Identification of Quantitative Trait Loci Influencing Traits Related to Energy Balance in Selection and Inbred Lines of Mice

Corresponding author: D. Pomp, Department of Animal Science, A218 Animal Science Bldg., University of Nebraska, Lincoln, NE 68583-0908., dpomp{at}unl.edu (E-mail)

Communicating editor: T. F. C. MACKAY

	ABSTRACT

TOP ABSTRACT MATERIALS AND METHODS RESULTS DISCUSSION LITERATURE CITED

Energy balance is a complex trait with relevance to the study of human obesity and maintenance energy requirements of livestock. The objective of this study was to identify, using unique mouse models, quantitative trait loci (QTL) influencing traits that contribute to variation in energy balance. Two F₂ resource populations were created from lines of mice differing in heat loss measured by direct calorimetry as an indicator of energy expenditure. The HB F₂ resource population originated from a cross between a noninbred line selected for high heat loss and an inbred line with low heat loss. Evidence for significant QTL influencing heat loss was found on chromosomes 1, 2, 3, and 7. Significant QTL influencing body weight and percentage gonadal fat, brown fat, liver, and heart were also identified. The LH F₂ resource population originated from noninbred lines of mice that had undergone divergent selection for heat loss. Chromosomes 1 and 3 were evaluated. The QTL for heat loss identified on chromosome 1 in the HB population was confirmed in the LH population, although the effect was smaller. The presence of a QTL influencing 6-wk weight was also confirmed. Suggestive evidence for additional QTL influencing heat loss, percentage subcutaneous fat, and percentage heart was found for chromosome 1.

ENERGY balance, or the difference between energy intake and energy expenditure, is a complex trait with important implications for human health and livestock production. Energy imbalance resulting in weight gain, and potentially leading to obesity, results when energy intake is greater than energy expenditure. The identification of low resting metabolic rate or total energy expenditure as risk factors for weight gain in certain human populations (RAVUSSIN et al. 1988 ; ROBERTS et al. 1988 ; GRIFFITHS et al. 1990 ) has led to the hypothesis that low energy expenditure provides a mechanism by which people susceptible to obesity can make excess energy available for weight gain (SALTZMAN and ROBERTS 1995 ). In contrast to the situation in humans in which low energy expenditure may be undesirable because of its potential contribution to unwanted weight gain and obesity, low energy expenditure and thus low maintenance energy requirements in livestock may be desirable to minimize the cost of feed required to maintain mature animals.

The genetic regulation of energy expenditure has been studied in mice through divergent selection for heat loss, which is measured using individual-animal direct calorimeters (NIELSEN et al. 1997A ). Because energy expenditure must equal energy intake at maintenance, and because energy is expended primarily as heat, heat loss was measured in mature mice as an indicator of energy expenditure and maintenance energy requirements. After 15 generations of selection, the difference in heat loss between lines selected in the high (MH) and low (ML) directions was 53% relative to the control line mean. The MH mice consumed significantly more feed, were leaner, more active, and had larger litters compared to ML mice (MOODY et al. 1997 ; NIELSEN et al. 1997A , NIELSEN et al. 1997B ; MOUSEL 1998 ). These results demonstrate that the regulation of heat loss has a significant genetic component, and that selection for heat loss successfully changed energy expenditure as well as other traits related to energy balance. In particular, decreased fatness of MH relative to ML mice despite increased feed intake makes them a unique model with which to study correlations among fat deposition, feed intake, and energy expenditure (see POMP and NIELSEN 1999 ).

Although several studies have identified quantitative trait loci (QTL) contributing to energy imbalance measured as adiposity (see POMP 1997 ), few have focused on component characteristics of energy balance, such as energy intake or energy expenditure. Because of the divergence in heat loss generated through selection, the MH and ML lines provide a valuable genetic resource with which to pursue such QTL. However, because the selection lines are outbred lines that originated from a common base population, they are likely to share common alleles at many molecular markers, which would make standard methods of QTL analysis for inbred lines inappropriate (LANDER and BOTSTEIN 1989 ; ZENG 1993 ). Many inbred lines with variable phenotypes are readily available, so that a resource population created from a cross between a selection and an inbred line with significantly different heat loss would allow QTL to be identified by using fully informative markers and statistical methods developed for inbred lines.

In a previous study, MOODY et al. 1997 compared traits related to energy balance among three inbred lines and the heat loss selection lines of mice. The greatest difference in heat loss between an inbred and selection line was found between MH and the inbred line C57BL/6J (BL). Average heat loss and feed intake of BL mice were similar to ML, but their body composition was similar to MH. Thus, even though BL and ML lines have similarly low heat loss, this phenotype may arise from different mechanisms. In the ML line, the mechanism leading to low heat loss may also facilitate deposition of excess energy as fat, whereas low heat loss in the BL line may not result in increased fat deposition. Thus, it was hypothesized that some QTL alleles responsible for low heat loss would be shared between BL and ML, while others influencing both heat loss and fat deposition would be unique within each of the two lines. The present study has identified QTL that influence heat loss and other phenotypes related to energy balance in two resource populations. A complete genome scan was conducted in a cross between MH and BL. Regions harboring QTL with the greatest effects were then evaluated in a cross between MH and ML.

	MATERIALS AND METHODS

TOP ABSTRACT MATERIALS AND METHODS RESULTS DISCUSSION LITERATURE CITED

Resource populations:
Two different resource populations were created to utilize the resources available from diverse selection and inbred lines of mice. Detailed descriptions of the MH and ML lines divergently selected for heat loss have been presented elsewhere (NIELSEN et al. 1997A , NIELSEN et al. 1997B ). Briefly, selection was initiated from a composite base population created from four outbred strains of mice including NIH and ICR acquired from Harlan Sprague Dawley and CF-1 and CFW acquired from Charles River (JONES et al. 1992 ). Selection was based on measurement of heat loss (kcal/kg^0.75/day) in 9- to 11-wk males using individual-animal direct calorimeters. Following 16 generations of selection, the lines were maintained by random mating with care taken to minimize inbreeding. The cumulative realized selection differentials for generation 15 were 136.9 and -106.6 kcal/kg^0.75/day for MH and ML selection, respectively. Realized heritability for heat loss was 0.28 based on divergence of MH and ML. The first resource population (HB; n = 560) was created from a cross between the MH selection line and the BL inbred line. Average heat loss and food intake were similar for BL and ML and significantly less than for MH (MOODY et al. 1997 ). A second F₂ intercross population (LH; n = 560) was created from a cross between the MH and ML lines. (See Table 1 for phenotypic means of BL, MH, and ML lines.) All MH and ML mice used in this study were sampled from the first of three replicated selection lines.

