SEARCH

SEARCH BY CITATION

Keywords:

  • calorie restriction;
  • longevity;
  • stress resistance;
  • type 5 adenylyl cyclase

Summary

  1. Top of page
  2. Summary
  3. Introduction
  4. Results
  5. Discussion
  6. Materials and methods
  7. Acknowledgments
  8. Author contributions
  9. References
  10. Supporting Information

Adenylyl cyclase type 5 knockout mice (AC5 KO) live longer and are stress resistant, similar to calorie restriction (CR). AC5 KO mice eat more, but actually weigh less and accumulate less fat compared with WT mice. CR applied to AC5 KO results in rapid decrease in body weight, metabolic deterioration, and death. These data suggest that despite restricted food intake in CR, but augmented food intake in AC5 KO, the two models affect longevity and metabolism similarly. To determine shared molecular mechanisms, mRNA expression was examined genome-wide for brain, heart, skeletal muscle, and liver. Significantly more genes were regulated commonly rather than oppositely in all the tissues in both models, indicating commonality between AC5 KO and CR. Gene ontology analysis identified many significantly regulated, tissue-specific pathways shared by the two models, including sensory perception in heart and brain, muscle function in skeletal muscle, and lipid metabolism in liver. Moreover, when comparing gene expression changes in the heart under stress, the glutathione regulatory pathway was consistently upregulated in the longevity models but downregulated with stress. In addition, AC5 and CR shared changes in genes and proteins involved in the regulation of longevity and stress resistance, including Sirt1, ApoD, and olfactory receptors in both young- and intermediate-age mice. Thus, the similarly regulated genes and pathways in AC5 KO and CR mice, particularly related to the metabolic phenotype, suggest a unified theory for longevity and stress resistance.


Introduction

  1. Top of page
  2. Summary
  3. Introduction
  4. Results
  5. Discussion
  6. Materials and methods
  7. Acknowledgments
  8. Author contributions
  9. References
  10. Supporting Information

Moderate calorie restriction (CR) extends lifespan and provides stress resistance in organisms ranging from yeast to primates (Sinclair, 2005; Canto & Auwerx, 2009; Houtkooper et al., 2010), as does disruption of type 5 adenylyl cyclase (AC5) (Okumura et al., 2003, 2007; Yan et al., 2007), which is one of the major AC isoforms expressed in the heart and brain (Hu et al., 2009) (Fig. S1). We tested the hypothesis that the molecular mechanisms mediating these two models of longevity and stress resistance differed and accordingly could be combined to obtain additive effects and induce super-longevity. This was the first goal of this investigation. However, when we superimposed CR on the adenylyl cyclase type 5 knockout (AC5 KO) mouse, the mice rapidly lost weight and died, most likely because the AC5 KO mouse already had lower glycogen and fat reserves prior to CR and thus could not tolerate the added burden of CR. Accordingly, we revised our hypothesis to determine whether the two models of aging and stress resistance had similar genotypes, such that the results might provide insight into a unified theory for aging and stress resistance. We found that both AC5 KO mice and CR-fed mice had similar physical profiles, such as a lower body weight, reduced fat and glycogen stores, and lower fasting blood glucose levels suggesting that many mechanisms, particularly relating to fuel utilization and energy metabolism, might be similar or amplified in these two models of longevity and stress resistance. We then examined gene expression profiles in the brain, heart, skeletal muscle, and liver, with the goal of revealing the mechanisms common to both CR and AC5 KO models of longevity and stress resistance. This study was conducted in 4-month-old young mice, which already showed stress resistance (Okumura et al., 2003, 2007) to identify the genotype at a young age that could be responsible for the extended longevity. However, to confirm that our results were not simply due to developmental responses, we compared the key findings in mice at an intermediate age (14 months), that is, in older mice, but at a time when aging cardiomyopathy had not developed.

Results

  1. Top of page
  2. Summary
  3. Introduction
  4. Results
  5. Discussion
  6. Materials and methods
  7. Acknowledgments
  8. Author contributions
  9. References
  10. Supporting Information

Adenylyl cyclase type 5 knockout and calorie restriction affect body weight and fat accumulation in a similar and additive fashion

Calorie restriction reduced body weight and fat accumulation in mice. We observed reduced body weight and fat accumulation in AC5 KO mice on an ad lib diet. Food intake was, however, more than that of WT mice (Fig. 1A,B). Correspondingly, the epididymal, retroperitoneal, and inguinal fat pads were smaller in AC5 KO mice as shown in Fig. 1D, resulting in a lower adiposity index (Fig. 1C). In addition, AC5 KO demonstrated a higher metabolic rate in terms of increased oxygen consumption (VO2), which was uniformly increased during the dark and light cycles (Fig. 1E).

image

Figure 1. Adenylyl cyclase type 5 knockout (AC5 KO) mice ingested more food but weighed less and accumulated less fat in their adipose tissues than WT mice. (A) Average ad lib food intake per day in WT and AC5 KO mice (n = 4 per group) during a 10-day monitoring period. AC5 KO mice ingested more food than WT mice. (B) Body weight in WT and AC5 KO mice (n = 26 per group). Even though AC5 KO mice ate more, the mice still weighed less than WT mice. (C) AC5 KO mice accumulate less fat in their adipose tissues. Adiposity index showed that the amount of total body fat in AC5 KO mice was less than that in WT mice at a similar age (6–7 months old, n = 4 per group). (D) The images for comparison of the fat pad size in AC5 KO vs. WT. (E) AC5 KO mice displayed a higher metabolic rate as reflected by higher basal oxygen consumption. Data were expressed as mean ± SEM. *P < 0.05 vs. WT.

