Gene transcripts associated with BMI in the motor cortex and caudate nucleus of calorie restricted rhesus monkeys

Genomics 99 (2012) 144–151

Contents lists available at SciVerse ScienceDirect

Genomics journal homepage: www.elsevier.com/locate/ygeno

Gene transcripts associated with BMI in the motor cortex and caudate nucleus of calorie restricted rhesus monkeys Amanda C. Mitchell a, b, Rehana K. Leak c, d, Michael J. Zigmond d, Judy L. Cameron e, f, Károly Mirnics a, b,⁎ a

Department of Psychiatry, Vanderbilt University, Nashville, TN, USA Vanderbilt Kennedy Center for Research on Human Development, Vanderbilt University, Nashville, TN, USA Division of Pharmaceutical Sciences, Mylan School of Pharmacy, Duquesne University, Pittsburgh, PA, USA d Department of Neurology, University of Pittsburgh, Pittsburgh, PA, USA e Department of Psychiatry, University of Pittsburgh, Pittsburgh, PA, USA f Oregon National Primate Research Center, Beaverton, OR, USA b c

a r t i c l e

i n f o

Article history: Received 7 November 2011 Accepted 20 December 2011 Available online 29 December 2011 Keywords: DNA microarray Rhesus monkey BMI Motor cortex Caudate nucleus Gene expression ERK pathway

a b s t r a c t Obesity affects over 500 million people worldwide, and has far reaching negative health effects. Given that high body mass index (BMI) and insulin resistance are associated with alterations in many regions of brain and that physical activity can decrease obesity, we hypothesized that in Rhesus monkeys (Macaca mulatta) fed a high fat diet and who subsequently received reduced calories BMI would be associated with a unique gene expression signature in motor regions of the brain implicated in neurodegenerative disorders. In the motor cortex with increased BMI we saw the upregulation of genes involved in apoptosis, altered gene expression in metabolic pathways, and the downregulation of pERK1/2 (MAPK1), a protein involved in cellular survival. In the caudate nucleus with increased BMI we saw the upregulation of known obesity related genes (the insulin receptor (INSR) and the glucagon-like peptide-2 receptor (GLP2R)), apoptosis related genes, and altered expression of genes involved in various metabolic processes. These studies suggest that the effects of high BMI on the brain transcriptome persist regardless of two months of calorie restriction. We hypothesize that active lifestyles with low BMIs together create a brain homeostasis more conducive to brain resiliency and neuronal survival. © 2011 Elsevier Inc. All rights reserved.

1. Introduction Since 1980 the number of people worldwide considered obese has doubled with over 1.5 billion people now considered overweight (body mass index, BMI, 25 or higher) and 500 million people considered obese (BMI 30 or higher) [1]. The negative effects of obesity on health are far reaching. As weight increases to levels defined as overweight and obese, the risks for the following conditions also increase: coronary heart disease, type 2 diabetes, cancers, hypertension, dyslipidemia, stroke, liver and gallbladder disease, sleep apnea, osteoarthritis, and gynecological problems [2–4]. According to the American Medical Association the best ways to decrease the risk of obesity are to eat a diet low in fat and to increase physical activity levels [5]. Indeed, modern sedentary life is associated with increased risks for obesity, as well as cardiovascular disease, type II diabetes [6], and depression [7,8]. In the hypothalamus of the central nervous system (CNS) both insulin and leptin control body weight and feeding behavior. Leptin, a hormone derived from fat circulating in the blood, informs the ⁎ Corresponding author at: Department of Psychiatry, Vanderbilt University, 8130A MRB III, 465 21st Avenue South, Nashville TN 37232, USA. E-mail address: [email protected] (K. Mirnics). 0888-7543/$ – see front matter © 2011 Elsevier Inc. All rights reserved. doi:10.1016/j.ygeno.2011.12.006

hypothalamus via JAK/STAT pathway about the energy status of peripheral tissues [9], and reduces the expression of neuropeptides (neuropeptide Y (NPY), agouti-related peptide (AgRP), and melanocyte stimulating hormone (MSH also known as POMC)) that stimulate feeding [9–11]. Insulin, another circulating hormone, serves as a feedback signal from the periphery to reduce appetite and body weight in the hypothalamus [12] and serves to cause peripheral tissues to take glucose up from the blood and store it as glycogen [13]. Furthermore, insulin receptors (receptor tyrosine kinases) located throughout the brain signal through both the PI3K/AKT and ERK1/2 survival signaling pathways [14]. In obesity resistance to leptin and insulin impairs proper signaling about energy status that results in increased body weight and feeding behaviors. The AKT and ERK1/2 signaling pathways are also downstream of many neurotrophic factor signaling pathways and activation of these pathways leads to neuroprotection in several regions of the brain [15,16]. Thus, insulin resistance, may lead to decreased activation of AKT and ERK1/2, increased apoptosis, and decreased survival of neurons. This could explain why type II diabetes (insulin resistance) is often linked to Alzheimer's disease, a neurodegenerative disorder, and could be even further detrimental to the total health of the brain [17]. Insulin resistance that results in impaired signaling also affects other regions of the brain [18]. The effects of insulin on cortical

