Adult-onset deficiency of acyl CoA:monoacylglycerol acyltransferase 2 protects mice from diet-induced obesity and glucose intolerance.

Acyl-CoA:monoacylglycerol acyltransferase (MGAT) 2 catalyzes triacylglycerol (TAG) synthesis, required in intestinal fat absorption. We previously demonstrated that mice without a functional MGAT2-coding gene (Mogat2(-/-)) exhibit increased energy expenditure and resistance to obesity induced by excess calories. One critical question raised is whether lacking MGAT2 during early development is required for the metabolic phenotypes in adult mice. In this study, we found that Mogat2(-/-) pups grew slower than wild-type littermates during the suckling period. To determine whether inactivating MGAT2 in adult mice is sufficient to confer resistance to diet-induced obesity, we generated mice with an inducible Mogat2-inactivating mutation. Mice with adult-onset MGAT2 deficiency (Mogat2(AKO)) exhibited a transient decrease in food intake like Mogat2(-/-) mice when fed a high-fat diet and a moderate increase in energy expenditure after acclimatization. They gained less weight than littermate controls, but the difference was smaller than that between wild-type and Mogat2(-/-) mice. The moderate reduction in weight gain was associated with reduced hepatic TAG and improved glucose tolerance. Similar protective effects were also observed in mice that had gained weight on a high-fat diet before inactivating MGAT2. These findings suggest that adult-onset MGAT2 deficiency mitigates metabolic disorders induced by high-fat feeding and that MGAT2 modulates early postnatal nutrition and may program metabolism later in life.

as an intestinal MGAT mediating fat absorption, constitutive global inactivation of the enzyme, through germ-line transmission of a null mutation in Mogat2 , greatly reduces intestinal MGAT activity and delays fat absorption ( 7 ). Interestingly, these Mogat2 Ϫ / Ϫ mice absorb a normal quantity of fat but are protected from obesity and other metabolic disorders induced by high-fat feeding. The underlying physiological mechanisms involve a transient decrease in food intake and a persistent increase in energy expenditure ( 7,8 ). Unexpectedly, the increase in energy expenditure does not require high-fat feeding, and MGAT2 defi ciency also protects Agouti mice from excess weight gain ( 9 ). Findings from both gain-and loss-of-function mouse models indicate that MGAT2 in the intestine is a major contributor but incompletely accounts for the Abstract Acyl-CoA:monoacylglycerol acyltransferase (MGAT) 2 catalyzes triacylglycerol (TAG) synthesis, required in intestinal fat absorption. We previously demonstrated that mice without a functional MGAT2-coding gene ( Mogat2 ؊ / ؊ ) exhibit increased energy expenditure and resistance to obesity induced by excess calories. One critical question raised is whether lacking MGAT2 during early development is required for the metabolic phenotypes in adult mice. In this study, we found that Mogat2 ؊ / ؊ pups grew slower than wildtype littermates during the suckling period. To determine whether inactivating MGAT2 in adult mice is suffi cient to confer resistance to diet-induced obesity, we generated mice with an inducible Mogat2 -inactivating mutation. Mice with adult-onset MGAT2 defi ciency ( Mogat2 AKO

) exhibited a transient decrease in food intake like Mogat2
؊ / ؊ mice when fed a high-fat diet and a moderate increase in energy expenditure after acclimatization. They gained less weight than littermate controls, but the difference was smaller than that between wild-type and Mogat2 ؊ / ؊ mice. The moderate reduction in weight gain was associated with reduced hepatic TAG and improved glucose tolerance. Similar protective effects were also observed in mice that had gained weight on a high-fat diet before inactivating MGAT2. These fi ndings suggest that adult-onset MGAT2 defi ciency mitigates metabolic disorders induced by high-fat feeding and that MGAT2 modulates early postnatal nutrition and may program metabolism later in life. Supplementary key words triacylglycerol • dietary fat • fat absorption • energy expenditure • metabolic programming • monoacylglycerol acyltransferase 2 • coenzyme A Triacylglycerol (TAG) synthesis is crucial for many physiological processes involving storage or delivery of fatty ac-Upon treatment with the estrogen analog tamoxifen, Cre translocates to the nucleus and excises exon 2 of the Mogat2 gene in Mogat2 im mice, resulting in inactivation of MGAT2 ( Mogat2 AKO mice). Mice were housed on a 12 h light/dark cycle. Weighing of mice and changes of diets and cages were performed between 3 PM and 6 PM. All animal procedures were approved by the University of Wisconsin-Madison Animal Care and Use Committee and were conducted in conformity with the Public Health Service Policy on Humane Care and Use of Laboratory Animals.

