Department of Anatomy and Neurobiology, Washington University School of Medicine, 660 S. Euclid Ave., St. Louis, MO 63110. E-mail: email@example.com
Objective: To examine the differential response of obesity- and diabetes-related traits to a high- or low-fat diet in LG/J and SM/J mice. We also examined food consumption in these strains.
Research Methods and Procedures: Mice were placed on a high- or low-fat diet after weaning. Animals were weighed once per week and subjected to glucose tolerance tests at 20 weeks. At sacrifice, fat pads and internal organs were removed along with serum samples. For food consumption, LG/J and SM/J mice of each sex were assigned to a high-fat or low-fat diet after reaching maturity. Mice were weighed three times per week, and food consumed was determined by subtraction.
Results: LG/J animals consume more total food, but SM/J animals consume more food per gram of body weight. LG/J mice grow faster to 10 weeks but slower from 10 to 20 weeks, have higher cholesterol and free fatty acid levels, and have lower basal glucose levels and better response to a glucose challenge than SM/J mice. For most traits, SM/J mice respond more strongly to a high-fat diet than LG/J mice, including body weight and growth, basal glucose levels, organ weights, fat distribution, and circulating triglycerides and cholesterol levels.
Discussion: Obesity-related phenotypes, as well as response to increased dietary fat, differ genetically between LG/J and SM/J and can, therefore, be mapped. This study indicates that the cross of SM/J and LG/J mice would be an excellent model system for the study of gene-by-diet interaction in obesity.
Considerable progress has been made in identifying single genes in humans and in mice whose disruption leads to obesity or obesity-related phenotypes (1). However, it is generally accepted that, although single gene mutations play a crucial role in elucidating the biochemical pathways leading to obesity, in most humans, obesity is caused by the interactions of many genes of individually small effect with one another (2), as well as with the environment (3). Although few novel single gene mutations affecting obesity were reported in rodent models during the past year (1, 4, 5, 6), this does not imply that all of the important genetic factors in obesity determination have already been characterized. Because of the nature of mutation mapping, it is likely that many of the genes involved in obesity remain undetected (7). Population mapping represents a way of identifying these heretofore hidden genes whose effects are generated through interactions with other genes as well as with the environment. Unfortunately, the statistical genetic study of obesity is very difficult in human subjects. This is due both to complications inherent in studying a multigenic phenotype in an outbred population (8) and to reporting errors and biases (9). Therefore, one potential strategy for discovering new obesity-related genes is to map the genes in model systems more amenable to powerful genetic analysis and follow-up with an examination of syntenic regions of the human genome (10).
With carefully controlled genetic lineages and environments, rodent model systems are ideal for elucidating the networks of genetic and environmental interactions contributing to the obese phenotype (10). Mouse models have been employed in the study of obesity for nearly a century (11) and have proven invaluable in elucidating the effects and causes of obesity and related phenotypes (12). In 1992, West et al. demonstrated that different inbred strains of mice responded differently to a high-fat (32.6% energy from fat) diet (13). Of the strains tested, AKR/J demonstrated the strongest response to a high level of dietary fat, whereas SWR/J showed the lowest response (13). West et al. followed up on this study by mapping quantitative trait loci (QTL)1 for adiposity, weight gain, and fat distribution in a cross of AKR/J and SWR/J (14). However, all of the animals in the later study were placed on a high-fat diet, so West et al. were not specifically mapping genes that differentially respond to levels of dietary fat (14).
This study uses large (LG/J) and small (SM/J) mouse strains. LG/J (15) and SM/J (16) were derived from separate experiments selecting for large and small body size at 60 days, respectively. There is about a 20-g difference in body weight between these strains at this age (17, 18). Both strains have been systematically inbred by brother-sister mating for over 100 generations (19), making them genetically homozygous aside from spontaneous mutations. This difference in size is caused by at least 18 different segregating genes of small to moderate effect (20). In 2001, Cheverud et al. used an F2 intercross of SM/J and LG/J to map regions affecting adiposity (21). To date, this cross has yielded more QTL affecting obesity and body size than any other (1).
