1 Department of Anatomy and Neurobiology
2 Departments of Pediatrics, Endocrinology, and Metabolism, Washington University School of Medicine, St. Louis, Missouri 63110
![]() |
ABSTRACT |
---|
![]() ![]() ![]() ![]() ![]() ![]() ![]() |
---|
genetics; dietary obesity; mice; fat pads; glucose levels
![]() |
INTRODUCTION |
---|
![]() ![]() ![]() ![]() ![]() ![]() ![]() |
---|
Much progress has been made in characterizing single-gene mutants leading to obesity in inbred mouse populations (24, 25). Recently, the molecular basis of the obese (ob; 57), diabetes (db; 7), fat (fat; 33), agouti yellow (Ay; 32), little (lit; 18) and tubby (tub; 25) mouse mutant phenotypes have been identified. In some cases, there has been a rapid application of these findings to understanding and treating human growth abnormalities (31, 47). The identification and mapping of these mutants are an important beginning in our understanding of obesity and provide a window into the control of weight and fat deposition in mammals. However, it is likely that there are many more genes involved in obesity control that have not produced gross mutations, either by chance or because of their positions in the network of control (29), and therefore remain undetected.
The human genetics of dietary obesity has been difficult to address through epidemiologic means because of the complexity of the trait (41). Body weight and body mass index are typically moderately heritable in humans (1), indicating some genetic basis for normal variability. Although it seems generally clear that individuals on a high-fat diet gain more weight than those on a low-fat diet, the results can be highly variable both within and between studies (41). Some individuals add fat in response to fat in the diet, whereas others do not. Also, it is difficult to obtain controlled weight and diet data in humans because of reporting errors and biases. For these reasons, inconsistent results have been produced from epidemiologic studies (41). It has been especially difficult to deal with the effects of family, diet, and their interaction. Even so, Heitmann and colleagues (22) found a positive association between dietary fat intake and weight gain only in women who were from families genetically predisposed to obesity. This finding indicates the possibility of family (genotype) x dietary fat intake interaction as a cause of obesity, an observation worthy of further research (41).
Dietary obesity is the difference in obesity observed when individuals are placed on diets with different levels of fat. Many different mammalian study systems exhibit weight gain and fat increase on higher fat diets, as recently reviewed by West and York (54). In contrast to human epidemiologic studies, other mammalian species allow carefully controlled experiments that can help in the generation and testing of hypotheses relevant to human populations. The most extensive of these studies have been performed on inbred lines in rodents (16, 39, 54). These studies clearly demonstrate a genetic basis for variation in relative weight increase on a high-fat diet.
In a series of studies on genetic variation in dietary obesity, West and colleagues (50) showed that mouse strains varied substantially in their response to a high-fat diet. Some strains, such as AKR/J, C57L/J, A/J, C3H/HeJ, DBA/2J, and C57BL/6J, responded relatively strongly to the high-fat diet by adding significant amounts of fat. Other strains failed to respond at all, gaining little or no additional weight or fat on a high-fat diet, including the SJL/J, I/STN, and SWR/J strains. West and colleagues (15, 42, 53) proceeded to investigate the diet/genes/obesity relationship using two strains with very different responses to a high-fat diet, AKR/J and SWR/J. They found that adipocytes from AKR/J mice had greater insulin-stimulated glucose transport than SWR/J mice (15) and confirmed that the AKR/J strain has a stronger response to dietary fat than SWR/J and that this difference increases with the level of fat in the diet (53). The difference in response between the strains was on the order of 1.0 within-strain standard deviations (SD) on a raw scale (grams). They also found that the two strains differ in macronutrient selection and feeding rate, with the AKR/J strain consuming 30% more calories and a much higher percentage of calories from fat (69% vs. 28%) than the SWR/J strain (42). These findings suggest that obesity level of the AKR/J strain in response to a high-fat diet is based, in part, on hyperphagy and a preference for calories from fat over calories from carbohydrates. Although West and colleagues (51, 52, 56) have mapped several genes affecting body weight and fat pad weight on a high-fat diet in the intercross of AKR/J and SWR/J, they did not specifically test for genes differentially responsive to dietary fat intake.