A.C. Mitchell et al. / Genomics 99 (2012) 144–151

activity are negatively related to the amount of body fat and the degree of peripheral insulin resistance [19]. Furthermore, decreased brain volumes have been seen in the frontal lobe, the temporal lobe, the anterior cingulate, the hippocampus, and the basal ganglia with high BMI [20]. Using correlations of BMI with brain volumes the same group showed that overweight and obese individual's brains age prematurely [21]. Other studies also link higher BMI with smaller regional brain volumes [21–24]. These findings provide evidence for cerebral insulin resistance in overweight humans. Given that high BMI is associated with insulin resistance, that insulin resistance is associated with specific brain alterations, and that physical activity can decrease obesity, we hypothesized that we could find genes associated with BMI in motor regions of the brain. We investigated transcriptome changes associated with BMI in the motor cortex and caudate nucleus of 14 Rhesus monkeys with DNA microarrays. We found a distinct BMI associated gene expression signature in both the motor cortex and the caudate nucleus, and decreased protein phosphorylation of a survival kinase, pERK1/2, in the motor cortex. Our studies indicate that obesity is also associated with molecular changes in motor regions of the brain that these changes include alterations in metabolic and apoptotic related genes and that increases in BMI are associated with decreases in the phosphorylation of a survival kinase. The same cohort of monkeys was used in a previous calorie restriction study, as numerous studies indicate that calorie restriction, like exercise, is neuroprotective. To mimic the typical American diet Rhesus monkeys were fed a diet high in fat for the last 3 years of their life. They also underwent one month of a 30% calorie reduction followed by one month of a 60% calorie reduction at the end of the study [25]. This is typical of a dieting person, and increases the production of BDNF and activates the AKT and ERK1/2 signaling and survival pathways in the brain [26]. While calorie restriction could theoretically represent a potential confound in our experiments as it may overcome some of the effects of a high fat diet and insulin resistance [27], this was not the case: decreased pERK1/2, increased CASP9 & FOXO3 mRNA, and an enrichment of apoptosis and metabolism related genes, all indicate decreased neuronal survival, suggesting that calorie restriction does not overcome the transcriptional signature of high BMIs in the brain.

145

2. Results

We identified 27 transcripts that were associated with BMI in the motor cortex (17 positively correlated and 10 negatively correlated) and 51 transcripts that were associated with BMI in the caudate nucleus (41 positively correlated and 10 negatively correlated) (Supplementary Table 1 and 2). All altered transcripts in the motor cortex and in the caudate nucleus were changed in the same direction on both human and monkey microarrays, suggesting a high degree of concordance between the two data sets, and a very low false discovery rate in our experiment. Our Pearson Hierarchal Clustering [32] revealed a progression of gene expression levels from monkeys with high BMI to monkeys with to low BMI (Fig. 2). Furthermore, we validated increased expression of caspase 9 (CASP9) in the motor cortex and increased expression of forkhead box O3 (FOXO3) in the caudate nucleus with increasing BMI by qPCR (Fig. 3). Next, we performed Gene Ontology (GO) Biological Process analysis (Supplementary Table 3) using the Gene Ontology Enrichment Analysis Software Toolkit (GOEAST) [33,34] and GSEA assessment using the Broad Institute's Gene Pattern Software [35] (Tables 1 and 2) to decipher which biological processes were involved with the BMI associated gene expression signature. GO Biological Process analysis searched our differentially expressed gene datasets (Supplementary Table 1 and 2) for significantly altered GO Biological Processes. While these analyses revealed no significant biological processes associated with significant gene expression changes in the motor cortex, in the caudate nucleus we identified a large number significantly altered biological processes that suggested alteration of gene networks related to metabolism, cell death, and cellular development (Supplementary Table 3). We also assessed our data for the enrichment of functional gene pathways using Gene Set Enrichment Analysis (GSEA) [35] using the gene sets in the Molecular Signatures Database (MSigDBv.3.0) for the human (HG) arrays. We found 9 gene sets positively enriched with BMI in the motor cortex (Table 1), and 3 gene sets negatively enriched with BMI in the caudate nucleus (Table 2). Prominently, in the motor cortex the BMI-associated changes included VEGFA targets (genes upregulated in HUVEC endothelium cells at 30 min after VEGFA stimulation [36]) and AKT phosphorylation targets in the cytosol, while in the caudate nucleus significant gene sets included HDAC targets (genes whose transcription is altered by histone deacetylase inhibitors [37]).

2.1. Microarray data analysis reveals a BMI-associated gene expression pattern

2.2. The observed gene expression changes are putatively a result of altered ERK1/2 signaling

We hypothesized that BMI would be associated with a unique gene expression signature in the motor cortex and caudate nucleus that may be of importance to neurodegenerative disorders given observed alterations in the brain in insulin resistance. To discover other alterations we performed whole genome expression profiling of the motor cortex of 14 female, calorie restricted rhesus monkeys with a wide range of BMIs using both Affymetrix Rhesus Macaque Genome (RhG) and Human Genome (HG) microarrays (Fig. 1). The rhesus and human genomes share 93% sequence identity [28] and human microarrays successfully have been used to query rhesus monkey gene expression changes in the past [29]. Whereas the Affymetrix RhG microarrays provided us with the most specific hybridization information for rhesus monkeys, the Affymetrix HG microarrays contained more extensive and higher quality annotations. Microarray data were log2 transformed, RMA-normalized [30,31] and linearly modeled with BMI. We focused on convergent findings between the RhG and HG microarrays, which enabled us to identify a strong and distinct BMI-associated gene expression signature in the motor cortex and caudate nucleus. Two types of data-mining approaches were employed. First, we identified the transcripts that were associated with BMI, followed by a molecular pathway analysis that revealed the most affected BMI-associated groups of interdependent genes.

