Quantitative trait loci (QTL) for carbohydrate (Mnic1) and total energy (Kcal2) intake on proximal mouse chromosome 17 were identified previously from a C57BL/6J (B6) X CAST/Ei (CAST) intercross. Here we report that a new congenic strain developed in our laboratory has confirmed this complex locus by recapitulating the original linked phenotypes: B6.CAST-17 homozygous congenic mice consumed more carbohydrate (27%) and total energy (17%) compared with littermate wild-type mice. Positional gene candidates with relevance to carbohydrate metabolism, glyoxalase I (Glo1) and glucagon-like peptide-1 receptor (Glp1r), were evaluated. Glo1 expression was upregulated in liver and hypothalamus of congenic mice when compared with B6 mice. Analyses of Glp1r mRNA and protein expression revealed tissue-specific strain differences in pancreas (congenic>B6) and stomach (B6>congenic). These results suggest the possibility of separate mechanisms for enhanced insulin synthesis and gastric accommodation in the presence of high carbohydrate intake and larger food volume, respectively. Sequence analysis of Glp1r found a G insert at nt position 1349, which results in earlier termination of the open reading frame, thus revealing an error in the public sequence. Consequently, the predicted length of GLP-1R is 463 aa compared with 489 aa, as previously reported. Also, we found a polymorphism in Glp1r between parental strains that alters the amino acid sequence. Variation in Glp1r could influence nutrient intake in this model through changes in the regulatory or protein coding regions of the gene. These congenic mice offer a powerful tool for investigating gene interactions in the control of food intake.
- genetic linkage
- food intake
inbred strains of mice show marked variation in nutrient intake in self-selection paradigms (33). For example, one study has shown that AKR/J, NZB/B1NJ, and C57BL/6J mice selected the highest proportion of fat, whereas the CAST/Ei, BALB/cByJ, and 129/J strains chose the lowest (33). Our previous analysis of mice from a cross between the C57BL/6J (B6) and CAST/Ei (CAST) inbred strains revealed six chromosomal loci controlling complex traits [quantitative trait loci (QTL)] for dietary fat and carbohydrate consumption, and three loci for total kilocalorie (kcal) intake (34). Nutrient intake phenotypes were determined by providing the choice between two experimental diets: a fat/protein (F/P) vs. a carbohydrate/protein (C/P) mixture containing 78%/22% of energy, respectively. These were the first QTL for macronutrient or energy intake to be mapped in the mouse.
One region on proximal chromosome 17 colocalized for two significant QTL: Mnic1 for increased macronutrient intake-carbohydrate and Kcal2 for total kilocalorie intake. The genetic linkage for each of these traits was strengthened considerably when body weight was used as a covariate (the logarithm of the odds score increased from 6.0 to 6.7 and from nonsignificant to 4.9, respectively) (34), suggesting that this complex locus has effects on both food intake and body weight but with effects in opposite direction (21). Both traits were associated with the CAST allele. QTLs require confirmation in mice possessing a uniform genetic background. Thus the next step toward identifying complex trait gene(s) was to develop a congenic strain in which a CAST donor segment approximating the QTL region was introgressed onto the B6 genome. Phenotypic comparisons between congenic mice and controls from the recipient strain were then used to determine whether or not the differential locus retained its effect. We now report the development and phenotypic characterization of a B6.CAST-17 congenic strain that eats significantly more carbohydrate and total calories, thus verifying the genetic linkage for these traits.
The genetic variation underlying the QTL could consist of polymorphisms either in the coding region, thus altering amino acid sequence of the translated protein, or in the regulatory region, affecting expression of a gene. Two positional candidates were screened for genetic differences: glyoxalase I (Glo1) at 16 cM and glucagon-like peptide 1 receptor (Glp1r) at 18 cM. Both of these genes approximate the peak positions of the Mnic1 QTL at 12 cM and Kcal2 at 16 cM and have relevance to carbohydrate metabolism and nutrient intake. Glo1 encodes an enzyme that catalyzes the detoxification of methylglyoxal, a cytotoxic byproduct of carbohydrate metabolism (39). GLP-1 secreted by the intestine controls glucose homeostasis and inhibits gastric emptying (11, 15–16). The results of expression profiling and sequence analyses support Glp1r and Glo1 as good candidates for the QTL gene(s) in this model.
