Sex-dependent Differences in Liver and Gut Metabolomic Profiles With Acarbose and Calorie Restriction in C57BL/6 Mice

Abstract

Acarbose, an alpha-glucosidase inhibitor used in treating type 2 diabetes, impairs complex carbohydrate digestion and absorption and extends life span in mice (without a requisite reduction in food intake). To assess sex-differential effects coincident with calorie restriction versus a nonrestricted longevity enhancing intervention, we evaluated the metabolite profiles (by liquid chromatography–mass spectroscopy) from livers and cecal contents of C57BL/6J mice (n = 4/sex/group), which were maintained for 10 months under one of the three diet treatments: ad libitum control diet (CON), ad libitum control diet containing 0.1% acarbose (ACA), or 40% calorie restriction using the control diet (CR). Principal component analysis revealed sex-differential profiles with ACA in livers. Of the identified metabolites (n = 621) in liver, CR significantly altered ~44% (males:187↑/131↓, females:74↑/148↓) compared with CON, in contrast with ACA (M:165↑/61↓, F:52↑/60↓). Dissimilarity in ACA-F liver metabolites was observed for ~50% of common metabolites from ACA-M and CR-M/F. CR resulted in fewer significant cecal metabolite differences (n = 615 metabolites; M:86↑/66↓, F:51↑/48↓ vs CON), relative to ACA treatment (M:32↑/189↓, F:36↑/137↓). Metabolomic profiling identifies sex-differential and tissue-specific effects with amino acid metabolism sub-pathways including those involving tryptophan, branch-chain and sulfur amino acids, and the urea cycle, as well as bile acid, porphyrin, and cofactor metabolism pathways.

Many of the chronic, noncommunicable diseases that increase in incidence with advancing age are found to be nutritionally modulated with the most highly reported intervention in preclinical models focused on reduced calorie intake (calorie restriction, CR) with adequate or optimal nutrition (1). Similarly, more recent nutritional studies have focused on the balance between macronutrients (eg, protein:carbohydrate) or individual components within a single macronutrient class as potential effectors of health and longevity, with or without the presence of calorie intake reductions (sometimes referred to more generally as dietary restriction, DR) (2,3). Despite these advances in knowledge, effort to translate CR or DR into long-term, sustainable applications remains challenging due, in part, to significant lifestyle modifications and practical application hurdles contributing to lack of compliance. As such, efforts have expanded to include the identification of compounds that mimic the metabolic, health, and longevity benefit of CR or DR without requiring the sustained, significant reductions in daily caloric intake, a growing field of calorie restriction mimetics (CRM) (4–6).

Over the last two decades, a number of compounds have been proposed and tested as potential CRM. One particularly promising category of CRM has included pharmaceutical agents used in glucoregulatory control of type 2 diabetes. Considering one of the hallmarks of CR is improved glucose homeostasis and insulin responsiveness/sensitivity, the significant similarities in expression profiling between CR and type 2 diabetes medications underscore the potential shared metabolomic responses that are present in conditions of improved aging (7,8). Furthermore, impaired glucose homeostasis has increased in the U.S. adult population, with approximately half or more of older adults (aged 65 or older) exhibiting quantifiable deficits (9). Although metformin, an oral biguanide with antihyperglycemic effects highly prescribed for type 2 diabetes in the United States, has received much interest for prolongevity and anticancer effects in preclinical models (10), other type 2 diabetes medications have been tested as potential CRM as well, including the alpha-glucosidase inhibitor acarbose.