Several of the BMI-dependent genes attracted further attention. Our data identified inversely correlated mRNA levels of forkhead box O3 (FOXO3) with BMI in the caudate nucleus. FOXO3 is activated via the growth factor pathway, and involved in apoptosis and neuronal survival [15]. Furthermore, as both AKT and ERK1/2 are effector signaling pathways of neurotrophic factors [15,16,26,38], we next investigated the association of the AKT and ERK protein phosphorylation state with BMI (Fig. 4) through Western analysis. While AKT levels and its phosphorylation state was unchanged, we identified a strong negative correlation of both ERK1/2 and pERK1/2 levels with BMI (pERK1/2/tubulin r = −0.663 with p = 0.010 and pERK1/2/ ERK1/2 of −0.595 with p = 0.025) (Fig. 4). As pERK1/2 phosphorylates CREB, a transcription factor in the nucleus to reduce apoptosis [38] and activate transcription of neuronal survival genes [15], the decreased phosphorylation of ERK1/2 with increasing BMIs suggests that this might be a critical signaling pathway mediating some of the detrimental effects of BMI on the gene expression in the brain. 3. Discussion In summary, we discovered a distinct gene expression signature associated with BMI after calorie restriction in the motor cortex and caudate nucleus. Our linear model allowed us to model BMI and

146

A.C. Mitchell et al. / Genomics 99 (2012) 144–151

Fig. 1. Experimental design. Motor cortex and caudate nucleus tissue was harvested from 14 female rhesus macaques with wide ranges of BMI. The motor cortex and caudate nucleus transcriptomes were profiled separately using both Affymetrix Rhesus Macaque Genome microarrays (RhG) and Affymetrix Human Genome U133 2.0 Plus microarrays (HG). We used a linear model to define gene expression associated with BMI (gene value = α + β (BMI)) for both microarrays separately. Probes were cross matched based on sequence similarity and significance was defined if both microarray models (i) were significant with p b 0.05, and (ii) had an interquartile range (IQR) > 0.263 corresponding to a 20% gene expression change. Significant genes are contained in Supplementary Tables 1 and 2, and clustered in Fig. 2. Pathway analysis was performed using both Gene Ontology (GO) contained in Supplementary Tables 3 and 4, and Gene Set Enrichment Analysis (GSEA) contained in Tables 1 and 2. Note that the human and monkey data are highly concordant for a set of genes both in the motor cortex and caudate nucleus.

gene expression continuously, and to find a gene expression signature with 100% concordance in directionality between significant probes on RhG and HG microarrays. We saw increases in apoptotic related genes and changes in metabolic related genes across both brain regions, suggesting that motor regions of the brain are affected at the transcriptional level by obesity. Decreased ERK1/2 phosphorylation was associated with increasing BMI. Activation of this prototypical survival kinase provides a potential explanation for the enrichment of pro-apoptotic genes in animals with high BMI and putatively

explains how obesity raises the risk for reduced brain volume [21–24]. Furthermore, leptin and insulin are both known to activate ERK1/2 in addition to the AKT pathway [14,39]. Although we did not find evidence of BMI-dependent changes to the AKT signaling pathway, we cannot rule out the possibility of cross talk between the AKT and ERK1/2 pathways [40–43] in the non-human primate brain. Furthermore, our previous data suggest that physical activity is positively correlated with pAKT protein levels in non-human primates [44], potentially leading to a double-risk to

A.C. Mitchell et al. / Genomics 99 (2012) 144–151

147

Fig. 2. Hierarchal clustering of gene expression associated with BMI. Hierarchical clustering was performed on log2-transformed expression level of genes associated with BMI. Vertical column labels indicate BMI in decreasing order. Genes, denoted by Affymetrix Human Probes and NCBI gene symbols, were clustered horizontally. Each colored square represents a normalized gene expression value with color intensity proportional to magnitude of change (blue indicates a decrease in expression and red an increase). (a) Motor Cortex Gene Expression Signature for Rhesus Genome Microarrays. (b) Motor Cortex Gene Expression Signature for Human Genome Microarrays. (c) Caudate Nucleus Gene Expression Signature for Rhesus Genome Microarrays. (d) Caudate Nucleus Gene Expression Signature for Human Genome Microarrays. Note that in both structures there is a progression of gene expression values from monkeys with low BMI to monkeys with high BMI, and the data are concordant across the human and monkey array platforms.

the brain: the sedentary lifestyle might decrease pAKT signaling, and increased BMI can repress the activation of the neuronal survival kinase ERK1/2. These two events, in concert might significantly increase brain susceptibility to various insults. Although it is tempting to hypothesize that both of these events are the result of a change in a growth factor signaling pathway, we found relatively meager

evidence to support this theory at the level of the transcriptome. However, it is known that growth factor signaling is profoundly regulated at the post-transcriptional level [16,45] and this should be a topic of further follow-up experiments. The positive correlation between the expression of apoptosisrelated genes and BMI is also noteworthy. Increased pro-apoptotic