The strains used in the present study are the large (LG/J) and small (SM/J) inbred mouse lines. The extreme genetic variability between these strains for body size and fatness (26) make them and the crosses between them excellent rodent models for multigenic obesity. The LG/J strain originated from a strain selected for large body size at 60 days (19, 20). The SM/J strain was derived from a separate experiment in which selection was for small body size at 60 days (30). In each case, the original selected lines were systematically inbred and maintained by brother-sister mating to the present (26). The studies of Chai (2, 3) showed extreme body weight differences between the strains (24 g difference at 60 days).
In a rodent model for multigenic obesity, body weight and fatness levels should be affected by many genes of nearly equal, relatively small effect, because this is considered to be the likely source of most genetic variation in human obesity (1), although some instances of obesity in humans may involve variation at only one or a few genes (34). A multigenetic obesity model should have genetically well-characterized inbred parental strains and/or inbred derivatives from these strains to allow the use of powerful experimental breeding designs for gene mapping. The use of inbred strains or their crosses also allows replication of studies through recreation of the experimental mapping populations. The LG/J x SM/J intercross populations meet these criteria fully. Although other model systems have been successfully used for studying the genetics of obesity and dietary response, such as the AKR/J x SWR/J intercross (51, 52, 56), the BSB mice (48), and C57BL/6J x A/J crosses (44, 45, 49), it is likely that much new information can be gained from a new model.
The LG/J and SM/J inbred strains have been maintained by brother-sister mating for nearly 50 years, making them fully homozygous except for new mutations. In our own recent experiment, after correcting for sex differences, we found an 20-g difference between strain body weights at 10 wk (26). We found that this difference is due to many loci of small but varying effect and their interactions (8, 10, 38). Eighteen potential quantitative trait loci (QTLs) for week 10 weight were identified in an earlier intercross experiment (10). Variation in this cross is due to more QTLs than any other mouse multigenic obesity model to date. The allele resulting in larger size tends to be dominant in this cross, although standardized measures of dominance for age-specific weight decline with age. We found considerable epistasis among genes affecting adult body weight (8, 38).
From previous studies (10, 26, 38; and J. M. Cheverud, T. T. Vaughn, L. S. Pletscher, A. Peripato, K. King-Ellison, E. Adams, and C. Erikson, unpublished observations), we know that the LG/J strain grows faster than the SM/J strain prior to 10 wk of age. At 10 wk, the LG/J strain has a longer tail, larger reproductive fat pad (J. M. Cheverud et al., unpublished observations), longer long bone lengths, and larger organ sizes (46) than the SM/J strain. These genetic differences between strains are substantial and have led to successful gene mapping experiments for the phenotypes in question (10, 46; and J. M. Cheverud et al., unpublished observations). Here we report the results of an experiment investigating the differential response of these strains to dietary fat. If increased dietary fat affects body size and composition differently in the two strains, then these differences have a genetic basis. The location and effects of such genes can then be mapped in further experiments.
![]() |
MATERIALS AND METHODS |
---|
![]() ![]() ![]() ![]() ![]() ![]() ![]() |
---|
Animals were obtained by in-crossing LG/J females with LG/J males and SM/J females with SM/J males. Source animals were obtained from the Jackson Laboratories. All pups were weaned at 3 wk of age, at which time they were randomly placed on either the low- or high-fat dietary regime. Animals were weighed weekly to 10 wk, and periodically thereafter. Animals were killed and necropsied at 27 to 45 wk. Within the month prior to necropsy, all animals were tested for fasting glucose levels. We used Hemocue glucose analyzers to measure glucose levels in blood drawn from the SM/J and LG/J mice after fasting for 4 h. Animals ranged from 23 to 41 wk in age at the time of testing.
