Articles in PresS. Physiol Genomics (March 4, 2003). 10.1152/physiolgenomics.00184.2002
1 Evaluation of hypothalamic gene expression in mice divergently selected for heat loss
Stephanie R. Wesolowski, Mark F. Allan, Merlyn K. Nielsen, and Daniel Pomp Department of Animal Science, University of Nebraska, Lincoln, Nebraska 68563-0908
Running head: Hypothalamic gene expression and energy balance
Corresponding author: Daniel Pomp Department of Animal Science University of Nebraska ANS A218h Lincoln, NE 68583-0908 Email:
[email protected]
Keywords: obesity, microarrays, real-time PCR
Copyright (c) 2003 by the American Physiological Society.
2 Abstract Mouse lines divergently selected for heat loss were evaluated for correlated responses in the hypothalamic transcriptome. High (MH) heat loss mice have ~50% greater heat loss, ~35% less body fat, ~20% greater feed intake, ~100% greater locomotor activity levels, and higher core body temperature compared to low (ML) heat loss mice. We evaluated hypothalamic expression between inbred lines derived from MH and ML lines (IH and IL, respectively) using cDNA microarrays and selected genes previously isolated in a large dd-PCR experiment. Northern analysis was used to confirm differences, revealing higher hypothalamic mRNA expression of oxytocin (Oxt) and tissue inhibitor of metalloproteinase 2 (Timp-2) in the IH line. Real-time PCR assays were developed for Oxt, Timp-2, and ribosomal protein L3 (Rpl3, previously found to be up-regulated in IL) and confirmed differential expression of these genes with potential physiological relevance in energy balance. These results provide information on correlated responses in the transcriptome of mice selected for high and low energy expenditure, and reveal new information regarding genetic regulation of energy balance.
Keywords: obesity, microarrays, real time PCR
3 Introduction Energy balance is the difference between energy intake and energy expenditure. Imbalances in either direction can lead to body weight disregulation with potentially serious health consequences. Obesity is an increasingly serious medical problem in today’s society as approximately 60% of the American population is overweight (15, 36, 38).
In domestic
livestock species, feed costs are the largest economic inputs in food production systems (10, 24). Improving production efficiency, conserving feed resources, and producing leaner products for consumers are key priorities.
Understanding the genetic regulation of energy balance will
facilitate development of diagnostic and therapeutic advances to improve human health and enhance efficiency and quality of food products (15, 32, 38). Energy balance is a complex trait regulated by the combined effects of numerous genes, environmental factors and multi-way interactions. Several monogenic rodent obesity models produce phenotypes indicative of altered energy metabolism (34). While several quantitative trait loci (QTL) for traits related to energy balance have been detected, none have been cloned and identified (33).
The complex genetic architecture of energy balance remains largely
uncharacterized. To contribute to an understanding of the complexity of energy balance at the polygenic level, we have employed gene expression techniques to find hypothalamic transcriptional differences between two unique mouse lines created by divergent selection based on direct calorimetry measurement of heat loss (29). The high heat loss line (MH) has ~50% higher heat loss, ~20% greater feed intake, ~35% less body fat, ~100% greater locomotor activity, and elevated core body temperature compared to the low line (ML) despite no differences in body weight (25, 27, 29, 30). These lines are unique models to dissect the genetic complexity of
4 energy balance and its related phenotypes on a population basis. The hypothalamus was selected for evaluation because of its role in controlling traits such as feeding behavior, weight gain, and adaptive thermogenesis (14, 22, 36). Our hypothesis was that response to selection for heat loss is mediated, to some extent, through direct and/or correlated responses in the hypothalamic transcriptome. Specifically, we used replicated cDNA microarray experiments to identify and prioritize candidate genes. Candidate genes were also selected from the literature based on known roles in energy balance regulation, and from a previous differential display project that evaluated hypothalamic expression differences between the IH and IL (inbred MH and ML, respectively) lines (1).
Confirmation of differential gene expression was accomplished using northern
hybridization, and real-time PCR assays were designed to create sensitive and high-throughput quantification assays.