148

A.C. Mitchell et al. / Genomics 99 (2012) 144–151

transcriptome profiling studies performed on monkeys that underwent dietary restriction [26]. The neuroprotective effect of dietary restriction reported increased production of neurotrophic factors and protein chaperones, resulting in protection against oxyradical production, stabilization of cellular calcium homeostasis, and inhibition of apoptosis. Finally, based on our studies and previous literature we conclude that active lifestyles with low BMI create a brain homeostasis more conducive to brain resiliency and neuronal survival. Brain health is clearly linked to body health, and, we now show the converse, that maintaining a health BMI is also essential for maintaining a healthy brain transcriptome. 4. Materials and methods 4.1. Subjects Fourteen adult female ovariectomized rhesus monkeys (Macaca mulatta) were allowed to eat a high fat diet freely for three years with care methods previously published [25,48,49]. Their activity and percent body fat were measured throughout the diet and were found to be extremely stable. As part of a different study they were then placed on a two month calorie reduction diet. Food composition was the same, but their calorie intake was reduced by 30% the first month and 60% the second month from their normal intake. Their BMI was measured in kilograms/crown of rump length squared (kg/ CRL 2) at the end of the study. Monkeys were deeply anesthetized with ketamine/pentobarbital and their entire brain was removed. The brain was then hemisected by a midline sagittal cut. The right hemisphere was cut into 5-mm thick coronal blocks, flash-frozen in isopentane over dry ice, and stored at −80 °C until use. 4.2. Sample preparation and hybridization

Fig. 3. Validation of selected microarray-reported changes by qPCR. (a) Body Mass Index (BMI) was measured in kilograms per crown of rump length squared (kg/CRL2) while CASP9 and FOXO3 relative expression levels were reported in threshold cycle difference relative to β-actin and peptidylprolyl isomerase D respectively (ΔCT, dCT). CASP9 and FOXO3 gene expressions were significantly positively correlated with BMI. (b) Scatter plot showing qPCR expression (1/dCT) versus BMI (CRL2) for CASP9. (c) Scatter plot showing qPCR expression (1/dCT) versus BMI (CRL2) for FOXO3. Note that the microarrayreported expression changes were successfully validated by qPCR.

gene expression can lead to increased vulnerability of neurons to various environmental insults, and the decreased brain volume observed in persons with high BMI [21–24] might be closed linked to the gene expression changes we uncovered. Since our monkeys were placed on a calorie restriction diet immediately prior to transcriptome analyses some of the deleterious effects associated with high BMI may have been masked. A number of previous studies have demonstrated many beneficial effects of calorie restriction on health, the rate of aging, and longevity [46]. These findings suggest important roles for the plasma membrane redox system in protecting brain cells against age-related increases in oxidative and metabolic stress, and these findings might be directly related to the BMI-correlated increases in pro-apoptotic gene expression [47]. Furthermore, our results are in also in concordance with previous

Motor cortex and caudate nucleus brain tissue was homogenized and total RNA isolated using TRIzol® reagent (Invitrogen, Carlsbad, CA) with RNA quality assessed via analysis on an Agilent 2100 Bioanalyzer. Only samples with an RIN > 7.0 were considered for further analysis. The samples were prepared using the Enzo BioArray™ Single-Round RNA Amplification and Biotin Labeling System. The samples were primed with a standard T7-oligo(dT) primer and cDNA synthesis was performed using 2 μg of total RNA according to the Affymetrix® manufacturer's protocol. Amplified antisense RNA (aRNA) was produced using in vitro transcription directed by T7 polymerase. Fifteen micrograms of the purified and fragmented aRNA were hybridized to GeneChip® Rhesus Macaque Genome (RhG) Arrays and GeneChip® Human Genome (HG) U133 Plus 2.0 Arrays. Image segmentation analysis and generation of DAT files was performed using Microarray Suite 5.0 (MAS5). All data is compliant with the Minimum Information About a Microarray Experiment (MIAME) guidelines and raw data is available at ftp://mirnicslab. vanderbilt.edu/BMI microarrays/. 4.3. Microarray data analysis Fourteen RhG arrays for the motor cortex and 14 RhG arrays for the caudate nucleus were scanned and analyzed. Scaling factors ranged from 1.58 to 3.11 with backgrounds from 29.8 to 85 and percent present calls from 45 to 52%. Fourteen HG arrays for the motor cortex and 14 HG arrays for the caudate nucleus were scanned and analyzed. Scaling factors ranged from 6.26 to 15.28, noise was between 1.2 and 1.8, and percent present calls were from 31.3% to 35.0%. Segmented images were normalized and log2 transformed using robust multiarray analysis (RMA) separately for Rhesus and Human arrays [50] with RMA normalized expression levels utilized for all of our subsequent analyses.

A.C. Mitchell et al. / Genomics 99 (2012) 144–151

149

Table 1 Molecular Signatures Database (MSigDB) significant gene sets in the motor cortex using human arrays (HG). Name

Size

FDR q-val

Enriched genes

Kuninger IGF1 vs. PDGFB targets up

41

0.104

One carbon compound metabolic process Module 440 ABE VEGFA targets 30 min Schuring STAT5A targets up Module 199

26 18 19 7 57

0.128 0.173 0.179 0.184 0.197

Reactome AKT phosphorylates targets in the cytosol S adenosylmethionine dependent methyltransferase activity Methyltransferase activity