Tamoxifen treatment to inactivate MGAT2
To inactivate MGAT2, adult Mogat2 im and Mogat2 f/f control mice were intragastrically administered a daily dose of tamoxifen (MP Biomedicals; 200 mg/kg body weight/day; 20 mg/ml in corn oil) for 5 days either at 4 months of age (prior to high-fat feeding) or at 5 months of age (after 7 weeks of high-fat feeding). Because tamoxifen treatment causes a transient decrease in body weight (data not shown), we allowed mice to recover for 1 month before performing energy balance experiments. Tamoxifeninduced weight loss was also observed in diet-induced obese mice .

Genotyping
Genotypes of mice were determined by PCR. To determine the presence of the Cre-recombinase transgene, the following four primers were used: Cre forward, 5 ′ -CCCGGCAAAACAGG-TAGTTA-3 ′ ; Cre reverse, 5 ′ -TGCCAGGATCAGGGTTAAG-3 ′ ; Positive control forward, 5 ′ -CCTTTAGCCTGGTCTAGGCA-GAG-3 ′ ; and Positive control reverse, 5 ′ -CAGCAAAGC CCCC-TCCTGAATCTCTC-3 ′ . This reaction produces a 194 bp amplicon and a 381 bp amplicon from the transgene and the internal control gene, respectively. To determine the presence of the wildtype or fl oxed Mogat2 allele, the following two primers were used: Mogat2 forward, 5 ′ -GTATGCCACCTGGTGGTAC-3 ′ ; and Mogat2 reverse, 5 ′ -GCAGTCCTATACCAGTACAG-3 ′ . This reaction produces a 478 bp amplicon and a 512 bp amplicon from the wildtype allele and the allele with the addition of a 34 bp loxP site, respectively. To confi rm Cre-recombinase-mediated deletion of Mogat2 following tamoxifen treatment we regenotyped mice using the same Mogat2 primers plus an additional primer, Target forward, 5 ′ -GAACTTCGTCGAGATAACTTCGT-3 ′ , specifi c for a region upstream of the fl oxed allele. This reaction produces 512 and 1,089 bp amplicons prior to excision, and a 191 bp amplicon after excision.

Diets
Mice were fed a complete, fi xed-formula chow (#8604; Teklad, Madison, WI). A series of semipurifi ed (defi ned) diets containing 10, 45, or 60% calories from fat (D12450B, D12451, and D12492; Research Diets, New Brunswick, NJ) were used to examine the effect of dietary fat on food intake and energy expenditure, as indicated. These defi ned diets contained 20% calories from protein (casein) and fi xed amounts of micronutrients and fi ber per calorie, but they varied in metabolizable energy (3.8, 4.7, or 5.2 kcal/g), corresponding to the fat content.

Real-time quantitative PCR analysis
The levels of Mogat2 mRNA in tissues collected from fetuses of embryonic day 19 or mice fed the 60 kcal% fat diet for 10 weeks were assessed as previously described ( 8 ). Cyclophilin B ( Cypb ) expression was used as an internal control. The primer sequences of the Cypb gene were 5 ′ TGCCGGAGTCGACAATGAT-3 ′ (forward) and 5 ′ -TGGAAGAGCACCAAGACAGACA-3 ′ (reverse). The primers to differences in body weight ( 8,10 ), suggesting complex mechanisms underlying the role of MGAT2 in the regulation of systemic energy balance.
MGAT2 in mice is expressed in embryos as early as 7 days postcoitus (dpc) and in the liver and intestine during the perinatal period [NCBI UniGeneID:307396, Gene Expression Omnibus GDS3764, and ( 11,12 )], suggesting a potential role of MGAT2 during early development. Several lines of evidence in both human and rodent studies support the concept that nutritional status during early development, including embryonic and fetal as well as early postnatal periods, programs metabolism and propensity to gain weight later in life (13)(14)(15)(16)(17)(18)(19). For example, in a historical cohort study, exposure to the Dutch famine of 1944-1945 during the fi rst half of pregnancy is associated with increased obesity rates in early adulthood, while exposure during the last trimester and the fi rst months of life is associated with reduced obesity rates ( 20 ). In mice, early postnatal undernutrition limits adiposity induced by high-fat diet in adulthood ( 21 ). Thus, we question whether MGAT2 determines metabolic effi ciency during early development and, more importantly, whether lacking MGAT2 during those critical periods is required for the protective effects against metabolic disorders seen in adult Mogat2 Ϫ / Ϫ mice.
In this study, to determine whether MGAT2 plays a role in overall energy balance during the intrauterine and the suckling period, we examined body weights of fetuses at late gestation and of pups before weaning. In addition, to determine whether adult-onset MGAT2 defi ciency is sufficient to modulate the responses to high-fat feeding, we developed inducible knockout mice in which MGAT2 could be ablated at specifi c developmental stages and examined their metabolic phenotypes in response to highfat feeding.

Mice
Germ-line MGAT2 knockout ( Mogat2 Ϫ / Ϫ ) mice with constitutive and global MGAT2 defi ciency were produced as described ( 7 ). To enable inactivation of Mogat2 in adult mice, a line of inducible MGAT2 knockout mice was generated using the Cre-loxP system with a ubiquitously expressed Cre recombinase that remains inactive until induction by tamoxifen treatment ( 22 ).