Materials and Methods Mice. Mice used in this study were from partially inbred lines derived from long-term selection for high (MH) and low (ML) heat loss (29). Briefly, selection was initiated from a composite base population using direct calorimetry measurements of heat loss in 9- to 11-wk males for 16 generations in three separate replicates. Means for heat loss in the MH and ML lines were 179.1 and 107.5 (kcal/kg0.75/day), respectively, and realized heritability was estimated to be 0.28. Full-sib matings were initiated at Generation 21 from Replicate 1 of each selection line to develop inbred high (IH) and inbred low (IL) lines. Mice used in this study were from Generation 32 and 34 of the IH and IL lines. Mice were weaned at 3 wk, caged by sex, and provided ad libitum access to water and feed (Teklad 8604 rodent chow). Temperature was
5 maintained at 22oC with relative humidity at 35-50%, and a 12:12 light-dark cycle with lights on at 0700 h was employed. All procedures and protocols were approved by the University of Nebraska-Lincoln Institutional Animal Care and Use Committee. Collection of phenotypic data and tissues. Heat loss was measured on male mice at 11 to 12 wk of age using direct, gradient-layer, individual-animal calorimeters (29). Briefly, mice were weighed and placed in calorimeters with 2.25 g feed at ~1700 h. Following a 30-min acclimation period, heat loss measurements were recorded every minute for a 15-h period. Data were adjusted for metabolic body weight and projected to a 1-d basis (kcal/kg0.75/day). At 14 weeks of age, mice were fasted 2 to 3 hr and decapitated at 1000-1100 h after brief exposure to CO2. Hypothalamic samples were immediately dissected and snap frozen in liquid nitrogen as previously described (1). RNA Extraction. Total RNA was isolated from hypothalamus samples using TRIZOL LS reagent (Invitrogen Life Technologies; Carlsbad, CA). Pools of hypothalamus (4-5 samples) or a single hypothalamus were extracted following manufacturer’s protocol. Single hypothalamic extractions were performed in the presence of 5 µl of glycogen (20 mg/ml) using phase-lock gel tubes (Eppendorf; Westbury, NY). Aliquots of RNA samples were loaded in agarose gels to visualize quality and integrity. RNA was quantified with a fluorometer (TD-700, Turner Designs, Sunnyvale, CA) or plate reader (Wallac Victor2, Perkin Elmer; Boston, MA) using the RiboGreen RNA Quantitation Kit (Molecular Probes; Eugene, OR). Microarrays. Two separate hybridizations were performed using the Mouse GEM 2.08 cDNA microarray (Incyte; St. Louis, MO). Twenty-five hypothalamic tissue samples collected from male mice in each line with the most extreme heat loss measurements were randomly allocated to six groups within line. Tissue samples within each group were pooled, extracted,
6 and checked for quality and quantity as described above. Pooled RNA samples were mixed within each line to yield two pools per line of total RNA. Samples were purified for mRNA using two passes through Oligotex mRNA Isolation Columns (Qiagen; Valencia, CA). Aliquots (600 ng) of each pooled sample were sent to Incyte for array experiments. In the first experiment (Replicate 1), the IH pooled sample was labeled with Cy3 dye and the IL pooled sample with Cy5. In Replicate 2, different IH and IL pools were used and dye labels were reversed. The array contained 9596 sequence verified cDNA clones.
The ratio of the two fluorescent
intensities, adjusted using controls, provided a quantitative measurement of gene expression (GemTools Software; Version 2.4, Incyte Genomics). Genes with significantly and suggestively different expression levels were defined to be those with absolute balanced differential expression (ABDE) ratios greater than or equal to 2.0 and 1.6, respectively. Candidate Genes. Numerous genes in various pathways have been implicated in the regulation of energy balance based on identification of loci responsible for single-gene obesity phenotypes, gene expression and transgenic model analyses (16, 34, 36). Genes that have been specifically shown to have altered hypothalamic mRNA expression resulting in phenotypic differences related to energy balance were selected as candidates for evaluation for hypothalamic transcriptional differences between the IH and IL lines. These genes included beacon (Beac) (5), neuropeptide Y (Npy) (36), metallothionein 1 (Mt1) (4), metallothionein 2 (Mt2) (4), Vgf nerve growth factor (Vgf) (34), and carboxypeptidase E (Cpe) (16). Genes were also selected from a differential display PCR experiment that evaluated differences in hypothalamic gene expression between the IH and IL lines (1). Differentially displayed fragments from this project had already been cloned and sequenced (GenBank accession numbers AW358547-AW358553 and BQ135237-BQ135326).
