and proatherogenic genes at early age. Contrary to our hypothesis, these alterations were not accompanied by robust changes in gene expression in the surrounding perivascular fat. These results suggest that if perivascular fat plays an important role in atherogenesis, it must do so at later stages of the development in adult animals.

GRANTS

Funding was provided by University of Missouri Mizzou Advantage (R. S. Rector, J. A. Ibdah, and F. W. Booth), AHA 11POST5080002 (J. Padilla), National Institutes of Health (NIH) Grants T32-AR-048523 (N. T. Jenkins), RO1HL-085119 (M. A. Hill), RO1HL-036088 (M. H. Laughlin), VA-CDA-IK2 BX-001299-01 (R. S. Rector), and a grant from the Allen Foundation (R. S. Rector). We acknowledge the support of NIH Grants RR-013223 and HL-062552 to M. Sturek and the Comparative Medicine Center of IUSM and Purdue University for the female Ossabaw swine used in this study. This work was supported with resources and the use of facilities at the Harry S. Truman Memorial Veterans Hospital in Columbia, MO. Finally, F. W. Booth conceived, organized, and submitted the Mizzou Advantage Grant. F. W. Booth also secured cross-matching funding from the College of Agriculture, Food and Natural Resources, Department of Biomedical Sciences, Department of Internal Medicine, and Dalton Cardiovacular Research Center at the University of Missouri.

DISCLOSURES

No conflicts of interest, financial or otherwise, are declared by the author(s).

AUTHOR CONTRIBUTIONS

Author contributions: J.P., N.T.J., M.Y.Z., C.Z., M.A.H., J.W.P., J.A.I., F.W.B., J.W.D., M.H.L., and R.S.R. conception and design of research; J.P., N.T.J., S.L., H.Z., J.C., J.W.P., and R.S.R. performed experiments; J.P., N.T.J., J.W.D., and R.S.R. analyzed data; J.P., N.T.J., S.L., M.Y.Z., C.Z., M.A.H., J.W.P., J.A.I., F.W.B., J.W.D., M.H.L., and R.S.R. interpreted results of experiments; J.P. and J.W.D. prepared figures; J.P. drafted manuscript; J.P., N.T.J., S.L., H.Z., J.C., M.Y.Z., M.A.H., J.W.P.I., J.A.I., F.W.B., J.W.D., M.H.L., and R.S.R. edited and revised manuscript; J.P., N.T.J., S.L., H.Z., J.C., M.Y.Z., M.A.H., J.W.P., J.A.I., F.W.B., J.W.D., M.H.L., and R.S.R. approved final version of manuscript.