Mogat2
f/f mice carrying both copies of "fl oxed" Mogat2 alleles with exon 2 fl anked by Cre-recombinase target sites (loxP) ( 8 ) were crossed with a line of transgenic mice ubiquitously expressing a tamoxifen-inducible Cre recombinase [B6;Cg-Tg(UBCcre/ERT2)1Ejb/J ( 23 )]. Under the control of the ubiquitin C promoter, the inducible Cre recombinase is a fusion protein between Cre and a mutant form of the estrogen receptor, sequestering the enzyme in the endoplasmic reticulum. Cre activity is inducible by tamoxifen but not estrogen ( 23 ), which in turn mediates the precise excision of exon 2 of the Mogat2 gene fl anked by two loxP sites, switching off the functional gene. Through cross-breeding, inducible mutant ( Mogat2 im ) mice carrying "fl oxed" Mogat2 alleles and one copy of the inducible Cre recombinase and their Mogat2 f/f littermate controls were produced. independent samples at least twice to confi rm reproducibility of the results. For comparisons between two groups, Student's t -tests were used. Differences measured over time were compared using repeated-measures two-way ANOVA to determine main effects of, and interactions between, time and genotype. To identify group differences, Bonferroni's posttests were performed. Analyses were conducted using GraphPad Prism statistical analysis software (version 5.01; GraphPad Software, La Jolla, CA).

Mouse pups with constitutive Mogat2 inactivation gain less weight during the suckling period
To explore a potential role of MGAT2 during early development, we fi rst assessed if its coding gene Mogat2 is expressed prenatally in fetal tissues collected on embryonic day 19 (dpc) using quantitative PCR. We found high levels of MGAT2 mRNA in intestine, similar to in adult mice. Relatively moderate levels of MGAT2 were also found in yolk sac, liver, and kidney, but not in placenta ( Fig. 1A ). We next compared body weights of fetuses at a late stage of gestation (19 dpc) from intercrosses of heterozygous mice ( Mogat2 +/ Ϫ ), bearing a constitutively null Mogat2 allele. We found no difference in intrauterine growth between wild-type, Mogat2 +/ Ϫ , and Mogat2 Ϫ / Ϫ fetuses ( Fig. 1B ), suggesting that MGAT2 does not play an indispensable role in the regulation of energy balance during the prenatal period. In contrast, after birth, pups with the germ-line null mutation ( Mogat2 Ϫ / Ϫ ) gained less weight than wild-type littermates throughout the suckling period, when pups nursed on milk rich in fat [ ‫ف‬ 20% by weight ( 25 )] ( Fig. 1C ). At weaning, their average weight was ‫ف‬ 80% that of controls. Consistent with previous reports ( 7,9 ), after weaning onto a standard low-fat chow, these mice exhibited compensatory growth and were able to reach a body weight similar to wild-type littermates by 3 months of age ( Fig. 1D ). When their feed was switched to a diet containing 60% calories from fat, wild-type mice gained signifi cantly more weight than did Mogat2 Ϫ / Ϫ littermates ( Fig. 1E ). During 10 weeks of high-fat feeding, wild-type mice gained on average 20.2 ± 1.6 g while Mogat2 Ϫ / Ϫ mice gained only 4.7 ± 0.3 g ( Fig. 1E ).

Generation of mice with tamoxifen-inducible Mogat2 inactivation
To determine whether MGAT2 defi ciency during early development is required for the obesity-resistant phenotype seen in Mogat2 Ϫ / Ϫ mice, we developed a mouse model in which Mogat2 can be inactivated in adulthood. Mice carrying a tamoxifen-inducible ubiquitin-Cre recombinase were crossed with mice carrying two loxP sites fl anking exon 2 of the Mogat2 gene ( 8 ). The offspring were intercrossed to generate mice carrying homozygous "fl oxed" Mogat2 alleles ( Mogat2 f/f ), and half of them were also hemizygous , carrying a copy of the transgene expressing the inducible Cre recombinase ( Mogat2 f/f , UBC-Cre + ). Before induction, these inducible mutants are designated Mogat2 im mice.

In vitro MGAT assays
MGAT activity assays were performed with total tissue homogenates as previously described ( 10 ). Reactions were started by adding total tissue homogenates to the assay mixture and were stopped by adding chloroform-methanol (2:1, v/v). The lipids were extracted, dried, and separated by TLC on silica gel G-60 TLC plates with hexane-diethyl ether-acetic acid (80:20:1, v/v/v). Lipid bands were visualized with iodine vapor, and products were identifi ed by comparison with the migration of lipid standards. The incorporation of radioactive substrates into lipid products was also visualized by an imaging scanner (Typhoon FLA 7000; GE Healthcare Life Sciences, Piscataway, NJ) followed by band scraping and counting in a scintillation counter analyzer (Packard Tri-Carb 2200 CA Liquid Scintillation Counter Analyzer; PerkinElmer, Santa Clara, CA).