All animal and phenotyping protocols used in the original mapping study have been described in detail elsewhere (34). Briefly, all experimental animals, including the progenitor (CAST, B6) and congenic strains, consisted of male mice bred and maintained at the Pennington Biomedical Research Center (PBRC; Baton Rouge, LA) in a colony generated from breeding pairs obtained from the Jackson Laboratory (Bar Harbor, ME). All animal protocols were approved by the Institutional Animal Care and Use Committee of PBRC.
Congenic strain development.
A congenic line was constructed using the “speed congenics” method (43) by which a target donor locus is introgressed onto the background of a recipient strain aided by marker-assisted selection. Selection at each generation was based on the presence of desired genomic segment and the absence of contaminating donor DNA.
The marker-assisted selection protocol produces congenic strains in less than half the number of generations required by the classic protocol (N12). By S5, a “speed congenic” line has < 0.5% contaminating donor genome.
In this manner, (CASTXB6)F1 mice were backcrossed to B6 females for five generations with selection at each generation for the chromosome 17 locus from CAST using microsatellite markers (D17Mit19 to D17Mit20). An additional 65 mit markers that lie outside this segment were used to select mice with the least contaminating donor region at S2-S5. This was followed by an intercross at S6 to generate mice homozygous for the chromosome 17 CAST locus. Once the congenic strain was completed, fine mapping was performed to define the QTL segment within a 1–2 cM interval. The resulting congenic strain B6.CAST(D17Mit19-D17Mit91) (B6.CAST-17) carries a CAST donor segment of ∼38 cM on the B6 background (Fig. 1).
Generation of experimental animals.
B6.CAST-17 S6 mice, heterozygous for the CAST genomic segment, were intercrossed, and their progeny were genotyped using mit markers D17Mit19, D17Mit198, D17Mit20, D17Mit89, D17Mit91, and D17Mit71. Three littermate genotypes were produced (recombinants between any markers were excluded).
The use of littermate controls to test the phenotypic effect of a congenic segment is extremely important for overcoming the problems of background heterogeneity as well as the effects of litter, age, or other unknown environmental variables.
DNA was extracted from tail snips of mice using a phenol/chloroform extraction. PCR amplification and gel electrophoresis were performed as previously reported (34).
Phenotyping methods have been previously described in detail (34). Mice had no prior ingestive experience except for the dam's milk and laboratory chow (cat. no. 5001; LabDiets, Richmond, IN) on which they were weaned. Only progeny from litters containing 4–10 pups were used for phenotyping. Pups were weaned at 24–26 days of age and housed with siblings in same-sex groups of 1–4. At 5–6 wk of age, male mice were weighed and singly-housed in Plexiglas cages until body weight returned to baseline. At 7–10 wk of age, mice were then transferred to individual housing in stainless steel hanging cages with wire mesh floors and polyvinylchloride nesting tubes to reduce time spent on wire flooring. After 7–10 days of adaptation to these new housing conditions, animals were provided a choice between two diets presented together: F/P (vegetable shortening and casein) and C/P (corn starch, powdered sugar, and casein). The protein composition of both diets was equivalent (22% energy) and the balance of calories for each was contributed by fat or carbohydrate (78%). Both diets were supplemented with vitamins, minerals, and cellulose (see Ref. 34). Diet intake and spillage were measured daily.
For the mRNA expression studies, singly-housed 10- to 12-wk-old mice were maintained on standard rodent chow (cat. no. 5001; LabDiets, Richmond, IN) and tap water ad libitum. Food was removed 4–6 h before animals were killed with an overdose of isoflurane inhalation in the midlight period. Immediately afterward, the hypothalamus and liver were extracted and placed in liquid nitrogen. Simultaneously, the pancreas was exposed by laparotomy, quickly removed, and processed in cold TRI reagent (Molecular Research Center, Cincinnati, OH). Next, fundal (ligated at the fundal ridge) and antral stomach were rinsed with PBS, quickly frozen in liquid nitrogen, and stored at −80°C. For the Western blot analysis of antral stomach, protein was extracted from the mucosal lining based on in situ hybridization experiments identifying receptor mRNA in the gastric pits of the stomach (4).
RNA isolation and cDNA synthesis.