Acarbose acts primarily through competitive inhibition of complex carbohydrate digestion and absorption, effectively lowering dietary exposure to postprandial glucose and increasing carbohydrate passage to the lower intestine where microbial fermentation occurs (11). In recent studies, 0.1% dietary acarbose treatment significantly extended the life span of mice, albeit with a larger longevity benefit in males than females (12). Due to the primary mechanism of action in the gut, acarbose has pronounced effects on dietary nutrient exposure through intestinal interactions. Although both acarbose treatment and CR alter glucose metabolism and homeostasis, both interventions also induce significant changes in gastrointestinal (GI) function. With CR generally reported to have positive glucoregulatory and longevity effects within studies in both sexes, albeit with exceptions (13–16), versus the sex-biased benefits of acarbose more toward males, we hypothesized differences in metabolomic profiles may contribute to these sex-differential effects. A high-dimensional metabolomics platform was utilized to assess treatment and sex effects on >600 metabolites using C57BL/6 inbred mice randomized to control, CR, or acarbose (0.1%, ACA) treatment, including both tissue-directed responses with liver samples and a measure of GI-related effects through cecal contents coincident with the treatment.

Methods

Animals

Male and female C57BL/6J mice were acquired at 6 weeks of age from Jackson Laboratories (Bar Harbor, ME; n = 12 of each sex) and housed at the University of Alabama at Birmingham. Animals were group housed on arrival for 3 weeks of acclimatization and then assigned via random number generation to experimental groups, at which point they were housed individually in polycarbonate cages with filter lids and woodchip bedding in individually ventilated racks at 22°C (Thoren Caging Systems, Hazleton, PA). Two paper tubes were provided per cage as enrichment, with cages changed every other week. Rooms were maintained in a 12:12 hour light:dark cycle with lights on at 6 am. Hydropac autoclaved water was provided ad libitum. Studies were carried out under the approval of the University of Alabama at Birmingham IACUC.

Diets

Animals were provided access to one of three diets. Control animals (CON; n = 4 mice/sex) were provided an NIH-31-based diet (LabDiet 5LG6; Supplementary Table S1) ad libitum. Diet for acarbose animals (ACA; LabDiet 5W5U, based on 5LG6; n = 4 mice/sex) was identical to CON with the addition of 0.1% acarbose (w/w; which has previously been shown to increase longevity) and was provided ad libitum. CR mice (n = 4 mice/sex) were provided LabDiet 5LG6 at 60% of CON animals’ average intake. Restriction was implemented stepwise with a 10% interval increase in restriction each week for 4 weeks. On reaching 40% restriction, food provided was adjusted weekly based on sex-specific CON average intake from the previous week. Food was provided to CR animals daily in one meal within 1-hour before lights off. Caloric densities of diets were equivalent (Supplementary Table S2).

Dissections

Following 10 months of diet treatment, animals were measured for body composition (fat and lean mass) by quantitative magnetic resonance, weighed to the nearest 0.01 g and sacrificed via decapitation for blood collection and serum separation. All dissections were performed by the same individuals, blinded to group, during the middle of the light phase (6–9 hours after lights on). Livers were removed whole, weighed, and samples snap-frozen in liquid nitrogen. Cecums were separated from large intestine, weighed, and then cecal contents were removed and snap-frozen in liquid nitrogen. Liver and cecal content samples were stored at −80°C until processed for metabolites.

Metabolite Profiling of Samples

Frozen samples were shipped to Metabolon (Morrisville, NC) where sample preparation and processing were completed (http://www.metabolon.com/). Sample extracts were analyzed using an untargeted approach on the Metabolon Platform. In brief, ultrahigh performance liquid chromatography–tandem mass spectroscopy (UPLC–MS/MS) was conducted using a Waters ACQUITY UPLC and a Thermo Scientific Q-Exactive high-resolution/accurate mass spectrometer interfaced with a heated electrospray ionization (HESI-II) source and Orbitrap mass analyzer operated at 35,000 mass resolution. Sample extracts were divided into fractions for analysis by reverse phase/UPLC–MS/MS in positive- and negative-ion mode electrospray ionization and by hydrophilic interaction chromatography (HILIC)/UPLC–MS/MS with negative-ion mode electrospray ionization. Raw data were extracted, peak-identified, and processed for quality control by Metabolon. Compounds were then identified by comparison to Metabolon’s library of authenticated standards using retention time index, mass to charge ratio, and MS/MS spectral data. Although detected compounds included both biochemical intermediates of metabolic pathways and various derivatives of xenobiotics not associated with metabolic pathways, we have used the more common designation of “metabolites” to describe all compounds assessed.

