To constitute a valuable resource to identify individual genes involved in the development of obesity, a novel mouse model, the Berlin Fat Mouse Inbred line 860 (BFMI860), was established. In order to characterize energy intake and energy expenditure in obese BFMI860 mice, we performed two independent sets of experiments in male BFMI860 and B6 control mice (10 per line). In experiment 1, we analyzed body fat content noninvasively by dual-energy X-ray absorptiometry and measured resting metabolic rate at thermoneutrality (RMRt) and respiratory quotient (RQ) in week 6, 10, and 18. In a second experiment, energy digested (energy intake minus fecal energy loss) was determined by bomb calorimetry from week 6 through week 12. BFMI860 mice were heavier and had higher fat mass (final body fat content was 24.7% compared with 14.6% in B6). They also showed fatty liver syndrome. High body fat accumulation in BFMI860 mice was restricted to weeks 6–10 and was accompanied by hyperphagia, higher energy digestion, higher RQs, and abnormally high blood triglyceride levels. Lean mass–adjusted RMRt was not altered between lines. These results indicate that in BFMI860 mice, the excessive accumulation of body fat is associated with altered lipid metabolism, high energy intake, and energy digestion. Assuming that BFMI860 mice and their obese phenotypes are of polygenic nature, this line is an excellent model for the study of obesity in humans, especially for juvenile obesity and hyperlipidemia.
There is accumulating evidence that genetically determined obesity is caused by an interaction of multiple gene loci which may predispose an individual to altered energy balance and thereby produce an extreme phenotype (1,2,3,4). Because the genetic analysis of such polygenic traits is difficult due to the effects of the environmental and genetic background, polygenic animal models are excellent for the search for genetic factors that contribute to obesity. To date, several different mouse lines exist which have been derived from heterogeneous base populations and selected either for body mass or fat content (5), for example, the lines LG/J, DU6, M16, NZO (New Zealand Obese), KK, and F (6,7,8,9,10,11). However, high body mass consists not only of fat, but also to a considerable proportion of body lean mass. It is therefore very likely that many quantitative trait loci and their underlying genes merely affecting fat accumulation are still unknown.
The proportion of excess energy stored as fat depends on energy intake in relation to energy expenditure (basal energy expenditure, thermogenesis, behavior). The metabolic component that constitutes the largest fraction of daily energy expenditure is basal metabolic rate (BMR). BMR accounts for 30–50% of daily energy turnover and is thought to reflect basal maintenance costs such as transmembrane ion pumping and protein synthesis (12). In humans, studies have shown that a low resting metabolic rate (RMR) is a risk factor for developing obesity (13,14,15).
Recently, we have established a new polygenic mouse model for obesity, the Berlin Fat Mouse Inbred 860 line (BFMI860) by selective breeding for a high-fat phenotype with low body mass (16). In our previous study, the BFMI860 line (at 20 weeks) disposed to higher body mass (∼1.7 times), body fat mass (6.9 times for males and 11.7 times for females), but also body lean mass (∼1.2 times) in comparison to the often used B6 strain. BFMI860 mice additionally responded to high-fat diet by a diet-induced body mass and fat mass gain. Preliminary results further indicated that 20-week-old BFMI860 mice had a higher feed efficiency, i.e., from an age of 8–20 weeks cumulative food intake was higher and the body mass gain per kJ food consumed during this period was greater. In other nonobese mouse strains, a link between food intake and RMR could be shown (17,18).
In this study, we have conducted a metabolic characterization of BFMI860 mice from an early age in order to investigate whether metabolic malfunctions lead to obesity in this polygenic mouse model. We assessed RMR at thermoneutrality (RMRt; called BMR when measured in postprandial, nongrowing animals) in relation to body composition, and monitored respiratory quotient (RQ) as an indicator of substrate fueling in BFMI860 and B6 control mice. In addition, we analyzed energy intake and energy digested.
Methods and Procedures
Animals and diets
We used male mice of the BFMI860 line in the 15th generation of inbreeding (inbreeding coefficient 0.98; A.W. and G.A.B., unpublished data). A detailed description of the BFMI860 line breeding history is outlined in ref. 16. In brief, the BFMI860 line was generated from an outbred population. Founders were originally purchased from pet shops, and subsequently selected firstly for low protein content secondly for low body mass and high-fat content and then for high fatness for 58 generations before inbreeding. The estimated inbreeding coefficient of the outbred population was 0.45. As control, C57BL/6NCrl mice (Charles River Laboratories, Sulzfeld, Germany), hereafter termed as B6, were used which were reproduced in our mouse facility at Berlin Humboldt University. Mice were kept at room temperature (22–24 °C) with a light–dark cycle of 12 h (lights on: 6 am Central European Time). After weaning at the age of 3 weeks, mice were maintained on standard maintenance diet which contained 12.8 kJ/g metabolizable energy with 9% of its energy from fat, 33% from protein content and 58% from carbohydrates (V1534-000 ssniff R/M-H; ssniff Spezialdiäten, Soest, Germany). Access to diet and water were ad libitum. All experimental procedures were approved by the German Animal Welfare Authorities (approval no. G0152/04 and V54-19c20/15c MR 17/1).