Total RNA was isolated from pancreas, antral stomach, hypothalamus, and liver tissues by a Trizol extraction procedure (TRI reagent; Molecular Research Center). The extracted RNA was incubated with DNAse I to digest traces of genomic DNA and purified using the RNeasy kit (Qiagen). RNA was quantified by spectrophotometry (model DU650; Beckman Coulter) and its integrity was checked on 1.5% formaldehyde-agarose gels. First-strand cDNA templates were synthesized by mixing 1–5 μg DNase-treated RNA and 5 μM of the (dT)11-primer. The mixtures were heated for 5 min at 70°C and immediately snapped on ice to cool. Then 1× RT buffer, 200 units Superscript RT (Invitrogen), 25 μM dNTP, and 10 mM DTT were added and incubated for 90 min at 42°C in a final volume of 20 μl. After heat inactivation of enzyme for 5 min at 70°C, the cDNA was diluted 1:5 with nanopure water and stored at −20°C for the subsequent qRT-PCR analysis and sequencing of Glo1 and Glp1r.
One-step (Glp1r) and two-step (Glo1) quantitative (q)RT-PCR reactions were carried out using Taqman 1000 RXN Gold and SYBR Green, respectively (Applied Biosystems, Foster City, CA). The primers and probes used for qPCR were designed to amplify regions that exclude any known expressed sequence single nucleotide polymorphisms (SNPs).
For Glp1r qRT-PCR, pancreas, antral stomach, and hypothalamus, RNA samples were used after diluting to 50 ng/μl in formamide and then to 1 ng/μl in deionized water just before use. Five microliters of diluted RNA was used in each 50-μl reaction with single-reporter measurement by using the Prism SDS 7900 HT (Applied Biosystems, Foster City, CA) detection system. Primers and probes were designed with Primer Express Software (Applied Biosystems) to span intron-exon boundaries. RNA from CAST pancreas was prepared and serially diluted to generate a standard curve. Both standards and samples were analyzed in triplicate and the expression values for Glp1r were normalized to values of an internal control gene (Ppia encoding mouse cyclophilin a). The sequences for the primers and probe (5′-FAM and nonfluorescent quencher in the 3′ end) were: Glp1r, forward, 5′-ACTTTCTTTCTCCGCCTTGGT-3′; reverse, 5′-TTCCTGGTGCAGTGCAAGTG-3′; probe, 5′ -CGCTTCAGCCATCCTTGTTGGCTTC-3′; Ppia, forward, 5′-AGGGTTCCTCCTTTCACAGAATTA-3′; reverse, 5′-CAGTGCCATTATGGCGTGTAAA-3′; probe, 5′-CCACCCTGGCACATGAATCCTGGA-3′.
For Glo1 qRT-PCR, cDNA synthesized from liver and hypothalamus RNA samples were used. One microliter of cDNA was used in each 30-μl reaction volume that consisted of 30 pmol of forward primer and 30 pmol of reverse primer in 2× SYBR Green master mix (Applied Biosystems). Primers were designed with Primer Express Software (Applied Biosystems). A standard curve was generated using a serial dilution of a pool of all cDNA samples. The samples were analyzed in duplicate, and the expression values for each sample were normalized with an internal control gene Ppia encoding cyclophilin-a. The primer sequences used were Cyclophilin forward: 5′-AGGGTTCCTCCTTTCACAGAATTA-3′; Cyclophilin reverse: 5′-AGTGCCATTATGGCGTGTAAA-3′; Glo1 forward: 5′-TCCAGAAAAGCCACCCTTGA-3′; and Glo1 reverse: 5′-CAGGAACGGCAATCCCAAT-3′.
Total RNA was extracted from the hypothalamus and liver of B6 and CAST mice and cDNA was synthesized. Full-length Glp1r and Glo1 cDNA was amplified by PCR using a proofreading DNA polymerase (Advantage; Clontech, Mountain View, CA) (primers: Glp1r forward, 5′-ggcagctatgacccagtcctga-3′ and reverse, 5′-ggtgtgcctgtgtccttcacatt-3′; Glo1 forward: 5′-CTAGGGCAGGTTGGTGATTC-3′ and reverse 5′-GGCAAAACAGAGGGGAGATT-3′), subcloned (pCR2.1 vector; Invitrogen, Carlsbad, CA), and sequenced. Sequencing was performed by the Genomic Core Facility of the PBRC on an Applied Biosystems 3100 using the BigDye Terminator Cycle Sequencing Kit version 3 (Perkin-Elmer Biosystems, Foster City, CA). The Sequencher software program (Genecodes, Ann Arbor, MI) was used to align DNA sequences between the two strains.