Statistical Analyses

For comparing body weight and composition among experimental groups, Student’s t tests were used comparing ACA and CR individually to CON. Both sexes were analyzed separately and uncorrected p values reported. Raw data on detected metabolites were log transformed, and missing values were imputed with the minimum observed value for the respective compound. Pearson correlation coefficients were determined for shared metabolites between liver and cecal content samples using the mean values of the four biological replicates, and significance of the correlation was determined using the “cor.test” function in R version 3.3.2. (p ≤ .05). Principle components analysis (PCA) was conducted incorporating group and sex to discern overall variation in the data set. Analysis of variance comparisons were used to identify compounds significantly altered compared with the CON group (p ≤ .05).

After the analyses to determine significantly altered metabolites, significantly altered pathways were determined by permutation technique using results from significantly altered metabolites. A new statistic was computed as a product of p values of significantly altered metabolites under a particular sub- or super-pathway. The relatedness among metabolites under a pathway makes it difficult to know the exact distribution of this new statistic. Hence, the empirical approach was taken to compute p values for the new statistic. Single metabolome analyses for the metabolites in a pathway were performed 1,000 times iteratively by randomly shuffling metabolome values in each analysis model. Then, the new statistics were computed for each iteration. These 1,000 statistics gave us distribution under null hypothesis of no association between metabolome and treatment. The new statistic (for unshuffled data) was then ranked in 1,000 statistics computed under null hypotheses to get an empirical p value for a pathway. The same method was used to compute empirical p values for sub- and super-pathways. Pathway significance based on the permutation technique was determined based on Bonferroni corrected alpha level of .0001.

In addition to permutation analysis, pathway enrichment analysis was performed using Metabolon’s pathway set enrichment pairwise comparisons. Enrichment value for a pathway was determined based on the quotient of the ratio of the number of significant metabolites in a pathway to the total number of detected metabolites in the pathway and the ratio of the total number of significant metabolites to the total number of detected metabolites. A value greater than one indicates “enrichment.”

Results

Body Weight and Composition After 10-Month Diet Feeding

Ten months of diet feeding resulted in significantly lower body weights in treated male mice versus CON (mean (g) ± SD; CON: 40.29 ± 3.71, ACA: 30.63 ± 1.82, CR: 21.58 ± 0.80; CONvACA: p = .007, CONvCR: p < .001), whereas only CR females had lower body weights (CON: 27.55 ± 1.93, ACA: 26.23 ± 0.79, CR: 16.29 ± 0.93; CONvACA: p = .31, CONvCR: p < .001). This difference in body weight reflects a reduction in body fat mass in ACA and CR treated mice in both sexes (males—CON: 14.95 ± 2.99, ACA: 6.40 ± 1.29, CR: 3.06 ± 0.19; CONvACA: p = .004, CONvCR: p < .001; females—CON: 8.26 ± 1.61, ACA: 5.37 ± 0.29, CR: 2.68 ± 0.35; CONvACA: p = .02, CONvCR: p = .001). However, lean mass was significantly reduced with CR, but not reduced with ACA relative to CON (males—CON: 23.96 ± 0.86, ACA: 22.68 ± 0.62, CR: 17.21 ± 0.56; CONvACA: p = 0.08, CONvCR: p < .001; females—CON: 17.86 ± 0.45, ACA: 19.46 ± 0.64, CR: 12.62 ± 0.52; CONvACA: p = .01, CONvCR: p = .001).

Correlation Among Metabolites From Liver and Cecal Content Samples

As the diets were uniform in composition but altered in amount fed or the addition of acarbose, it allowed for the assessment of GI-related changes on tissue-level changes in metabolites through parallel assays of cecal content and liver samples from the same animals. A total of 621 named metabolites were detected in liver samples and 615 in cecal content samples (Supplementary Table S3). Of these metabolites, 427 were common to both liver and cecal contents (Supplementary Table S3). The prevalence of these shared metabolites demonstrated a significant, positive correlation between the cecal and liver samples (p < .0001, r = .56, Figure 1, Supplementary Table S4). Metabolites associated with the amino acid, lipid, nucleotide, carbohydrate, and peptide super-pathways contributed to this significant correlation (Figure 1; Supplementary Table S4), further refined by group and sex for the liver and cecal content metabolites (Figure 1; Supplementary Table S5). Of the total metabolites detected, 194 were unique to the liver, of which 42% were from lipid and 16% from amino acid super-pathways (Supplementary Table S3). In the cecal contents, 182 metabolites were unique with 36% from lipid, 24% from amino acid, and 21% from xenobiotic super-pathways (Supplementary Table S3).

