DAF-16 induces a switch to low metabolic rate whereas DAF-12 controls dauer morphogenesis
To distinguish which signaling pathways control metabolism in the dauer state, we investigated the metabolic activities of wild-type dauers, as well as of mutants of key dauer regulatory factors. Dauer larvae have substantially diminished metabolic rate and heat production compared to other larval stages [10]. For that reason, we first examined the heat flow produced by wild-type worms undergoing reproductive development or dauer formation from the L1 larval stage onwards using time-resolved isothermal microcalorimetry. To induce a synchronous dauer formation, we grew worms on 4-methylated sterol (4-MS), which blocks the production of DAs [26]. After an initial increase of heat flow by both groups, the two trends diverged after about 24 h, increasing further in worms in the reproductive mode, while decreasing in animals that underwent dauer formation (Fig. 1b and Additional file 1: Fig. S1a). A similar trend was observed in the TGF-β Daf-c mutant daf-7, which forms dauer larvae but reproduces when DA is added [19, 38] (Additional file 1: Fig. S1b).
Next, we set out to determine how the insulin and steroid hormone pathways contribute to the switch. To disentangle the pathways, we chose conditions under which DAF-16 is active but not DAF-12. We made use of a group of Daf-c alleles of daf-2, designated class II, that is not fully suppressed by Daf-d mutations of daf-12 or by the addition of DAs [19, 25]. One such allele is daf-2(e1370). Worms bearing this mutation reproduce at the permissive temperature of 15 °C, whereas at the restrictive temperature of 25 °C, they form dauers [25]. We compared this strain to a double mutant daf-2(e1370);daf-12(rh61rh411) or DA-treated daf-2(e1370) that arrest the development at an L3-like larval stage at 25 °C due to DAF-16 activity (Fig. 1c) [19, 25]. The metabolism and morphology of these larvae have not been characterized in detail. Unlike daf-7 on DA, at 25 °C, daf-2;daf-12 and daf-2 on DA shifted the metabolic mode to low heat production after 24 h (Fig. 1d and Additional file 1: Fig. S1b and S1c). In contrast, a double mutant daf-2(e1370);daf-16(mu86) that undergoes reproductive development at 25 °C [25] displayed high heat production (Fig. 1d and Additional file 1: Fig. S1c). Thus, DAF-12 is not required for the switch to low metabolic activity during dauer formation when DAF-16 is active.
We further asked whether DAF-16 or DAF-12 determines the morphology of dauer larvae and whether metabolic state and morphology are interconnected. Our previous studies indicated that DAF-12 could induce morphological features of dauer larvae in the absence of DAF-16 [26]. However, it was not clear whether the activation of DAF-16 alone could promote dauer morphology. To test this, we performed electron microscopy on daf-2;daf-12 and DA-treated daf-2 larvae grown at 25 °C. Interestingly, similar to L3 larvae, they had a large body diameter and an elongated gut lumen with long, densely packed microvilli, and lacked characteristic features of the dauer cuticle such as alae and a striated layer (Fig. 1e and Additional file 2: Fig. S2) [39]. However, similar to dauers, daf-2;daf-12 and DA-fed daf-2 animals deposited numerous lipid droplets (LDs) (Fig. 1e and Additional file 2: Fig. S2). Thus, DAF-12 controls dauer morphogenesis, whereas DAF-16 has no direct influence on this process but appears to affect the dauer-associated metabolic changes that culminate in low metabolic rate and high LD accumulation.
DAF-16 controls catabolism and, together with DAF-12, promotes a shift from TCA cycle-driven metabolism to gluconeogenesis
Cells produce heat almost exclusively through catabolic reactions [40]. Thus, low heat production in dauers indicated that they have decreased catabolism. To determine which pathways regulate this process, we compared the amounts of heat that daf-2 dauers and daf-2;daf-12 L3-like larvae produce after entering the arrested state. Food was omitted to exclude heat generation by bacteria. daf-2 dauers and daf-2;daf-12 larvae generated similar amounts of heat (Fig. 2a and Additional file 3: Fig. S3a), suggesting that DAF-16 does not require DAF-12 activity to regulate the energy expenditure. We next asked how the loss of DAF-16 activity would influence the metabolic rate. For the inactivation of DAF-16, we used daf-16(mu86) mutants grown on 4-MS. Under these conditions, DAF-12 promotes the dauer program, but DAF-16 is absent and worms arrest as dauer-like animals [26] (Fig. 2b). Compared to wild-type dauers, 4-MS-treated daf-16 animals displayed higher heat production in their arrested state (Fig. 2c and Additional file 3: Fig. S3b), suggesting that DAF-16 suppresses metabolic rate and the catabolism of energy stores of dauers.