Genes selected from this analysis for use in our study included
7 glutamine synthetase (Glns) and ribosomal protein S11 (Rps11), which had greater expression in the IL line, and ribosomal protein S10 (Rps10) and tissue inhibitor of metalloproteinase 2 (Timp2), which had greater expression in the IH line. Northern Hybridization. Northern blots were made with four pools of RNA samples (two per line). Five or 10 µg of total RNA were used following standard procedures (1). Probes for northern hybridization were prepared for genes of interest using the Megaprime DNA random labeling kit (Amersham). PCR products for each target gene were amplified using primers designed from available GenBank sequences (Table 1) and cDNA prepared from pooled hypothalamic RNA samples. Products were cloned and sequenced to verify gene sequence specificity (University of Nebraska Core Sequencing Facility). For genes identified from the previous dd-PCR experiment, fragments of interest were amplified from the cloned vector using the respective primers that generated the product.
Briefly, cDNA was synthesized using
Superscript II reverse transcriptase (Invitrogen) in the presence of RNasin (Promega; Madison, WI) following manufacturer’s protocols. PCR products were purified using QiaQuick PCR purification kit (Qiagen) and 12.5 ng of purified product were used in a 25 µl labeling reaction following the manufacturer’s protocol. procedures (1).
Hybridizations were done according to standard
Evaluation of northern results was accomplished using images of
autoradiographs and densitometric values obtained using the Hitachi CCD Bio System and GeneTools Software (Hitachi Genetic Systems; Alameda, CA) or PhosphorImager System SF (Molecular Dynamics). Membranes were stripped with 0.1% SDS at 95oC, and re-probed with glyceraldehyde phosphate dehydrogenase (Gapdh) for standardization. Real-Time PCR.
Real-time Taqman PCR assays were designed and optimized for
quantification of oxytocin (Oxt) and ribosomal protein L3 (Rpl3) gene expression. Reagents and
8 conditions previously reported for the tissue inhibitor of metalloproteinase 2 (Timp-2) gene were used (44). RNA samples from single hypothalamic extractions (500 ng) were treated with DNase I (Invitrogen) to remove any contaminating genomic DNA. Reactions contained 1X DNase I Buffer and 100 U DNase I in a 10 µl volume. Reverse transcription was accomplished in a 20 µl volume using 200 ng RNA from DNase treatment, 1X RT buffer, 5.5 mM MgCl2, 500 µM dNTP mixture, 2.5 µM oligo-d(T)16 primer, 8 U RNase Inhibitor, and 25 U MultiScribe enzyme (Applied Biosystems, Inc., Foster City, CA). PCR quality checks for cDNA were performed using primers for D4Mit190 microsatellite marker to ensure no genomic DNA was present, and primers for growth hormone releasing hormone (Ghrh) to ensure hypothalamic transcripts were extracted. Real-time PCR reactions were performed in a 25 µl volume using 12.5 µl of 2X Universal Master Mix without AmpErase UNG (Applied Biosystems), appropriate amounts of primers and probe, and 1 ng cDNA. Reactions were amplified in the ABI Prism 7700 Sequence Detector System (Applied Biosystems). PCR conditions were 50oC for 5 min, followed by 40 cycles with denaturation at 95oC for 15 sec and a combined annealing and extension step at 60oC for 1 min. Primer and probe sequences were designed using Primer Express Software (Version 1.5a, Applied Biosystems) and Genbank sequences GI:200167 and GI:6113383537 for Oxt and Rpl3, respectively (Table 2).
Amplicon specificity was verified by sequence analysis
(University of Nebraska Core Sequencing Facility). Target gene probes were synthesized by Applied Biosystems with the reporter dye FAM attached to the 5’-end. The Rodent Gapdh Control Reagents (Applied Biosystems) with VIC reporter dye were used to amplify Gapdh to evaluate and adjust for sample loading differences.
9 Optimal monoplex reaction conditions for Oxt, Rpl3, and Timp-2 were achieved with 200 nM of forward and reverse primers with 100 nM probe. Standard Gapdh concentrations of 100 nM forward and reverse and 200 nM probe were used. Duplex reactions for each target gene (Oxt, Rpl3, Timp-2) and Gapdh were optimized. For Oxt and Rpl3 assays, 150 nM of forward and reverse primers and 100 nM of probe were used with 75 nM Gapdh forward primer, 100 nM Gapdh reverse primer, and 200 nM Gapdh probe. For Timp-2, 200 nM of forward and reverse primers and 100 nM probe were duplexed with 50 nM Gapdh forward, 75 nM Gapdh reverse, and 200 nM Gapdh probe. To make accurate comparisons among samples and adjust target gene data for the control gene (Gapdh), the efficiency of the two PCR reactions must be similar. Dilutions of cDNA were run in triplicate using the optimal monoplex primer and probe concentrations. The average threshold cycle (CT) value for each dilution was plotted against the log of the input cDNA amount to evaluate linearity of the assay. To find the relative efficiency, the difference between average target gene CT and average control gene CT ( CT) for each dilution was plotted against the log input cDNA amounts. A slope with an absolute value IL Equal Equal IH > IL Equal
The mouse chromosomal location and centimorgan (cM) position. NA indicates map location for the gene is unknown.