Western blot analysis.
Protein was extracted from stomach and pancreas. Tissue (500 mg) was pulverized in liquid nitrogen, homogenized and lysed in 500 μl of RIPA buffer [100 mM Tris·HCl (pH 8.0), 0.1% SDS, 150 mM NaCl, 50 mM EDTA, 0.1% Nonidet P-40, 0.1% sodium deoxycholate, 100 mM β-glycerophosphate and 50 μl of protease inhibitor cocktail (Sigma Aldrich)], sonicated and centrifuged. About 200 μg of protein in lammelli loading buffer (Bio-Rad, Hercules, CA) was separated in 15% SDS-polyacrylamide gel and transferred to PVDF membrane (Bio-Rad). The membranes were incubated overnight at 4°C with antibodies to GLP-1R or mouse β-actin (Abcam, Cambridge, MA), followed by incubation with fluorescent-labeled secondary antibody (IRDye800 or Cy5.5) according to the manufacturer's protocol. Bands were visualized using the Odyssey imaging system (LI-COR; Bioscience, Lincoln, NE).
The mean and standard error of each phenotype was determined for the parental strains, homozygous and heterozygous B6.CAST-17 S6, and B6.CAST-17 wild type. The effect of strain on nutrient intake and body weight phenotypes was determined by ANOVA. Diet intakes expressed as kilocalories by day were analyzed by ANOVA with repeated measures using a mixed model, with strain as the between-group factor and day as the within-group factor. When a main effect was observed, individual comparisons were evaluated using Tukey's protected t-test.
Nutrient intake phenotypes in progenitor strains.
Phenotypic data for the progenitor strains were published previously (34) but were not expressed per body weight as shown now in Table 1. Notably, the CAST strain exhibited, per body weight, a significantly higher mean intake of total energy (1.4-fold, P < 0.0001) and C/P kcal (2.3-fold, P < 0.0001) compared with B6 mice, over the 10-day phenotyping period.
Body weight in congenic strains.
At the beginning (baseline) of the 10-day phenotyping period, 8-wk-old homozygous B6.CAST-17 congenic mice weighed significantly less than either the heterozygous congenics or the wild-type B6 mice [F(1, 68) = 24.33, P < 0.0001] (Table 2). After selecting their preferred level of fat vs. carbohydrate intake (protein content was fixed at 22% of total energy) for 10 days, body weight was equivalent among the three strains due to the fact that the homozygous and heterozygous B6.CAST-17 congenic mice gained significantly more weight than the wild-type B6 [F(2, 67) = 8.64, P < 0.001, data not shown].
Nutrient intake phenotypes in congenic strains.
As shown in Table 2, the cumulative calorie intake over 10 days, adjusted for body weight, was HOMO > HET = WT, despite the smaller size of the congenic strain, where HOMO is homozygous, HET is heterozygous, and WT is wild type. Overall, there were significant effects of genotype on C/P kcal [F(2,67) = 5.99, P < 0.005] and total kcal [F(2,67) = 17.50, P < 0.0001], but not on F/P kcal [F(2,67) = 0.59, P = 0.56]. Specifically, the homozygous B6.CAST-17 strain consumed, per body weight, 17% more total calories and 27% more C/P calories when compared with wild-type B6 mice. The suggestive linkage for F/P kcal and significant linkage for F/P preference (%F/P intake) identified in the previous mapping study (34) were not confirmed in the B6.CAST-17 strain, perhaps due to the congenic design, i.e., these particular traits were linked to the B6 allele and the congenic donor segment was contributed by the CAST strain.
Nutrient intake patterns in congenic strains.