Metabolic phenotyping studies
To assess phenotypes related to acute energy balance, mice were housed in a metabolic phenotyping system with housing and wood chip bedding similar to the home cage environment (LabMaster modular animal monitoring system; TSE, Chesterfi eld, MO). Adult male mice were acclimated to individual housing and metabolic cages for 1 week before experiments and were fed indicated diets sequentially for 3 days each. Data collection and analysis were performed as previously reported ( 9, 10 ).

Body weight response to chow or high-fat feeding
We monitored weight gains daily before weaning and weekly afterward to assess long-term energy balance. To examine long-term body weight responses to low-fat chow and high-fat diet, male mice were fed a regular mixed meal chow at weaning (3 weeks) and then switched to a 60% diet at 4 months of age and for an additional 10 weeks. To investigate the effect of MGAT2 inactivation in dietinduced obese mice, male mice were switched to the 60 kcal% fat diet at 2-3 months of age, prior to inactivation of MGAT2.

Biochemical assays
Plasma lipids were assayed using enzymatic kits (Infi nity TM Triglycerides Lipid Stable Reagent, Thermo Fisher; Cholesterol E, Non Esterifi ed Fatty Acid assay, Wako Diagnostics). Hepatic and fecal lipids were extracted in chloroform-methanol (2:1, v/v) following homogenization in PBS. Blood samples were collected from the orbital plexus, and plasma was separated by centrifugation. Hepatic TAG and glycogen were measured as described previously ( 8 ). Protein concentration was measured by Pierce BCA Protein Assay Kit (Thermo, Rockford, IL). To examine glucose metabolism following high-fat feeding, mice were fed 60 kcal% fat diet for ‫ف‬ 8 weeks, then a glucose tolerance test was performed. Briefl y, male mice were fasted for 6 h beginning at 7 AM and then injected with glucose (1 g/kg bodyweight, ip, 10% glucose in PBS). Blood glucose was measured immediately before and at defi ned intervals after glucose injection using a handheld glucose monitor (OneTouch Ultra; LifeScan Inc., Milpitas, CA). Plasma insulin was measured by ELISA (Crystal Chem Inc., Downers Grove, IL).

Statistical analyses
All data are presented as mean ± SEM. P < 0.05 was considered statistically signifi cant. Each experiment was performed with tamoxifen treatment was effi cient in activating Cre recombinase and causing Mogat2 ablation in the intestinal stem cells, which normally replace intestinal epithelia every 3-5 days ( 26 ).
Mogat2 mRNA levels correlated with MGAT activity in the small intestine. Before tamoxifen treatment, Mogat2 f/f and Mogat2 im mice showed MGAT activity similar to that of wild-type mice ( Fig. 2B ). Ablation of MGAT2 expression in adulthood by tamoxifen treatment reduced intestinal MGAT activity in Mogat2 AKO mice to levels similar to those in constitutive germ-line Mogat2 Ϫ / Ϫ mice ( Fig. 2B ). These results indicate that tamoxifen treatment effectively transforms adult inducible mutant Mogat2 im mice into adultonset MGAT2-defi cient Mogat2 AKO mice.