Download figure to PowerPoint

When CR was superimposed on AC5 KO, all mice died within 1 month (Fig. 2A), the weight loss was greater in the AC5 KO on CR than in WT mice on CR (Fig. 2B and Fig. S2A), and this was associated with reduced fat deposits (Fig. 2C). Compared with WT mice on CR, the organ weights, as assessed by the ratio of the organ weight to tibia length (TL), were significantly reduced in the left ventricle (LV), gastrocnemius muscle (GM), and spleen with a nonsignificant decrease in the kidney and no change in the brain (Fig. S2B). After 2 weeks of CR, insulin was significantly lower in both the AC5 KO and WT group (Fig. 2G), but fasting glucose (Fig. 2D) and the HOMA insulin resistance index (Fig. 2H) were significantly lower in AC5 KO mice compared with WT, and free fatty acids (FFAs) tended to be lower, albeit not significantly (Fig. 2F). Interestingly, higher levels of ketone bodies were present in the serum of AC5 KO on control diet (CON), but decreased in AC5 KO on CR (Fig. 2E). Liver glycogen tended to be decreased by CR in WT, but was depleted in AC5 KO and could not be depleted further with superimposition of CR on AC5 KO (Fig. 2I,J). Hence, AC5 KO mice on CON phenocopy WT mice on CR in terms of body weight (BW) change and adiposity index, but with much lower glycogen stores. These data support our concept that the two models share similar mechanisms in metabolism and that metabolic reserves are already depleted in both models.

image

Figure 2. The effects of calorie restriction (CR) superimposed on adenylyl cyclase type 5 knockout (AC5 KO) on survival, body weight, fat, and metabolism. (A) Survival of AC5 KO mice under CR or control diet (CON). 25% CR at the first 2 weeks and 40% CR thereafter were superimposed on AC5 KO. CR superimposed on AC5 KO was lethal within a month (male, 3–5 months old, n = 20/group). (B) The weight loss in WT and AC5 KO mice after 3 weeks of CON and CR (n = 11–12 per group). (Note that the body weight of WT on the control diet for longer periods of time, that is, 4–6 weeks, which was used for the microarray studies, was not significantly different from those on ad lib diet.) WT with CR induced more weight loss than WT with CON. AC5 KO mice lost more weight than WT when both placed on CR. (C) Measurement of adiposity index (n = 11–12 per group). (D–H) Metabolic analyses of AC5 KO and WT after 3 weeks of CON or CR following 6-h fasting, including fasting glucose (D), ketone bodies (E), free fatty acid (F), insulin (G), and insulin resistance (H), n = 10–11 per group. (I) Examples of liver glycogen, measured with periodic acid-schiff staining. (J) Glycogen was quantified as a percentage of total staining area. Glycogen was significantly decreased to a similar extent in AC5 KO and CR mice (n = 4–5 per group). Data were expressed as mean ± SEM. *P < 0.05 vs. WT on CON, P < 0.05 vs. AC5 KO on CON, and P < 0.05 vs. WT on CR.

Download figure to PowerPoint

Microarray analysis revealed a similarity between the changes in gene expression between adenylyl cyclase type 5 knockout and calorie restriction mice

To explore whether these physiological similarities between AC5 KO and CR mice were also tied to similar molecular changes, microarray analysis of gene expression was performed in brain, heart, skeletal muscle, and liver of AC5 KO and CR mice. Surprisingly, the gene density heat maps showed that AC5 KO had a remarkable similarity to CR mice in all four tissues (Fig. 3A, left panels). Notably, a higher correlation in muscle tissues (skeletal muscle and heart) was found than in other tissues (brain and liver). We summarized the similarities and differences between AC5 KO and CR mice using Venn diagrams (Fig. 3A, right panels), which indicated the number of genes that are commonly regulated in both models. In brain, 662 genes (19%) were upregulated and 184 genes (12%) were downregulated commonly, P = 0.27. In heart, 1236 genes (28%) were upregulated and 307 genes (23%) were downregulated commonly, P = 0.52. In skeletal muscle, 866 genes (31%) were upregulated and 255 genes (24%) were downregulated commonly, P = 0.56. In liver, 189 genes (13%) were upregulated and 515 genes (22%) were downregulated commonly, P = 0.31. Conversely, few genes were regulated oppositely between the two models. In addition, there were more upregulated genes than downregulated genes in brain, heart, and skeletal muscle, while there were more downregulated genes than upregulated genes in liver.

