Deficiency of MGAT2 increases energy expenditure without high-fat feeding and protects genetically obese mice from excessive weight gain.

Acyl CoA:monoacylglycerol acyltransferase 2 (MGAT2) is thought to be crucial for dietary fat absorption. Indeed, mice lacking the enzyme (Mogat2−/−) are resistant to obesity and other metabolic disorders induced by high-fat feeding. However, these mice absorb normal quantities of fat. To explore whether a high level of dietary fat is an essential part of the underlying mechanism(s), we examined metabolic responses of Mogat2−/− mice to diets containing varying levels of fat. Mogat2−/− mice exhibited 10−15% increases in energy expenditure compared with wild-type littermates; although high levels of dietary fat exacerbated the effect, this phenotype was expressed even on a fat-free diet. When deprived of food, Mogat2−/− mice expended energy and lost weight like wild-type controls. To determine whether MGAT2 deficiency protects against obesity in the absence of high-fat feeding, we crossed Mogat2−/− mice with genetically obese Agouti mice. MGAT2 deficiency increased energy expenditure and prevented these mice from gaining excess weight. Our results suggest that MGAT2 modulates energy expenditure through multiple mechanisms, including one independent of dietary fat; these findings also raise the prospect of inhibiting MGAT2 as a strategy for combating obesity and related metabolic disorders resulting from excessive calorie intake.

sumption and carbon dioxide production were measured, and the metabolic rates were calculated from a modifi ed Weir equation: kcal = 3.941 × O 2 (L) + 1.106 × CO 2 (L) ( 19 ). No adjustment for protein metabolism was included, because errors introduced by a moderate protein diet are practically negligible ( 19 ) and because the defi ned diets contained a constant level of protein calories. Consumption of food and drink also were monitored continuously. Male mice (2 Ϫ 4 months old) were acclimated to individual housing and metabolic cages for 1 week before experiments. Most mice were acclimatized in the metabolic cages within 24 h, as indicated by characteristic circadian rhythms of O 2 consumption and CO 2 production (data not shown).
For studies on acute metabolic responses to low-versus highfat feeding, mice were fed indicated diets sequentially for 3 days each. For studies on metabolic responses to fat-free feeding and fasting, mice were fed 10% fat, fat-free, and again the 10% fat diet for 3 days each, followed by a 48 h fast and another 10% fat refeeding for 3 days. Indirect calorimetry measurements (VO 2 and VCO 2 ) from the same time on days 2 and 3 of each diet treatment were averaged. During the 48 h fast, measurements from both days were averaged. Body weight measurements were made at the start of each diet treatment. To examine whether there was a genotype effect on oxygen consumption, total oxygen consumption per day, calculated by area under the curve, was plotted against baseline body weight for each mouse on each diet. In addition, VO 2 and metabolic rate were calculated and analyzed both with and without adjusting for body weight (20)(21)(22).

Pair feeding
To determine whether increases in energy expenditure of Mogat2 Ϫ / Ϫ mice are obligatory or are in response to increased food intake, we pair-fed weight-stable Mogat2 Ϫ / Ϫ mice (1 year-old males) to their wild-type littermates. Mouse body weight and food intake (chow) were measured daily during acclimation to individual housing. Pair-feeding began after body weight and food intake stabilized. Food intake of wild-type mice was determined each day; the average amount of food consumed by wild-type mice was fed to each of the Mogat2 Ϫ / Ϫ mice.

Statistical analyses
All data are presented as mean ± SEM. P < 0.05 was considered statistically signifi cant. For oxygen consumption per day, differences between genotypes were assessed by the mixed-effects model with repeated measures and adjusted for body weight within the four-diet strata using the PROC MIXED procedure in SAS (SAS, Inc.; Cary, NC) ( 23 ). Likewise, differences between diets were assessed in parallel within the two genotype strata. For weight patterns, differences between genotype were assessed by repeated measures ANOVA, followed by Tukey's multiple comparison test (Prism 5.01, GraphPad Inc.; La Jolla, CA). For all other parameters, differences between genotypes were determined by one-way ANOVA followed by the protected least significant differences technique (Prism 5.01). To identify changes in body weight after each diet treatment, we performed paired t -tests (Prism 5.01).