We measured growth in two periods, postweaning growth from 3 to 10 wk and later, adult growth from 10 wk to necropsy. To standardize for differences in age at necropsy, the body weight difference between necropsy and 10-wk weight was divided by the number of days between 10 wk and necropsy. Qualitatively similar results were obtained using adult growth without correction for age at necropsy. Ideally, all animals should be necropsied at the same age. By 10 wk, skeletal growth in mice is complete. Further growth consists of adding soft tissue mass. A series of measurements were collected at necropsy, including total body weight, tail length, heart, kidney, spleen, and liver weights and the weight of the reproductive (parametrial in females and epididymal in males), kidney, mesenteric, and inguinal fat pads. Scores for all bilateral structures represent the sum of the two sides. The reproductive fat pad is well delineated in its own mesentery. The kidney fat pad consisted of the perirenal fat. This fat pad was located on the peritoneal faces of the kidney. The mesenteric fat pad was obtained by removing the gut between the stomach and rectum and separating the dorsal mesentery from the posterior body wall. The mesentery and its contents were then stripped from the gut and weighed. The inguinal fat pad was defined as the subcutaneous fat in the perineal and inguinal regions. Fat extending around the proximal thigh was also included. All data except glucose level were transformed to the natural logarithmic scale to prevent spurious results due to scale. Glucose level was not transformed because its within-group distribution is normally distributed on a raw scale.
The data were analyzed using ANOVA and multivariate ANOVA (MANOVA) methods where appropriate (43). ANOVA was used for analyses of individual phenotypes, and MANOVA was used for groups of phenotypes, such as the various fat pad weights. These analyses include the factors of sex (male vs. female), diet (high fat vs. low fat), and strain (LG/J vs. SM/J) and their two- and three-way interactions. The major effects of interest here are the strain differences, which are indicative of genetic effects on the phenotypes themselves, and the strain x diet interactions, which indicate differential response of the strains to added dietary fat.
The results are reported in terms of the number of within-group standard deviations represented by the contrast. This value is obtained by dividing the difference in means between factor categories by the square root of the residual variance. Reporting results on a standardized scale allows us to compare results for different phenotypes and for the same phenotypes in different studies.
![]() |
RESULTS |
---|
![]() ![]() ![]() ![]() ![]() ![]() ![]() |
---|
|
|
Daily weight gain from 10 wk to necropsy shows a very different pattern of results. Again, the three main effects are all significant, but the relative rankings of the sexes and strains are reversed. Females grow 1.15 SD faster than males (P = 8.0 x 10-6), the high-fat diet leads to 2.25 SD faster growth than the low-fat diet (P = 2.2 x 10-11), and the SM/J animals grow faster than the LG/J animals (0.59 SD; P = 0.017). The sex x diet (1.49 SD, P = 0.0028) and strain x diet (1.29 SD, P = 0.0096) interactions are both significant. Diet had a greater effect in females than in males and a greater effect in the SM/J strain than in the LG/J strain.
Necropsy weight showed significant differences for diet (high-fat diet animals were 1.15 SD larger than low-fat animals; P = 5.0 x 10-6) and strain (LG/J were 2.70 SD larger than SM/J; P = 2.1 x 10-11), but not for sex. Males and females were the same weight at necropsy, in strong contrast to results for 10-wk weight for which males are over 2.0 SD units larger than females (10, 26). The relative growth deficit of females prior to 10 wk is compensated for by growth enhancement after 10 wk. The only significant interaction term is the strain x diet interaction (high-fat diet had a 1.27 SD greater effect on SM/J than on LG/J; P = 0.009). The sex x diet interaction was not statistically significant but followed the same pattern as found for adult growth.
Most of the internal organs (heart, kidney, spleen) exhibited sex and diet differences but no strain difference or strain x diet interaction. The liver displayed significant strain effects (LG/J was 1.0 SD larger than SM/J) and a significant strain x diet interaction (high-fat diet had a 0.99 SD larger effect in SM/J than in LG/J; P = 0.04). This interaction is much more prominent in females than in males. The main effects of diet and sex were not statistically significant for liver weight. Livers of SM/J animals fed a high-fat diet were mottled with fat in contrast to LG/J animals fed either diet and SM/J animals fed the low-fat diet (unpublished data).