The temporal patterns of nutrient intake over the 10-day phenotyping period were examined. Evidence of initial hyperphagia, followed by a gradual decline in calorie intake, was observed (Fig. 2) as indicated by a main effect of day [F(9,571) = 19.32, P < 0.0001] on total kilocalories. Over the course of the study, the homozygous B6.CAST-17 mice consumed 27% more total daily calories than the heterozygous congenic or wild-type strains [F(2,102) = 12.72, P < 0.0001], due to a genotype effect on consumption of the C/P diet [F(2,75.8) = 3.39, P < 0.05] but not the F/P diet [F(2,78.1) = 0.87, P = 0.42]. Notably, a homozygous CAST genomic segment at the congenic locus resulted in the preferential consumption of carbohydrate (vs. fat) which was established from day 1 of diet selection. By contrast, wild-type B6 (P < 0.0001) and heterozygous congenic mice (P < 0.0001) ate more C/P calories on day 1, but then significantly reduced their carbohydrate intake beginning on day 2 (Fig. 2). However, there were no differences in C/P intake of the homozygous B6.CAST-17 mice (P = 0.81) or in F/P intake for any of the three strains on day 1 when compared with day 2. These results recapitulate the original linkage data (34), i.e., an absence of genetic linkage for C/P kcal on day 1 but not on days 2-10 at the Mnic1 QTL locus (D17Mit100) that was due to the B6 allele.
Extent of the B6.CAST-17 congenic donor region.
The minimum size of the CAST donor region is 60.3 Mb, extending from the undefined starting marker D17Mit19 at 4.7 Mb to distal marker D17Mit91 at 63.5 Mb. The nearest marker proximal to D17Mit91 is D17mit71, which was not polymorphic, indicating an indeterminate region between 63.5 and 65.9 Mb. This congenic segment corresponds to ∼3.0–37.8 cM on the genetic linkage map (Fig. 1) and contains an estimated 1138 genes (NCBI-Build 36.1).
Glp1r expression analyses.
Using qRT-PCR, we found that Glp1r was differentially expressed in tissues selected for their relevance to GLP-1R function (Fig. 3A). In hypothalamus of the parental strains, Glp1r expression was downregulated in CAST compared with B6. Glp1r expression in nonglandular fundal stomach was negligible, but in antral stomach was significantly downregulated (by a factor of 2). By contrast, Glp1r expression in CAST pancreas was significantly upregulated (by a factor of 2). Western blot analyses of pancreas and stomach tissue yielded a single ∼50-kDa band in both B6 and CAST corresponding to the known molecular weight of the human GLP-1R (Fig. 4B). Expression of GLP1-R protein in both pancreas and stomach paralleled the observed strain and tissue-specific differences in gene expression by approximately twofold (Fig. 3B).
Similarly in homozygous B6.CAST-17 congenic and wild-type B6 mice, qPCR verified the strain- and tissue-specific patterns of Glp1r gene expression with respect to the CAST congenic segment (Fig. 4A). Specifically, Glp1r mRNA expression was downregulated in congenic stomach compared with B6 and significantly upregulated in the pancreas of congenic mice (Fig. 4A). The strain difference in hypothalamic Glp1r mRNA was not reproduced in the congenic strains. Notably, expression of GLP1-R protein in pancreas and stomach paralleled the observed congenic strain differences in Glp1r mRNA (Fig. 4B).
Sequencing and identification of SNPs in Glp1r.
The published mouse GLP-1R sequence (GenBank accession nos. AJ001692 and NM021332) predicts 489 amino acids. However, our sequence (GenBank accession no. DQ0932297) differs from the published mouse GLP-1R sequence due to a “G” insert at nt position 1349 of our sequence, which results in earlier termination of the open reading frame (ORF). Consequently, our data show the predicted length of GLP-1R is 463 aa compared with 489 aa as previously reported. The corrected protein sequence increases the mouse GLP-1R homology to rat and human GLP-1R, in the COOH terminal region, from 450 to 463 amino acids.
To look for variation that could predict a functional consequence, the coding region of Glp1r was sequenced from the B6 (GenBank accession no. DQ0932297) and CAST (GenBank accession no. DQ093398) inbred strains. Four SNPs were found, three in the ORF and one in the 5′UTR (Table 3). Only one of the three SNPs within the ORF leads to a change in an amino acid. This SNP, located in exon 13 (G1247A, nucleotide), predicts a nonconserved amino acid difference (C416Y, amino acid residue) between the two strains. A possible functional effect of this nonsynonymous SNP in Glp1r was evaluated by in silico analysis using the Sorting Intolerant From Tolerant program (SIFT). Although the results failed to predict a deleterious effect of this sequence variant, this analysis is insufficient alone to rule out altered function of the receptor protein, e.g., its ability to activate cAMP, in the absence of functional tests.
Glo1 mRNA expression and sequence analyses.