14 22 34

0.197 0.198 0.216

TPM2, CKM, MYBPH, MYOG, RYR1, USP2, MYL1, TNNT2, TNNI1, ENAH, PYGM, C1QTNF3, PPFIA4, NCAM1, NPNT, SGCA, PKIA, ATP2A1, SMTN NSUN2, DMAP1, PRMT5, NSD1, CARM1, DNMT3A, DNMT3B, EHMT1, PRM17, FOS, PRMT2 CKM, OAT, ASL, GLUD1, BBOX1, CKB, ODC1, ACY1 HDC, DBC1, C2, EGR2, EGR3, CYR61, NR4A1, TRIB1 LIF, TRAF4, PIM1 EPHB6, TEK, IL4R, KIT, KDR, PTK7, EPHB1, TIE1, IL2RB, MERTK, EPHA4, CCL2, MYLK, AXL, CDK5R1, ERBB2, EPHA7, MAP2K7, PDK4, FGFR3, IL13RA1, IL10RA, ROR2, EFNB3, PDGFRA AKT2, CASP9, AKT1, TSC2 METTL1, NSUN2, NSD1, DNMT3A, DNMT3B, EHMT1, PEMT, PRMT7 ATIC, METTL1, NSUN2, PRMT5, NSD1, CARM1, DNMT3A, DNMT3B, EHMT1, PBMT, PRMT7, SUV39H1, GART

We analyzed the microarrays for transcripts associated with BMI. We fit data from HG and RhG arrays using the model: log transformed gene expression value = α + β(BMI) [51]. RhG and HG probe sets were matched on the basis of sequence similarity using the Affymetrix HG-U133 Plus 2.0 to Rhesus Best Match Spreadsheet. Genes were considered correlated BMI if (a) p b 0.05, where H0: β = 0 for RhG arrays; (b) p b 0.05, where H0: β = 0 dfor HG arrays; (c) if the interquartile range, IQR > 0.263 on the RhG arrays, measuring the difference between the third and first quartiles was less than 0.263 (representing a 20% change in gene expression values); and (d) if the interquartile range, IQR > 0.263 on the HG arrays, measuring the difference between the third and first quartiles was less than 0.263 (representing a 20% change in gene expression values). The strength of our model was tested for false discovery using a binomial probability test for the direction of change, where the H0: p = 0.50. We tested our model using a binomial probability test with a null hypothesis that p = 0.50, which was rejected. Hierarchical clustering of the replicating data was performed using GenePattern software [32]. Clustering was performed on log2 transformed GC-RMA normalized expression levels using row (gene) centering and Pearson correlation. 4.4. Quantitative real-time PCR validation As previously reported [44,49], we used the High Capacity cDNA Archive Kit® (Applied Biosystems, Foster City, CA) to synthesize cDNA from 500 ng of the same total RNA used for microarray analysis in 11 monkeys. Priming was performed with random hexamers. For each sample, amplified product differences were measured with three independent replicates using SYBR Green chemistry-based detection [30]. β-actin was used as the endogenous reference gene for caspase 9 (CASP9) and peptidylprolyl isomerase D was used as the endogenous reference gene for forkhead box O3 (FOXO3). Primer sequences for β-actin were 5′ GAT GTG GAT CAG CAA GCA 3′ and 5′ AGA AAG GCT GTA ACG CAA CTA 3′, for CASP9 were 5′ GAG TGG CTC CTG GTA CGT TG 3′ and 5′ CCT TTC ACC GAA ACA GCA TT 3′, for peptidylprolyl isomerase D (cyclophilin D, CYC) were 5′ GCA GAC AAG GTC CCA AAG 3′ and 5′ GAA GTC ACC ACC CTG ACA C 3′, and for forkhead box O3 (FOXO3) were 5′ GCA TTG TGT GTG TTG CTT CC 3′ and 5′ CTG AAA TCA CAC CCT GAG CA 3′. The efficiency for each primer set was assessed prior to qPCR measurements, and a primer set was considered valid if its efficiency was >90%. The qPCR reactions were carried out on an ABI Prism 7300 thermal cycler (Applied Biosystems Inc.), quantified using ABI Prism 7300 SDS software (with the auto baseline and auto threshold detection options selected) and statistically analyzed using Pearson correlations. 4.5. Gene ontology (GO) and gene Set enrichment analysis (GSEA) We assessed our data for the enrichment of functional gene pathways using Gene Ontology (GO) Biological Process terms [33,34] and Gene Set

Enrichment Analysis (GSEA) [35]. First, we analyzed the GO Biological Process terms for all of our transcripts significantly associated with BMI in the motor cortex and caudate nucleus using the Gene Ontology Enrichment Analysis Software Toolkit (GOEAST). We reported significant GO biological processes using the hypergeometric test statistic, with a minimum of 5 genes, and a Benjamini–Yekutieli FDRb 0.1 [52]. Second, we performed GSEA [35] using all of the gene sets in the Molecular Signatures Database (MSigDBv.3.0) for the human (HG) arrays. We report enriched gene sets with a nominal pvalueb 0.01 and a false discovery rate (FDR)b 0.25. Module 199 is most strongly associated with leukemia and Module 440 is most strongly with associated liver cancer [53]. 4.6. Western blots Infrared imaging of immunoblots was used for assessing phosphorylation and total protein state of ERK1/2 and Akt (Li-Cor Biosciences, Lincoln, NE). Samples were sonicated in a Triton-based lysis buffer (Cell Signaling, Danvers, MA) and subjected to standard SDS-page immunoblotting procedures. Blots were then blocked in 5% bovine serum albumin in Tris-buffered saline (TBS) for 1 h prior to incubation overnight at 4 °C in primary antibody made in the same blocking solution with 0.1% Tween. Antibodies used were: rabbit anti-phosphoERK1/2 (Cell Signaling, Cat. no. 9101), mouse anti-total ERK1/2 (Millipore, Cat. no. 05–1152), rabbit anti-phospho-Akt (Cell Signaling, Cat. no. 9271), mouse anti-total Akt (Cell Signaling, Cat. no. 2920). Mouse α-tubulin (Sigma-Aldrich, Cat. no. T5168) was used concurrently to control for differences in protein loading. After overnight incubation with primary antibody, blots were washed in TBS-tween and incubated with goat secondary antibodies raised against the appropriate IgG fluorescing in the infrared range (700 and 800 nm, Li-Cor Biosciences). After further washing, immunostained protein bands on the blots were visualized on an Odyssey Infrared Imager and quantified with Odyssey software (Li-Cor Biosciences). Acknowledgments This work was supported by National Institutes of Health (NIH) R01 MH067234 and MH079299. Training support for Amanda C. Table 2 Molecular Signatures Database (MSigDB) significant gene sets in the caudate nucleus on the human arrays (HG). Gene Set Enrichment Analysis (GSEA) was performed on the caudate nucleus data set. Three gene sets were positively correlated with BMI, had a nominal p-value b 0.001, and a false discovery rate (FDR) b 0.25. Name