Figure 1.

Pearson correlation analysis of shared metabolites detected in cecal content and liver samples separated by super-pathway. Heatmap colors represent correlation coefficients between less than 0 (red) and 0.75 (purple). ACA = 0.1% acarbose; CON = control; CR = calorie restricted.

Significantly Altered Metabolites From Liver and Cecal Content Samples

In assessing the overall pattern of metabolite changes, PCA for liver metabolites revealed that sex and diet groups were the major sources of variation in detected metabolites (Figure 2A). The number of significantly altered metabolites of the total detected metabolites in liver samples, particularly for “up” regulated metabolites, was higher with CR in males (51%) (M:187↑/131↓) than females (36%) (F:74↑/148↓) relative to the sex-matched CON. Although significantly altered metabolites with ACA treatment were more modest in number for “down” regulated metabolites, ACA males had almost three times as many “up” regulated metabolites (36%) (M:165↑/61↓) as females (18%) (F: 52↑/60↓) for liver sample metabolites. Among the significantly altered liver metabolites in greater abundance compared with the CON (n = 264), ACA and CR groups shared less than half (n = 117) in common, with approximately one fourth (n = 62) unique to ACA (with males having a greater number of sex-specific, significantly higher metabolites [45 vs 6]), and an additional one third (n = 85) unique to CR (with males having a greater number [M vs F; 52 vs 13]; Figure 3A). When examining those metabolites, which were significantly lower compared with the CON (n = 236), ACA and CR groups shared a lower proportion of approximately one fifth (n = 56 metabolites), 35 of which were unique to ACA (with females having a greater number of sex-specific, significantly higher metabolites [10, 18]), and 145 metabolites were unique to CR (with females having a greater number [42, 61]; Figure 3B).

Figure 2.

Principle component analysis of detected metabolites. (A) Metabolites detected in the liver samples demonstrated clustering by diet group and by sex. (B) Metabolites detected in the cecal content samples clustered by diet group. ACA = 0.1% acarbose; CON = control; CR = calorie restricted.

Figure 3.

Venn diagram illustrating overlap in significantly altered metabolites of liver (A and B) and cecal content (C and D) samples. (A and C) Metabolites in greater prevalence compared with control. (B and D) Metabolites lower than found in control. ACA = 0.1% acarbose; CON = control; CR = calorie restricted.

Using a similar assessment as described above for liver samples, PCA for cecal content metabolites revealed that group was the main source of variation in detected metabolites, and sex did not demonstrate a distinct effect on the clustering in contrast with liver results (Figure 2B). CR treatment significantly altered 25% (M:86↑/66↓) and 16% (F:51↑/48↓) of the total detected metabolites in the cecal content samples relative to sex-matched CON, whereas ACA treatment resulted in slightly more significantly altered metabolites, with 36% (M:32↑/189↓) and 28% (F:36↑/137↓) of the cecal content metabolites altered. Among the significantly altered cecal content metabolites in greater abundance compared with the CON (n = 133), ACA and CR samples shared only 12 metabolites, 33 were unique to ACA (with an equal number of sex-specific, significantly higher metabolites in males and females [8,8]), and 88 were unique to CR (with males having a greater number [45, 9]; Figure 3C). On examination of the significantly reduced metabolites compared with the CON (n = 272), ACA and CR held 58 in common, 174 metabolites were unique to ACA (with males having a greater number of sex-specific, significantly higher metabolites [62, 39]), and 40 metabolites were unique to CR (with females having a greater number [8, 21]; Figure 3D).