We, therefore, monitored the breakdown of lipids, sugars, and amino acids in daf-2;daf-12 and daf-16 on 4-MS. Because food was omitted, only the catabolism of internal energy reserves would account for any observed change. Storage triglycerides (TGs) were visualized by coherent anti-Stokes Raman scattering (CARS) microscopy of LDs and by thin-layer chromatography (TLC). daf-2 dauers and, to a slightly lesser extent, daf-2;daf-12 arrested larvae and wild-type dauers on 4-MS, conserved their TGs over time (Fig. 2d and Additional file 4: Fig. S4a). In contrast, daf-16 larvae on 4-MS were depleted of TGs only after 2 days (Fig. 2d and Additional file 4: Fig. S4a). Phospholipids were preserved in all animals, suggesting that no substantial degradation of membranes occurred (Additional file 4: Fig. S4a). Furthermore, sugars and amino acids were maintained at high levels in wild-type dauers on 4-MS, daf-2, and daf-2;daf-12 larvae, but were rapidly degraded in daf-16 on 4-MS (Additional file 4: Fig. S4a and S4b). These results show that, in the absence of DAF-16, catabolism becomes misregulated and the energy depot is dramatically depleted.
The above findings suggest that faster depletion of energy reserves might reduce survival in daf-16 larvae. Indeed, the viability of these animals declined rapidly and they perished after 12 days, while almost 100% of dauers and daf-2;daf-12 arrested larvae remained viable (Fig. 2e). Thus, DAF-16 regulates the survival of dauer larvae by controlling energy expenditure.
The switch to a lower catabolic rate in dauers is accompanied by a shift from TCA cycle-driven metabolism to gluconeogenesis, leading to the accumulation of the disaccharide trehalose [9]. To determine whether DAF-16 or DAF-12 is responsible for this transition, we used 2D-TLC to trace the metabolism of 14C-radiolabeled acetate. Carbon atoms of acetate are only incorporated into trehalose if the glyoxylate shunt and the gluconeogenesis are active [9]. This labeling strategy mimics the usage of endogenous lipids as a carbon source for gluconeogenesis because both the lipid catabolism and the external acetate provide acetyl-CoA that enters the TCA or the glyoxylate cycle. As shown before [9], daf-2 dauers at 25 °C displayed stronger accumulation of labeled trehalose than L3 larvae at 15 °C (Fig. 2f). High incorporation of acetate into trehalose was also observed in daf-2;daf-12 arrested L3 larvae (Fig. 2f). Thus, activation of DAF-16 is sufficient to trigger gluconeogenesis. Surprisingly, we also detected higher levels of labeled trehalose in daf-16 mutants cultured on 4-MS, suggesting that DAF-12 can promote a gluconeogenic mode in the absence of DAF-16 (Fig. 2g). Besides, we noticed that all developmentally arrested stages displayed lower levels of a band corresponding to the amino acids alanine and threonine compared to the related L3 larvae (Fig. 2f, g). Because these amino acids can be derived from glycolysis/gluconeogenesis and the TCA/glyoxylate cycle (from pyruvate and oxaloacetate, respectively), we concluded that their synthesis might be downregulated redundantly by DAF-16 or DAF-12 to ensure higher availability of substrates for gluconeogenesis. In this context, DAF-16 may exert more stringent regulation because animals bearing daf-2(e1370) showed lower levels of these compounds in both growing and arrested state compared to the wild-type or daf-16 worms. It is known that even at the permissive temperature, DAF-16 is nuclear in daf-2(e1370) but not in the wild-type worms [18], suggesting that this trait may be transcriptionally promoted by DAF-16.
Together, our results demonstrate that DAF-16 alone maintains low catabolism, whereas DAF-16 and DAF-12 separately promote a shift from TCA cycle-driven metabolism to gluconeogenesis. Besides, the low catabolism and the gluconeogenic mode are independent metabolic modules that can be uncoupled under conditions of low DAF-16 but high DAF-12 activity.