30 Table 5. Results of real time PCR (Taqman) gene expression quantification assays. Genea Monoplex Oxt Rpl3 Timp-2 Duplex Oxt Rpl3 Timp-2
Line
Fold Differenceb
Normalizedc
Covariated
IH IL IH IL IH IL
1.18 0.91 * 0.87 1.17 *** 1.04 0.97
1.0947 1.1140 * 1.1241 1.1065 *** 1.2103 1.2175
25.73 26.17 * 26.42 25.98 *** 28.42 28.56
IH IL IH IL IH IL
1.12 0.95 ** 0.91 1.14**** 1.07 0.98
1.0515 1.0616 1.0614 1.0473 **** 1.1723 1.1777
24.74 25.14 **** 25.59 25.29 **** 28.14 28.44 **
*P < 0.10, ** P < 0.05, *** P < 0.01, **** P < 0.001 a
Monoplex data were collected on 8 males within in each line. Duplex data were pooled and analyzed from two runs of each assay. The first run contained 16 males and 4 females from each line and the second run contained the same mice plus four additional males per line. Statistical significance is indicated for each comparison. b The fold difference in expression was calculated using the formula for Comparative CT Method. c Normalized values were calculated by dividing target gene by control gene expression. d Covariate analysis used control gene as covariate for each target sample in analysis of variance.
31 Figure Legends Figure 1. Microarray expression results for oxytocin (Oxt) and tissue inhibitor of metalloproteinase 2 (Timp-2) genes. Oxt had an ABDE of 2.6 in Replicate 1 and 1.3 in Replicate 2 with both indicating greater expression in the IH line. Timp-2 had an ABDE of 1.0 in Replicate 1 and 2.0 in Replicate 2 indicating higher expression in the IH line. The color scale represents relative levels of expression. Figure 2. Northern hybridization confirmation of differential hypothalamic gene expression for Oxt (A) and Timp-2 (B) between the heat loss selection lines. Upper transcripts represent Gapdh and lower represents either Oxt (A) or Timp-2 (B). The left two lanes in each blot contain different pooled IL samples and the right two lanes contain pooled IH samples. Figure 3. Northern hybridization results for candidate genes. No differences were detected in hypothalamic mRNA expression between the IH and IL line. The left two lanes in each blot contain different pooled IL samples and the right two lanes contain pooled IH samples. Figure 4. Validation of real time PCR Taqman assays. (a) Linear dynamic range and efficiency of real-time PCR assays. Dilutions of a pooled cDNA sample were run in triplicate at 5, 2, 1, 0.5, 0.2 and 0.1 ng amounts. The average CT value at each dilution was plotted against the log of the input cDNA amount. A slope of –3.3 represents ~100% PCR efficiency. (b) Relative efficiency of real time PCR assays for each target gene compared with Gapdh control gene. Dilutions of cDNA were run for both target and control gene assays. At each dilution, the average CT was plotted against the log of the input cDNA amount. A slope with an absolute value of less than 0.1 indicates equal PCR efficiencies.
32 Figure 1: Replicate 1 IL Oxt
Timp-2
IH
Replicate 2 IL
IH
33 Figure 2: IL
IH Gapdh
A Oxt
Gapdh
B Timp-2
34 Figure 3: IL
IH
IL
Beac
IH
IL Npy
Itm2 Cpe
Rps1 0 Kif1a
Fgf1
Gln s
Rps11 Mt-1 Gapdh
Hp1bp 3
Mt-2
IH
35 Figure 4:
A
CT
30.0 25.0 20.0 -1.5
-1.0
-0.5
0.0
0.5
1.0
Log input cDNA amount (ng) Gapdh; slo pe = -3.56
Rpl3; slope = -3.34
Timp-2; slope = -3.60
Oxt; slo pe = -3.58
CT
7.0
B
5.0 3.0 1.0 -1.5
-1.0
-0.5
0.0
0.5
1.0
Log input cDNA amount (ng) Rpl3; slo pe = 0.088 Oxt; slo pe = -0.021
Timp-2; slo pe = -0.044