image

Figure 3. Microarray analysis demonstrated that adenylyl cyclase type 5 knockout (AC5 KO) and calorie restriction (CR) mice share alterations of many genes. (A) Gene density heat map for gene expression changes in AC5 KO and CR mice brain, heart, skeletal muscle, and liver. Genes were ranked according to gene expression changes and then displayed in a heat map with color representing gene density (left panel). Density of genes is shown in color according to the scale shown in the graph. Blue and red represent depletion and enrichment of genes, respectively. x-axis, CR vs. CON; y-axis, AC5 KO vs. WT. Spearman rank correlation coefficient (Spearman's rho, ρ) is represented with each heat map. Venn diagram (right panel) summarizes the similarities and differences between AC5 KO and CR mice in brain, heart, skeletal muscle, and liver using significant genes. Fisher's exact test was used to assess significance of bias in genes overlapped between CR and AC5 KO. (B and C) Significant gene ontology (GO) terms with regulated genes between AC5 KO and CR mice. Using cumulative distribution function-based method, a group of up/downregulated genes in each data set was statistically tested by their association with each GO entry. The resulting significance scores were combined and visualized in the heat map consisting of GO term (row) and condition (column). They are marked by red for upregulated genes and by green for downregulated genes. P-values are presented in a heat map according to the color scheme shown in the graph. In panel B, we showed all the significant GO terms (P < 0.001 in any condition) in two categories of GO (Biological process and Cellular component). In panel C, we selected the most significant GO terms for each tissue and eliminated redundant GO terms based on the level of overlapping genes with another GO term with higher significance.

Download figure to PowerPoint

To determine the biological functions of the commonly regulated genes expressed in AC5 KO and CR mice, we subsequently analyzed these commonly regulated genes in CR and AC5 KO mice by their gene ontology (GO) annotations in Biological Process (BP) and Cellular Components (CC) categories. Highly significant GO entries in brain, heart, skeletal muscle, and liver in both models are shown in Fig. 3B,C, and Table S1 (Supporting information). Figure 3B shows all the significant GO terms (< 0.001 in any condition) in two categories of GO (BP and Cellular Component), and Fig. 3C and Table S1 (Supporting information) show the most significant GO terms for each tissue after redundant GO terms were eliminated. Our results indicated that the majority of commonly regulated genes in the brain were involved in sensory perception. Mitochondrial function is another major GO commonly regulated in the brain of AC5 KO and CR mice. The majority of commonly regulated genes in the heart are involved in G-protein-coupled receptor activity as well as protein processing including degradation, transport, and post-translational modification. The majority of commonly regulated genes in the skeletal muscle are involved in the regulation of muscle function, and the majority of commonly regulated genes in the liver of AC5 KO and CR mice are involved in metabolism, chromatin modification, and organization.

Furthermore, we analyzed the shared significant gene expression in different organs of AC5 KO and CR mice (Table S2). The major gene alterations in liver were related to metabolism, which was very different from other organs. We therefore focused on brain, heart, and skeletal muscle. Within 662, 1236, and 866 commonly upregulated genes (Fig. 3A, right panels), respectively, in the brain, heart, and skeletal muscle of both models, there were 30 genes shared in both heart and brain, 114 genes shared in both heart and skeletal muscle, and 13 genes shared in both brain and skeletal muscle (Table S2). Within 184, 307, and 255 commonly downregulated genes (Fig. 3A, right panels) in the brain, heart, and skeletal muscle of both models, there were eight genes shared in both brain and heart, 13 genes shared in both heart and skeletal muscle, and three genes shared in both brain and skeletal muscle (Table S2).

Adenylyl cyclase type 5 knockout and calorie restriction mice share mechanisms in different organs

Brain

The gene analysis in each GO entry from brain microarray (Fig. 3C and Table S1) revealed that AC5 KO and CR had a major impact on the expression of genes related to sensory perception. The receptors responding to smell, taste, and chemical stimulus such as olfactory receptors, taste receptors, and vomeronasal receptors were significantly downregulated. In addition, the gene set involved in the oxidation–reduction pathway was upregulated. Other interesting regulatory genes included histone-encoding genes (Hist2h3c1), different isoforms of ribosomal proteins (Rpl26, Rpl31, Rpl35a, Rpl4, Rps27l), and apolipoproteins (ApoD, ApoL 7b, ApoL 7c, and ApoL 9b).

Heart

Significant GO terms shown in Fig. 3C and Table S1 (Supporting information) revealed that both the protein ubiquitination and protein translation pathways were the most significantly upregulated pathways shared by CR and AC5 KO mice. In the protein ubiquitination pathway, the genes Uba52, Ubxn2a, Uchl1, Usp17I5, Usp48, Usp53, and Usp6nl were significantly upregulated in both CR and AC5 KO mice. Several stress response genes in the protein translation GO term such as Dnajb11 and Dnac16 (Hsp40) were commonly upregulated. These genes may involve protection in the hearts of both CR and AC5 KO mice.