MGAT2 defi ciency increases oxygen consumption, an indicator of energy expenditure, throughout all diets
To determine if a high level of dietary fat is required for the increases in energy expenditure induced by MGAT2 defi ciency, Mogat2 Ϫ / Ϫ mice and their wild-type littermates These fi ndings raise the possibility that inhibiting MGAT2 could be a strategy for combating human metabolic disorders caused by excess fat accumulation. However, obesity in humans often results from excess caloric intake but not necessarily high levels of dietary fat. To determine whether a high level of dietary fat is required for the increases in energy expenditure induced by MGAT2 defi ciency, we examined acute metabolic responses of Mogat2 Ϫ / Ϫ mice to diets containing different levels of fat. In addition, to determine whether the protection against excessive weight gain is limited to obesity induced by a high-fat diet, we examined the effects of MGAT2 defi ciency in hyperphagic Agouti mice ( 18 ), a genetic model of obesity.

Diets
A series of semi-purifi ed (defi ned) diets containing 10, 45, or 60% calories from fat were used to examine the effect of replacing carbohydrate with dietary fat (D12450B, D12451, and D12492; Research Diets, New Brunswick, NJ). In these diets, protein calories (from casein) were held constant at 20%, as were micronutrient and fi ber contents, while fat (lard) replaced carbohydrate (corn starch and sucrose) to increase fat calories from 10 to 45 or 60%. Each diet contained 3.8, 4.7, and 5.2 kcal/g metabolizable energy, respectively (formulas are available at www.researchdiets.com). Chow diet (#8604, Teklad; Madison, WI), on which mice were normally maintained, was a complete fi xed-formula diet providing approximately 33% energy from crude protein (24.5% w/w) and 14% energy from fat (4.4% w/w). Chow diet contains 3.9 kcal/g gross energy and 3.1 kcal/g metabolizable energy. To determine whether dietary fat is required, a fat-free diet and its low-fat (10%) control diet were made in our laboratory to match protein, micronutrient, and fiber contents (maintaining the same nutrient-to-calorie ratios) of defi ned diet D12450B using the same ingredients. Fat was replaced by cornstarch.

Metabolic phenotyping studies
A metabolic phenotyping system (LabMaster modular animal monitoring system; TSE Systems, Chesterfi eld, MO) with housing and wood chip bedding similar to the home cage environment was used to continuously monitor phenotypes related to acute energy balance. As indicators of energy expenditure, oxygen con-mice maintained similar levels of oxygen consumption and their body weights throughout. As a result, with chow feeding, the body weights did not differ between genotypes. After consuming each of the defi ned diets, wild-type mice weighed 3.6, 9.1, and 11.8% more than they did at the start of the study ( P < 0.05 for all comparisons). Mogat2 Ϫ / Ϫ mice, on the other hand, maintained their weights throughout the experiment and weighed 15.6% less than wild-type mice did at the end of the study ( Fig. 1B ).

The increase in oxygen consumption is augmented during feeding periods
When rates of oxygen consumption were plotted over time, both groups of mice exhibited a clear circadian rhythm. As expected, oxygen consumption rates were higher during the dark phase of the light cycle ( Fig. 2A ), when mice are active and feeding ( Fig. 2B ). This pattern was present for wild-type and Mogat2 Ϫ / Ϫ mice across each of the diets. Compared with wild-type mice, Mogat2 Ϫ / Ϫ mice consumed more oxygen per mouse, as indicated by greater areas under the curves across diets ( Fig. 2A ; genotype main effect, P < 0.05). The differences between genotypes were most evident during the dark phase. For example, on chow, the oxygen consumption of Mogat2 Ϫ / Ϫ mice was 7.4% higher than that of controls during the dark phase, but not signifi cantly different during the light phase. Likewise, when consuming each of the defi ned diets, oxygen consumption of Mogat2 Ϫ / Ϫ mice increased 5.4-7.0% over that of the wild-type. Although the increases when consuming 45% and 60% diets did not reach statistical significance ( P = 0.09 and 0.15, respectively), the differences were remarkable because wild-type, but not Mogat2 Ϫ / Ϫ , mice gained weight when fed the high-fat diets ( Fig. 1B ).