The SM/J strain has been previously characterized as having an unusually high lipid level (22). They have also been characterized as being susceptible to diet-induced atherosclerosis (23), although a more recent study has shown that SM/J-strain animals have relatively low cholesterol and plasma phospholipid transferase protein activity relative to their small body size (24). In addition, whereas LG/J animals tend to grow faster during the first 10 weeks of skeletal growth, SM/J animals grow faster during the next 10 weeks (19), when most body weight is added in the form of adipose tissue (25).
In 1999, Cheverud et al. (25) found that LG/J and SM/J mice differ at least as much in response to dietary fat as the AKR/J and SWR/J strains used by West et al. Cheverud et al. compared animals on a moderately high-fat diet (21% energy from fat) with animals on a low-fat diet (12% energy from fat) (25). The present study confirms and extends earlier work with LG/J and SM/J by comparing animals on a more extreme high-fat (43% energy from fat) diet with animals on a low-fat (15% energy from fat) diet and by incorporating more obesity-related phenotypic traits into the analysis.
It has previously been shown that some rodent model systems will display hyperphagy when fed a moderately high-fat diet ad libitum, and this overfeeding may have a genetic basis (26). In response to the possibility that differential response to a high-fat diet may be caused primarily by feeding differences, we also examine the differential feeding response to high- and low-fat food by SM/J and LG/J mice.
Research Methods and Procedures
SM/J and LG/J source animals were obtained from Jackson Laboratories (Bar Harbor, ME) and in-crossed to obtain LG/J and SM/J pups. For the animals used in this study, the average litter size of the SM/J strain was 5, and the average litter size of the LG/J strain was 6.1. Pups were housed with their mothers until 3 weeks of age, at which time they were placed in single-sex cages and segregated by diet. The animal facility operates on a 12-hour light/dark cycle and is maintained at a constant temperature of 21 °C.
For the obesity traits study, there were 8 animals in each category of strain, sex, and diet, for a total of 64 animals. All animals were fed ad libitum. Maternal diet was PicoLab Rodent Chow 20 (5053; Ralston Purina Corporation, St. Louis, MO), which derives 12% of its energy from fat. The high-fat diet (catalog TD88137, Harlan Teklad, Madison, WI) and low-fat diet (catalog D12284; Research Diets, New Brunswick, NJ) compositions are shown in Table 1.
Table 1. Composition of diets
Energy from fat
High fat (42%)
Low fat (15%)
The diets were chosen to be as similar as possible with the exception of fat content. The high-fat diet derives its fat content from anhydrous milkfat, whereas the low-fat diet derives its fat content from corn and coconut oil.
Hydrogenated coconut oil
Kilojoules per gram
All animals were weighed once per week on the day of the week of their birth from 1 to 20 weeks of age. This period was divided into three growth periods, early growth from 1 to 3 weeks, middle growth from 3 to 10 weeks, and late growth from 10 to 20 weeks. Growth was analyzed relative to the starting value by the log10 of the ratio of final to the starting weight. The early growth period of first 3 weeks is strongly influenced by the maternal environment and is completed before dietary treatment starts. The middle growth period consists of the period of skeletal growth. The late growth period represents a period in which soft tissue growth and fat deposition continue.
For the feeding study, adult mice were sorted into treatment groups of five mice each and caged singly. At this time, the diets of one-half of the mice were switched, resulting in these four categories: early high-fat diet/treatment high-fat diet, early high-fat diet/treatment low-fat diet, early low-fat diet/treatment low-fat diet, and early low-fat diet/treatment high-fat diet. The SM/J early low-fat diet/treatment low-fat diet category had four mice rather than five. The average age of the adult mice was 13 weeks. For a period of 3 weeks, the food consumption and weight gain of each mouse were recorded every other day. Virtually all mice lost weight after being housed singly at the start of the feeding study, which is consistent with previous studies (27). Therefore, the analysis excludes the first week of the study.