The fat pads were analyzed with MANOVA because of their relatively high intercorrelations (within-group Pearson's r 0.7 to 0.8). Significant effects were found for all three main effects and for the strain x sex and strain x diet interactions (see Table 3). Females were heavier than males for all four fat pads, and those fed a high-fat diet had more fat than those fed a low-fat diet. The LG/J strain had more fat than the SM/J strain for the reproductive, kidney, and mesenteric fat pads, but not for the inguinal fat pad. The significant sex x strain interaction indicates that strain differences are much more pronounced in females than in males but this effect is restricted to the reproductive fat pad. Significant strain x diet interactions indicate that the effects of additional dietary fat have a much larger effect on the SM/J strain than on the LG/J strain. The reproductive, kidney, and mesenteric fat pads all show this significant pattern, whereas there is no significant effect on the inguinal fat pad.
|
![]() |
DISCUSSION |
---|
![]() ![]() ![]() ![]() ![]() ![]() ![]() |
---|
Strains also differ dramatically in tail length (4.96 SD), necropsy weight (2.70 SD), liver weight (1.00 SD), and fat pad weights (1.25 SD). In each case, the LG/J strain weighs more than the SM/J strain. However, strain differences in necropsy weight are small relative to differences in 10-wk weight. The LG/J strain is 6.45 SD larger than the SM/J strain at 10 wk and only 2.70 SD heavier than SM/J at necropsy. This reduction in difference is due to the faster growth of the SM/J strain in the adult period. Fasting glucose levels also differ between strains in a characteristic fashion, with the SM/J strain having 1.76 SD higher levels than the LG/J strain. Interstrain differences in all these phenotypes, postweaning and adult growth, organ weights, fat pad weights, and glucose levels make it possible to map genes for these traits in various experimental crosses of the LG/J and SM/J strains. Gene mapping studies of some of these phenotypes, such as postweaning growth, tail length, and reproductive fat pad weight, have already been completed or are presently ongoing (811, 27, 38; and J. M. Cheverud et al., unpublished observations). The present study indicates that the genetic basis of adult weight gain and glucose levels could also be addressed using crosses between these two strains. In future experiments, we should be able to map genes with an approximate effect size of
0.25 SD, which would be the average effect of four to six genes.
Strain x diet interactions measure genetic variation between strains in their response to varying levels of dietary fat. These interactions were ubiquitous for body composition phenotypes and glucose levels. The LG/J strain responded much more dramatically to higher levels of dietary fat during the postweaning growth period, adding 2.05 SD more in weight than added by the SM/J strain. Perhaps the higher energy intake can be utilized for increased growth in the LG/J strain but not in the SM/J strain during this period. The effects of dietary fat differences on adult growth are reversed relative to those observed for the earlier period. Dietary fat has a much larger effect on adult growth in the SM/J strain than in the LG/J strain. This reversal of dietary effects over the life cycle indicates that genes promoting growth on a high-fat diet from 3 to 10 wk also reduce growth after 10 wk or that different sets of genes affect pre- and post-10-wk growth on a high-fat diet (10).
Neither diet nor strain x diet interaction was significant for tail length, indicating that levels of dietary fat do not influence skeletal growth. However, necropsy, liver, and fat pad weights show significant strain x diet interactions, indicating that increased dietary fat has a much larger effect in the SM/J strain than in the LG/J strain for body size and obesity. Furthermore, glucose level showed a significant three-way interaction between sex, diet, and strain. These results indicate that a high-fat diet has a larger effect on glucose levels in the SM/J strain than in the LG/J strain but that this difference is restricted to the males. In each of these traits, SM/J responds much more strongly than LG/J to increased levels of dietary fat. Genes responsible for this difference could be mapped in the LG/J x SM/J intercross.
The results obtained from the LG/J and SM/J strains are reminiscent of the pattern predicted by the "thrifty" phenotype hypothesis (21, 28, 36, 55). According to this hypothesis, disease states in adults depend, in part, on prenatal and early postnatal environmental conditions. The fetus adapts to early environmental deficits by efficient use of nutrients causing a life-long programming of the physiological system. If later in life the individual is presented with an enriched environment (high-fat diet), then their preset physiological system overreacts, potentially resulting in a series of chronic diseases, including diabetes, heart disease, and hypertension (28). Epidemiologic data in humans support this hypothesis, but the confounding effects of uncontrolled factors make interpretation difficult (37). Because of the uncontrolled covariates in human epidemiologic studies, tests of this hypothesis in model systems are particularly important (21). As in the "thrifty" phenotype hypothesis, SM/J mothers produce relatively small offspring due to an effect of maternal environment (2, 3) and are smaller than LG/J animals at weaning but are more responsive to high-fat diets as adults. If this pattern is confirmed, then it will be possible to map genes affecting both maternal environment and later adult phenotypes to test the "thrifty" phenotype hypothesis with this pair of strains.