Grouping the significantly altered metabolites based on associated super-pathways revealed more than half were associated with the amino acid and lipid super-pathways for both liver and cecal contents (Figure 4). In the liver, amino acid, lipid, and nucleotide super-pathways were major contributors to super-pathway profiles of increased metabolites relative to CON for ACA males, CR males, and CR females (Figure 4). For ACA females, the amino acid super-pathway dominated. Metabolites from the lipid super-pathway were primarily those reduced relative to the CON for all groups (Figure 4). The observed proportions of metabolites in super-pathways did not significantly differ from expected proportions for increased (χ² = 19.12, df = 21, p = 0.58) or reduced metabolites (χ² = 21.24, df = 21, p = .44).

Figure 4.

Significantly altered metabolites grouped by associated super-pathway from liver and cecal content samples. Values represent number of metabolites. ACA = 0.1% acarbose; CON = control; CR = calorie restricted.

Despite similarity among super-pathway profiles of the liver, those of the cecal contents showed variation among groups (Figure 4). The top three super-pathways increased relative to CON for the ACA males and females were the lipid, carbohydrate (Supplementary Figure S1i), and xenobiotic super-pathways (Figure 4), and in the CR males and females, amino acid, lipid, and peptide super-pathways made up the major proportion (Figure 4; χ² = 61.36, df = 21, p < .0001). ACA males and females had lower levels of metabolites in the amino acid, lipid, and cofactor/vitamin super-pathways compared with CON, whereas CR males and females had lower levels of metabolites representatives of lipid and nucleotide super-pathways (Figure 4; χ² = 67.4, df = 21, p < .0001).

Pathway Permutation and Enrichment Analyses

The large number of significantly altered metabolites was captured in the permutation analysis where the significant involvement of all super-pathways for one or more groups and sex, and group-specific involvement of sub-pathways for both liver and cecal content samples were identified (Supplementary Table S6). In amino acid sub-pathways, histidine metabolism was significant for CR males and females as well as glutamate metabolism, which was also significant for ACA males, but not females (Supplementary Table S6). Pathway enrichment analyses confirmed significant involvement of amino acid and lipid super-pathways (Supplementary Table S6). Within these lipid and amino acid metabolism pathways, examination of significant sub-pathways containing three or more significant metabolites revealed the long-chain fatty acid sub-pathway was significant for CR males and females and ACA males (but not females) and the monoacylglycerol sub-pathway was unique to the CR group (Supplementary Figure S1a–j). In addition, CR males and ACA males shared significance with the sub-pathways from alanine and aspartate metabolism, sphingolipid metabolism, and phospholipid metabolism (Supplementary Table S6).

Discussion

By testing two distinct interventions that have been demonstrated to improve glucoregulatory control and promote longevity of preclinical models, metabolomics profiles of liver tissue and cecal content were assessed to identify metabolites differentially altered in abundance across group treatments and biological sexes. Calorie restriction and acarbose supplementation represent unique dietary approaches to address the underlying bases of aging in that CR is routinely applied through the overall reduction in calories provided across all macronutrient groups, whereas acarbose treatment works through selective inhibition of carbohydrate metabolism while allowing for caloric compensation in intake amounts (8). Considering these disparate macronutrient approaches have been observed to promote health and longevity in some laboratory model organisms, a comprehensive assessment of the GI changes resulting from each intervention was compared with the circulating metabolites present in the liver, which was chosen for its role in metabolism including glucose, lipids, amino acids, and detoxification pathways (15,17,18).