At 20 weeks of age, the animals received a glucose tolerance test. Animals were fasted for 4 hours, at which time a basal glucose reading was obtained using a Glucometer Dex blood glucose meter (Bayer Corp., Leverkusen, Germany). The animals were intraperitoneally injected with 0.01 mL of a 10% glucose solution for every gram of body weight. Further glucose readings were obtained 15, 30, 60, and 120 minutes after the injection. Time-specific glucose readings were used to calculate the area under the glucose response curve (AUC) as a measure of response to glucose challenge. Lower AUC values reflect more efficient glucose clearance.
Within a week of the glucose test, animals were again fasted for 4 hours. Animals were anesthetized with sodium pentobarbital, and a blood sample was obtained through cardiac puncture as a terminal procedure. Blood plasma was separated and analyzed for free fatty acids, triglycerides, and cholesterol. During necropsy, internal organs including heart, spleen, kidneys, and liver were removed and weighed, as well as the reproductive (epididymal in males and parametrial in females), renal, mesenteric, and inguinal fat depots. Fat distribution was measured by the log10 of the ratio of specific fat depot weight to the total weight of dissected fat. These data were converted back to percentages for ease of interpretation.
The feeding study and the obesity study were both analyzed using ANOVA models. For the feeding study, neither early diet nor any of its interactions were significant, so they were dropped from further analysis. Therefore, feeding study data collected were analyzed using the following ANOVA model:
where μ is the mean and Yijk is the measurement on the kth individual of strain i with treatment diet j.
Obesity traits were analyzed using the following ANOVA model (28):
where μ is the mean and Yijkl is the measurement on the lth individual of strain i and sex j, raised on diet k. Strain effects reflect genetic differences between the strains, whereas strain × diet and strain × sex × diet interactions measure genetic variation in responses to a high-fat diet. Significant strain interactions indicate that the strains respond differently to high-fat dietary challenge. This is the primary factor of interest in our study and involves a comparison between paired high-fat and low-fat cohorts for the two strains. The presence of significant strain and strain interaction effects indicate that individual genes responsible for these effects and interactions can be mapped in LG/J by SM/J intercross populations.
If a higher-order interaction is not significant, it is dropped from analysis, and a lower-order interaction is considered (29). For example, a strain × diet × sex interaction indicates sexual dimorphism in the differential response of the strains to a high-fat diet. If this higher-order interaction is not significant, there is no evidence for sexual dimorphism in the differential response of the strains to a high-fat diet. In this situation, the strain × diet interaction is examined with the sexes pooled. It is possible that the strain × diet interaction is significant in one sex but not in the other. However, the results in males and females must be pooled unless there is a significant strain × diet × sex interaction (29).
We began our examination of obesity-related traits in LG/J and SM/J mice with a feeding study to determine if one strain consumed more of one type of food than the other. Animals of each strain consumed the same amount of food on the high- and low-fat diets. LG/J animals are considerably heavier and consumed significantly more food than SM/J mice (p = 3.4 × 10−4) but consumed less food per gram of body weight (p = 2.1 × 10−11). The diets differed slightly in energy density (kilojoules per gram of food; see Table 1). When also adjusted for energy content, the same amount of energy was obtained from each diet, and SM/J mice consumed 50% more energy per gram body weight than LG/J mice (p = 2.1 × 10−11; see Figure 1).
Dietary Response Study
Multivariate probabilities for sets of related traits obtained from the ANOVA model are presented in Table 2. These probabilities explicitly account for the number of traits tested in each set. All trait sets show significant strain, sex, diet, and strain × diet interaction effects. The significant strain × diet interactions indicate that the strains respond differently to the high-fat diet.