Each resource population was produced across four time periods separated by 4 wk. Matings in periods 1 and 2 were repeated in periods 3 and 4 such that each mating produced ~16 full-sibs in two time periods. BL males (n = 5) and MH females (n = 15) were mated to produce F₁ progeny, and male (n = 22) and female (n = 36) F₁ progeny were crossed to produce the HB resource population (n = 560). Similarly, males (n = 12) and females (n = 12) representing the MH and ML lines produced an F₁ generation, and F₁ males (n = 22) and females (n = 37) were crossed to produce the LH resource population (n = 560). All litters were standardized to eight pups at birth when possible. Pups were weaned at 3 wk and housed 3–4 per cage with ad libitum access to feed (Teklad 8604 Rodent Chow) and water until energy balance phenotypic measurement began at 10–12 wk. All mice were housed in stainless steel cages with wood-chip bedding and maintained at 22°, 35–50% relative humidity, and a light:dark cycle of 12:12 hr beginning at 0700 hr.

Measurement of phenotypes:
Body weights were measured at 3, 6, and 10 wk. Direct calorimeters were used to measure heat loss at 10–12 wk of age as previously described (NIELSEN et al. 1997A ). Mice were placed into 1 of 10 individual-animal calorimetry units (Thermonetics Corporation, San Diego, CA) with a 3.5-g pellet of food at ~1630 hr. After a 30-min acclimation period, heat loss was measured every minute for a continuous 15-hr period. The average heat loss was adjusted for metabolic body weight and expressed on a 24-hr basis (kcal/kg^0.75/day). Data were discarded for mice that failed to consume at least 2.5 g of the food pellet because of a reduction in heat loss (n = 16 and 11 mice in the HB and LH populations, respectively). Following heat loss measurement, mice were housed in individual cages with hanging wire feeders. Food intake was measured by difference over a 14-day period, adjusted for metabolic body size, and expressed on a 24-hr basis (g/kg^0.75/day). Data were discarded for mice with significant food spillage (15 and 2 mice in the HB and LH populations, respectively). Otherwise, spillage was minimal and assumed to be randomly distributed. Body weights were measured and mice were killed 1 to 3 days after the food intake measurement period. Subcutaneous, gonadal (epididymal or perimetrial pads of males and females, respectively), and brown adipose pads, livers, and hearts were dissected and weighed. All adipose and organ weights were expressed as a percentage of body weight. The percentages of combined subcutaneous and gonadal fat pads were also evaluated as an indicator of total body fatness. All phenotypes were also measured on mice (n = 20 males and 20 females) representing each parental line (ML, MH, and BL) and F₁ generation (MH x ML [n = 70] and MH x BL [n = 40]).

Genetic Analysis
Genotyping of HB population: Standard methods were used to extract DNA from tail clips. Fully informative markers were identified by screening microsatellite markers (MapPairs; Research Genetics, Huntsville, AL) in grandparents of the HB population to identify markers with different alleles in the MH compared to BL line. Genotypes were determined by standard PCR and agarose gel electrophoresis protocols. Genotypes were scored as B, M, or H representing BL allele homozygotes, MH allele homozygotes, or heterozygotes, respectively. Genotypes were scored twice and discrepancies between the two scores were rectified.

Genotyping of the HB population was completed in three phases. In phase 1, mice with the highest and lowest heat loss within each full-sib family were selected after adjusting for effects of sex and period. Additional mice were selected until a total of 46 for each criterion (high or low heat loss) was identified with approximately equal representation of each time period and sex. This selected group was genotyped for a total of 62 markers representing each chromosome at 20- to 40-cM spacing. In phase 2, the complete HB population was genotyped for 19 markers identified in phase 1 as having potential linkage to QTL influencing heat loss (see Statistical Analysis section). Phase 3 involved genotyping the complete HB population for additional markers located on chromosomes harboring markers identified in phase 2 as having significant effects on heat loss (see Statistical Analysis section).

Genotyping of LH population: Genotypes were determined by standard PCR using an infrared fluorescent dye-labeled forward primer followed by electrophoresis and analysis on the Li-Cor (Lincoln, NE) Model 4200 IR² system. Two reactions using primers with different labels (IRD700 and IRD800) were combined after PCR. Gels were analyzed using Gene ImagIR analysis software (Li-Cor) to determine allele sizes of each individual. The LH population was evaluated for chromosomes 1 and 3, which contained QTL having the largest effects on heat loss in the HB population. Markers on these chromosomes were screened in MH and ML grandparents of the LH population and selected if more than one allele was found. A total of eight and six markers on chromosomes 1 and 3, respectively, were genotyped in all MH and ML grandparents, F₁ parents, and LH F₂ mice.

Statistical Analysis
Description of phenotypes: The means and standard deviations of traits were determined for the HB and LH populations, as well as for mice representing each parental and F₁ line. Phenotypic differences among MH, ML, BL, and both F₁ populations were evaluated using the generalized least-squares procedure of SAS (1988) with fixed effects of line, sex, and line-by-sex interaction. Significant line effects were further evaluated by defining contrasts to test differences between MH and ML, MH and BL, and between each F₁ population and the average of its two parental lines.

HB population: Genetic distances between markers were determined using the Mapmaker 3.0 (LINCOLN et al. 1992 ) analysis program with distances reported in Haldane centimorgan units.