We further compared the genes shared in AC5 KO and CR mice with those published data from pressure overload hypertrophy induced by 1-week transverse aortic constriction (TAC) (Park et al., 2011). As shown in Fig. 4A, there were 156 genes significantly upregulated in both AC5 KO and CR mice, but downregulated in TAC model. These oppositely regulated genes were mainly involved in mitochondrial regulation, cellular amine metabolic process, regulation of heart contraction, and peroxisomal regulation (Fig. 4B). Interestingly, the cardiac protective genes, which regulate GTPase activator activity (e.g., Rgs2, Arhgap19, Arhgap29) and glutathione transferase activity (e.g., Gstk1, Gstm) were found downregulated in the TAC model, but upregulated in AC5 KO and CR models. The genes that were commonly upregulated in AC5 and CR hearts, but downregulated in TAC model, could be involved in the cardioprotection of AC5 KO and CR mice, based on previous studies.

image

Figure 4. Comparison of gene regulation between hearts from adenylyl cyclase type 5 knockout (AC5 KO)/calorie restriction (CR) mice and hearts subjected to stress and after aging. We selected commonly regulated genes in AC5 KO and CR hearts and compared them with genes regulated in mouse pressure overload hypertrophy induced by 1-week transverse aortic constriction (TAC). Data from mice with 1-week TAC were compared with sham-operated mice (GSE24242). Significantly regulated genes are those with log2(ratio) > 1*standard deviation (SD) of all expressed genes. (A) The Venn diagram shows that a considerable number of genes are upregulated in AC5 KO and CR and downregulated in the stress heart model. Significant gene ontology (GO) terms are selected in those genes using hypergeometric test. (B) Genes with opposite expression patterns between CR/AC5 KO and TAC are shown. The genes form a common set of two overlapping regions marked by an arrow in the Venn diagram. The heat map is generated with gene expression changes (log2 ratio) from microarray data. Red indicates genes that are upregulated and green indicates those genes that are downregulated.

Download figure to PowerPoint

Skeletal muscle

As shown in Fig. 3C and Table S1 (Supporting information), AC5 KO and CR mice demonstrated highly similar GO changes involved in muscle function in the skeletal muscle. The gene sets mediating angiogenesis, muscle contraction, muscle development, circulatory system, and actin filament-based process were significantly upregulated in the skeletal muscle of both AC5 KO and CR mice.

Liver

We further identified commonly upregulated and downregulated metabolic genes in each GO entry in liver (Fig. 3C and Table S1). The selective shared genes are shown with their expression changes in Fig. S3A (Supporting information), and representative gene expression was validated by qPCR (Fig. S3B). The upregulated genes were related to gluconeogenesis, glycolysis, and lipid metabolism, and the downregulation of genes relate to sterol and steroid biosynthesis and metabolism, cholesterol biosynthesis and metabolism, and fatty acid metabolism. These data support the phenotype observed in AC5 KO and CR mice, for example, reduced body weight, fat accumulation, and glycogen stores.

Gene regulation shared by adenylyl cyclase type 5 knockout and calorie restriction mice in longevity and stress resistance

Although the regulation of olfactory receptors throughout lifespan is unknown in mammals, it has been reported that the olfactory receptors modulate longevity in Caenorhabditis elegans (Alcedo & Kenyon, 2004) and Drosophila melanogaster (Libert et al., 2007). Our microarray data analysis (Fig. 3C) demonstrated that downregulation of the G-protein-coupled receptor protein signaling pathway, particularly the olfactory receptors, was the most significant pathway shared by AC5 KO and CR mice in heart, brain, and skeletal muscle in both young- (4 months) (Fig. 5A) and intermediate-aged (14 months) groups (Fig. S6A), suggesting the olfactory receptor pathway may be involved in the regulation of longevity in AC5 KO and CR mice.

image

Figure 5. Common gene pathways shared by different organs in both adenylyl cyclase type 5 knockout (AC5 KO) and calorie restriction (CR) mice. (A) Downregulated olfactory receptors in different organs (brain, heart, and skeletal muscle) were determined by quantitative PCR. Red frame shows the same direction of alterations in different organs (n = 4 per group). (B) Western blots of Sirt1 in liver and heart. Sirt1 was significantly induced in both liver and heart of AC 5KO and CR mice (n = 4 per group). (C) Quantitative PCR analysis of ApoD. ApoD was significantly increased in brain, heart, and skeletal muscle of AC5 KO and CR mice (n = 4 per group). Data were expressed as mean ± SEM. *P < 0.05 vs. WT on CON.

Download figure to PowerPoint

In addition, we found that the stress resistance regulator in CR, Sirt1, was upregulated in liver and heart in both AC5 KO and CR mice at both young (4 months) (Fig. 5B) and intermediate age (14 months) (Fig. S6B), indicating AC5 KO mice share this important mechanism with CR. To examine the regulatory mechanism of transcription, we investigated transcription factor binding sites (TFBS) for known transcriptional factors in the promoter region of regulated genes in the brain, heart, skeletal muscle, and liver of both CR and AC5 KO mice. Interestingly, by TFBS analysis (Table S3 and S4), the regulatory genes in both CR and AC5 KO mice frequently contained the binding sites for the transcription factor such as FoxO (3 and 4), a key target of Sirt1 on chromatin. These results suggest that FoxO transcription factors are involved in common pathways shared by AC5 KO and CR mice in the regulation of longevity and stress resistance.