Size NOM FDR Correlation Enriched genes p-val q-val with BMI

Cell 15 maturation Developmental 17 maturation MARKS HDAC 18 targets up

0.000 0.137 – 0.000 0.202 – 0.000 0.238 –

TP53, GADD45B, FAS, GSN, CASP3, GLRX, BAD, CCNE1, TXN, MCM3 KCNIP2, MYH11, ACTA1, KRT19, EREG, IL21, PICK1, MYOZ1 MYH11, ACTA1, KRT19, EREG, IL21, PICK1, MYOZ1

150

A.C. Mitchell et al. / Genomics 99 (2012) 144–151

Fig. 4. PhosphoERK1/2 levels are negatively correlated with BMI. Figure denotes a scatterplot showing BMI (kg/CRL2) versus protein ratios for pERK1/2/tubulin (blue diamond) and for pERK1/2/ERK1/2 (red square). There is a significant negative correlation between BMI and pERK1/2/tubulin protein ratio (r = − 0.663; p = 0.010) and a significant negative correlation of BMI and protein ratio for phospho ERK1/2/ total ERK1/2 (r = − 0.595; p = 0.025). Inset provides examples of ERK1/2 protein levels in four monkeys with various BMI, with BMI levels denoted above the images.

Mitchell was provided by National Institutes of Health (NIH) T32 MH064913-07. Additionally, we are thankful to Lily Wang for Biostatistics advice. Appendix A. Supplementary data Supplementary data to this article can be found online at doi:10. 1016/j.ygeno.2011.12.006. References [1] World Health Organization, Obesity and OverweightURL:, http://www.who.int/ mediacentre/factsheets/fs311/en/index.html2008. [2] NIH and NHLBI Obesity Education Initiative, Clinical Guidelines on the Identification, Evaluation, and Treatment of Overweight and Obesity in Adults, NIH Publication No. 98–4083, 1998. [3] H. Yatsuya, K. Yamagishi, K.E. North, F.L. Brancati, J. Stevens, A.R. Folsom, Associations of obesity measures with subtypes of ischemic stroke in the ARIC Study, J. Epidemiol. 20 (5) (2010) 347–354. [4] D. Falkstedt, T. Hemmingsson, F. Rasmussen, I. Lundberg, Body mass index in late adolescence and its association with coronary heart disease and stroke in middle age among Swedish men, Int. J. Obes. (Lond) 31 (5) (2007) 777–783. [5] R.F. Kushner, Roadmaps for Clinical Practice: Case Studies in Disease Prevention and Health Promotion—Assessment and Management of Adult Obesity: A Primer for Physicians, American Medical Association, 2003. [6] T.H. Marwick, M.D. Hordern, T. Miller, D.A. Chyun, A.G. Bertoni, R.S. Blumenthal, G. Philippides, A. Rocchini, Exercise training for type 2 diabetes mellitus: impact on cardiovascular risk: a scientific statement from the American Heart Association, Circulation 119 (25) (2009) 3244–3262. [7] D.A. Lawlor, S.W. Hopker, The effectiveness of exercise as an intervention in the management of depression: systematic review and meta-regression analysis of randomised controlled trials, BMJ 322 (7289) (2001) 763–767. [8] D.E. Vance, V.G. Wadley, K.K. Ball, D.L. Roenker, M. Rizzo, The effects of physical activity and sedentary behavior on cognitive health in older adults, J. Aging Phys. Activ. 13 (3) (2005) 294–313. [9] J.M. Friedman, J.L. Halaas, Leptin and the regulation of body weight in mammals, Nature 395 (6704) (1998) 763–770. [10] R.S. Ahima, S.M. Hileman, Postnatal regulation of hypothalamic neuropeptide expression by leptin: implications for energy balance and body weight regulation, Regul. Pept. 92 (1–3) (2000) 1–7. [11] J.S. Flier, E. Maratos-Flier, Lasker lauds leptin, Cell 143 (1) (2010) 9–12. [12] G.J. Morton, Hypothalamic leptin regulation of energy homeostasis and glucose metabolism, J. Physiol. 583 (Pt 2) (2007) 437–443. [13] D. Hubble, Insulin resistance, Br. Med. J. 2 (4845) (1954) 1022–1024. [14] Y. Kido, J. Nakae, D. Accili, Clinical review 125: the insulin receptor and its cellular targets, J. Clin. Endocrinol. Metab. 86 (3) (2001) 972–979. [15] A. Brunet, A. Bonni, M.J. Zigmond, M.Z. Lin, P. Juo, L.S. Hu, M.J. Anderson, K.C. Arden, J. Blenis, M.E. Greenberg, Akt promotes cell survival by phosphorylating and inhibiting a Forkhead transcription factor, Cell 96 (6) (1999) 857–868.