Analysis of phase 1 and phase 2 genotypes: In phase 1, genotypic frequencies of selectively genotyped mice were evaluated using a chi-square analysis to test for equal allelic frequencies between high and low selected groups, which is expected under the null hypothesis of no linked QTL. Markers with a chi-square test statistic >2.71 (P < 0.1; 1 d.f.) were identified for further evaluation by whole-population genotyping. In phase 2, the effect of marker genotype on heat loss phenotype was evaluated by analysis of variance using the PROC MIXED procedure of SAS (1988). Fixed effects included time period, sex, marker genotype, and genotype-by-sex interaction, with sire-dam included as a random effect to account for effects that may be attributed to the genetic background of a specific sire and dam combination. Eight chromosomes with markers having nominally significant (P < 0.05) effects on heat loss were identified for evaluation by interval analysis in phase 3.

Interval analysis of phase 3 genotypes: Interval analysis (LANDER and BOTSTEIN 1989 ) was used to better define the location and effects of QTL influencing all traits for the eight chromosomes identified in phase 2. Given the assumption that the MH and BL lines were fixed for alternate QTL alleles (Q and q) and the effects of QQ, Qq, and qq are a, d, and -a, respectively (FALCONER and MACKAY 1996 ), the expected value of an F₂ individual may be described in terms of a and d,

(1)

where m is a constant determined by fixed effects; c_ai is the coefficient for the additive component for individual i at the given location, defined as Prob(QQ) - Prob(qq); c_di is the coefficient for the dominance component for individual i at the given location, defined as Prob(Qq); and e_i is the residual error. Equation 1 may be expanded to account for fixed effects and effects of conditioning markers located on other chromosomes. The inclusion of conditioning markers has been recommended as a way to account for effects of background genes and other QTL (ZENG 1993 ). The expanded equation may be described in matrix notation as

(2)

where y is a vector of phenotypes; b is a vector of fixed effects; X is an incidence matrix relating fixed effects to individuals; m is a vector of effects attributed to conditioning markers; C is a matrix of indicators that relate effects of genotyped conditioning markers to y; e represents the residual error; and s and t are vectors of c_ai and c_di, the coefficients of a and d, respectively, from Equation 1.

Analysis programs were developed to calculate the c_ai and c_di coefficients at 2-cM intervals on the basis of genotypes and locations of flanking markers (see HALEY and KNOTT 1992 ), and the maximum-likelihood estimate for the residual variance for Equation 2 was obtained. The difference between the resulting -2 log-likelihoods of the full model (Equation 2, including a and d) and the reduced model (ignoring a and d) provided a likelihood-ratio test statistic (LR) that was converted to a LOD score (LOD = LR/4.601) for each 2-cM interval.

Fixed effects of sex, time period, and sire-dam were included for all traits. Litter size was also included as a fixed effect for 3- and 6-wk weights and percentage liver. Sex was accounted for as a fixed effect because no line-by-sex interaction was found when phenotypes of the MH, ML, and BL lines were compared for these traits (MOODY et al. 1997 ) and because no marker by sex interaction was found for heat loss in the analysis of variance completed for phase 2. Conditioning markers were selected for each trait in a two-step process using the backward selection option of the PROC REG procedure of SAS (1988). In the first step, models including all markers on a single chromosome were evaluated. Markers remained in the model if their effect approached significance (P < 0.1). In the second step, markers identified from all chromosomes in the first step were included in a single model. Conditioning markers included in the QTL analysis were those that remained in the final regression model (P < 0.05). Conditioning markers were omitted from analyses of chromosomes on which they were located.

Effects of QTL: Effects of QTL are presented as the additional percentage of residual phenotypic variance explained by the QTL and the a and d effects of each QTL. The percentage variance was calculated as [(residual variance of reduced model - residual variance of full model)/residual variance of reduced model] x 100. Solutions and standard errors for a and d were obtained using option 4 of the MTDFREML programs (BOLDMAN et al. 1995 ) with c_ai and c_di at the QTL location determined by the peak LOD score included in the model as covariates along with the same fixed effects and conditioning markers used in the interval analysis. Standard errors were calculated using the REML estimate of residual variance.

Confidence regions: One-LOD confidence regions are presented for significant QTL as the chromosomal region with LOD scores greater than or equal to one less than the peak LOD score at the QTL position.

Significance thresholds: Permutation testing described by CHURCHILL and DOERGE 1994 was used to determine empirically derived significance thresholds specific to the HB data. Permuted heat loss data were analyzed 1000 times with fixed effects and conditioning marker genotypes permuted along with associated phenotypes such that these effects were accounted for with each permutation. The maximum LOD for each chromosome, as well as the maximum across all chromosomes, was identified for each permuted analysis. The experiment-wise significance threshold level was defined as the 50th highest LOD among the 1000 maxima identified across all chromosomes, corresponding to an experiment-wise type 1 error rate of 0.05. Suggestive linkage threshold values were defined for each chromosome as the 50th highest LOD among the 1000 maxima for the respective chromosome.

LH population: Linkage maps for chromosomes 1 and 3 were constructed using Cri-map (GREEN and CROOKS 1990 ) with results reported in Haldane centimorgan units. Interval analysis in the LH population was completed as described for the HB population, except that c_ai and c_di were calculated for 1-cM intervals conditional on genotypes in the three-generation pedigree for all markers in a linkage group (HALEY et al. 1994 ). This method allows all available information to be used to calculate the probability of a QTL genotype at a given location. Fixed effects were the same as for the HB population, but no conditioning markers were included in the analysis of the LH population.

Because only specific regions of chromosomes 1 and 3 were evaluated in the LH population to test for the presence of significant QTL identified from the HB population, the problem of multiple testing was reduced. Significance of QTL in these specific regions (20–30 cM) was determined by P < 0.01 as suggested by LANDER and KRUGLYAK 1995 . However, because no previous evidence for QTL in other regions of these chromosomes exists, identification of QTL in these regions was handled as if they were evaluated as part of a complete genome scan. Permutation testing was carried out as described for the HB population to determine suggestive linkage threshold levels. Because only two chromosomes were genotyped, permutation testing could not be used to determine genome-wide significant threshold levels.