Experiment 1: RMRt and body composition at weeks 6, 10, and 18
In this experiment, we sought to investigate the possibility that altered BMR contributed to the obese phenotype in BFMI860 mice. Mice included in the experiment were randomly chosen from five litters (two mice per litter) of the BFMI860 and B6 lines at the age of 5 weeks. They were maintained in single cages throughout the entire experimental period. Body mass was measured weekly. For determination of RMRt at the age of 6, 10, and 18 weeks, individual mice were transferred into metabolic cages (1.8 l) inside a temperature controlled cabinet. Ambient temperature was set to 30 °C, which corresponds to previous estimates of thermoneutrality in a number of inbred laboratory mouse strains (19,20,21). All measurements commenced at 8 am Central European Time. During 4 h, individual oxygen consumption (ΔVol% O2) and carbon dioxide production (ΔVol% CO2) were recorded using an electrochemical O2 analyzer (S-3AII; Ametek, Pittsburgh, PA) and a CO2 analyzer (UNOR 6N; Sick-Maihak, Reute, Germany, and Uras 14; Hartmann & Braun, ABB, Zurich, Switzerland, respectively). The principal setup of the respirometric system has been published previously (22). In each measurement, two or three mice and one empty reference channel were measured in 1-min intervals, yielding a resolution of readings of 3 or 4 min per mouse. O2-readings were converted to metabolic rate according to the following equation: MR (ml O2 × h−1) = ΔVol% O2 × flow (l × h−1) × 10, at a flow rate of ∼35 l/h. To adjust for differences in flow rates in air leaving and entering the metabolic cages, RQ (volumes of CO2 produced/volumes of O2 consumed) was used.
By definition, BMR is determined under a set of standard conditions (23). These require the animal to be resting, postprandial, normothermic, and not allocating energy into overall growth. Furthermore, measurements have to be performed at an ambient temperature which does not induce thermoregulatory heat production (i.e., a thermoneutral environment). In practice, it is difficult to meet the requirement that mice are truly postabsorptive because prolonged starvation may induce mechanisms (changes in body temperature or increased drive to search for food) that interfere with BMR standard conditions. Furthermore, we also determined metabolic rates in growing mice. To account for these differences in measurement conditions with respect to BMR standards, we adopted the term RMRt from ref. 24. In our study, RMRt is equivalent to the lowest mean of three consecutive O2-readings (moving average, covering 9 min (for two mice and one reference channel) and 12 min (for three mice and one reference channel), which yielded lowest variability (assessed by CV, coefficient of variation). RMRt values were obtained in the second half of the measurement period, i.e., between 10 am and 12 noon Central European Time.
After the respirometric measurements mice were weighed and body fat content was determined using dual-energy X-ray absorptiometry. (PIXImus II, version 1.46.007; GE Medical Systems, Madison, WI). The head of the animals was not included in the analysis, as recommended by the manufacturer. An additional measurement of body fat content was performed in week 15 (without determination of RMRt). Lean mass was calculated by subtracting measured fat mass from body mass.
At the age of 20 weeks, mice were fasted for 2 h, anesthetized with carbon dioxide and immediately killed by cutting the Vena cava inferior. Blood was collected from the thoracal cavity and organs were dissected. Serum was recovered by centrifugation for 15 min at 600 g and stored at −20 °C until analysis. Prior to analysis samples were thawed to room temperature and vortexed vigorously to resuspend any precipitated lipids. Serum triglycerides and total cholesterol were determined using the commercially available Fluitest TG (Triglyceride GPO-PAP) and Fluitest Chol (Cholesterin CHOD-PAP) kits (Biocon, Vőhl-Marienhagen, Germany), respectively, that have been optimized for human plasma. The measurements were made using the protocol from Biocon. Upon dissection, livers were shock frozen on dry ice. Cryosections (20 µm) were fixed in 10% formaldehyde-calcium, incubated with 0.3% oil-red O for 10 min, washed with isopropanol, and counterstained with hematoxylin.