High-fat feeding exacerbates differences in oxygen consumption
Differences in oxygen consumption between wild-type and Mogat2 Ϫ / Ϫ mice were accentuated when adjusted for body weight. When mice were fed purifi ed diets, compared with chow, the differences in oxygen consumption were more pronounced; moreover, the differences increased as dietary fat content increased ( Fig. 2C ). When mice were fed chow, the difference between genotypes was 10.2%, whereas the differences were 11.2, 12.3, and 15.7% on the 10, 45, and 60% diets, respectively ( P < 0.05 for all comparisons). The magnitude of differences cannot be explained fully by the differences in body weights at the beginning of the high-fat diet treatments.

Mogat2 ؊ / ؊ mice oxidize more fat than do wild-type mice when consuming the defi ned diets
Mice consistently showed an increase in respiratory exchange ratio (RER) during the dark phase ( Fig. 2D ), indicating a shift from utilizing body fat stores to utilizing dietary carbohydrate.
In keeping with oxidation of more carbohydrates than fat, the average RERs over 24 h were higher when mice were switched from chow (53% carbohydrate) to the 10% diet high in refi ned carbohydrate (70%; from 0.89 ± 0.06 to 0.93 ± 0.06). Consistently, RERs decreased step-wise, were housed in the metabolic chambers and fed sequentially for 3 days on each of the four diets: a regular mixedmeal chow and a defi ned diet containing 10, 45, or 60% of calories from fat (referred to as chow, 10%, 45%, and 60% diet, respectively). When total oxygen consumption per day was plotted against body weight of each mouse, the regression lines of Mogat2 Ϫ / Ϫ mice were signifi cantly elevated over those of wild-type controls fed the same diets ( Fig. 1A ), indicating higher energy expenditure in Mogat2 Ϫ / Ϫ mice than in wild-type controls. Based on the mixed-effects model within the four-diet strata, when fed chow, Mogat2 Ϫ / Ϫ mice consumed approximately 8% more oxygen than did controls at any given body weight, and 10, 11, and 13% more when fed 10, 45, and 60% diets, respectively (main effect of genotype, P < 0.05). When stratifi ed by genotype, as calories from fat increased in purifi ed diets, wild-type, but not Mogat2 Ϫ / Ϫ , mice signifi cantly decreased their oxygen consumption (main effect of diet, P < 0.05). For example, wild-type mice when fed the 60% diet consumed approximately 5% less oxygen than when fed the 10% diet, whereas there was no difference in Mogat2 Ϫ / Ϫ mice.

Increased oxygen consumption is associated with resistance to weight gain
Associated with the decreased oxygen consumption, wild-type mice increased body weight after consuming the calorie-dense defi ned diets ( Fig. 1B ). In contrast, Mogat2 Ϫ / Ϫ the light phase. The decreases reached statistical significance when mice were fed the 45% and 60% diets ( Fig.  2D ). However, the differences diminished after dark, when RER rose in both groups, presumably due to increased oxidation of carbohydrate coming from the diet.