	RESULTS

TOP ABSTRACT MATERIALS AND METHODS RESULTS DISCUSSION LITERATURE CITED

Description of phenotypes:
The mean and standard deviation for each trait in the original BL, MH, and ML lines, as well as in the F₁ crosses, are shown in Table 1. The mean of the MH line was significantly greater than that of BL and ML for heat loss and food intake. Means of traits related to adiposity were greater in ML than in MH but similar in BL and MH. Body weight of MH was significantly greater than that of BL but similar to that of ML. Heterosis for heat loss was observed as the mean of both F₁ populations was less than the average of their two parental lines. The BL/MH F₁ mean was similar to the average of BL and MH for the remaining traits, except for brown adipose, liver, and 10-wk weights where the F₁ mean exceeded the mean of either of its parental lines. In contrast, mean of the ML/MH F₁ was different from the average of ML and MH for all traits except food intake. The ML/MH F₁ mice were significantly larger and had greater subcutaneous, gonadal, and combined fat than either of their parental lines, but their brown adipose, liver, and heart weights were less than the average of MH and ML. A large degree of phenotypic variation was generated in both F₂ populations for all traits evaluated (Table 2).

HB population:
All markers genotyped in the HB population and their chromosomal locations are listed in Table 3. In general, positions of markers determined from the HB population agreed with those found in the Mouse Genome Database. However, three markers (D1Mit234, D9Mit243, and D10Mit44) were unlinked to their respective chromosomes in the HB population and were omitted from further analyses.

Phase 1 and 2 analyses: Markers with preliminary evidence of QTL influencing heat loss on the basis of selective genotyping are indicated in Table 3. Nineteen markers representing 11 chromosomes were identified as having potential linkage with QTL (P < 0.1) and were genotyped in the complete population. Of these, markers on chromosomes 5, 6, and 9 failed to influence heat loss on the basis of analysis of variance (P > 0.05) and were not included in subsequent interval analyses (Table 3).

Interval analysis: The experiment-wise 5% LOD threshold value based on permutation analysis across eight chromosomes was 3.28, which was defined as a threshold for declaring significant evidence for linkage of QTL. This value was essentially identical for the three traits subjected to permutation analysis (heat loss, subcutaneous fat, and 10-wk weight). Suggestive evidence for QTL was defined as the chromosome-wise 5% LOD threshold value. Because of constraints on computational time, the highest threshold value among the three traits on a particular chromosome was used as the threshold level for all traits on that chromosome. Suggestive threshold values ranged from 2.05 (chromosome 11) to 2.39 (chromosome 3). There was little variation in threshold values across the three traits within a chromosome.

Locations of peak LOD scores and effects of QTL at these locations for all traits are presented in Table 4, Table 5, and Table 6. Nine QTL influencing heat loss were identified including five (Hlq1, Hlq2, Hlq3, Hlq4, and Hlq5) that exceeded the significant linkage threshold level (Figure 1). Although two peaks are present on chromosome 3 (Figure 1), the confidence intervals for Hlq3 and Hlq4 overlapped, indicating they may represent a single QTL. The QTL that exceeded the significant threshold level each accounted for 3.1 to 4.7% of residual variance while those exceeding the suggestive level accounted for 2.0 to 2.7%. Together, these nine QTL explained 27.7% of residual variance for heat loss, after accounting for variation due to time period, sex, sire-dam, and conditioning markers. With the exception of the suggestive QTL region on chromosome 4, the effects of these QTL were primarily additive, ranging from 2.1 to 4.4 kcal/kg^0.75/day. The MH allele resulted in increased heat loss for all QTL.

View larger version (23K): In this window In a new window Download PPT slide   Figure 1. Significant QTL influencing heat loss (HLOSS) and percentage brown fat (BAT) identified by interval analysis in the HB population are shown for chromosomes 1, 2, 3, and 7. Significant and suggestive threshold levels are shown as horizontal solid and dotted lines, respectively, for each chromosome. One-LOD confidence intervals are shown as lines below the x-axis for each QTL.

Four QTL influencing brown adipose tissue weight were identified (Table 5), and each of them were within confidence intervals identified for heat loss QTL (Figure 1). The brown adipose QTL accounted for 1.8 to 3.3% of the residual variance and had additive effects ranging from 0.012 to 0.024%. The MH allele resulted in increased brown adipose for Batq1, Batq2, and for the suggestive QTL on chromosome 11, but the BL allele caused increased brown adipose for the suggestive QTL located on chromosome 17.

Significant evidence was found for one QTL influencing gonadal and combined fat (Fatq1) on chromosome 1 (Figure 2), while suggestive evidence (Table 5) was found for QTL on chromosomes 1 and 4 (subcutaneous), 7 and 12 (all three fat traits), and 11 (gonadal and combined). The MH allele resulted in increased fatness for QTL on chromosomes 1, 11, and 12, but decreased fatness for QTL on chromosomes 4 and 7. Fatq1 on chromosome 1 explained 5.4 and 5.9% of the residual variance of combined and gonadal fat, respectively, while the remaining QTL each accounted for 2.0 to 2.7%.

View larger version (18K): In this window In a new window Download PPT slide   Figure 2. Significant QTL influencing percentage subcutaneous (SUB), gonadal (GON), and subcutaneous + gonadal (FAT) fat pad weights identified by interval analysis in the HB population are shown for chromosome 1. Significant and suggestive threshold levels are shown as horizontal solid and dotted lines, respectively. One-LOD confidence intervals are shown as lines below the x-axis for each QTL.