Before inactivation of MGAT2, inducible mutant mice exhibit normal energy balance and gain weight like controls
Before tamoxifen treatment, the inducible mutant Mo-gat2 im pups gained weight normally during the suckling period like their Mogat2 f/f littermates ( Fig. 3A ). After weaning, when mice were fed a low-fat chow, these mice continued to grow at a similar rate as control mice did into To inactivate Mogat2 in adulthood, we treated Mogat2 im mice with tamoxifen at 12 weeks of age. Tamoxifen treatment resulted in deletion of Mogat2 exon 2 in Mogat2 im mice, but not in their Mogat2 f/f littermate controls, as confi rmed by PCR using genomic DNA (data not shown). These mice with Mogat2 inactivated in adulthood by tamoxifen-mediated Cre-recombinase induction were designated "adult-onset knockout" ( Mogat2 AKO ). To assess the specifi city as well as effi ciency of Mogat2 ablation, we measured Mogat2 mRNA expression level in these mice. Before tamoxifen treatment, both Mogat2 f/f and Mogat2 im mice expressed levels of Mogat2 mRNA in small intestine similar to that found in wild-type mice ( Fig. 2A ). As reported previously ( 7,8 ), Mogat2 Ϫ / Ϫ mice had no detectable Mogat2 mRNA ( Fig. 2A ). In contrast, following tamoxifen treatment, Mogat2 f/f mice maintained their Mogat2 mRNA levels while Mogat2 AKO mice showed very low levels of expression in the small intestine ( Fig. 2A ). Similar relative expression levels among genotypes were also found in the kidney and white adipose tissue (supplementary Fig. 1), confi rming the effi ciency of Mogat2 deletion by the inducible Cre recombinase. Tissues examined were collected from mice after 10 weeks of high-fat feeding. The lasting MGAT2 defi ciency indicates that the same four diets. Mogat2 AKO mice were produced by treating 3-month-old chow-fed Mogat2 im mice with tamoxifen to induce Mogat2 deletion. As controls, their Mogat2 f/f littermates were also subjected to the same treatment. Indirect calorimetry was performed 1 month after treatment to allow for recovery of stable body weight. When fed chow or the semipurifi ed diet containing 10 kcal% fat, Mogat2 AKO mice exhibited levels of food intake, respiratory exchange ratio (RER), and energy expenditure similar to those of Mogat2 f/f controls ( Fig. 4A -C ). Though all levels trended higher in Mogat2 AKO mice, the differences did not reach statistical signifi cance. When fi rst exposed to high-fat diets (45 kcal% and 60 kcal% fat for 3 days each), Mogat2 AKO mice reduced food intake signifi cantly as compared with Mogat2 f/f controls ( Fig. 4A ). The difference disappeared after 1 week of high-fat feeding ( Fig. 4A ), similar to that reported in mice with constitutive global MGAT2 deficiency as well as those with intestine-specifi c defi ciency ( 8,9 ). Associated with decreases in food intake, Mogat2 AKO mice exhibited a lower RER than Mogat2 f/f controls when fi rst exposed to high-fat feeding ( Fig. 4B ), suggesting an increase in fat oxidation and/or a decrease in fat accretion. Mogat2 AKO mice did not exhibit signifi cant elevations in energy expenditure until they were acclimated to highfat feeding. They showed a signifi cant 13% increase in 24 h energy expenditure compared with Mogat2 f/f mice when food intake was similar between groups ( Fig. 4A, C ).
Mogat2 AKO and Mogat2 f/f mice did not differ in energy balance, as indicated by similar body weights when fed chow or the 10% fat diet ( Fig. 4D ). When fed the high-fat diets, Mogat2 AKO mice maintained their body weight, while Mogat2 f/f mice gained weight during the metabolic chamber studies ( Fig. 4D ). After acclimatization to high-fat feeding, Mogat2 AKO mice absorbed similar amounts of dietary fat as Mogat2 f/f mice did, as indicated by similar levels of fecal output and fecal fat ( Fig. 4E ).
The differences in body weights increased throughout the following 10 weeks of high-fat feeding ( Fig. 5A ). Mo-gat2 AKO mice gained only 65% as much as Mogat2 f/f controls (12 g vs. 19 g, respectively). The differences were largely due to differences in fat mass. White and brown adipose tissues were signifi cantly smaller in Mogat2 AKO mice compared with Mogat2 f/f controls (both ‫ف‬ 70% of controls; Fig. 5B ). Lean mass, represented by calf and heart, was not different across genotypes ( Fig. 5B ).

Adult-onset MGAT2 defi ciency protects mice from diet-induced hepatic steatosis and glucose intolerance
Constitutive MGAT2 defi ciency protects Mogat2 Ϫ / Ϫ mice from several metabolic disorders induced by high-fat feeding ( 7 ). Thus, we next examined whether inactivation of MGAT2 in adulthood was suffi cient to protect Mogat2 AKO mice from comorbidities associated with obesity. After 10 weeks of high-fat feeding, Mogat2 AKO mice had similar levels of TAG and free fatty acids but a moderately lower level of total cholesterol in fasting plasma, compared with Mogat2 f/f mice ( Fig. 6A -C ). Like wild-type mice, after longterm high-fat feeding, Mogat2 f/f mice developed hepatic adulthood ( Fig. 3B ). To further confi rm that the inducible Cre recombinase does not affect energy metabolism prior to induction, we characterized acute energy balance of these mice in response to a regular low-fat chow diet and three semipurifi ed diets containing 10, 45, or 60% of calories from fat sequentially by indirect calorimetry. Both food intake and energy expenditure, refl ected by oxygen consumption (VO 2 ), were similar between Mogat2 f/f and Mogat2 im mice ( Fig. 3C, D ), resulting in parallel body weight changes during each of the diet treatments ( Fig.  3E ). Both groups of mice gained 2.5 g of body weight after being fed the semipurifi ed diets for 9 days. Consistently, both Mogat2 f/f and Mogat2 im mice gained on average 22.3 g of excess body weight after 10 weeks of high-fat feeding ( Fig. 3F ). These data indicate that expressing UBC-Cre recombinase has no effect on energy metabolism before tamoxifen-induced activation.