Energy expenditure is elevated in
To determine levels of energy expenditure as metabolic rates, we used a modifi ed Weir equation that takes both from 0.93 to 0.83 and then 0.78, when wild-type mice were fed diets with increasing amounts of fat replacing carbohydrate, from 10% to 45% and then 60%.
Like wild-types, Mogat2 Ϫ / Ϫ mice also showed the expected step-wise decreases in RER, from 0.92 to 0.79 and then 0.75, when dietary carbohydrate was replaced by fat. Interestingly, Mogat2 Ϫ / Ϫ mice oxidized higher proportions of fat than did wild-type mice when fed the defi ned diets. Their RERs were 3 Ϫ 7% lower than those of controls during their similar levels of energy expenditure, Mogat2 Ϫ / Ϫ mice and their wild-type controls lost a similar percent of their body weights during the 2 day fast (17.4 ± 0.6 vs. 16.6 ± 0.6%, P = 0.365; Fig. 3B ).
Mogat2 ؊ / ؊ mice maintain energy balance despite increased energy expenditure When Mogat2 Ϫ / Ϫ mice were fed chow, their weight curves were not signifi cantly different from those of wildtype littermates for at least 1 year [( 17 ); data not shown]. Thus, the increases in energy expenditure of Mogat2 Ϫ / Ϫ mice in the absence of high-fat feeding were not expected. Because body weight is determined by a balance of energy intake and energy expenditure, we found that when fed chow, Mogat2 Ϫ / Ϫ mice consumed 8.4% more per day than did wild-type mice (4.64 ± 0.15 vs. 4.28 ± 0.11 g/day; Fig.  2B ). By pair-feeding Mogat2 Ϫ / Ϫ mice to wild-type controls, we next tested whether MGAT2 defi ciency induces an obligatory increase in metabolic rate or if increased metabolic rate is secondary to increased food intake. When their chow intake was restricted to the level of controls, Mogat2 Ϫ / Ϫ mice lost a signifi cant 4.5% of their weight within a day and the negative energy balance continued until they reached a new equilibrium after 6 days of pairfeeding ( Fig. 4 ). This fi nding suggests that increased energy expenditure of Mogat2 Ϫ / Ϫ mice consuming a chow diet is obligatory and is normally compensated for by increases in food intake to maintain energy homeostasis.
When fed the defi ned 10% diet, the 5% increase in food consumption of Mogat2 Ϫ / Ϫ mice as compared with wildtype littermates (3.84 ± 0.13 vs. 3.65 ± 0.10 g/day) did not reach statistical signifi cance. Mogat2 Ϫ / Ϫ mice maintained energy balance, whereas wild-type mice gained 0.8 ± 0.1 g in 3 days when switched to the calorie-dense diet rich in oxygen consumption and RER into account after adjusting for body weight ( 19 ). In parallel with oxygen consumption, metabolic rates of wild-type mice were higher during the dark phase as compared with during the light phase, and the rates decreased with high-fat feeding ( Fig. 2E ). Mogat2 Ϫ / Ϫ mice exhibited signifi cantly higher metabolic rates compared with wild-type mice on each of the four diets (12.6, 10.7, 12.3, and 16.2%, respectively; P < 0.05). These increases were associated with food intake and remained high during the postprandial state. The magnitudes of differences were similar, but not identical, to those of oxygen consumption described above, consistent with differences in substrate utilization as indicated by RERs.

Dietary fat is not required for MGAT2 defi ciency-induced increases in energy expenditure
Because Mogat2 Ϫ / Ϫ mice exhibited increased metabolic rates after feeding on a diet containing as low as 10% fat, we next determined whether dietary fat is essential for the underlying mechanisms. We did so by examining the metabolic responses of Mogat2 Ϫ / Ϫ mice to a fat-free diet and to fasting. When switched to a defi ned fat-free diet, Mogat2 Ϫ / Ϫ mice maintained their increased energy expenditure and food intake to degrees similar to those seen when they were fed a comparable 10% fat diet ( Fig. 3A ). During the dark phase, when mice were most active and feeding, metabolic rate in Mogat2 Ϫ / Ϫ mice was elevated 10.9% over that of wild-type mice on a fat-free diet, as compared with a 10.5% elevation when they were fed the 10% diet.
Even though dietary fat is not essential for expression of the phenotype, the presence of some dietary component is. When fasted, Mogat2 Ϫ / Ϫ mice reduced metabolic rate ( Fig. 3A ) to the same levels as their wild-type controls, both before and after adjustment for body weight. Consistent with protective effects upon chronic high-fat feeding ( 17 ) are mainly conferred by the increases in energy expenditure.