The magnitudes of the strain x diet interaction for postweaning and adult growth, necropsy weight, liver weight, and fat pad weight are typically slightly larger than 1.0 SD units. This is the same magnitude of effect as observed by West and colleagues (42, 5054, 56) in their studies on genetic differences in response to dietary fat in the AKR/J and SWR/J strains of mice. They chose these two strains for their work because they showed a strong response to dietary fat (AKR/J) and a lack of such response (SWR/J), respectively. The LG/J and SM/J strains are as different in response to dietary fat as the AKR/J and SWR/J strains. This difference was also obtained under a smaller difference in dietary fat levels in the present study. As West and colleagues (53) showed, larger proportions of dietary fat lead to larger differential strain effects. We expect that the differences in response between the LG/J and SM/J strains would be further magnified if the diets used were more distinct in their levels of dietary fat.
Other measured effects are of less direct importance for mapping genes in populations derived from the LG/J and SM/J strains. Males grew faster than females during the postweaning period but grew slower than females during the adult period. Female fat pads were larger than male fat pads and this was especially true in the LG/J strain. Males also have higher glucose levels than females.
Dietary effects are also as expected in that animals fed a relatively high-fat diet grow faster to a larger body weight with heavier fat pads and livers than animals fed a low-fat diet. Neither tail length nor glucose level exhibited a direct effect of diet. However, glucose level is different for SM/J males on low- and high-fat diets.
The contrast between LG/J and SM/J strains provides an excellent polygenic model for the genetics of body size, obesity, glucose level, and differential genetic response to dietary fat. As with complex diseases in human populations, differences between these strains are due to many genes of relatively small effect (10) and their epistatic interactions (8, 38). Furthermore, these strains are genetically variable in their response to dietary fat. The SM/J strain responds more strongly to increased levels of dietary fat than the LG/J strain for obesity-related characteristics. This makes populations formed from the cross of these strains a potent resource for mapping genes affecting the variable response to dietary environment.
We are presently developing a large series of recombinant inbred strains from the LG/J x SM/J F2 intercross. These strains will provide an efficient means for mapping quantitative trait loci affecting obesity-related phenotypes and responses to dietary fat (13). An advantage in using recombinant inbred strains for initial gene mapping purposes is that phenotypes measured at different times from different animals can be analyzed together; also, there is no need to genotype every individual, because animals within a strain are genotypically identical. Once a general map position is obtained from analysis of recombinant inbred strains, further fine-scale mapping can be pursued in an advanced intercross line (13, 14). We have maintained an advanced intercross line of more than 120 individuals each generation by random mating from the F2 generation. If necessary, replicate F2 populations can be generated again from the parental strains. The combination of these genetic mapping resources with strains showing variability in growth, obesity, and glucose levels, in addition to their differential response to a high-fat diet, makes the LG/J x SM/J cross an excellent model system for the polygenic basis of complex traits.
![]() |
ACKNOWLEDGMENTS |
---|
We gratefully acknowledge the National Institute of Diabetes and Digestive and Kidney Diseases (Grant DK-52514) and the Washington University Diabetes Research and Training Center for support to J. M. Cheverud during this research. B. Marshall was supported by National Institute of Diabetes and Digestive and Kidney Diseases Grant DK-02339 and by the Hardison Family Foundation and is a scholar of the Child Health Research Center of Excellence in Developmental Biology at Washington University School of Medicine (National Institutes of Health Grant HD-33688).
Address for reprint requests and other correspondence: J. M. Cheverud, Dept. of Anatomy and Neurobiology, Washington Univ. School of Medicine, 660 S. Euclid Ave., St. Louis, MO 63110 (E-mail: cheverud{at}thalamus.wustl.edu).
![]() |
FOOTNOTES |
---|
![]() |
REFERENCES |
---|
![]() ![]() ![]() ![]() ![]() ![]() ![]() |
---|
HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
Visit Other APS Journals Online |