As shown in Table S2 (Supporting information), ApoD, another longevity regulator in Drosophila, was the only gene that was significantly upregulated in the three tissues, brain, heart, and skeletal muscle, in both AC5 KO and CR mice by microarray analysis. The gene was further validated by qPCR at both young (4 months) (Fig. 5C) and intermediate age (14 months) (Fig. S6C). This finding suggests ApoD could also regulate longevity in mammals.

Discussion

  1. Top of page
  2. Summary
  3. Introduction
  4. Results
  5. Discussion
  6. Materials and methods
  7. Acknowledgments
  8. Author contributions
  9. References
  10. Supporting Information

Calorie restriction and AC5 KO mice are two models of longevity and stress resistance (Sinclair, 2005; Yan et al., 2007; Canto & Auwerx, 2009), but as summarized in Fig. 6, the AC5 KO model differs from CR in that food intake and metabolic rate are reduced in CR as reported previously (Ferguson et al., 2007), but increased in AC5 KO mice. However, AC5 KO, like CR, mice are characterized by altered metabolism and reduced body fat, weight, glycogen stores, and fasting glucose levels. The initial goal of this study was to determine whether superimposing CR on the AC5 KO mice increased longevity and stress resistance further. This was not observed. When CR was superimposed on the AC5 KO mice, the animals rapidly lost weight and died within a month from a starvation syndrome, because of lack of adequate reserves of fat or glycogen to sustain the increased metabolic rate in this combined model. We then sought to define the common molecular pathways shared by AC5 KO and CR mice that may lead to a unified theory for aging and stress resistance.

image

Figure 6. Summarized differences and similarities between adenylyl cyclase type 5 knockout (AC5 KO) and calorie restriction (CR) mice. This figure summarizes differences and similarities between AC5 KO and CR mice in metabolism and the common pathways in liver, heart, skeletal muscle, and brain with respect to anti-aging and stress resistance.

Download figure to PowerPoint

The current studies were performed in WT and AC5 KO mice on a defined control diet, which contained 10% less calories than the ad lib diet (Dhahbi et al., 2004) compared with WT on CR, to insure that the controls for CR were not obese. To verify that that the control diet did not affect WT data significantly, we compared subgroups of WT on an ad lib diet and control diets for some key metabolic parameters in Fig. 2 and selective genes and proteins in Fig. 5 and confirmed that the data were essentially identical in these two control groups. Furthermore, although there was a small weight loss in WT when the mice were initially placed on the control diet (Fig. 2B), potentially because the control diet is different in color and shape, within a few weeks thereafter (at a time when microarray studies were performed), there was no statistical difference in body weights between these two subgroups, most likely because the mice became accustomed to the new food. It is most important to note that the major comparisons for the current study, between CR and AC5 KO, were conducted with both WT and AC5 KO on the control diet, so that this could not be a confounding factor.

Adenylyl cyclase type 5 knockout mice share many of the major compensatory metabolic mechanisms engaged in CR, which involve increased glucose utilization, enhanced ketogenesis, decreased glycogen stores, upregulated enzymes involved in gluconeogenesis, and increased insulin sensitivity (Dhahbi et al., 2001). Other models of longevity share these features. For example, mutant mice with growth hormone deficiency, including Ames (Murakami et al., 2003), Snell dwarf mice (Brown-Borg et al., 1996), and Little mice (Flurkey et al., 2001), all live longer and have a reduction in aging-associated phenotypes. A number of single-gene mutations on GH/IGF-1 pathway or its effectors, including growth hormone receptor/binding protein knockout (GHR/BP−/−) mice (Coschigano et al., 2000), IGF-1 receptor knockout (Igf1r+/−) mice (Holzenberger et al., 2003), fat-specific insulin receptor knockout (FIRKO) mice (Bluher et al., 2003), and hormone Klotho overexpression mice (Kurosu et al., 2005), also significantly extend lifespan. Most of these longevity models share metabolic similarities with CR and AC5 KO mice, including reduced body weight, decreased plasma glucose and insulin levels, as well as reduced growth hormone levels – all indications of insulin sensitization. Importantly, AC5 KO mice do not have reduced calorie intake; they even eat more and have increased metabolic rate. These data imply that inhibition of AC5 not only improves longevity but also could be a novel mechanism to treat obesity and diabetes.