AAK-2 is required for the DAF-16-mediated metabolic switch and developmental arrest
We postulated that under conditions of high DAF-16 but low DAF-12 activity (Fig. 1c), the disruption of a hypothetical factor required for the DAF-16-mediated metabolic switch could promote higher catabolism and prevent the gluconeogenic mode. Moreover, if the metabolic and developmental transitions are coupled, one prediction would be that such an intervention may rescue the developmental arrest caused by DAF-16. Thus, it was of high importance to identify such a factor. The uncontrolled catabolism and mortality in daf-16 on 4-MS were very similar to those observed in dauers with loss of activity of the AMPK α-subunit AAK-2 [41]. Thus, DAF-16 and AAK-2 may jointly control the metabolic state of dauers, making AMPK a potential candidate for this factor. We first asked whether aak-2 mutant dauers lose TGs, sugars, and amino acids similar to daf-16 on 4-MS. We generated a daf-2(e1370);aak-2(gt33) strain harboring a large deletion in aak-2. At 25 °C, almost all animals formed dauer larvae with typical dauer morphology (Fig. 3a). Curiously, although a previous study that used a strain daf-2(e1370);aak-2(ok524), bearing a different deletion in aak-2, showed that at 25 °C, the animals spontaneously exited from dauer state and produced adults within 5 days [42], dauers of daf-2(e1370);aak-2(gt33) grown on a solid medium with ample food for 5 days at 25 °C did not undergo spontaneous exit from dauer state. Almost all worms survived the treatment with the detergent SDS, which is a hallmark of dauer larvae [43] (Additional file 5: Fig. S5a and S5b). The dauer-specific alae and striated layer of the cuticle were also preserved over time (Fig. 3a). Hence, using daf-2(e1370);aak-2(gt33) is a very suitable model to study the metabolic control in the dauer state.
Electron micrographs suggested that after 5 days, daf-2;aak-2 dauers enter a state of starvation characterized by an extreme decrease of the cellular volume of the hypodermis, expansion of the body cavities, and deterioration of the mitochondria (Fig. 3a). In line with these observations, TGs and trehalose (Fig. 3b, c), as well as amino acids (Additional file 5: Fig. S5c), were rapidly depleted. Phospholipids were less affected (Fig. 3b, c). To test the cellular response to starvation in AMPK mutants, we monitored FIB-1, a small nucleolar ribonucleoprotein (snoRNP) whose localization to the nucleolus is reduced during starvation or loss of TOR activity [44]. Dauers without Daf mutations isolated from overcrowded plates had nucleolar FIB-1 consistent with a non-starved state (Additional file 5: Fig. S5d). daf-2 and daf-2;aak-2 dauers, as well as daf-2 DA-fed larvae, also displayed nucleolar FIB-1 shortly after the arrest (Additional file 5: Fig. S5e). This localization remained unchanged over time in daf-2 dauers and daf-2 DA-fed arrested L3 larvae (Additional file 5: Fig. S5e). In daf-2;aak-2 dauers, however, FIB-1 formed granular structures in the nucleoplasm after 4 days and was almost completely dispersed in the nucleoplasm of cells after 7 days (Additional file 5: Fig. S5e). Thus, daf-16 and aak-2 mutants have highly related phenotypes in terms of catabolism in the dauer state and may act in the same pathway. Moreover, DAF-16 could prevent a TOR-dependent starvation response in the absence of DAF-12, but not of AAK-2, suggesting that AMPK might be required for the DAF-16-mediated maintenance of the energy reserves.
Since the disruption of aak-2 enhanced catabolism in daf-2 dauers, we asked if it could also abolish the gluconeogenic mode. 14C-acetate labeling in daf-2;aak-2 at 25 °C showed a pronounced gluconeogenic mode (Fig. 3d). Because DAF-12 could activate gluconeogenesis in the absence of DAF-16, we asked whether it could perform this activity also in the absence of AAK-2. Remarkably, when we inhibited DAF-12 in daf-2;aak-2 by adding DA, the gluconeogenesis was abolished (Fig. 3d). Furthermore, daf-2;aak-2 showed higher glycine/serine levels at 15 °C and 25 °C with DA and lower at 25 °C without DA (Fig. 3d). Since glycine can be produced from glyoxylate, and serine from the glycolysis/gluconeogenesis intermediate 3-phosphoglycerate, the production of these amino acids from acetate may be kept low by AAK-2 and DAF-12 so that glyoxylate and 3-phosphoglycerate are more efficiently used in trehalose synthesis. Hence, AAK-2 fulfills the criteria for a factor required for the switch to low energy expenditure and to gluconeogenesis induced by DAF-16 when DAF-12 is inhibited. To determine whether AAK-2 is also necessary for the DAF-16-induced growth arrest, we monitored the development of daf-2;aak-2 worms at 25 °C with DA. Astoundingly, these worms completely bypassed dauer arrest and developed into adults (Fig. 3e, f). Thus, in daf-2 mutants, AAK-2 is essential for the DAF-16-mediated growth and metabolic transition in the absence of DAF-12 activity.