Glo1 is widely distributed in a number of tissues, but available data show some of the highest expression in liver and brain (http://symatlas.gnf.org/SymAtlas/). Using qPCR, we found that Glo1 was significantly upregulated in both liver (1.5-fold) and hypothalamus (2.7-fold) of homozygous B6.CAST-17 compared with B6 (Fig. 5). The coding region of Glo1 was sequenced from the B6 and CAST inbred strains and one silent polymorphism was found, located in the ORF (Table 3).
Phenotyping studies of a new congenic strain developed in our laboratory have confirmed the significant genetic influence of two chromosome 17 QTL: Mnic1 and Kcal2 for carbohydrate and total energy intake, respectively (34). The results showed that the B6.CAST-17 congenic strain when compared with wild-type B6 self-selected 27% more calories from carbohydrate and 17% more total calories per body weight while consuming equivalent amounts of fat. Therefore, this congenic strain retained the phenotypes of the original, CAST-linked QTL. The lack of a difference in fat intake between the congenic and wild-type strains underscores the nutrient-specific effects of this complex locus. Of interest, several human QTL for total energy and one for dietary fat intake have been identified recently (6, 7), but none of them correspond to the mouse chromosome 17 region that was characterized in the present study. These mouse nutrient intake QTLs and congenics are the first to be reported in mammals, and they provide a unique model for uncovering new genes and pathways influencing the control of food intake.
The mechanisms underlying phenotypic strain differences in diet selection and intake could involve genes affecting peripheral sensory responses, postingestive and metabolic effects of food, and/or central nervous system integration of external and internal cues (2). For example, rodents learn to associate the perceived flavor of food, i.e., the combination of taste, smell, and texture cues, with their postingestive, metabolic, or energetic consequences through a form of classical conditioning (26). Alternately, genetic differences in taste or oral sensation could be primary determinants of nutrient selection (22, 25). Therefore, genes encoding proteins involved in these processes are likely candidates for modifying the self-selected intake of macronutrient diets.
The pattern of fat vs. carbohydrate selection observed on day 1 of diet exposure in the present study is key because it recapitulates our original linkage data showing an absence of genetic linkage for carbohydrate intake on day 1 but not on days 2–10 at the Mnic1 QTL locus (data not shown). This temporal change in linkage was dependent on the B6 allele (34), i.e., the carbohydrate intake of F2 mice homozygous for the B6 allele was not different from that of homozygous CAST at D17Mit100 on day 1. This finding suggests a strain by diet interaction in the initial response to the experimental diets. In our genetic model, it is unclear why the CAST preference for carbohydrate is established from day 1, yet the fat appetite shown by B6 mice does not take effect until day 2 (Fig. 2). One possibility is that sweetness of the C/P diet, which contains 21% powdered sugar by weight, stimulates intake in naive B6 mice. A strong behavioral (27, 28) and gustatory nerve (17) response to sweeteners has been well-documented in C57BL/6J mice when compared with 129P3/J mice. Thus we propose that the carbohydrate preference and overconsumption displayed by B6 mice on day 1 is based on their avidity for the taste of sucrose. The diminished carbohydrate intake observed on days 2–10, compared with day 1, suggests the possibility of overriding postoral factors in the B6 strain's response to one of the two diet mixtures. For example, a reduced capacity to utilize carbohydrate could play a role in the animals' preferred choice of nutrient. Less is known about sweet appetite in the CAST strain although their preference for saccharin solutions in two bottle tests is similar to that of B6 mice (Smith Richards BK, unpublished observation and Ref. 25). Whether orosensory or postoral factors prevail in the CAST strain's choice between these diets is not yet clear.
The coincidence of the two genetic loci on chromosome 17 suggests that a common gene(s) may be contributing to both traits. The peaks of Mnic1 at 12 cM and Kcal2 at 16 cM approximate the location of Glo1 at 16 cM and Glp1r at 18 cM, positional gene candidates with physiological relevance to nutrient intake. Therefore, expression profiling and sequencing were performed to screen for strain differences that could signal the possible involvement of these obvious candidates.