MGAT2 defi ciency increases energy expenditure in genetically obese Agouti mice
To rigorously test the biological signifi cance of MGAT2 defi ciency on energy balance without high-fat feeding, we introduced the Mogat2 -targeted allele into genetically obese Agouti mice ( Ay/a ). We fi rst examined their metabolic responses to the same four diets. On chow and 10% diet, Ay/a, Mogat2 Ϫ / Ϫ mice exhibited 10.6% and 11.7% increases, respectively, in metabolic rates, compared with Agouti mice ( Fig. 6A ). When fi rst exposed to the 45% high-fat diet, however, Ay/a, Mogat2 Ϫ / Ϫ mice showed only an 8.8% increase in metabolic rate. The increase was probably subdued by a 30% lower food intake of Ay/a, Mogat2 Ϫ / Ϫ mice than that of Agouti mice ( Fig. 6B ; 2.96 ± 0.28 vs. 4.17 ± 0.14 g/day). On 60% diet, these mice had 16.6% higher metabolic rates than Agouti mice despite 32% lower food intake (2.49 ± 0.18 vs. 3.64 ± 0.19 g/day). Like mice without the Agouti mutation, Ay/a, Mogat2 Ϫ / Ϫ mice decreased their food intakes when fed the more-calorie-dense diets; in contrast, Agouti mice had elevated food intakes and consumed 32, 54, and 50% more calories when fed 10, 45, and 60% diets, respectively, than when fed chow. Accordingly, during the metabolic phenotyping studies, Agouti mice gained weight rapidly, especially when fed the defi ned diets, whereas MGAT2 defi ciency strongly attenuated the acute increases in body weight to the levels of wild-type mice without the Agouti mutation ( Fig. 6C ).

Agouti mice defi cient in MGAT2 are protected from excessive weight gain
To determine the long-term effect of MGAT2 defi ciency on weight gain induced by the Agouti mutation, we fed Ay/a, Mogat2 Ϫ / Ϫ mice and their littermate controls chow diet for 40 weeks. In the presence of MGAT2, male Agouti mice ( Ay/a, Mogat2 +/+ ) weighed signifi cantly more than did their wild-type littermates ( a/a, Mogat2 +/+ ) by 8 weeks of age (25.2 ± 0.6 vs. 22.5 ± 0.1 g; Fig. 7A ). In the absence of MGAT2, Agouti mice ( Ay/a, Mogat2 Ϫ / Ϫ ) were protected from excessive weight gain at all time points monitored. After 40 weeks of chow feeding, Ay/a, Mogat2 +/+ mice weighed 52% more than did wild-type littermates, whereas Ay/a, Mogat2 Ϫ / Ϫ mice had body weights similar to wild-type controls without the Agouti mutation throughout the study ( Fig. 7A ). As seen in obesity induced by a high-fat diet, heterozygous ( Ay/a, Mogat2 ϩ/ Ϫ ) mice exhibited an intermediate phenotype, and fat mass accounted for most of the differences in weights (data not shown). Similar results were observed in females. After chow feeding for 40 weeks, Agouti mice ( Ay/a, Mogat2 +/+ ) weighed 59% more than wild-type controls. Ay/a, Mogat2 Ϫ / Ϫ female mice weighed signifi cantly less than Agouti mice at all time points, resulting in 23% lower body weight after 40 weeks (28.8 ± 1.6 vs. 37.6 ± 2.4 g; Fig. 7B ).