Table 2. Multivariate probabilities associated with independent factors based on ANOVA of Strain (Str), Sex (Sx), Diet (Dt), and their interactions
Sx × Str
Dt × Str
Dt × Sx
Dt × Sx × Str
1.59 × 10−34
4.82 × 10−12
Basal glucose and AUC
2.54 × 10−05
1.70 × 10−08
2.78 × 10−12
2.92 × 10−26
7.20 × 10−04
4.20 × 10−04
5.00 × 10−04
Fat pad ratio
1.40 × 10−04
3.54 × 10−16
3.10 × 10−08
Body Weight and Growth
Weekly weights for each strain by sex × diet combination are presented in Figure 2. The early growth period is from birth to 3 weeks of age, at which time the mice were weaned. In the early growth period (Figure 2), LG/J animals grew faster than SM/J animals (p = 5.40 × 10−4), and males grew faster than females (p = 0.004). The difference in growth rate between males and females was greater in LG/J than in SM/J mice, leading to a sex × strain interaction (p = 0.022). Because animals were not placed on the specialized diets until the end of this period, there was no effect because of diet.
The middle growth period is from 3 to 10 weeks, by which time skeletal growth is complete. In the middle growth period (Figure 2), LG/J animals put on weight much faster than SM/J animals (p = 1.46 × 10−11), and males put on weight much faster than females (p = 2.19 × 10−11). As with the early growth period, the difference in rate of growth between males and females was more pronounced in LG/J animals than SM/J animals, leading to a sex × strain interaction (p = 0.007). Animals on a high-fat diet grew faster than animals on a low-fat diet (p = 0.021), with SM/J mice showing a greater response to a high-fat diet than LG/J mice (p = 0.038).
The late growth period is from 10 to 20 weeks and mostly reflects added soft tissue. In the late growth period (Figure 2), SM/J animals grew faster than LG/J animals (p = 0.02), reversing earlier patterns in growth. Also, unlike earlier periods, there was no significant difference in the rate of growth between males and females. Earlier trends in diet were amplified, however, with animals on a high-fat diet growing much faster than animals on a low-fat diet (p = 4.39 × 10−4), regardless of strain or sex.
Fat and Fat Distribution
The reproductive, renal, mesenteric, and inguinal fat pads of the mice were removed and weighed at time of necropsy. LG/J animals had more total fat than SM/J animals. However, as a percentage of body weight, total fat removed vs. body mass (adiposity) differed little between the strains examined. The total weight of fat pads removed from LG/J animals on a low-fat diet averaged 15% of total body mass, whereas SM/J animals on a low-fat diet averaged 19% of total body mass as excised fat pads. On a high-fat diet, the total mass of fat removed expressed as a percentage of body mass increased to 21% for LG/J animals and 23% for SM/J animals. Statistically, adiposity differed significantly only by diet, with animals on a high-fat diet having a higher percentage of their body weight comprised of fat (22% vs. 17%; p = 0.016).
As a percentage of total fat removed, the reproductive fat pad demonstrated variation based on strain, sex, and diet (Figure 3A). The reproductive fat pad was comprised of a greater percentage of total fat in females than in males (p = 2.00 × 10−11). The strain × sex × diet interaction (p = 5.42 × 10−4) was primarily caused by the pattern observed in SM/J males. Unlike all other cohorts, SM/J males stored proportionally less fat in the reproductive fat pad on a high-fat diet vs. a low-fat diet. SM/J males seem to store proportionally more fat in the renal fat pad in response to a high-fat diet (Figure 3B), a result that is borderline significant for a strain × sex × diet interaction (p = 0.061) but contributes to significant strain (p = 0.003) and sex × strain (p = 0.007) effects.
The mesenteric fat pad (Figure 3C) was comprised of a greater proportion of total fat in LG/J animals than in SM/J animals (p = 1.44 × 10−5). LG/J animals seemed to store proportionally less fat in the mesenteric fat pad on a high-fat diet, although this result was not statistically significant (p = 0.088). The proportion of fat stored in the inguinal fat pad differs by sex (Figure 3D), with males storing a greater percentage of fat here than females (p = 2.29 × 10−11).