Body weight QTL (Figure 3) were identified on chromosomes 1 (3, 6, and 10 wk), 3 (10 wk), 11 (6 and 10 wk), and 17 (3 wk). Suggestive evidence was also found for additional QTL on these chromosomes, as well as on chromosomes 7 and 4 (Table 6). The QTL with the largest effect was on chromosome 1, where Wt3q1 accounted for 8% of the residual variance or 0.6 g of body weight at 3 wk. The remaining QTL explained from 1.9 to 5.0% of the residual variance (Table 5). Although most of these QTL demonstrated additive gene action with the MH allele causing increased body weight, the MH allele resulted in decreased body weight at Wt3q3 and Wt10q1, and significant dominance effects were identified for Wt3q3, Wt6q1, Wt10q2 and for the suggestive QTL influencing 3-wk weight on chromosome 7.

View larger version (19K): In this window In a new window Download PPT slide   Figure 3. Significant QTL influencing body weight measured at 3 (WT3), 6 (WT6), and 10 (WT10) wk identified by interval analysis in the HB population are shown for chromosomes 1, 3, 11, and 17. Significant and suggestive threshold levels are shown as horizontal solid and dotted lines, respectively, for each chromosome. One-LOD confidence intervals are shown as lines below the x-axis for each QTL.

Seven QTL that influence liver and heart weights were found, each accounting for 2.1 to 5.2% of residual variance. Livq1 is located in a similar region of chromosome 7 as Hlq5 (Table 6). Both additive and dominance effects were observed, and the MH allele resulted in both increases and decreases in liver and heart depending on the QTL. Interval analysis for food intake failed to identify QTL surpassing either the significant or suggestive threshold levels on the eight chromosomes evaluated.

LH population:
Locations of the 14 markers genotyped in the LH population are shown in Table 7. The markers genotyped and total length of chromosome 1 for the HB and LH populations were different (135 and 103 cM, respectively), making it difficult to identify the expected location of significant QTL from the HB population. However, the confidence interval for Hlq1 spanned ~20 cM on the distal end of chromosome 1 in the HB population. Therefore, the most distal 20 cM of chromosome 1 in the LH population was considered as the region where Hlq1 would reside. Interval analysis in this region revealed a peak LOD score of 1.45 at 90 cM, which corresponds to P < 0.01 for a single position test. Following the guidelines proposed by LANDER and KRUGLYAK 1995 , this is sufficient evidence to confirm the presence of Hlq1 in the LH population. However, the effect of Hlq1 was much smaller in LH than in HB, explaining only 1.2% of the residual variance with an additive effect of 2.01 kcal/kg^0.75/day. Likewise, the confidence interval for the location of Wt6q1 in the HB population included the proximal end of chromosome 1 from 9 through 40 cM. The peak LOD score at the proximal end of chromosome 1 (36.9 cM) in the LH population was 3.44, corresponding to P < 0.0001. Thus, Wt6q1 was also confirmed in the LH population. Other QTL with significant evidence for linkage on chromosomes 1 (Fatq1, Batq1, Wt3q1, Wt3q2, Wt10q1, and Hrtq1) and 3 (Hlq3, Hlq4, Batq2, and Wt10q2) in HB were not confirmed in the LH population.

Complete interval analyses of chromosome 1 revealed suggestive evidence (LOD > 2.13) for three additional QTL on chromosome 1 (Figure 4). These QTL influencing heat loss, subcutaneous fat, and heart weights accounted for 2.1 to 2.5% of residual variance (Table 8). The QTL influencing subcutaneous fat in the LH population was located in the same region of chromosome 1 where suggestive evidence for a QTL was found in the HB population. No QTL influencing any of the measured traits were found on chromosome 3 in the LH population.

View larger version (13K): In this window In a new window Download PPT slide   Figure 4. Suggestive QTL influencing heat loss (HLOSS), percentage subcutaneous fat (SUB) and heart (HRT), and body weight measured at 6 wk (WT6) identified by interval analysis in the LH population are shown for chromosome 1. The suggestive threshold level is shown as a horizontal dotted line.

	DISCUSSION

TOP ABSTRACT MATERIALS AND METHODS RESULTS DISCUSSION LITERATURE CITED

The development of the HB resource population from a cross between MH and an inbred line enabled fully informative markers to be identified and used for genotyping, which increased the power to detect QTL and simplified the genotyping process. Crossing MH with ML provided a second population in which to confirm QTL and provided a model with which to study the relationship between QTL influencing heat loss and fat deposition. Loci identified in both resource populations may represent QTL where similar alleles producing low heat loss were contributed by the ML and BL lines. Alternatively, a QTL effect observed in both resource populations may result from a high heat loss QTL allele originating from the MH line in both resource populations. Evidence for such unique MH alleles may be provided by the asymmetric response to selection observed in the development of MH and ML (NIELSEN et al. 1997A ). Additional heat loss QTL unique to each resource population were also expected, and it was hypothesized that these QTL may contribute to the similar heat loss but different fatness observed in ML and BL. Loci influencing heat loss in HB but not LH may represent genes that influence heat loss independently of fat deposition, while QTL unique to LH may have pleiotropic effects on heat loss and fatness. A similar model has been proposed by RICE et al. 1996 to describe the shared genetic regulation of resting metabolic rate, fat free mass, and fat mass in humans.

Evidence for QTL influencing heat loss was found on seven of the eight chromosomes evaluated by interval analysis. Of the five QTL exceeding the significance threshold level established by permutation testing, two (Hlq1 and Hlq4) also exceeded the significance threshold (LOD > 4.3) suggested by LANDER and KRUGLYAK 1995 . The presence of Hlq1 but not Hlq4 was confirmed in the LH population. Confirmation of Hlq1 may indicate the presence of similar alleles responsible for low heat loss in both ML and BL, with a larger effect on heat loss resulting from the BL allele. The lack of confirmation of Hlq4 may indicate that the allele responsible for the observed variation in heat loss originated from BL and that alleles at this locus did not differ between ML and MH. Alternatively, different effects observed at Hlq1 and Hlq4 in the two resource populations may result from the interaction of these QTL with different genetic backgrounds.