Both CR (Gredilla & Barja, 2005) and AC5 KO mice (Yan et al., 2007) share mechanisms involving protection against oxidative stress. Genes related to oxidative stress regulation were observed in this analysis, but those in brain, skeletal muscle, and liver did not reach the same level of significance of other known and novel pathways. However, in the heart, increased expression of two genes encoding glutathione S-transferase (Gstk1, Gstm), which were downregulated in hypertrophic hearts (Fig. 4), seems to be important, as glutathione S-transferase is known to have an antioxidant role by conjugating glutathione on a variety of substrates, which may protect the heart against stress via reducing damage induced by reactive oxygen species (ROS) (Cho et al., 2003; Xiong et al., 2011). Interestingly, many of these molecular pathways involved in defense against antioxidant and other forms of stress are evolutionarily conserved and controlled by a handful of equally conserved transcriptional networks, involving the sirtuin and FoxO families of transcriptional regulators (Longo & Finch, 2003; Houtkooper et al., 2010). Our data demonstrated that Sirt1 was upregulated in AC5 KO mice similar to that observed in CR mice in both young- and intermediate-age groups. TFBS analysis also suggests that FoxO transcription factors, key stress response regulators and targets of Sirt1 on chromatin, were involved in common pathways shared by AC5 KO and CR mice. The sirtuin (NAD-dependent protein deacetylase) protein family (Sirt1-7) has received much attention for its regulatory role, mainly in metabolism and aging. Overexpression of sirtuins has been reported to increase lifespan in yeast (Kaeberlein et al., 1999), C. elegans (Tissenbaum & Guarente, 2001; Berdichevsky et al., 2006), and Drosophila (Rogina & Helfand, 2004) and has been suggested to be required for extending longevity in CR (Lin et al., 2000; Guarente & Picard, 2005). However, the extent to which sirtuins mediate longevity in mammals is controversial and was recently challenged (Burnett et al., 2011). Although the regulation of sirtuins on lifespan is controversial, the beneficial roles of sirtuins, particularly Sirt1 in promoting healthful aging, are generally accepted (Banks et al., 2008; Firestein et al., 2008; de Oliveira et al., 2010; Srivastava & Haigis, 2011; Sundaresan et al., 2011). Our data demonstrated that the level of Sirt1 was upregulated in both liver and heart not only in CR but also in the AC5 KO model.

In addition to these known stress and longevity pathways, our work also identified novel pathways shared by AC5 KO and CR mice, which could mediate longevity and stress resistance. For example, apolipoprotein D (ApoD) was also upregulated in both AC5 KO and CR mice. Importantly, the ApoD gene was the only gene that was upregulated in the three organs studied, that is, heart, brain, and skeletal muscle. ApoD was found previously to be elevated in neurodegenerative diseases and in aging brain (Rassart et al., 2000). However, it was reported recently that ApoD played an important role in lifespan extension and stress resistance in Drosophila and protected against oxidative stress in mice (Sanchez et al., 2006; Walker et al., 2006; Ganfornina et al., 2008; Muffat et al., 2008). Overexpression of either a Drosophila homolog of ApoD (Walker et al., 2006) or human ApoD (Muffat et al., 2008) leads to increased lifespan and stress resistance in Drosophila. Conversely, loss of the homolog of ApoD reduces lifespan and stress resistance in flies (Sanchez et al., 2006). More interestingly, overexpression of human ApoD in the mouse protects against oxidative stress, and ApoD knockout mice show more sensitivity to oxidative stress (Ganfornina et al., 2008). Thus, our data, supported by the literature, suggest that ApoD is a conserved protective mechanism in the regulation of lifespan and stress resistance.

Another novel finding that we found was significant downregulation of olfactory receptors in both CR and AC5 KO mice in both young- and intermediate-age groups. Although the regulation of lifespan by olfactory receptors has not been investigated in mammals, it has been found that odorants were sufficient to modulate lifespan in C. elegans (Alcedo & Kenyon, 2004) and Drosophila (Libert et al., 2007). Therefore, our findings suggest that olfactory receptors could also mediate longevity in mammals. The current study focuses on demonstrating the mechanisms shared by CR and AC5 KO. The genes identified in this study will need to be studied in further depth in future studies, by our laboratory or others.

In summary (Fig. 6), AC5 KO and CR mice share many tissue-specific pathways in the regulation of longevity and stress resistance and many molecular pathways in the brain, skeletal muscle, heart, and liver, with a unifying feature of altered regulation of metabolism, resulting in lower body weight (Canto & Auwerx, 2009; Houtkooper et al., 2010). The major similarities between CR and AC5 KO are that Sirt1 and Apo D are upregulated in both models, whereas olfactory receptors are downregulated in both models. We also found an inverse pattern in the heart between genes identified in AC5 KO and CR vs. genes regulated by stress, that is, pressure overload to the heart (TAC) in WT mice. Cardioprotective genes are upregulated in hearts from AC5 KO and CR mice compared with downregulation of these genes in TAC.

These pathways may also help to understand the mechanisms underlying the inverse relationship between body weight and aging in humans. The regulation of several genes, which have previously been reported to affect lifespan and stress resistance in lower organisms upon CR, such as Sirt1/FoxO, ApoD, and olfactory receptors, is shared by CR and AC5 KO mice. The demonstration that two models of increased longevity and stress resistance share metabolic phenotypes and also genotypes, and also with aging models from lower organisms, will help to provide a basis for a unified theory for longevity and stress resistance.