[16] N. Lindgren, R.K. Leak, K.M. Carlson, A.D. Smith, M.J. Zigmond, Activation of the extracellular signal-regulated kinases 1 and 2 by glial cell line-derived neurotrophic factor and its relation to neuroprotection in a mouse model of Parkinson's disease, J. Neurosci. Res. 86 (9) (2008) 2039–2049. [17] Z. Arvanitakis, R.S. Wilson, J.L. Bienias, D.A. Evans, D.A. Bennett, Diabetes mellitus and risk of Alzheimer disease and decline in cognitive function, Arch. Neurol. 61 (5) (2004) 661–666. [18] M. Hallschmid, B. Schultes, Central nervous insulin resistance: a promising target in the treatment of metabolic and cognitive disorders? Diabetologia 52 (11) (2009) 2264–2269. [19] O. Tschritter, H. Preissl, A.M. Hennige, M. Stumvoll, K. Porubska, R. Frost, H. Marx, B. Klosel, W. Lutzenberger, N. Birbaumer, et al., The cerebrocortical response to hyperinsulinemia is reduced in overweight humans: a magnetoencephalographic study, Proc. Natl. Acad. Sci. U. S. A. 103 (32) (2006) 12103–12108. [20] A.J. Ho, C.A. Raji, J.T. Becker, O.L. Lopez, L.H. Kuller, X. Hua, S. Lee, D. Hibar, I.D. Dinov, J.L. Stein, et al., Obesity is linked with lower brain volume in 700 AD and MCI patients, Neurobiol. Aging 31 (8) (2010) 1326–1339. [21] C.A. Raji, A.J. Ho, N.N. Parikshak, J.T. Becker, O.L. Lopez, L.H. Kuller, X. Hua, A.D. Leow, A.W. Toga, P.M. Thompson, Brain structure and obesity, Hum. Brain Mapp. 31 (3) (2010) 353–364. [22] D. Gustafson, L. Lissner, C. Bengtsson, C. Bjorkelund, I. Skoog, A 24-year followup of body mass index and cerebral atrophy, Neurology 63 (10) (2004) 1876–1881. [23] N. Pannacciulli, D.S. Le, K. Chen, E.M. Reiman, J. Krakoff, Relationships between plasma leptin concentrations and human brain structure: a voxel-based morphometric study, Neurosci. Lett. 412 (3) (2007) 248–253. [24] Y. Taki, S. Kinomura, K. Sato, K. Inoue, R. Goto, K. Okada, S. Uchida, R. Kawashima, H. Fukuda, Relationship between body mass index and gray matter volume in 1,428 healthy individuals, Obesity (Silver Spring) 16 (1) (2008) 119–124. [25] E.L. Sullivan, F.H. Koegler, J.L. Cameron, Individual differences in physical activity are closely associated with changes in body weight in adult female rhesus monkeys (Macaca mulatta), Am. J. Physiol. Regul. Integr. Comp. Physiol. 291 (3) (2006) R633–R642. [26] T.A. Prolla, M.P. Mattson, Molecular mechanisms of brain aging and neurodegenerative disorders: lessons from dietary restriction, Trends Neurosci. 24 (11 Suppl.) (2001) S21–S31. [27] G. Valdez, J.C. Tapia, H. Kang, G.D. Clemenson Jr., F.H. Gage, J.W. Lichtman, J.R. Sanes, Attenuation of age-related changes in mouse neuromuscular synapses by caloric restriction and exercise, Proc. Natl. Acad. Sci. U. S. A. 107 (33) (2010) 14863–14868. [28] R.A. Gibbs, J. Rogers, M.G. Katze, R. Bumgarner, G.M. Weinstock, E.R. Mardis, K.A. Remington, R.L. Strausberg, J.C. Venter, R.K. Wilson, et al., Evolutionary and biomedical insights from the rhesus macaque genome, Science 316 (5822) (2007) 222–234. [29] M.J. Sabatini, P. Ebert, D.A. Lewis, P. Levitt, J.L. Cameron, K. Mirnics, Amygdala gene expression correlates of social behavior in monkeys experiencing maternal separation, J. Neurosci. 27 (12) (2007) 3295–3304. [30] D. Arion, T. Unger, D.A. Lewis, P. Levitt, K. Mirnics, Molecular evidence for increased expression of genes related to immune and chaperone function in the prefrontal cortex in schizophrenia, Biol. Psychiatry 62 (7) (2007) 711–721. [31] D. Arion, T. Unger, D.A. Lewis, K. Mirnics, Molecular markers distinguishing supragranular and infragranular layers in the human prefrontal cortex, Eur. J. Neurosci. 25 (6) (2007) 1843–1854. [32] A. Subramanian, P. Tamayo, V.K. Mootha, S. Mukherjee, B.L. Ebert, M.A. Gillette, A. Paulovich, S.L. Pomeroy, T.R. Golub, E.S. Lander, et al., Gene set enrichment analysis: a knowledge-based approach for interpreting genome-wide expression profiles, Proc. Natl. Acad. Sci. U. S. A. 102 (43) (2005) 15545–15550. [33] M. Ashburner, C.A. Ball, J.A. Blake, D. Botstein, H. Butler, J.M. Cherry, A.P. Davis, K. Dolinski, S.S. Dwight, J.T. Eppig, et al., Gene ontology: tool for the unification of biology. The Gene Ontology Consortium, Nat. Genet. 25 (1) (2000) 25–29. [34] Q. Zheng, X.J. Wang, GOEAST: a web-based software toolkit for Gene Ontology enrichment analysis, Nucleic Acids Res. 36 (Web Server issue) (2008) W358–W363. [35] A. Subramanian, H. Kuehn, J. Gould, P. Tamayo, J.P. Mesirov, GSEA-P: a desktop application for Gene Set Enrichment Analysis, Bioinformatics 23 (23) (2007) 3251–3253. [36] M. Abe, Y. Sato, cDNA microarray analysis of the gene expression profile of VEGFactivated human umbilical vein endothelial cells, Angiogenesis 4 (4) (2001) 289–298. [37] P.A. Marks, Discovery and development of SAHA as an anticancer agent, Oncogene 26 (9) (2007) 1351–1356. [38] E.M. Park, S. Cho, Enhanced ERK dependent CREB activation reduces apoptosis in staurosporine-treated human neuroblastoma SK-N-BE(2)C cells, Neurosci. Lett. 402 (1–2) (2006) 190–194. [39] A.S. Banks, S.M. Davis, S.H. Bates, M.G. Myers Jr., Activation of downstream signals by the long form of the leptin receptor, J. Biol. Chem. 275 (19) (2000) 14563–14572. [40] M.K. Tang, H.Y. Zhou, J.W. Yam, A.S. Wong, c-Met overexpression contributes to the acquired apoptotic resistance of nonadherent ovarian cancer cells through a cross talk mediated by phosphatidylinositol 3-kinase and extracellular signalregulated kinase 1/2, Neoplasia 12 (2) (2010) 128–138. [41] R. Dai, R. Chen, H. Li, Cross-talk between PI3K/Akt and MEK/ERK pathways mediates endoplasmic reticulum stress-induced cell cycle progression and cell death in human hepatocellular carcinoma cells, Int. J. Oncol. 34 (6) (2009) 1749–1757. [42] J.G. Shelton, W.L. Blalock, E.R. White, L.S. Steelman, J.A. McCubrey, Ability of the activated PI3K/Akt oncoproteins to synergize with MEK1 and induce cell cycle progression and abrogate the cytokine-dependence of hematopoietic cells, Cell Cycle 3 (4) (2004) 503–512. [43] D.J. Levinthal, D.B. DeFranco, Transient phosphatidylinositol 3-kinase inhibition protects immature primary cortical neurons from oxidative toxicity via