Suggestive evidence was found for QTL influencing subcutaneous fat and heat loss proximal to Hlq1 in the LH population. Because these regions were linked and the presence of the ML allele resulted in decreased heat loss and increased fatness, these regions may contribute to the negative correlation between heat loss and subcutaneous fat found in the LH population (data not shown). The region of chromosome 1 influencing subcutaneous fat in the LH population corresponds to the region containing Fatq1 in the HB population, which also had a suggestive effect on subcutaneous fat. Thus, even though the effect of Fatq1 was not confirmed in the LH population, suggestive evidence of a QTL influencing subcutaneous fat was found in both resource populations in a similar chromosomal region. In the HB population, the MH allele resulted in increased fatness with greater effects observed for gonadal and combined fat compared to subcutaneous fat. In the LH population, the ML allele did not cause additional increases in gonadal or combined fat, but did have a suggestive effect of increased subcutaneous fat. Thus, this region may represent a locus responsible for differences in fat regulation between BL and the selection lines. The suggestive effect on subcutaneous fat observed in LH may be due to different alleles contributed by MH and ML or it may be caused by pleiotropic effects of heat loss QTL linked to a QTL influencing fatness.

Previous studies have identified QTL influencing adiposity using several different types of resource populations (see POMP 1997 ). One QTL has been reported on chromosome 1 (Obq2; TAYLOR and PHILLIPS 1996 ), but it appears to be distal to Fatq1. Several QTL influencing adiposity have been reported in other regions (see POMP 1997 ). Regions containing Dob1 on chromosome 4 (WEST et al. 1995 ), Dob4 (WEST et al. 1995 ) and Obq1 (TAYLOR and PHILLIPS 1996 ) on chromosome 7, and Mob3 (WARDEN et al. 1995 ) on chromosome 12 correspond to regions with suggestive evidence for QTL influencing fatness in the HB population.

An interesting result of this study was the identification of two significant and two suggestive QTL influencing percentage weight of brown adipose tissue. Each of these regions was closely linked to regions containing QTL for heat loss, suggesting a potential pleiotropic effect of these loci. Although the MH allele was associated with increased brown adipose for all QTL except chromosome 17, the effect of brown adipose differences on heat loss and energy expenditure is unclear. Increased brown adipose weight may indicate a proliferation of tissue to facilitate greater heat loss and energy expenditure. However, increased weight of brown adipose depots due to accumulation of lipid stores resulting from reduced brown adipose tissue activity has also been observed (ENERBACK et al. 1997 ; THOMAS and PALMITER 1997 ). Increases observed in ML relative to MH support the latter hypothesis, but the presence of QTL causing both increased heat loss and increased brown adipose supports the former. Further investigation of brown adipose composition and activity is needed to understand the impact of these QTL on the regulation of energy balance.

The chromosomal regions containing QTL identified in this study are relatively large and contain many known and unknown genes. Although the specific genes responsible for these QTL effects cannot yet be identified, the presence of intriguing candidate genes within these QTL regions deserves mention. Genes encoding the ß-subunit of thyroid stimulating hormone (Tshb) and the neuropeptide Y receptor Y2 (Npy2r) are located in the region of chromosome 3 containing Hlq4 and Batq2 (NAYLOR et al. 1986 ; NAKAMURA et al. 1996 ). Both of these genes are involved in the regulation of several metabolic processes and have been implicated in the specific regulation of brown adipose activity (HIMMS-HAGEN 1989 ; CASSARD-DOULCIER et al. 1994 ; see WOODS et al. 1998 ). Genes encoding uncoupling proteins 2 and 3 are located on chromosome 7 within the region containing Hlq5 (BOSS et al. 1997 ; FLEURY et al. 1997 ; VIDAL-PUIG et al. 1997 ). Uncoupling proteins facilitate the dissipation of energy as heat and have been actively investigated as candidate genes for obesity-related phenotypes (BOUCHARD et al. 1997 ).

Two heat loss QTL regions are homologous to regions in other species that harbor QTL for traits involved in energy balance. NORMAN et al. 1998 reported results of a genome scan for QTL influencing obesity and energy metabolism in Pima Indians. Evidence for a QTL influencing the ratio of carbohydrate oxidation to fat oxidation (24RQ) was found on human chromosome 1p22-p12, which is homologous to the region of mouse chromosome 3 containing Hlq4. Evidence for a QTL influencing percentage body fat was found on human chromosome 11 in the same study, but this region does not appear to be homologous to the mouse region harboring Hlq5. In pigs, ANDERSSON et al. 1994 identified a QTL for fatness on porcine chromosome 4 in a region with homology to Hlq1.

In summary, several QTL influencing heat loss and other components of energy balance were identified. As more is learned about the genetic regulation of specific component characteristics that define energy balance, more factors explaining variation in this complex polygenic trait will be identified. Ultimately, it will be possible to study these individual factors and specific interactions among them to gain a more thorough understanding of the regulation of energy balance. This knowledge will be critical to the continued development of new methods to enable increased efficiency of livestock production and improved diagnosis and treatment of human obesity.

	ACKNOWLEDGMENTS

The authors are very grateful to Tong Zou, Louise Baskin, Lori Messer, Mark Allan, and Edward Cargill for assistance with data collection and to Jeryl Hauptman for expert care of the experimental animals. This manuscript is published as paper number 12336 in the Journal Series of the Agricultural Research Division, University of Nebraska, Lincoln, NE 68583-0908.

Manuscript received September 8, 1998; Accepted for publication February 24, 1999.

	LITERATURE CITED

TOP ABSTRACT MATERIALS AND METHODS RESULTS DISCUSSION LITERATURE CITED

ANDERSSON, L., C. S. HALEY, H. ELLEGREN, S. A. KNOTT, and M. JOHANSSON et al., 1994  Genetic mapping of quantitative trait loci for growth and fatness in pigs. Science 263:1771-1774[Abstract/Free Full Text].