As has been noted for previous interorgan metabolomics comparisons, there was a significant correlation between cecal metabolites, representing both the remaining dietary contents in the lower GI tract as well as bacterial metabolic by-products, vitamins and xenobiotic compounds, and tissue-level metabolites in the liver across diet groups and sexes. This raises the question of whether part of the liver-specific metabolomics profiles observed for intergroup comparisons with controls derives from GI differences in specific macronutrient digestion and absorption dynamics present with ACA and CR and resulting lower GI metabolic by-product absorption as the primary cause or whether alterations in major metabolic pathways in the liver ultimately influences upper-middle GI physiology and induces the alterations in cecal contents of the lower GI (17,19). Multiple, significant diet group differences were observed in cecal content profiles despite all mice being fed diets of the same macronutrient and micronutrient composition, highlighting the significant metabolic differences occurring in the GI tract, either prior to or concomitant with nutrient absorption and in vivo tissue distributions. For instance, with the CR group, animals were fed the same diet pellets as the control (therefore the same macronutrient and micronutrient ratios), just as a reduced daily provision, suggesting changes in cecal metabolites (and ultimately many of the liver metabolites given the significant correlations of abundances) reflect to some extent an interaction between the alteration in digestion, absorption and secondary metabolite production. However, directly assessing the primary site of action relevant to the in vivo metabolite profiles (occurring primarily or secondarily as a result of changes in the gut or in the liver, or some combination thereof) requires further testing in future studies. Acute feeding studies with multiple sampling across the upper, middle, and lower GI at the initiation of the intervention may help address these questions of the direction of effect for the cecal content-liver metabolites. Additional studies with germ-free mice may also be valuable, in particular, to address the importance of the lower GI, which is important for bacterial metabolic by-product production and absorption including short-chain fatty acids and vitamin/cofactor. Considering previous studies with Caenorhabditis elegans, wherein the longevity benefits of metformin, another diabetes medication with reported health and longevity benefits (8,10,20), were dependent on both the growth of a bacterial culture and the metabolites produced by particular bacteria (21,22), there potentially remains much to be tested and understood in the microbial effects on mammalian models of aging as hypothesized over a century ago (23). Reasons for the number of significant changes in the metabolite abundances beyond the diet composition equivalence could include the timing and duration of food intake related to other dietary practices which have received increasing attention for health or longevity effects in laboratory models (eg, intermittent fasting, time of day feeding, circadian misalignment, etc.) (24–28).

When considering the overall profiles of cecal content samples, PCA identified “diet group” as the prominent grouping, with sexes responding similarly to each other within a given diet group. In contrast, in liver samples, PCA identified “sex”-related differences were more characteristic of group separation, with CR reducing this separation between sexes. ACA treatment in males shifted liver profiles in the direction of female controls and CR, whereas ACA in females had little directional effect (Figures 2–4; Supplementary Tables S3 and S6). Across both cecal contents and liver tissue profiles, CR exhibits a unique comprehensive signature that appears to be only partially recapitulated with ACA treatment profiles.

To clarify which metabolites were shared between these interventions, we further compared individual metabolites and pathways. A large proportion (~1/3) of detected metabolites were at significantly increased or decreased abundance levels relative to controls in liver and cecal samples (Figure 3 and 4; Supplementary Table S3). Two observations were somewhat counter-intuitively related to the cecal and liver profiles. First, with the similarity of cecal profiles for “diet group” effects, in both sexes the number of metabolites with significantly increased abundance was higher than the number of metabolites with decreased abundance in CR compared with control—despite the “reduced” daily caloric provision. Thus, when fewer macronutrients are consumed, cecal abundances of multiple metabolites are actually increased. With ACA treatment where diet consumption is ad libitum and actually observed to increase relative to controls, the opposite profiles of a greater number of decreased abundances of metabolite (four times more) than increased metabolites occurs. In both CR and ACA diet groups, cofactor and vitamin abundances were reduced relative to CON. Second, the “sex” profile effect in the distribution of the significantly altered metabolites across metabolite types (super-pathways) highlights the impact biological “sex” has in modifying the GI exposures and effects to metabolite profiles, presumably through metabolic and physiologic differences in vivo that are sex-hormone related. For instance, although a similar number of metabolites were significantly altered with ACA treatment in cecal contents in both sexes, the number of significantly altered metabolites in liver samples were much fewer in ACA females than males, with ACA females having the lowest number of significant changes in liver of any group (Figures 3 and 4; Supplementary Table S3). Whether these cecal abundance differences reflect changes in GI function related to transit time, nutrient concentration or dilution, rheological properties of the food bolus, fecal output, or some other parameter is not determined with this study.