Materials and methods

  1. Top of page
  2. Summary
  3. Introduction
  4. Results
  5. Discussion
  6. Materials and methods
  7. Acknowledgments
  8. Author contributions
  9. References
  10. Supporting Information

Animal model and diets

The majority of experiments were performed in 3- to 4-month-old male AC5 KO and their WT littermates on control diet (CON) and CR. AC5 KO mice were generated as described previously (Okumura et al., 2003). Key findings from these mice were confirmed in 14-month-old WT, AC5 KO, and CR mice. These mice have been backcrossed into 129/Sv background, by mating with 129S2/SvPasCrl (Strian Code: 476 from Charles River) for more than seven generations. Starting at 8 weeks of age, all mice were randomized to either CON or CR. The CON group was fed 93 kcal per week of chemically defined control diet (AIN-93M, Diet No. F05312; Bioserv, Frenchtown, NJ, USA) as described previously (Dhahbi et al., 2004), which provided ~10% fewer calories than ad lib, but was apparently sufficient as the food was frequently not completely consumed. This allowed comparison of nonobese controls with CR mice. To confirm that the control diet does not introduce a confounding variable, we compared WT on a control diet and WT on an ad lib diet for some of the key metabolic parameters in Fig. 2 and selective genes and proteins we observed in Fig. 5. The data were essentially identical in these two control groups (Figs S4 and S5). We also compared body weight changes in new groups of mice (n = 6/group). Using ANOVA, body weights were not statistically different in mice on ad lib and control diets, after 4–6 weeks when the mice became accustomed to the new diet. It was at this point that microarrays were conducted. In Fig. 2B, there was a small decrease in body weight in mice on the control diet, most likely because the mice were only studied for 3 weeks, at a time which they were not yet accustomed to the change in diet; the control diet differs from the ad lib diet in both shape and color of the pellets. CR was introduced by reducing calories to 77 kcal per week of chemically defined diet for first 2 weeks (AIN-93M 25% restricted, Diet No. F05314; Bioserv), followed by 52.2 kcal per week of CR diet thereafter (AIN-93M 40% restricted, Diet No. F05314; Bioserv). 40% of CR was widely used in previous aging and longevity studies (Dhahbi et al., 2004; Liao et al., 2010). All animals were housed individually in ventilated cages with the temperature between 68–75 °C and humidity between 40–60%, and with 12-h light/12-h dark. Body weight was measured twice a week starting at day 0. Body weight change and food intake as a function of time were statistically analyzed to detect significant weight differences between the groups. Food intake was obtained by subtracting the weight of the remaining food from the initially supplied food. Both control and CR diets were cold-packed into 1-g pellets. The restricted diet is enriched in protein, vitamins, and minerals to avoid malnutrition. All mice were fed daily in a similar time of the morning and given one-seventh of the weekly allotment of food on each day. Cages were checked for pieces of food, and these pieces were transferred to new cages when the bedding was changed. The breeders were fed an autoclavable mouse breeder chow 5021 from LabDiet. This study was approved by the Institutional Animal Care and Use Committee at New Jersey Medical School. All investigations conformed to the National Institutes of Health Guide for the Care and Use of Laboratory Animals.

Metabolic measurements

Mice were fasted for 6 h prior to blood being drawn via venous tail puncture. From these samples, blood glucose was measured with a glucose meter. Total ketone bodies and free fatty acid levels in blood were measured using commercially available kits from Wako Diagnostics. Insulin levels were measured by ELISA (Crystal Chem, Dovers Grove, IL, USA). Liver glycogen was assessed histologically using periodic acid-schiff (PAS) staining (Saitoh et al., 2010).

Calculation of adiposity index

Epididymal, retroperitoneal, and inguinal fat pads were isolated and weighed for calculation of adiposity index (total adipose depot weight/live body weight × 100) (Chiu et al., 2007).

Microarray

Total RNA was extracted from liver, heart, and whole brain of each mouse with phenol–chloroform and purified with RNeasy® Mini Kit (QIAGEN, Valencia, CA, USA). Heart, brain, and skeletal muscle samples were hybridized to Affymetrix GeneChip® Mouse Exon 1.0 ST arrays, and liver samples were hybridized to Affymetrix GeneChip® Mouse Genome 430 2.0 arrays. The preparation of the samples (double-strand cDNA synthesis, and synthetic RNA synthesis and labeling) as well as hybridization and scanning of the microarrays were performed in the Microarray Facility at Yale University.