Experiment 2: Body composition and digestion efficiency (week 6–12)
For the estimation of the amount of energy digested, food intake and the production of feces were determined. For the phenotypic characterization, 10 randomly chosen male mice per line were weighed weekly on the basis of their day of birth from the age of 5 to 12 weeks. Body fat mass was determined in nonanesthetized animals by quantitative magnetic resonance analysis using the EchoMRI whole body composition analyzer (Echo Medical Systems, Houston, TX) (25,26) in week 6, 8, 10, and 12. Animals were measured twice and the mean was used for further analyses. Lean mass was calculated by subtracting measured fat mass from body mass.
Food intake was estimated as the difference between the offered and the remnant amount. Food pellets were specifically pressed for low spillage by the food providing company and possible residual spillage was not considered. Cumulative food intake was measured weekly from week 5 and then calculated on a per-day basis. In week 6, 8, 10, and 12 feces were collected twice after 48 h, dried to constant weight (±0.001 g) at 60 °C, and energy content of dried feces (kJ/g) was analyzed for each individual using a bomb calorimeter (IKA C7000, Staufen, Germany). Gross energy content of dried food (assessed by bomb calorimetry) was 18.2 kJ/g, and this value was subsequently used to determine individual energy intake (Ein) per day. The amount of energy digested (Edig) per day was calculated from the difference between Ein and energy lost with feces (Efec). Digestion efficiency is the percentage of energy lost with feces (Efec) relative to Ein.
Statistical and data analyses
In order to compare the data of phenotypic and metabolic parameters between the lines BFMI860 with B6, repeated measurement ANOVA was used. Significant line differences were subsequently identified by t-test, with alpha adjusted for multiple comparisons according to Bonferroni. For endpoint line comparisons (e.g., TG levels, body fat content at week 20, or body mass gain), we also used t-tests.
In each line and at each point of measurement, RMRt correlated with body mass or lean mass at P < 0.1 (n = 10 per group). To account for the line specific effects of body mass or lean mass in the comparison of RMRt, body mass or lean mass were used as covariates in a two-factorial regression model (line, week, line × week) to predict individual RMRts at a common body mass (32.5 g) or lean mass (26.1 g), respectively. These adjusted RMRt values were subsequently analyzed by repeated measurement ANOVA. In contrast, body mass or lean mass were poor predictors for variations in energy intake or energy digested within line groups of different ages (P > 0.1 for most correlations). The amount of energy digested per day (Edig) was adjusted per gram lean mass.
All calculations were performed using the SPSS version 10.0 (SPSS, Chicago, IL). The level of significance was set to P < 0.05. In figures and tables, means ± s.d. are indicated.
Initially, RMRt in relation to body composition and RQs were compared between BFMI860 and B6 control mice (experiment 1). Subsequently, digestion efficiency was analyzed in both lines in a second experiment. In each experiment, mice of both lines showed similar growth curves for body mass, body fat mass, and body lean mass. Figure 1 shows the development of body mass, fat and lean mass of both lines of experiment 1. As previously noted, juvenile BFMI860 mice were visibly larger and heavier and they remained so until the end of the experimental periods. At week 18, BFMI860 mice weighed ∼20 g more than B6 mice (49.6 ± 4.0 g and 29.7 ± 2.6 g, t-test P < 0.001; Figure 1a). Approximately 52% (∼10 g) of excess body mass in 18-week-old BFMI860 mice had already accumulated at the age of 6 weeks. Between week 6 and 12, mice of either line grew heavier, but body mass gain in BFMI860 mice was higher than in B6 controls (17.6 ± 3.4 g in BFMI860 compared with 9.0 ± 2.0 g in B6, t-test P < 0.001). From week 12 to 18 both lines gained body mass at the same rate (cumulative body mass gain was 3.1 ± 1.5 g in B6 vs. 4.1 ± 2.8 g in BFMI860 mice; t-test P = 0.3).
The age related increase in body mass in both lines was accompanied by changes in body composition. In B6 mice, lean mass was consistently lower than in BFMI860 mice, but the rate of increase in this parameter was comparable in both lines (Figure 1b). As expected, body fat content was higher in BFMI860 mice at all time points (Figure 1c). On average, 18-week-old BFMI860 mice had a final body fat content of 24.7% compared with 14.6% in B6 (t-test P < 0.001). Livers from the 20-week-old BFMI mice had a yellowish color in situ, while livers from the B6 mice had a normal appearance. Subsequent oil-Red-O staining confirmed the presence of enlarged lipid droplets in liver sections from the BFMI mice, indicating fatty liver syndrome (Figure 1e). B6 mice accumulated body fat at an almost constant rate throughout the experimental period, but the temporal pattern of increase in body fat content was markedly different in BFMI860 mice. In this line, virtually all fat accumulated during the experiment was gained between week 6 and week 10 (Figure 1c).