DISCUSSION
We have previously reported that Mogat2 Ϫ / Ϫ mice exhibit increased energy expenditure and are protected from obesity and other metabolic disorders induced by refi ned carbohydrates ( Fig. 1B ). In an independent longerterm study, wild-type mice gained a signifi cant amount of weight after 1 week on the 10% diet and continued to gain weight throughout the nearly 2 month experiment, resulting in a 14.5% weight gain ( Fig. 5 ). Mogat2 Ϫ / Ϫ mice, on the other hand, maintained their baseline body weights.
At the end of the study, wild-type mice weighed 20% more than did Mogat2 Ϫ / Ϫ mice. When switched to 45% and 60% diets, Mogat2 Ϫ / Ϫ mice also were protected from weight gains seen in controls ( Fig. 1B ). Interestingly, Mogat2 Ϫ / Ϫ mice transiently decreased caloric intake when fed high-fat diets, compared with when they were fed the 10% diet; whereas wild-type mice slightly increased their caloric intake over the fi rst 3 days of high-fat feeding (data not shown). The effects of MGAT2 defi ciency on intake of high-fat diets were transient, inasmuch as food intakes between genotypes were not signifi cantly different after 3 days of high-fat feeding (data not shown). In contrast, the effects on energy expenditure persisted after mice were fed the high-fat diets for 10 weeks (data not shown), suggesting that the reported ; open circles; n = 7) and wildtype littermates (WT; black squares; n = 11) were measured daily. Average food intake for WT mice was calculated daily, and the same amount was fed to each Mogat2 Ϫ / Ϫ mouse during the pairfeeding period (between dotted lines). Values are means ± SEM. Bracket includes data points that differ from body weight at the start of pair-feeding in Mogat2 Ϫ / Ϫ mice. ; open circles; n = 7) and wild-type littermates (WT; black squares; n = 12) were measured weekly after switching from chow to 10% diet. Values are means ± SEM. Bracket includes data points that differ from body weight at the start of 10% feeding in wild-type mice. *, P < 0.05 versus diet-matched WT. excessive weight gain in a genetic model of obesity without high-fat feeding.
On the basis of the established role of intestinal MGAT in fat absorption, we expected that the effects of MGAT2 defi ciency would require high levels of fat intake. Indeed, the body weights of Mogat2 Ϫ / Ϫ mice did not differ from those of wild-type littermates during chow feeding; and in our previous report, the increases in energy expenditure, measured using a different metabolic phenotyping system, did not reach statistical signifi cance until mice were fed a 60% fat diet ( 17 ). In this study, using a new metabolic phenotyping system that resembles their home cage environment and allows for bedding, we found that Mogat2 Ϫ / Ϫ mice exhibited increases in energy expenditure on all tested diets containing a wide range (0 Ϫ 60%) of fat. The increases in energy expenditure between genotypes are evidenced by increases in oxygen consumption as well as the calculated metabolic rate after considering substrate utilization. Increases in oxygen consumption were identifi ed high-fat feeding ( 17 ). In the present study, we show that defi ciency of MGAT2 in mice also increases energy expenditure, decreases the propensity to gain weight, and protects against obesity in the absence of high-fat feeding. In acute feeding experiments, we found that Mogat2 Ϫ / Ϫ mice had 10 Ϫ 15% higher metabolic rates than did their wildtype littermates when they were fed diets containing 10 Ϫ 60% calories from fat. Unexpectedly, when fed a diet without fat, Mogat2 Ϫ / Ϫ mice still exhibited increases in metabolic rates to a similar degree as when mice were fed 10% fat diet. This effect was associated with dietary intake, and it disappeared during fasting, implicating an essential role of MGAT2 in effi cient assimilation of nutrients and in regulating an obligatory component of diet-induced thermogenesis. The functional signifi cance of these increases in energy expenditure was illustrated by the lower propensity of Mogat2 Ϫ / Ϫ mice to gain weight, as compared with wild-type littermates, when fed a low-fat diet rich in refi ned carbohydrate. Furthermore, MGAT2 defi ciency prevents also refl ect the increases in fat mass, which expends less energy than lean mass. Mogat2 Ϫ / Ϫ mice exhibited increased energy expenditure in response to a meal, even in the absence of dietary fat, indicating that at least one underlying mechanism is independent of fat intake. These mice showed a similar 10% higher metabolic rate over wild-type littermates in response to a fat-free or a 10% fat diet ( Fig. 3 ). Increasing dietary fat from 10% to 60% exacerbated the differences only by an additional 5% ( Fig. 2E ), suggesting the fatindependent mechanism has a dominating impact. Fasting eliminated this effect. In mice, food consumption and physical activity are closely correlated. We postulate that the increase in energy expenditure of Mogat2 Ϫ / Ϫ mice refl ects a response to meals, rather than a change in physical activity, because there was no difference detected in ambulatory activity ( 17 ), energy expenditure during fasting ( Fig. 3A ), or weight loss after a 48 h fast ( Fig. 3B ). These mice can still respond to food deprivation properly by reducing metabolic rate and slowing the rate of weight loss to the same extent as wild-type mice. During fasting, Mogat2 Ϫ / Ϫ mice presumably remained active in food-seeking behaviors, like wild-type mice, as indicated by the elevation in metabolic rate during the dark phase, as compared with the light phase. The differences in energy expenditure between genotypes were seen again upon refeeding ( Fig. 3 ). At the systems level, MGAT2 modulates energy balance most likely through its role in the intestine, because MGAT2 is highly expressed only in the intestine in mice ( 11 ) and because the effects of MGAT2 defi ciency on energy expenditure require food intake ( Fig. 3 ). Intestinal MGAT2 is probably an integral part of the processes that enhance metabolic effi ciency by reducing the cost of nutrient assimilation and suppressing thermogenesis. In contrast to the well-established role of the intestine in regulating food intake and nutrient absorption, the phenotype of Mogat2 Ϫ / Ϫ mice suggests a less-recognized role of the intestine in modulating energy expenditure. This conjecture is supported also by fi ndings from mice deficient in diacylglycerol acyltransferase 1 (DGAT1), another triacylglycerol synthesis enzyme highly expressed in the small intestine ( 2,32 ). Mice defi cient in DGAT1 exhibit increased energy expenditure and are protected from diet-induced obesity ( 33 ). Reintroducing DGAT1 specifically in the intestine is suffi cient to reverse the weight protection phenotype ( 34 ). These fi ndings suggest that MGAT2-and DGAT1-mediated lipid metabolism in the intestine plays a pivotal role in coordinating systemic responses to maximize metabolic effi ciency and propensity to gain weight. At the cellular level, the absorptive enterocytes (where MGAT2 is highly expressed) are most likely responsible through their impacts on delivery of lipid substrates to peripheral tissue and on secretion of gut hormones from the enteroendocrine cells ( 16,17 ). Levels of several postprandial hormones differ between Mogat2 Ϫ / Ϫ and wild-type mice ( 17,35 ). Because both its substrate monoacylglycerols and its product diacylglycerols can serve as signaling within the four diet strata using the mixed-effects model as well as using a ratio method adjusting for body weight ( Figs. 1, 2 ). Mogat2 Ϫ / Ϫ mice consistently consumed higher absolute amounts of oxygen than did wild-type littermates. These increases were remarkable, inasmuch as Mogat2 Ϫ / Ϫ mice tended to weigh less. The differences in energy expenditure between Mogat2 Ϫ / Ϫ mice and wild-type littermates were most pronounced during the feeding period, suggesting that MGAT2 modulates diet-induced thermogenesis. Also known as the thermic effect of food, this increase in metabolic rate after a meal is estimated to account for 10% of total energy expenditure in adult humans, in addition to physical activity and basal metabolism ( 24,25 ). Diet-induced thermogenesis is considered an obligatory energy cost for digestion, absorption, transport, and other metabolic activity needed for the assimilation of nutrients. It has also been proposed as a facultative part of adaptive thermogenesis responding to excess food intake, like nonshivering thermogensis responding to cold, that involves the sympathetic nervous system and brown adipose tissues (26)(27)(28)(29)(30).
Dietary fat induces the least thermogenesis (estimated at 0 Ϫ 5%), whereas protein induces the most (20 Ϫ 30%) among the energy-yielding nutrients ( 31 ). The lower thermic effect of fat compared with carbohydrate may explain the weight gains upon high-fat feeding ( 4,5 ). In our study, wild-type, but not Mogat2 Ϫ / Ϫ , mice have decreased postprandial metabolic rates when fed high-fat diets, suggesting that MGAT2 reduces the energy cost of fat assimilation. Because wild-type mice were gaining weight on high-fat diets, the decreases in weight-adjusted metabolic rate may Ϫ / Ϫ mice with the Agouti mutation ( Ay/a ,