Glo1 is part of the glyoxalase system, present in the cytosol of all mammalian cells. It detoxifies dicarbonyl metabolites, such as methylglyoxal which is the major substrate of the glyoxalase system (24). If not catalyzed by the glyoxalase system, methylglyoxal participates in the glycation of proteins and nucleotides, resulting in the formation of advanced glycation end products (AGEs) that can lead to mutagenesis, apoptosis, protein degradation and induction of proinflammatory cytokines (38). An excess accumulation of methylglyoxal can occur during the hyperglycemia associated with diabetes mellitus (42). In this condition, for reasons that are unclear, glyoxalase upregulation may be inadequate to normalize methylglyoxal levels and prevent chemical modifications leading to AGEs (39), as shown recently in diabetic mouse lenses (36). However, in normal cultured mouse lenses cells treated with high glucose, glyoxalase I activity seems to be effective in metabolizing accumulated methylglyoxal (36). In our model system, it is likely that an accumulation of triose phosphates and a high production of methylglyoxal occur as a result of high carbohydrate consumption. We speculate that the upregulation of glyoxalase I observed in CAST and B6.CAST-17 mice supports their need to protect against dietary oxidants, given the nutrient intake traits exhibited by these strains.
Sequence analysis of Glo1 detected a silent polymorphism that is unlikely to have functional significance but does not rule it out as a QTL gene, e.g., Glo1 could be differentially regulated by an upstream mechanism. The upregulation of Glo1 expression in both liver and hypothalamus of CAST and B6.CAST-17 mice indicates a positive association between Glo1 expression and high carbohydrate intake. Whether or not changes in Glo1 activity are involved in producing the phenotype of increased carbohydrate intake remains to be demonstrated.
The GLP-1R is a putative seven-transmembrane domain receptor belonging to the family of G-protein-coupled receptors; the receptor binds specifically GLP-1 (3, 37). In rodents, GLP-1R mRNA and high-affinity GLP-1 binding sites have been identified in pancreatic islets, stomach, intestine, lung, kidney, heart, fat, skeletal muscle, and liver and throughout the brain (4, 5, 12, 14). The satiety-promoting and glucose-homeostasis effects of GLP-1 clearly implicate Glp1r as a physiologic gene candidate for ingestive behavior phenotypes. In both humans and animals, GLP-1 stimulates insulin secretion from pancreatic β-cells in response to glucose ingestion and inhibits gastric acid secretion and gastric emptying (11, 15, 16). GLP-1 potently decreases food intake when injected into the brain; by contrast, the GLP-1R antagonist exendin 9-39 stimulates food intake (9, 40). GLP-1 may also mediate the response to visceral illness in rats (32) but not in mice (18). A role for the GLP-1R in learning and memory has also been proposed (13).
GLP-1 exerts effects on β-cells to stimulate glucose-dependent insulin secretion (e.g., see Ref. 23) and improves glucose tolerance both through its insulinotropic action and by increasing glucose effectiveness (10). The effects of gastrointestinal peptides on insulin secretion and glucose homeostasis are an essential component of the processes regulating carbohydrate metabolism, accounting for ∼50% of postprandial insulin secretion (1, 41). In pancreas, GLP-1R protein and mRNA expression was twofold higher in the CAST inbred strain and homozygous B6.CAST-17 congenics when compared with B6, suggesting a possible mechanism for increased insulin synthesis in the presence of high carbohydrate consumption. In stomach we found that GLP-1R protein and mRNA expression was 2-fold lower in carbohydrate-preferring CAST and B6.CAST-17 mice when compared with B6, a finding consistent with enhanced gastric accomodation. Downregulation of Glp1r in stomach could support higher carbohydrate consumption with respect to this nutrient's larger volume and lower calorie density relative to fat.
Glp1r expression was downregulated in hypothalamus of CAST mice, consistent with their higher energy intake and the possibility of a reduced satiety effect. However, in the B6.CAST-17 congenics, there was no reduction in hypothalamic Glp1r expression, suggesting the existence of modifiers of Glp1r expression outside the congenic region. A previous study of GLP1R −/− mice found no effects of the targeted GLP1R mutation on food intake (29). This result could be due to a variety of genetic or environmental factors (19), including the satiety model that was used (20-h fast) or effects of the genetic background that was composed of 129 and CD1 genome (30) but contained no C57BL/6J or CAST/Ei DNA.