Microarray data analysis

Probe set signal intensity was estimated by the robust multi-chip analysis (RMA) method. Gene density maps were used to compare gene expression changes in AC5 KO and CR mice. All expressed genes were evenly divided into 20 groups based on fold change in each comparison and then distributed in a 20 × 20 grid. Each cell has a defined range of expression ratios for each condition. The number of genes in each cell was normalized to an expected number derived from randomized data. The colors showed enrichment (red) or depletion (blue). Venn diagram showed the total number of genes that were regulated (either up- or downregulated) in both models as compared to the corresponding WT mice. Significantly regulated genes (> 1× standard deviation of all genes) were selected. We identified the functional significance of commonly regulated genes between AC5 KO and CR mice using gene sets previously defined based on biological knowledge. The GO annotations for genes were obtained from the gene database of NCBI. The relationship between transcription factor and the target genes was retrieved from the Molecular Signatures Database (MSigDB) (Xie et al., 2005). The association between each gene set and regulated genes was assessed using hypergeometric test or cumulative distribution function (CDF) as previously described (Park et al., 2011). The statistical testing was separately achieved with up- and downregulated genes.

Quantitative RT–PCR

Specific primers and probes (derived with FAM and TAMRA, ordered from IDT DNA Company) were designed for the transcripts of interest. The optimal combination of primers and probes for a qPCR assay was determined with the Primer Express software (Applied Biosystems, Foster City, CA, USA). Following reverse transcription of the mRNA of interest from 50 ng of total RNA, the cDNA was used for quantitative PCR (qPCR) (40 cycles of a 10-s step at 95 °C and a 1-min step at 60 °C) using the SybrGreen method on a 7700 ABI-Prizm Sequence Detector (Applied Biosystems). Values are reported per cyclophilin transcript to correct for sample-to-sample RNA loading variations. The primer sequences are provided in Table S5 (Supporting information).

Western blotting

Proteins separated by SDS–PAGE were transferred to nitrocellulose membranes. The membranes were probed with Sirt1 rabbit polyclonal antibody (1:1000 dilution) purchased from Millipore (Billerica, MA, USA) (Cat. No. 07-131) at 4 °C overnight. The bands were visualized using chemiluminescence reagents. The linear range of detection for different proteins and band intensities were determined by densitometry. Blots were re-probed with GAPDH to equalize sample loading.

Acknowledgments

  1. Top of page
  2. Summary
  3. Introduction
  4. Results
  5. Discussion
  6. Materials and methods
  7. Acknowledgments
  8. Author contributions
  9. References
  10. Supporting Information

We thank Ms. Hui Ge for technical support of Western blotting analysis and Dr. Yimin Tian for immunohistology analysis. This work was supported by National Institutes of Health grants 5P01AG027211, 5R21HL097264, 1R01HL102472, 5R01HL033107, 5T32HL069752, 5R01HL095888, 5P01HL069020, 5R01HL091781, R01HL106511, R01HL093481, DK059820, the École Polytechnique Fédérale de Lausanne, the Swiss National Science Foundation, the Velux Foundation, and the EU ideas program (ERC-2008-AdG-23118).

Author contributions

  1. Top of page
  2. Summary
  3. Introduction
  4. Results
  5. Discussion
  6. Materials and methods
  7. Acknowledgments
  8. Author contributions
  9. References
  10. Supporting Information

Drs. Stephen Vatner and Lin Yan designed the study, analyzed and interpreted the data, and prepared the manuscript. Drs. Johan Auwerx, Dorothy Vatner, William Stanley, and Bin Tian advised on design and revised the manuscript. Dr. Ji Yeon Park performed microarray data analysis. Dr. David Ho interpreted the data and summarized the literature. Dr. Jean-Guillaume Dillinger, Dr. Mariana De Lorenzo, Dr. Chujun Yuan, Ms. Lo Lai, and Dr. Chunbo Wang performed the majority of experiments.

References

  1. Top of page
  2. Summary
  3. Introduction
  4. Results
  5. Discussion
  6. Materials and methods
  7. Acknowledgments
  8. Author contributions
  9. References
  10. Supporting Information

Supporting Information

  1. Top of page
  2. Summary
  3. Introduction
  4. Results
  5. Discussion
  6. Materials and methods
  7. Acknowledgments
  8. Author contributions
  9. References
  10. Supporting Information
FilenameFormatSizeDescription
acel12013-sup-0001-FigS1-S6-TableS1-S5.pdfapplication/PDF282K

Fig. S1 AC5 expression in mouse and human tissue expression panel derived from public database.

Fig. S2 Change of body weight and organ weight change after CR.

Fig. S3 Genes shared by AC5 KO and CR in liver.

Fig. S4 Comparison of control diet vs. ad lib diet on metabolism.

Fig. S5 Comparison of control diet vs. ad lib diet (n = 6/group) on the selective genes of olfactory receptors (Olfrs) and Sirt1 which were commonly shared by AC5 and CR (Fig. 5).

Fig. S6 Common gene pathways shared by different organs in 14 month old AC5 KO and CR.

Table S1 Similarities in gene ontology (GO) changes between AC5 KO and CR.

Table S2 Shared gene expression in different tissues of AC5 KO and CR.

Table S3 Upregulated and downregulated genes in both CR and AC5 KO that frequently contained binding sites for the following transcription factors.

Table S4 Significant TFBS.

Table S5 Primer sequences for qPCR.

Please note: Wiley Blackwell is not responsible for the content or functionality of any supporting information supplied by the authors. Any queries (other than missing content) should be directed to the corresponding author for the article.