Mogat2
Ϫ / Ϫ ; gray hexagons) were fed chow after weaning through 40 weeks. n = 5-9 mice /per group. Values are means ± SEM. Error bars not shown are smaller than symbols. ture after meals. Although the differences are exacerbated by high-fat feeding, the underlying mechanisms are dominated by one independent of dietary fat. The protective effect of MGAT2 defi ciency against excessive weight gain is not limited to obesity induced by excessive fat intake. Mice defi cient in MGAT2 also were protected from weight gain induced by a low-fat diet rich in refi ned carbohydrate and from genetic obesity induced by the Agouti mutation. Because MGAT2 is highly expressed in human intestine (11,39), these fi ndings also raise the prospect of inhibiting MGAT2 as a strategy for combating obesity and related metabolic disorders resulting from overconsumption of calories in humans. molecules, MGAT2 may directly modulate the secretion of enteroendocrine cells in response to luminal nutrients. It remains to be explored whether MGAT2 is expressed in enteroendocrine cells in vivo. Additionally, a few tissues, such as brown adipose tissue, express low levels of MGAT2 (less than 0.1% of that found in the small intestine). At this time, the possibility cannot be excluded that these tissues might modulate energy balance in a cell-autonomous fashion through mechanisms such as substrate cycling between de novo lipogenesis and FA oxidation.
The effects of MGAT2 on food intake depend on dietary composition. Mogat2 Ϫ / Ϫ mice compensated for the increased energy expenditure by consuming more chow to maintain their weight. When chow intake levels were restricted to those of wild-types, they lost weight ( Fig. 4 ), confi rming that the constant increase in food intake is an adaptive response to an obligatory increase in energy expenditure. The transient suppression of food intake, seen in these mice when fi rst exposed to a high-fat diet, may refl ect a change in factors that are involved in short-term intake regulation, such as cholecystokinin, peptide YY, and glucagon-like peptide-1 ( 17,36 ). These factors may be overridden by factors that regulate long-term energy balance ( 37 ), inasmuch as intake levels of Mogat2 Ϫ / Ϫ mice returned to normal within 3 days (data not shown). Unlike wild-type mice, Mogat2 Ϫ / Ϫ mice are protected from obesity after prolonged high-fat feeding ( 17 ). This protective effect is probably due to the increases in energy expenditure, which persisted over 2 months, whereas the decreases in food intake were transient (data not shown). The levels of fat absorption and the amounts and energy content of feces do not differ between genotypes ( 17 ). The protective effect of MGAT2 defi ciency is broader than high-fat-diet-induced obesity, because inactivating MGAT2 also protects against weight gains induced by a diet rich in refi ned carbohydrate and by the Agouti mutation. In addition to increased food intake, Agouti mice are more metabolically effi cient, allowing them to gain more weight than controls when fed chow, a high-fat, or a highsucrose diet ( 38 ). Consistent with their higher metabolic effi ciency, we found that the metabolic rates in Agouti mice trended lower than in wild-type littermates without the Agouti mutation, although the difference did not reach statistical signifi cance during the metabolic phenotyping study. Nonetheless, they were hyperphagic and gained weight much faster than did their wild-type littermates, especially when fed each of the defi ned diets ( Fig.  6B ). In contrast, inactivation of MGAT2 in Agouti mice increased energy expenditure, attenuated hyperphagia, and minimized weight gain. These differences were most pronounced during acute high-fat feeding ( Fig. 6 ). In the long-term chow feeding study, Agouti mice defi cient in MGAT2 were protected from the adult-onset obesity seen in Agouti mice ( Fig. 7 ). The underlying mechanisms remain to be determined.
In conclusion, the results of our study indicate that MGAT2 normally modulates diet-induced thermogenesis, promotes metabolic effi ciency, and favors positive energy balance. Defi ciency of MGAT2 enhances energy expendi-