The hearts, livers, kidneys, and spleens of the animals were removed and weighed at time of necropsy. Among cohorts, heart (Figure 4A) and liver (Figure 4B) weights mirrored total body weight, with organ weights higher for LG/J than for SM/J mice (p = 3.84 × 10−4 and 2.41 × 10−5, respectively) and higher for males than for females (p = 0.005 and 1.19 × 10−4, respectively). Kidney weight (Figure 4C) did not differ significantly by strain but did differ greatly by sex, with males having substantially larger kidneys than females (p = 1.61 × 10−11). Interestingly, all organ weights showed a significant strain × diet interaction, with SM/J animals on a high-fat diet having larger hearts, livers, kidneys, and spleens (Figure 4D; p = 0.001, 9.06 × 10−5, 0.003, and 0.014, respectively) than SM/J animals on a low-fat diet compared with no significant organ weight differences because of diet among LG/J animals.
Animals were fasted for 4 hours, after which a basal glucose level was obtained. Animals were injected with glucose solution, and readings were obtained at 15, 30, 60, and 120 minutes after injection. With the exception of LG/J females, animals on a high-fat diet had higher basal glucose (Figure 5A) levels than animals on a low-fat diet (p = 2.15 × 10−6). When considered separately by sex, females demonstrated a significantly different dietary response for the two strains (strain × diet interaction effect; p = 0.003), whereas males did not (p = 0.205). Even though the strain difference in male dietary response was not significant, SM/J males did respond twice as much to the high-fat diet as the LG/J males. Moreover, there was no significant strain × diet × sex interaction (p = 0.546), so there was no positive evidence for sexual dimorphism in dietary response. We pooled the results for the two sexes because of the lack of a significant strain × diet × sex interaction; basal glucose levels demonstrated a significant strain × diet effect (p = 0.007), indicating that the interaction of genes and dietary fat in influencing this trait should be amenable to mapping.
The AUC represents an integration of the animal's response to glucose challenge: a higher value represents a diminished ability to respond to glucose (Figure 5B). Response to glucose challenge was diminished in all cohorts on a high-fat diet (p = 5.00 × 10−9), and this diet-based difference was greater in males than in females (p = 0.034). SM/J animals did not respond as well to glucose challenge as LG/J animals (p = 0.001). Strain differences were much more pronounced in females (p = 1.4 × 10−5) than in males (p = 0.40), but this difference in strain effects between the sexes was not statistically significant (p = 0.06). As with basal glucose, differences in dietary response for the two strains was much more pronounced in females (p = 0.011) than in males (p = 0.94), although again there was no significant strain × sex × diet interaction (p = 0.16).
A serum sample was obtained at time of necropsy and analyzed for free fatty acids, cholesterol, and triglycerides. LG/J animals had higher serum levels of both free fatty acids (Figure 6A) and cholesterol (Figure 6B) than SM/J animals (p = 4.20 × 10−4 and 8.42 × 10−7, respectively). Females had higher serum levels of free fatty acids (p = 0.021), whereas males had higher levels of cholesterol (p = 0.003). Whereas diet had no significant effect on serum free fatty acid levels, it did affect serum cholesterol (p = 0.003). This is to be expected because cholesterol was added to the high-fat diet but not the low-fat diet (Table 1). The serum triglyceride levels of LG/J males rose on a high-fat diet, whereas they tended to fall slightly for all other cohorts (Figure 6C), a distinction that was borderline statistically significant (p = 0.054).
Crosses of LG/J and SM/J inbred strain mice have been previously demonstrated to be an excellent model system for the study of a complex polygenic traits related to body size, skeletal morphology, and obesity (20, 21, 25, 30, 31). This study establishes new baseline phenotypic measurements for these strains for a variety of obesity-related traits on high- and low-fat diets.
Whereas our previous studies have documented strain differences for body size, growth, and body composition, the battery of phenotypes tested here is far more comprehensive with regard to obesity and diabetes related traits. We have documented additional strain differences in basal glucose levels, response to a glucose challenge, organ weights, fat pad distribution, and circulating cholesterol and free fatty acid concentrations. We have also shown that the SM/J strain responds much more strongly to a high-fat diet than does the LG/J strain and consumes considerably more food per gram of body weight. Traits showing an increased response to diet in the SM/J strain include body weight and growth, basal glucose levels, organ weights, fat distribution, and circulating triglycerides and cholesterol levels.