Collectively, the current findings suggest a potential role of GLP-1 signaling in this genetic model to regulate appetite and satiety by incorporating both the physical and metabolic characteristics of food. Additional support for Glp1r as a relevant candidate for the chromosome 17 locus in this strain intercross derives from a constellation of related phenotypes indicating a more active metabolic state in the CAST. Specifically, compared with B6, the CAST strain is exceptionally lean and physically active (20), expends 43% more energy (31), eats ∼33% more total calories per body weight (Table 1), and consumes more carbohydrate than fat in a diet choice paradigm (33–35). The mechanism by which the GLP-1R protein could influence the food intake traits in this mouse model is not yet known but could result from changes in the regulatory region of the gene, as suggested by the observed strain differences in expression, or from changes in the protein coding region. Of the three nucleotide differences between B6 and CAST, only one is predicted to cause an amino acid substitution. The position of the Cys(B6) to Tyr(CAST) polymorphism at AA416, located in the COOH terminal intracellular domain (AA409–463), could have functional implications in the downstream regulation of protein activity. The polymorphism is in proximity to a putative G protein-coupling region located within the third intracellular loop (3). Interestingly, both the human and rat Glp1r sequences predict a Ser AA at the equivalent position to AA416. As both Ser and Tyr residues can potentially be phosphorylated by kinases, but Cys residues cannot, this region may be implicated as a target of kinase action.
The tissue- and strain-specific patterns of Glp1r gene expression in this model indicate the possible existence of splice variants. RNA splicing resulting in receptor isoforms with distinct functions has not been previously demonstrated for GLP-1R (3). In the present study, full-length Glp1r transcripts and additional mRNA splice variants were identified through RT-PCR and subcloning of Glp1r cDNA transcripts using primers situated in the known 5′ and 3′ UTR sequences, and oligo(dT)-primed cDNA from pancreas and stomach. Preliminary analyses have identified three variants, all of which have the first six exons and the last exon (13th exon) conserved. Interestingly, all three forms contain the hormone/ligand-binding domain but lack the intracellular domain for activation/stimulation of G proteins for downstream regulation. Whether or not these splice variants differentially regulate Glp1r expression or activity remains to be determined.
In summary, a novel congenic strain with high carbohydrate and high energy intake per body weight has confirmed two chromosome 17 QTL previously identified in a B6 X CAST intercross. In light of current knowledge on the biological roles of Glo1 and Glp1r, along with the genetic and genomic findings reported here, Glp1r and Glo1 warrant further testing for possible functional roles in the control of nutrient intake in mice. Based on the size of the QTL interval, the likelihood of detecting additional candidate genes is high. Other positional candidates that might contribute to the observed phenotypes are pancreatic colipase (Clps), apolipoprotein M (ApoM), and peroxisome proliferator-activated receptor-delta (Ppard) (http://www.ncbi.nlm.nih.gov). Pancreatic colipase encodes for a cofactor involved in fat digestion while ApoM appears to play a role in high-density lipoprotein metabolism. Multiple functions of ppard have been described including adipocyte physiology and transcriptional control of fatty acid oxidation in skeletal muscle. A comparison of the complete coding sequences of these genes derived from B6 and CAST found no amino acid substitutions (34; Kumar KG and Smith Richards BK, unpublished observation). These genes, as well as others, will be further evaluated in the process of expanding candidate gene identification. Strategies will include comprehensive expression profiling, molecular genetic technologies, and available bioinformatics tools, e.g., SNP database analyses. The SNP approach to candidate gene identification is limited by the current availability of CAST genome sequence. Nevertheless, future candidate gene analyses will be enhanced by newly compiled CAST sequence as it emerges from the large-scale, multistrain sequencing project now underway (http://mouse.perlegen.com/mouse/index.html). Importantly, our congenic model and prospective sublines will be used to narrow and refine the differential locus on chromosome 17, as well as to examine gene interactions and subphenotypes in the control of food intake (8), with the ultimate goal of identifying the set of genes responsible for these complex traits.
This research was supported by National Institute of Diabetes and Digestive and Kidney Diseases Grant DK-53113 (to B. K. Smith Richards).
The authors thank Drs. Jong-Seop Rim and Hans-Rudolf Berthoud for assistance with the Western blots and Brenda Belton for colony management and technical assistance in the animal studies.
A preliminary report of this study was published in abstract form (35).
The costs of publication of this article were defrayed in part by the payment of page charges. The article must therefore be hereby marked “advertisement” in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.
- Copyright © 2007 the American Physiological Society