Both ACA males and females share a significant increase in cecal carbohydrate metabolites (structured polysaccharides, eg, maltose-related compounds; Supplementary Figure S1i), verifying the primary mechanism of action as a glucosidase inhibitor of ACA (29,30). Sucrose was also increased in liver samples with ACA in both sexes (Supplementary Figure S1j), which has been previously noted with CR, although the direct connection to ACA and changes in in vivo production, absorption, or secretion were not assessed (19). It is thought that short-chain fatty acids such as butyrate, a microbial by-product of carbohydrate fermentation (31), increase intestinal barrier function by decreasing intestinal tight junction permeability (32–34), which would be expected to reduce any “leak” of disaccharides from the gut into the circulation. However, di- and tri-saccharides, such as trehalose, raffinose, and sucrose, have been shown to stimulate autophagy in cell model (35,36); thus, the GI absorption of sucrose or the potential in vivo, endogenous production of sucrose (37) might be postulated to have beneficial effects in this regard and should be considered in future studies. Also related to carbohydrate metabolism and blood glucose (BG) levels, 1,5-anhydroglucitol has been proposed as a biomarker of BG variability (38–40), with 1,5-AG levels inversely related to the extent of hyperglycemia (as BG goes up particularly over 180 mg/dL, 1,5-AG is decreased). In contrast with the glucoregulatory function of ACA in lowering postprandial glucose and stabilizing daily BG values, 1,5-AG was significantly decreased in ACA males and females (predicting higher BG levels than controls), whereas CR samples had elevated levels of 1,5-AG (Supplementary Figure S1j), indicating a more comprehensive assessment of glucose metabolism and time series measures of BG levels will be important for clarifying the effects of habitual CR or ACA treatment on a proposed biomarker and mediator of these two diet/drug interventions. In agreement with previous studies, lipid and amino acid pathways were identified as differentially expressed metabolites in pathway analyses (17,19). In liver samples, multiple lipid species were reduced with both sexes with CR, whereas males responded more strongly with significant reductions in ACA treatment compared with females that were intermediate and “trended” toward significance or were not changed. It is interesting to note the reduced levels is in the same direction for both CR and ACA, given the differences in fasting versus feeding paradigms with the two interventions as well as the dietary carbohydrate targeting with ACA treatment.

Metabolites of the gamma-glutamyl amino acid pathway, which are important for the synthesis, degradation, and regeneration of glutathione and xenobiotic detoxification pathways (41–43), were sex differential in their abundance within treatments. In cecal samples, over half the detected gamma-glutamyl amino acid metabolites were increased with CR, whereas none were significant with ACA. However, both CR and ACA males had significantly increased the abundance of multiple pathway members, CR females had half as many increased metabolites in the pathway as either diet group in males, and ACA females having only one metabolite with a “trend” toward increased. Thus, connections to this pathway (particularly related to detoxification and xenobiotic metabolism), which are reported to contribute to processes implicated in aging through multiple longevity interventions, is sex differential for ACA (44–46). Whether this marked difference contributes to deficits of protection against oxidant damage, detoxification, and ultimately longevity remains to be tested.