For body size and growth, SM/J mice show an earlier and stronger response in body weight to dietary fat than LG/J mice. The primary difference occurs during the middle growth period, from weeks 3 to 10. SM/J animals on a high-fat diet grow faster during this period than SM/J animals on a low-fat diet. The rate of growth in LG/J mice, in contrast, is unchanged by dietary influences. The rate of growth for both strains is affected by dietary fat during the late growth period, from weeks 10 to 20. However, SM/J animals on a high-fat diet maintain their earlier rate of growth during this period, whereas the rate of growth among LG/J animals slows dramatically. SM/J mice gain proportionally twice as much weight in response to a high-fat diet as LG/J mice. These findings agree with earlier studies demonstrating that LG/J animals grow faster initially, whereas SM/J animals grow faster during the late growth period when primarily adipose tissue is being added (25). This study demonstrates that SM/J mice also show a stronger response to dietary fat than LG/J animals. Weight gain in response to dietary fat may be trait amenable to being mapped by population genetic methods in an SM/J × LG/J cross of mice, possibly leading to the identification of genes with different phenotypic effects in response to differing levels of dietary fat.
Adiposity (total fat depot weight/total body weight) did not differ significantly between the two strains. This finding is in contrast to earlier work that was able to map QTL for adiposity in the F2 intercross of these strains (21), as well as to similar work with other strains (14). The lack of significant strain differences here could be because of several factors. For five of the eight adiposity QTL, the LG/J-derived allele had a positive effect, but it had a negative effect at two loci, and the other locus showed overdominance rather than a strain-specific allele of origin effect. The positive and negative effects of unlinked LG/J alleles at different loci will cancel each other out to some extent in a comparison of the strains. Adiposity QTL were shown to have significant epistatic interactions in the F2 that would not contribute to strain differences but can contribute to the detection of QTL. F2 animals in the earlier study were also bred before measurement of reproductive fat pad weight, and breeding can affect reproductive fat pad weight. None of the animals in this study were used for breeding.
Furthermore, the earlier work with SM/J and LG/J mice defined adiposity solely on the weight of the reproductive fat pads relative to total body weight (21). This is a widely accepted definition for adiposity in mice (14, 21, 32, 33). However, adiposity in this study is represented by the sum of four distinct fat pads (reproductive, renal, mesenteric, and inguinal). Contrasts between LG/J and SM/J mice for adiposity differ for different fat pads. This indicates strain differences in fat distribution (percent contribution of an individual fat pad to total fat removed). SM/J animals concentrate fat in the renal and reproductive fat pads, whereas LG/J animals concentrate proportionally more fat in the mesentery fat pads. LG/J animals also have higher serum levels of cholesterol and free fatty acids than SM/J mice, which may be related to the higher levels of fat in the mesentery fat pads. In humans, higher levels of fat in the mesentery fat pads are correlated with higher levels of serum cholesterol as well as greater risk for cardiovascular disease (34). Mapping genes affecting fat distribution in a cross of LG/J and SM/J mice may lead to the elucidation of new genes or pathways involved in this important phenotype.
Fat distribution differs by strain, sex, and diet. Female animals concentrate proportionally more fat in the reproductive fat pad than male animals, and the level of fat in the reproductive pad proportionally rises somewhat in response to an increased level of dietary fat in LG/J animals and SM/J females. However, the proportion of fat stored in the reproductive pad is reduced dramatically among SM/J males on a high-fat diet. Whereas changes in the distribution of body fat have been reported in response to exercise (35), to the best of our knowledge this represents the first time these changes have been reported in response to levels of dietary fat. Finding the genes responsible for this phenomenon could yield important insights into the mechanisms of fat distribution in response to diet.