Adult-onset MGAT2 defi ciency modulates energy balance and protects mice against diet-induced weight gain
We next determined the effect of MGAT2 inactivation in adulthood on acute energy balance in response to the Ϫ

Inactivation of MGAT2 in obese mice reduces body weight and improves glucose tolerance
To examine the effects of inactivating MGAT2 in dietinduced obese mice, we fed 3-month-old adult Mogat2 f/f and Mogat2 im mice a high-fat diet for 7 weeks. Both Mogat2 f/f and inducible Mogat2 im mice gained weight before tamoxifen treatment ( Fig. 7A ). During the 5-day tamoxifen treatment, both groups of mice lost similar amounts of body weight. Whereas Mogat2 f/f mice resumed high-fat dietinduced weight gain within 2 weeks of tamoxifen treatment, Mogat2 AKO mice continued to lose weight for an additional week. Mogat2 AKO mice weighed significantly less than Mogat2 f/f littermates, and the difference increased throughout the remainder of the study ( Fig. 7A ). The reduction in steatosis as indicated by increased liver mass and TAG content ( Fig. 6D, E ). Meanwhile, Mogat2 AKO mice were protected and had smaller livers and lower TAG content than Mogat2 f/f controls. Hepatic glycogen was not different across genotypes ( Fig. 6F ). In parallel with accumulation of liver TAG, Mogat2 f/f mice developed impaired glucose tolerance, while Mogat2 AKO mice were protected. After 8 weeks of highfat feeding, Mogat2 AKO mice had lower fasting glucose than Mogat2 f/f controls (191 mg/dl vs. 238 mg/dl) and a blunted increase in blood glucose following an intraperitoneal glucose challenge (AUC, 80% of controls; Fig. 6G ). Further, Mogat2 AKO mice had lower plasma insulin concentrations right before as well as after glucose challenge ( Fig. 6H ), which was consistent with greater insulin sensitivity. , green circles) mice during suckling (A) and chow feeding (B). n = 9-19 mice per group. C-E: Three-monthold mice sequentially fed chow or defi ned diets containing 10, 45, or 60% calories from fat for 3 days per diet. Cumulative food intake (C) and oxygen consumption rates (D) adjusted for baseline body weights of each mouse at the start of each diet treatment. Data from each mouse were pooled from the same time of the day of the same diet treatment. Graphs represent average days. Gray areas mark dark phase of the light cycle (6 PM to 6 AM). E: Body weight of mice during 12-day metabolic phenotyping experiment. n = 8 per group. F: Body weight of mice during 10 weeks of high-fat feeding. n = 9 or 13. 12 mg/dl) and insulin levels (1.9 ± 0.4 ng/ml vs. 2.5 ± 0.5 ng/ml), respectively. Response to an oral glucose challenge was similar in both groups of mice ( Fig. 7C , left panel). When the same test was administered 3 weeks after Mogat2 inactivation, both levels of fasting blood glucose and insulin were lower in Mogat2 AKO than those of controls ( Fig.  7C , middle panel; data not shown). In addition, the blood weight gain was associated with signifi cantly reduced liver and white adipose mass ( Fig. 7B ). Lean mass (as indicated by calf and heart) was not different between genotypes ( Fig. 7B ).
After 6 weeks of high-fat feeding, glucose metabolism was impaired in both Mogat2 im and Mogat2 f/f mice as illustrated by high fasting glucose (219 ± 9 mg/dl vs. 218 ± Fig. 4. Inactivation of MGAT2 in adulthood alters energy balance upon high-fat feeding. A-C: Fourmonth-old tamoxifen-treated Mogat2 f/f ( M2 f/f , gray) and tamoxifen-treated adult-onset MGAT2-defi cient ( M2 AKO , red) mice sequentially fed chow or defi ned diets containing 10, 45, or 60% calories from fat for 3 days per diet. Mice were fed the 60 kcal% diet for an additional 6 days. Cumulative food intake (A), RER (B), and oxygen consumption rates (C) adjusted for baseline body weights of each mouse at the start of each diet treatment. Data from each mouse were pooled from the same time of the day of the same diet treatment. Graphs represent average days. Gray areas mark dark phase of the light cycle (6 PM to 6 AM). D: Body weight of mice during 18-day metabolic phenotyping experiment. n = 7 or 6. E: Fecal output and lipid content of mice after acclimatization to high-fat feeding. Error bars represent SEM. Error bars not shown are smaller than the symbols. * P < 0.05 versus littermate controls by t -test [total cumulative mass for food intake, 24 h average for RER, and area under the curve (AUC) for VO 2 ]. # P < 0.05 versus littermate controls by repeatedmeasures ANOVA. adulthood is suffi cient to decrease metabolic effi ciency and the propensity to gain weight upon high-fat feeding. Before inducing the deletion of MGAT2, Mogat2 im mice exhibited energy balance similar to wild-type mice during the suckling period as well as upon high-fat feeding. After the ablation of MGAT2 in adulthood, Mogat2 AKO mice exhibited a transient decrease in food intake when fed a high-fat diet and a persistent increase in energy expenditure after acclimatization to high-fat feeding. These fi ndings indicate that inactivating glucose levels in Mogat2 AKO mice did not rise as high as in tamoxifen-treated Mogat2 f/f controls ( Fig. 7C , middle panel). After adjusting for blood glucose levels right before the challenge, the net AUC in Mogat2 AKO mice was 66% of that in controls ( Fig. 7C , middle panel, inset). The differences were even more pronounced after prolonged high-fat feeding (10 weeks after Mogat2 inactivation). Blood glucose levels rose to a lower level and declined faster in Mogat2 AKO mice than in Mogat2 f/f controls ( Fig.  7C , right panel).