At each point of measurement, mean RMRt in BFMI860 mice was significantly higher than in B6 mice (Figure 1d; P < 0.001). Despite the marked change in body composition of BFMI860 mice between 6 and 12 weeks mean RMRt was not significantly different between 6- and 10-week-old mice. Instead, RMRt was significantly higher in week 18 compared with week 6 and 10 (P < 0.001). In B6 mice there were no differences in mean RMRt at different ages.
BMR, and hence RMRt, scales with body mass (27) and we therefore initially included this parameter to compare RMRt, i.e., we accounted for the differences in body mass between lines at each of the three measurements. There was a significant time effect on body mass–adjusted RMRt, indicating that body mass–adjusted RMRt tended to decrease with age, and a weak interaction of time × line on body mass–adjusted RMRt (P = 0.05), i.e., body mass–adjusted RMRt was lower in 10-week-old BFMI860 compared with B6 mice (graph not shown). Because lean mass was another major predictor for changes in body mass (r2 = 0.97, n = 60), we also explored RMRt when adjusted for lean mass (Figure 2). Contrasting the model adjusting for body mass, the weak interaction of time × line was lost, i.e., lean mass–adjusted RMRt was not different between BFMI860 and B6 mice at any point of measurement.
During the 4 h in which mice were kept in the metabolic cage for determination of RMRt the parallel measurement of O2 consumption and CO2 production allowed to monitor for specific differences in RQ and RQ time courses (Figure 3). In each trial, RQ was higher in BFMI860 mice, and the decrease of RQ observed in B6 with time spent in the metabolic cage was less pronounced (week 6 and 10) or absent (week 18, P = 0.017) in BFMI860 mice. Because these changes in RQ are indicative of substrate fueling biased toward lipolysis we measured serum TG levels and total cholesterol after mice were killed at the age of 20 weeks. At week 20, plasma triglyceride levels, but not total cholesterol, were markedly elevated in the BFMI860 line (Table 1).
Table 1. Serum triglyceride (TG) levels and total cholesterol in B6 and BFMI860 mice
To assess whether the steep increase in postpubertal body fat content of BFMI860 mice observed between week 6 and week 10 was due to altered energy partitioning, we investigated a second set of mice of both lines with respect to body mass, body composition, energy intake (Ein), and energy content of feces (Efec), in more detail (experiment 2). Within the period between the age of 6 and 12 weeks, mice of the BFMI860 line consumed significantly more food and correspondingly more energy than B6 mice (Pline < 0.001; Figure 4a). The Ein of mice within line B6 did not vary much during the entire period, however mice of line BFMI860 reduced their Ein from week 9. This decrease in Ein corresponded with the onset of attenuated fat accumulation from week 10 (Figure 4b). The simultaneous measurements of food intake and feces production for 48 h in weeks 6, 8, 10, and 12 showed that mice of line BFMI860 also produced more feces than mice of line B6 (data not shown). Energy content of feces was identical between lines (mean Efec B6: 16.6 kJ/g; BFMI860:16.6 kJ/g). The percentage of energy digested per gram food consumed, however, was consistently higher in BFMI860 mice compared with B6 (Pline < 0.001; Figure 4c). During weeks 6–10, BFMI860 mice retained significantly more energy from food per gram lean mass compared with B6 mice (Figure 4d).
In concordance with our previous results (16), the BFMI860 line disposed to higher body mass, body fat mass, and body lean mass than the B6 line, independently of age. Here, we show that 50% of excess body mass gain in line BFMI860 developed before puberty (by week 6), while the remaining body mass gain (50% of which was fat) occurred between week 6 and 12, without a concomitant accelerated increase in lean mass. In expanding our previous findings, these results demonstrate that the BFMI860 mouse line has two body mass related phenotypes: an overall accelerated growth phenotype associated with higher lean mass and higher fat mass in juveniles, and an adult obesity phenotype.
To further identify components of altered energy balance in the postpubertal mice, we compared energy intake and energy expenditure (RMRt) in more detail. In BFMI860 mice, virtually all metabolic parameters were numerically higher compared with B6 mice, but the former were also larger and fatter. Since nonadipose tissues such as liver, brain, kidney and skeletal muscle are major contributors to (basal) energy expenditure (28,29), we focused on eliminating the effect of differences in lean mass, in order to compare metabolic data between lines of different body mass and composition.