The similarity between sexes within treatments for cecal content profiles contrasting with the sex-differential liver profiles under the same “diet group” suggests biological sex, with any number of hormonal, physiologic, and metabolic differences this entails, has potential modulating effects on prohealth and prolongevity interventions. Although several “sex-differential, diet group-specific” differences have been identified here, whether these or other metabolite and pathway modifications modulate alternative prohealth and prolongevity interventions remains to be demonstrated. As described herein, this study has several strengths including the large number of metabolites identified, the rigor of the experimental design, sample collection and measurements performed blinded, and the duration of the treatment prior to sample collection and assessment. Nevertheless, several limitations bear consideration. For instance, a well-described, long-lived mouse strain was utilized for these studies, but of a single inbred genotype and level of CR (40% restriction). Although 40% CR has been shown to increase life span in some but not all study conditions reported (1,13,16,47), a more recent report indicated a nonlinear, sex-dependent dose-effect differential with two levels of CR (20% vs 40% restriction), with 20% CR being superior for longevity benefits under the conditions tested (15). However, improvements in glucose homeostasis have been reported with levels of CR up to 40% (14). Additionally, only one concentration of acarbose (0.1%), which has been reported to positively affect life span in a sex-dependent manner, was used in this study (12,48). Any dose-dependent influence on changes in levels of metabolites as has been reported with CR, and ultimately, their relationship to longevity is thus not known. These results also potentially reflect a drug by diet interaction with ACA and the diet formulation used. Although we chose to use this diet based on the health and longevity profiles identified from previous studies, the high-carbohydrate content and balance of macronutrients may affect the metabolite results as macronutrient balance has been shown to influence life span across multiple studies (2,3,49). Whether ACA has a carbohydrate-independent effect on longevity particularly is currently unknown. Future studies should consider the dietary macronutrient content and complexity of carbohydrates (simple monosaccharides vs more complex sugars) to help clarify the effect and mechanism of ACA longevity benefit. Particularly interesting in that light would be low-protein/high-carbohydrate diet, which has reported prolongevity effects with the addition of ACA, and/or a very low-carbohydrate/high-protein diet, which would limit the primary GI mechanism of action of ACA, particularly if using a monosaccharide as the carbohydrate source. Finally, these results should be considered hypothesis generating due to the inability to make causal conclusions regarding the findings of specific pathways or metabolite due to the nature of the study design, the single age of assessment, the “unknown” metabolites not reported, and the static measures of metabolite abundances rather than “flux” analyses.

Despite the disparate “caloric” approaches to health-promoting interventions, both CR and ACA produce significant alterations in metabolic profiles for cecal (GI) contents and liver tissue samples, which are significantly correlated across multiple metabolite classes and pathways. Feeding either a bacteria-derived compound known to disrupt carbohydrate digestion and absorption or the same macronutrient composition but only one time a day results in multiple significant metabolites and pathway differences from ad libitum feeding of the same diet composition. Diet group and sex differences in metabolite profiles and individual pathways indicate lipids, amino acids, and cofactor/vitamin pathways correlate with treatment and sex in the direction of expected effects on health and longevity, further supporting the characterization of ACA as a CRM. Although the underlying molecular basis of CR for health and longevity remains disputed almost a century after reports of disease protection and longevity promotion, the identification of compounds, which can mimic some of these CR-related benefits, has never been more needed particularly in light of the current health epidemic of energy imbalance and excess body weight. In contrast to focusing on a single enzyme or pathway within the body or a particular tissue, metabolomics profiling and more recent dietary macronutrient balance studies suggest targeting individual dietary components and/or altering their metabolite balances may be alternative strategies that alleviate the necessity of the improbable adoption of chronic caloric restriction. Furthermore, the presence of these shared changes in lipids, amino acids, and cofactor/vitamin pathways raises important questions about the functional significance of the lower GI in promoting health and longevity. Future studies with older animals and identifying ways to target specific metabolic pathways will be necessary to help address the causal role the observed changes may play in disease prevention and longevity extension.

Supplementary Material

Supplementary data is available at The Journals of Gerontology, Series A: Biological Sciences and Medical Sciences online.

Funding

This work was support by the Ellison Medical Foundation New Scholar in Aging award and the University of Alabama at Birmingham Nathan Shock Center of Excellence in the Biology of Aging (P30AG050886). Preparation of this manuscript was supported in part by the National Institute on Aging of the National Institutes of Health under award number R01AG043972 and by the National Institute of Diabetes and Digestive and Kidney Diseases of the National Institutes of Health training grant T32DK062710 UAB Obesity Post-Doctoral Training Program to V.K.G.

Conflict of Interest

The views and opinions expressed herein are those of the authors and do not necessarily reflect the official policy or position of any funding source or agency with which the authors are affiliated. R.A.B. is a current employee of USANA Health Sciences, Salt Lake City, UT. D.L.S. and V.K.G. report no conflicts of interest.