DISCUSSION
We have previously reported that constitutive inactivation of MGAT2 through a null mutation in the germ line protects Mogat2 Ϫ / Ϫ mice against obesity and associated metabolic disorders induced by high-fat feeding as well as by the Agouti mutation ( 7,9 ). In this study, we showed that Mogat2 Ϫ / Ϫ mice experience undernutrition during the early postnatal period, as suckling Mogat2 Ϫ / Ϫ pups gained less weight than their wildtype littermates. To determine whether MGAT2 defi ciency during early development is essential for the effects on energy balance, we generated mice with adult-onset MGAT2 defi ciency. We found that MGAT2 inactivation during  , red) mice fed 60 kcal% fat diet for 10 weeks. n = 6 or 9. Bars represent mean ± SEM. Blood glucose (G) and plasma insulin (H) in 8-week high-fat fed mice before and at indicated times after an intraperitoneal injection of glucose (1 mg/g body weight, 10% glucose in PBS). n = 9 or 11 per group. * P < 0.05 versus littermate controls by t -test (mean for plasma cholesterol, liver mass and TAG, and AUC for blood glucose). ᭜ P < 0.05 versus time-matched littermate controls by t -test. postnatal life. However, the null mutation of Mogat2 in the germ line did not lead to apparent intrauterine growth defects, as Mogat2 Ϫ / Ϫ fetuses appeared normal and had body weights similar to wild-type and heterozygous littermates at E19. The role of MGAT2 during embryogenesis, if any, appears dispensable. In contrast, MGAT2 is required for normal growth in early postnatal life. Mogat2 Ϫ / Ϫ pups grew slower than their wild-type littermates when pups relied on fat-rich mouse milk for calories, consistent with its role in enhancing metabolic effi ciency upon highfat feeding ( 7,9 ).
The effects of nutrition during early postnatal development on long-term energy balance have been demonstrated in several mammalian species ( 16 ). Thus, limited weight gain during the suckling period and the compensatory growth after weaning seen in Mogat2 Ϫ / Ϫ mice could contribute to the obesity-resistance observed later in life. More importantly, from the prospective of designing an MGAT2 in adult mice is suffi cient to confer resistance to obesity induced by diet. However, the protection against weight gain observed in Mogat2 AKO mice was to a lesser extent than in Mogat2 Ϫ / Ϫ mice, raising the possibility that the early postnatal nutrition status may modulate energy balance phenotypes observed in Mogat2 Ϫ / Ϫ mice later in life.
The expression of Mogat2 in mice has been reported during early development in 7-day embryos [RIKEN fulllength enriched library ( 27 ); GenBank accession number: AK049560.1]. Its expression in fetal intestine is relatively low 6 days before birth but reaches levels as high as those in adults during perinatal periods ( 12 ). In tissues from E19 fetuses, we found a high level of Mogat2 expression in the intestine ( Fig. 1A ), consistent with a role of MGAT2 in mediating the absorption of milk fat soon after birth. Interestingly, we also found Mogat2 expression in the yolk sac, which plays an important role in supplying nutrients for early embryos ( 28 ), analogous to the intestine in  phenotypes intermediate between wild-type and global knockout mice ( 8,10 ). These possibilities remain to be tested experimentally. Interestingly, the difference in energy expenditure between Mogat2 AKO mice and their littermate controls ( Fig. 4C ) appeared smaller than the difference between Mogat2 Ϫ / Ϫ and wild-type littermates ( 8 ), which was consistent with the observation that Mogat2 AKO mice were protected from diet-induced weight gain to a lesser extent than were Mogat2 Ϫ / Ϫ mice ( Fig. 1E and Fig. 5A ). These fi ndings suggest that lacking MGAT2 during early development may have lasting effects later in life, including increases in energy expenditure. The underlying mechanisms might involve the hypothalamic pathways controlling energy balance, as implicated by fi ndings from studies in which mouse pups are raised in varying litter sizes as well as pups nursed by dams fed different fatty acids ( 15,29,30 ). On the other hand, even though the levels of Mogat2 mRNA and MGAT activity in tissues from Mogat2 AKO mice examined were reduced to levels similar to tissues from Mogat2 Ϫ / Ϫ mice, we cannot exclude the possibility that some minor population of cells could retain MGAT2 expression and contribute to the decreased phenotypic differences.
Despite an intermediate effect on prevention of dietinduced weight gain, inactivation of Mogat2 in adulthood in mice had robust effects on preventing hepatic TAG accumulation. The moderate reduction in weight gain was also associated with improved glucose homeostasis, as indicated by lower levels of fasting glucose and insulin as well as enhanced glucose tolerance compared with controls ( Fig. 6 ), similar to that seen in Mogat2 Ϫ / Ϫ mice ( 7,8 ). It is unclear if the role of MGAT2 in hepatic lipid metabolism and glucose homeostasis is completely dependent on its effect on systemic energy balance.