BOLDMAN, K. G., L. A. KRIESE, L. D. VAN VLECK, C. P. VAN TASSELL and S. D. KACHMAN, 1995 A manual for use of MTDFREML. A set of programs to obtain estimates of variances and covariances [DRAFT]. U. S. Department of Agriculture, Agricultural Research Service, Clay Center, NE.

BOSS, O., S. SAMEC, A. PAOLONI-GIACOBINO, C. ROSSIER, and A. DULLOO et al., 1997  Uncoupling protein-3: a new member of the mitochondrial carrier family with tissue-specific expression. FEBS Lett. 408:39-42[Medline].

BOUCHARD, C., L. PERUSSE, Y. C. CHAGNON, C. WARDEN, and D. RICQUIER, 1997  Linkage between markers in the vicinity of the uncoupling protein 2 gene and resting metabolic rate in humans. Hum. Mol. Genet. 6:1887-1889[Abstract/Free Full Text].

CASSARD-DOULCIER, A. M., M. LAROSE, J. C. MATAMALA, O. CHAMPIGNY, and F. BOUILLAUD et al., 1994  In vitro interactions between nuclear proteins and uncoupling protein gene promoter reveal several putative transactivating factors including Ets1, retinoid X receptor, thyroid hormone receptor, and a CACCC box-binding protein. J. Biol. Chem. 269:24335-24342[Abstract/Free Full Text].

CHURCHILL, G. A. and R. W. DOERGE, 1994  Empirical threshold values for quantitative trait mapping. Genetics 138:963-971[Abstract].

ENERBACK, S., A. JACOBSSON, E. M. SIMPSON, C. GUERRA, and H. YAMASHITA et al., 1997  Mice lacking mitochondrial uncoupling protein are cold-sensitive but not obese. Nature 387:90-94[Medline].

FALCONER, D. S., and T. F. C. MACKAY, 1996 Introduction to Quantitative Genetics. Longman Group Limited, Essex, England.

FLEURY, C., M. NEVEROVA, S. COLLINS, S. RAIMBAULT, and O. CHAMPIGNY et al., 1997  Uncoupling protein-2: a novel gene linked to obesity and hyperinsulinemia. Nat. Genet. 15:269-272[Medline].

GREEN, P. K., and S. CROOKS, 1990 Documentation for CRI-MAP, version 2.4. Washington University School of Medicine, St. Louis.

GRIFFITHS, M., P. R. PAYNE, A. J. STUNKARD, J. P. W. RIVERS, and M. COX, 1990  Metabolic rate and physical development in children at risk of obesity. Lancet 336:76-78[Medline].

HALEY, C. S. and S. A. KNOTT, 1992  A simple regression method for interval mapping in line crosses. Heredity 69:315-324[Medline].

HALEY, C. S., S. A. KNOTT, and J. M. ELSON, 1994  Mapping quantitative trait loci in crosses between outbred lines using least squares. Genetics 136:1195-1207[Abstract].

HIMMS-HAGEN, J., 1989  Role of thermogenesis in the regulation of energy balance in relation to obesity. Can. J. Physiol. Pharmacol. 67:394-401[Medline].

JONES, L. D., M. K. NIELSEN, and R. A. BRITTON, 1992  Genetic variation in liver mass, body mass, and liver:body mass in mice. J. Anim. Sci. 70:2999-3006[Abstract].

LANDER, E. S. and D. BOTSTEIN, 1989  Mapping Mendelian factors underlying quantitative traits using RFLP maps. Genetics 121:185-199[Abstract/Free Full Text].

LANDER, E. S. and L. KRUGLYAK, 1995  Genetic dissection of complex traits—guidelines for interpreting and reporting linkage results. Nat. Genet. 11:241-247[Medline].

LINCOLN, S., M. DALY and E. S. LANDER, 1992 Constructing genetic maps with MAPMAKER/EXP 3.0. Technical report. Whitehead Institute Technical Report, Cambridge, MA.

MOODY, D. E., D. POMP, and M. K. NIELSEN, 1997  Variability in metabolic rate, feed intake and fatness among selection and inbred lines of mice. Genet. Res. 70:225-235[Medline].

MOUSEL, M. R., 1998 Daily activity and core body temperature of mice selected for high and low heat loss. M.S. Thesis, University of Nebraska, Lincoln.

NAKAMURA, M., Y. AOKI, and D. HIRANO, 1996  Cloning and functional expression of a cDNA encoding a mouse type 2 neuropeptide Y receptor. Biochim. Biophys. Acta 1284:134-137[Medline].

NAYLOR, S. L., A. Y. SAKAGUCHI, L. MCDONALD, S. TODD, and P. A. LALLEY et al., 1986  Mapping thyrotropin beta subunit gene in mouse and man. Somat. Cell Mol. Genet. 12:307-311[Medline].

NIELSEN, M. K., L. D. JONES, B. A. FREKING, and J. A. DESHAZER, 1997a  Divergent selection for heat loss in mice: I. Selection applied and direct response through fifteen generations. J. Anim. Sci. 75:1461-1468[Abstract/Free Full Text].

NIELSEN, M. K., B. A. FREKING, L. D. JONES, S. M. NELSON, and T. L. VORDERSTRASSE et al., 1997b  Divergent selection for heat loss in mice: II. Correlated responses in feed intake, body mass, body composition, and number born through fifteen generations. J. Anim. Sci. 75:1469-1476[Abstract/Free Full Text].

NORMAN, R. A., P. A. TATARANNI, R. PRATLEY, D. B. THOMPSON, and R. L. HANSON et al., 1998  Autosomal genome scan for loci linked to obesity and energy metabolism in Pima Indians. Am. J. Hum. Genet. 62:659-668[Medline].

POMP, D., 1997  Genetic dissection of obesity in polygenic animal models. Behav. Genet. 27:285-305[Medline].

POMP, D. and M. K. NIELSEN, 1999  Quantitative genetics of energy balance—lessons from animal models. Obesity Res. 7:106-110[Medline].

RAVUSSIN, E., S. LILLIOJA, W. C. KNOWLER, L. CHRISTIN, and D. FREYMOND et al., 1988  Reduced rate of energy expenditure as a risk factor for body-weight gain. N. Engl. J. Med. 318:467-472[Abstract].