A.C. Mitchell et al. / Genomics 99 (2012) 144–151

[44]

[45]

[46] [47]

suppression of extracellular signal-regulated kinase activation, J. Biol. Chem. 279 (12) (2004) 11206–11213. A.C. Mitchell, R.K. Leak, K. Garbett, M.J. Zigmond, J.L. Cameron, K, Mirnics, Physical activity-associated gene expression signature in nonhuman primate motor cortex, Obesity (Silver Spring), 2011. M.J. Chen, A.A. Russo-Neustadt, Running exercise- and antidepressant-induced increases in growth and survival-associated signaling molecules are IGFdependent, Growth Factors 25 (2) (2007) 118–131. J.W. Kemnitz, Calorie restriction and aging in nonhuman primates, ILAR J. 52 (1) (2011) 66–77. D.H. Hyun, S.S. Emerson, D.G. Jo, M.P. Mattson, R. de Cabo, Calorie restriction upregulates the plasma membrane redox system in brain cells and suppresses oxidative stress during aging, Proc. Natl. Acad. Sci. U. S. A. 103 (52) (2006) 19908–19912.

151

[48] E.L. Sullivan, J.L. Cameron, A rapidly occurring compensatory decrease in physical activity counteracts diet-induced weight loss in female monkeys, Am. J. Physiol. Regul. Integr. Comp. Physiol. 298 (4) (2010) R1068–R1074. [49] A.C. Mitchell, G. Aldridge, S. Kohler, G. Stanton, E. Sullivan, K. Garbett, G. Faludi, K. Mirnics, J.L. Cameron, W. Greenough, Molecular correlates of spontaneous activity in non-human primates, J. Neural Transm. 117 (12) (2010) 1353–1358. [50] Z. Wu, R.A. Irizarry, Preprocessing of oligonucleotide array data, Nat. Biotechnol. 22 (6) (2004) 656–658 author reply 658. [51] E.J. Zabawski Jr., C.J. Cockerell, A middle aged female with a refractory comedone of the upper lip, Dermatol. Online J. 4 (1) (1998) 4. [52] Y. Benjamini, D. Drai, G. Elmer, N. Kafkafi, I. Golani, Controlling the false discovery rate in behavior genetics research, Behav. Brain Res. 125 (1–2) (2001) 279–284. [53] E. Segal, N. Friedman, D. Koller, A. Regev, A module map showing conditional activity of expression modules in cancer, Nat. Genet. 36 (10) (2004) 1090–1098.