Some of the most striking results of these studies are that, whereas LG/J animals have heavier internal organs than SM/J animals, organ weight responds much more dramatically to diet in the SM/J strain. Substantial differences between the strains observed on a low-fat diet are obliterated on a high-fat diet. Visual inspection indicates that much of the increased organ weight in SM/J animals fed a high-fat diet was because of fat stored in the organs.
Other studies have shown that SM/J mice tend to have high levels of blood glucose (22), and this study confirms these findings. Basal glucose levels and AUC differ significantly by strain between SM/J and LG/J mice. Phenotypes related to glucose tolerance are, therefore, candidates for genetic mapping with these strains. Basal glucose in SM/J and LG/J mice is also affected in different ways by dietary fat. Whereas basal glucose increases in both strains in response to dietary fat, it does so to a significantly greater degree in SM/J mice. This indicates that genes regulating basal glucose in SM/J may respond differently to high levels of dietary fat or that different genes may affect basal glucose levels depending on the dietary environment. Mapping of these genes may help to identify which of these possibilities is responsible for poor basal glucose regulation in SM/J and could also lead to the identification of genes relating to glucose tolerance and insulin resistance that are induced by levels of dietary fat.
Our results indicate that we will be able to map genes affecting traits showing significant strain differences, including basal glucose levels, response to a glucose challenge, organ weights, fat distribution, and circulating cholesterol and free fatty acid concentrations. We will also be able to map genes affecting dietary response for those traits having significant strain × diet and/or strain × diet × sex interactions, including body weight and growth, basal glucose levels, organ weights, fat distribution, and circulating triglycerides and cholesterol levels. The ability to successfully map these phenotypes does not depend on the raw or percent difference between strains but rather on how large that difference is relative to residual variation. With pairwise strain differences, this is usually measured by either the broad-sense heritability [genetic variance/(genetic variance + residual variance)] or the number of within-strain SDs separating strain means [(LG/J − SM/J)/SD]. The significant results reported here typically have heritabilities over 20% and show mean differences between strains >1.0 SD units. For example, the heritability of the dietary response of the reproductive fat pad mass as a percentage of total fat pad mass in males is 80% and represents a difference between the strains of over 4.0 SD units. Similar results were obtained for other fat distribution measures, such as the renal and mesenteric fat masses. Our past experience mapping genes in the F2 intercross of these strains indicates that we should be successful in mapping multiple QTL for traits with heritabilities of 20% and strain differences even <1.0 SD (18, 20, 31, 36). Therefore, we should be able to map genes affecting these obesity-related traits in the LG/J × SM/J intercross.
It is generally accepted that, in humans, genetic background plays a large role in variation of obesity among individuals (12). Estimates for the heritability of obesity-related traits are as high as 70% in some twin studies (12), although lower figures calculated from family studies are probably more accurate (37). It has also been noted that the prevalence of obesity in the United States is increasing too rapidly to be explained by changes in gene frequencies, indicating that the environment must play a central role (38). In addition, even in the present “obesigenic” environment (39), not all individuals become obese (9). Therefore, a study of how genes interact with aspects of the environment such as dietary fat will become increasingly important in efforts to understand and combat the spread of obesity. For example, peroxisome proliferator-activated receptor γ has been shown in human studies to interact with the ratio of dietary polyunsaturated fat to saturated fat in regulating insulin resistance and blood pressure (3). By demonstrating strain-by-diet interactions in a number of obesity-related phenotypic traits, this study establishes the SM/J × LG/J intercross as an excellent model system for the study of gene-by-environment interactions in the field of obesity.
NIH grants DK55736, DK52514, and RR15116, Washington University's Clinical Nutrition Research Unit (DK56341), and its Diabetes Research Training Center (2 P60 DK20579) supported this project. We thank Dr. Clay Semenkovich and Trey Coleman for help with the analysis of serum samples and Drs. Semenkovich and Sam Klein for critical comments on this manuscript.
Nonstandard abbreviations: QTL, quantitative trait loci; AUC, area under the curve.