Overall energy intake was higher in BFMI860 compared with B6 mice from week 6 through week 12, but we also observed a decrease in Ein of BFMI860 mice from week 10. A lowered food intake, and hence, Ein from week 10 coincided with normalizing rates of fat accumulation in BFMI860 mice. In addition, digestion efficiency was consistently higher in BFMI860 mice, and lean mass–adjusted energy extracted from food (Edig) was higher during the period of high-fat gain (week 6 through week 10). From Figure 4d we can calculate that the seemingly small line difference in daily Edig per gram lean mass (∼0.3 kJ 24 h−1 g lean mass−1) amounts to ∼8.4 kJ in 4 weeks. Hence, ∼228 kJ of excess energy could be directed toward lipid accumulation during this period. If we further assume that 1 g of lipid has an energy content of 38 kJ, ∼5.9 g of fat could be gained in 4 weeks (i.e., from week 6 through week 10), which is what we measured in vivo (Figures 1c and 4b). These results support that hyperphagia and higher digestion efficiency during early adulthood promoted obesity in BFMI860 mice.
We also investigated the possibility that altered energy expenditure (measured as RMRt) contributed to the obese phenotype in BFMI860 mice. Even though mean RMRts were numerically higher in BFMI860 mice, these differences between lines could be fully accounted for when differences in lean mass, i.e., metabolically active mass were considered. We therefore conclude that altered BMR does not contribute to the obese phenotype in BFMI860 mice. This is in accordance with findings where no correlation between RMR and weight gain could be found in nonobese mouse strains (17,18). However, the RQ remained consistently higher in BFMI860 mice compared with B6 mice, although mean values only reached significance at an age of 18 weeks. Notably, BFMI860 mice not only had higher mean RQs throughout the entire measurement period, but also showed less variable RQs, i.e., 18-week-old mice were able to respire for ∼4 h in the metabolic cage at an almost constant RQ of ∼1. In mice and humans, a RQ between 0.9 and 1 is generally observed in a fed state, reflecting metabolic fueling primarily derived from carbohydrate oxidation. In addition, release of CO2 from the pentose phosphate pathway associated with generating NADPH for de novo lipogenesis may contribute to high RQs in vivo. In the postprandial state, lipids are normally recruited for oxidation and RQ subsequently decreases to <0.8, as seen in the B6 control mice. The high postprandial RQs observed in BFMI860 mice are therefore likely to result from decreased lipid mobilization, e.g., by impaired lipolysis through blunted activity of key lipolytic enzymes or decreased fat oxidation (30,31,32).
Interestingly, we found that livers of 20-week-old BFMI860 were visibly hypertrophic with fat, and that serum triglyceride levels were markedly elevated. In our previous work, we have already shown that glucose levels in BFMI860 mice were normal from 8 through 20 weeks of age (16). High circulating fasting and prandial triglyceride levels (hyperlipidemia) and normal blood glucose levels have been associated with insulin resistance and the development of cardiovascular disease (33). For example, Jensen et al. (34) have recently shown that transgenic mice selectively overexpressing lipoprotein lipase in muscle oxidize more fat than controls. Conversely, loss of lipoprotein lipase expression is associated with reduced lipid metabolism (35). In BFMI860 mice, in vivo energy expenditure was clearly less tuned toward oxidation of fat, and the presence of hyperlipidemia further supports altered lipid metabolism in the body. The precise mechanisms underlying altered substrate partitioning in this line remain to be elucidated.
In conclusion, adult BFMI860 male mice are heavier due to increased fat and lean mass. The majority of their fat mass was accumulated between 6 and 10 weeks of age and was associated with hyperphagia in young BFMI860 mice. In addition, BFMI860 mice have consistently higher digestion efficiencies, higher RQs, and abnormally high blood TGs, suggesting that higher energy conversion and decreased lipid oxidation for metabolic fueling further promotes the accumulation of body fat. Assuming that the obese phenotype in BFMI860 mice is of polygenic nature this line is not only an excellent model for the study of obesity in humans, but also a valuable resource for the identification of new mechanisms leading to obesity, especially at the juvenile age. Furthermore, these mice may be important as animal models of obesity during pregnancy.
We thank Reinhard Schiefler, Katrin Beck, and Ulf Kiesling for assistance with mouse husbandry and phenotypic data collection. Maria Kutschke and Stefanie Kunst helped us with bomb calorimetry. This research was supported by grants of the German National Genome Research Network (NGFN-2 grants #01GS0486 to G.A.B., #01GS0483 to M.K. and G.H., and #01GS0822 to M.K.) and by grants of the German Research Foundation (GK1209 to G.A.B.).