RICE, T., A. TREMBLAY, O. DERIAZ, L. PERUSSE, and D. C. RAO et al., 1996  Genetic pleiotropy for resting metabolic rate with fat-free mass and fat mass: the Quebec family study. Obesity Res. 4:125-131[Medline].

ROBERTS, S. B., J. SAVAGE, W. A. COWARD, B. CHEW, and A. LUCAS, 1988  Energy expenditure and intake in infants born to lean and overweight mothers. N. Engl. J. Med. 318:461-466[Abstract].

SALTZMAN, E. and S. B. ROBERTS, 1995  The role of energy expenditure in energy regulation: findings from a decade of research. Nutr. Rev. 53:209-220[Medline].

SAS, 1988 SAS/STAT User's Guide, Release 6.03. SAS Institute, Cary, NC.

TAYLOR, B. A. and S. J. PHILLIPS, 1996  Detection of obesity QTLs on mouse chromosomes 1 and 7 by selective DNA pooling. Genomics 34:389-398[Medline].

THOMAS, S. A. and R. D. PALMITER, 1997  Thermoregulatory and metabolic phenotypes of mice lacking noradrenaline and adrenaline. Nature 387:94-97[Medline].

VIDAL-PUIG, A., G. SOLANES, D. GRUJIC, J. S. FLIER, and B. B. LOWELL, 1997  UCP3: an uncoupling protein homologue expressed preferentially and abundantly in skeletal muscle and brown adipose tissue. Biochem. Biophys. Res. Commun. 235:79-82[Medline].

WARDEN, C. H., J. S. FISLER, S. M. SHOEMAKER, P. Z. WEN, and K. L. SVEBSIB et al., 1995  Identification of four chromosomal loci determining obesity in a multifactorial mouse model. J. Clin. Invest. 95:1545-1552.

WEST, D. B., G. E. TRUETT, J. GOUDEY-LEFEVRE, and K. B. YORK, 1995  Loci on chromosomes 4, 7, 9, 12, & 15 control a significant proportion of the dietary obesity phenotype in the mouse. FASEB J. 9(3):A722.

WOODS, S. C., R. J. SEELEY, D. PORTE, JR., and M. W. SCHWARTZ, 1998  Signals that regulate food intake and energy homeostasis. Science 280:1378-1383[Abstract/Free Full Text].

ZENG, Z-B., 1993  Theoretical basis for separation of multiple linked gene effects in mapping quantitative trait loci. Proc. Natl. Acad. Sci. USA 90:10972-10976[Abstract/Free Full Text].

This article has been cited by other articles:

				Z. Zhang, R. J. Todhunter, E. S. Buckler, and L. D. Van Vleck Technical note: Use of marker-based relationships with multiple-trait derivative-free restricted maximal likelihood J Anim Sci, April 1, 2007; 85(4): 881 - 885. [Abstract] [Full Text] [PDF]

				Z. Su, Y. Li, J. C. James, A. H. Matsumoto, G. A. Helm, A. J. Lusis, and W. Shi Genetic linkage of hyperglycemia, body weight and serum amyloid-P in an intercross between C57BL/6 and C3H apolipoprotein E-deficient mice Hum. Mol. Genet., May 15, 2006; 15(10): 1650 - 1658. [Abstract] [Full Text] [PDF]

				Z. Su, Y. Li, J. C. James, M. McDuffie, A. H. Matsumoto, G. A. Helm, J. L. Weber, A. J. Lusis, and W. Shi Quantitative Trait Locus Analysis of Atherosclerosis in an Intercross Between C57BL/6 and C3H Mice Carrying the Mutant Apolipoprotein E Gene Genetics, March 1, 2006; 172(3): 1799 - 1807. [Abstract] [Full Text] [PDF]

				M. F. Allan, E. J. Eisen, and D. Pomp Genomic Mapping of Direct and Correlated Responses to Long-Term Selection for Rapid Growth Rate in Mice Genetics, August 1, 2005; 170(4): 1863 - 1877. [Abstract] [Full Text] [PDF]

				N. C. Jerez-Timaure, E. J. Eisen, and D. Pomp Fine mapping of a QTL region with large effects on growth and fatness on mouse chromosome 2 Physiol Genomics, May 11, 2005; 21(3): 411 - 422. [Abstract] [Full Text] [PDF]

				C. G. Tankersley and K. W. Broman Interactions in hypoxic and hypercapnic breathing are genetically linked to mouse chromosomes 1 and 5 J Appl Physiol, July 1, 2004; 97(1): 77 - 84. [Abstract] [Full Text] [PDF]

				W.-D. Li, C. Dong, D. Li, H. Zhao, and R. A. Price An Obesity-Related Locus in Chromosome Region 12q23-24 Diabetes, March 1, 2004; 53(3): 812 - 820. [Abstract] [Full Text] [PDF]

				J. P. Stoehr, J. E. Byers, S. M. Clee, H. Lan, I. V. Boronenkov, K. L. Schueler, B. S. Yandell, and A. D. Attie Identification of Major Quantitative Trait Loci Controlling Body Weight Variation in ob/ob Mice Diabetes, January 1, 2004; 53(1): 245 - 249. [Abstract] [Full Text] [PDF]

				S. R. Wesolowski, M. F. Allan, M. K. Nielsen, and D. Pomp Evaluation of hypothalamic gene expression in mice divergently selected for heat loss Physiol Genomics, April 16, 2003; 13(2): 129 - 137. [Abstract] [Full Text] [PDF]

				B. K. Smith Richards, B. N. Belton, A. C. Poole, J. J. Mancuso, G. A. Churchill, R. Li, J. Volaufova, A. Zuberi, and B. York QTL analysis of self-selected macronutrient diet intake: fat, carbohydrate, and total kilocalories Physiol Genomics, December 3, 2002; 11(3): 205 - 217. [Abstract] [Full Text] [PDF]