Acute inactivation of MGAT2 in mice that had already gained weight on a high-fat diet also diminished their ability to gain more weight and enhanced their glucose tolerance ( Fig. 7 ); their blood glucose peaked at a lower level, and it declined faster than those of controls after being challenged intragastrically with the same dose of glucose. Further, the differences in glucose tolerance were exacerbated over time. In concert with previous fi ndings that Mo-gat2 IKO mice have improved glucose homeostasis ( 8 ), these fi ndings indicate that acute inactivation of intestinal MGAT2 may enhance glucose homeostasis, similar to immediate metabolic improvements following some of the bariatric procedures ( 31 ).
In summary, our fi ndings indicate that MGAT2 enhances metabolic effi ciency and promotes positive energy balance during early development in mice. Nonetheless, inactivation of MGAT2 in adulthood is suffi cient to decrease the propensity to gain weight and prevent glucose intolerance when challenged with a calorie-dense diet, even in mice that had already accumulated excess body fat. These fi ndings imply the potential effi cacy of MGAT2 as an intervention target for combating obesity and related metabolic diseases caused by chronic caloric surplus. However, the reduction in weight gain is diminished as compared with mice born intervention, these observations raised the question whether inactivating MGAT2 in adulthood can have the same protective effects. Our data suggest that lacking MGAT2 during early development could contribute to obesity resistance later in life; nonetheless, inactivating MGAT2 in adulthood is suffi cient to reduce metabolic effi ciency and protect mice from metabolic disorders induced by high-fat feeding.
The energy balance phenotypes of Mogat2 AKO mice resembled those of Mogat2 Ϫ / Ϫ mice as well as mice with an intestine-specifi c MGAT2 defi ciency ( Mogat2 IKO ) ( 7,8 ). When fed chow or the semipurifi ed diet containing 10 kcal% fat, these MGAT2-defi cient mice tended to show increased energy expenditure as well as food intake. Compared with their respective controls, the increases were less pronounced in Mogat2 AKO and Mogat2 IKO mice than in Mogat2 Ϫ / Ϫ mice; however, all maintained energy balance similar to controls, as refl ected by similar body weights during short-term diet challenge in the metabolic chambers. The marked increase in energy expenditure protects Mogat2 Ϫ / Ϫ mice against weight gain-induced by a high-refined carbohydrate diet as well as by the Agouti mutation without high-fat feeding ( 9 ). It remains to be determined if adult-onset MGAT2 defi ciency also reduces propensity to gain weight independent of fat content in diet.
When switched to the high-fat diets, Mogat2 AKO mice, like Mogat2 IKO and Mogat2 Ϫ / Ϫ mice, were able to adjust to the high caloric content and reduced food intake, with a pronounced reduction on the fi rst day [ Fig. 4A ( 8 ) and data not shown]. These fi ndings suggest that MGAT2 in the intestine likely modulates short-term food intake regulation in the hypothalamus. The reduction in food intake of Mogat2 AKO mice contributed to the immediate differences in the initial weight gain to the same extent as observed in Mogat2 IKO and Mogat2 Ϫ / Ϫ mice ( 8 ). As control mice acclimatized to high-fat feeding and decreased their food intake, the differences in food intake disappeared ( Fig. 4A ). In the meantime, the increase in energy expenditure of Mogat2 AKO mice over control littermates became signifi cant and could explain the increasing differences in weight gain over time.
The molecular mechanisms underlying the effects of inactivating MGAT2 in adults on long-term energy balance are not clear. Intestinal MGAT2 is likely a major contributor. Its expression levels determine the rate of MAG uptake and esterifi cation in enterocytes ( 8,10 ). Defi ciency of MGAT2 in the intestine leads to changes in temporal and spatial distribution of fat absorption, which may result in blunted postprandial triglyceridemia as seen in both Mogat2 Ϫ / Ϫ and Mogat2 IKO mice ( 7,8,10 ). The associated changes in neural and hormonal signals may also contribute to the alterations in food intake, partitioning of energy-yielding nutrients, and systemic energy metabolism. Nonetheless, MGAT2 expressed in other tissues, such as the adipose tissues, may further contribute to the energy balance phenotypes, because both mice expressing MGAT2 and those lacking MGAT2 in an intestine-specifi c manner exhibit an energy balance without the enzyme, suggesting that MGAT2 can modulate nutritional status in early postnatal life and its deficiency during critical periods may program metabolism later in life.
The authors thank Dr. Guy Groblewski for insightful comments.