An interaction between daf-2 and aak-2 has also been observed in the context of adult longevity: the lifespan extension characteristic for daf-2 animals is fully suppressed by aak-2 mutations [45]. Thus, the metabolic mode associated with increased longevity in adult daf-2 mutants [46] could depend on AAK-2. To assess the gluconeogenesis, we labeled daf-2 and daf-2;aak-2 with 14C-acetate and grew them at 15 °C until the L4 stage to bypass dauer formation. From this point on, we either kept them at 15 °C to maintain the DAF-2 activity high or shifted them to 25 °C to suppress DAF-2. This temperature shift doubles the lifespan of daf-2 adults compared to the wild-type worms [47]. Twenty-four hours later, we extracted the metabolites and observed much higher accumulation of labeled trehalose in daf-2 worms at 25 °C as compared to 15 °C (Additional file 5: Fig. S5f). daf-2;aak-2 displayed lower incorporation of 14C-acetate into trehalose at both 15 °C and 25 °C in comparison with daf-2. This observation shows that AAK-2 is required for the full extent of the metabolic switch in daf-2 adults. A small elevation of labeled trehalose in daf-2;aak-2 at 25 °C compared to 15 °C (Additional file 5: Fig. S5f) suggests that in adults, DAF-16 could also promote gluconeogenesis to a very limited degree in an AAK-2-independent manner. One limitation of our study is that the 2D-TLC does not allow for quantitative evaluation of intermediate phenotypes such as the one observed in adult worms of daf-2;aak-2 at 25 °C. Future studies will be required to evaluate the exact flux through the metabolic pathway. Based on our observations, we conclude that the metabolic switch does not only determine dauer diapause but is also activated in adult worms with reduced insulin signaling and correlates with the lifespan extension typical for them.
Gluconeogenesis is turned on by a shift in the molar ratios of key metabolic enzymes
To gain insight into the molecular mechanism underlying the switch to gluconeogenesis and how AAK-2 and DAF-12 control it, we employed the LC-MS/MS method of MS Western [48] to quantify the absolute (molar) amount of 43 individual enzymes or subunits of enzymatic complexes involved in TCA cycle, glyoxylate shunt, glycolysis, gluconeogenesis, and mitochondrial pyruvate metabolism. The molar abundance of each protein was determined by comparing individual abundances of several (typically, 2 to 5) quantitypic peptides with 13C, 15N-isotopically labeled peptide standards (Additional file 6: Fig. S6a-f). Standard peptides were concatenated into a protein chimera (Additional file 6: Fig. S6g) that was in-gel co-digested with target proteins separated by one-dimensional SDS-PAGE from a whole animal lysate. The molar amount of chimera protein was referenced to the standard of BSA and quantified in the same LC-MS/MS experiment. MS Western quantification was highly concordant. The median coefficient of variation of molar abundances of proteins determined using alternative standard peptides was less than 10% (Additional file 6: Fig. S6h) with better than 0.99 Pearson coefficient of correlation between technical replicas (Additional file 6: Fig. S6i). The molar abundances of individual proteins were normalized to the total protein content in each animal lysate and could be directly compared between all biological conditions without metabolic or chemical labeling of target proteins.
We first analyzed the enzyme levels in dauers (daf-2 at 25 °C) and L3 larvae (daf-2 at 15 °C). The 43 enzymes were detected in a wide range of 1 to nearly 160 fmol/μg of the total protein (Fig. 4a, Additional files 7, 8, and 9: Fig. S7, S8a, and Tab. S1). Although a global metabolic perturbation would be expected, we found that the balance of molar abundances of members of different pathways between the two groups was not perturbed. However, in dauers, the enzymes of glycolysis/gluconeogenesis were slightly more prevalent with respect to other pathways indicating enhanced gluconeogenesis (Additional file 8: Fig. S8b). Interestingly, in total, dauers were 1.7-fold more enriched in metabolic enzymes compared to L3 larvae, despite that dauer is a metabolically reduced stage (Additional file 8: Fig. S8c). Glycolysis/gluconeogenesis enzymes were enriched to a greater extent in dauers compared to L3 larvae (Additional file 8: Fig. S8d). Hence, the overall architecture of the metabolic network was preserved in both developmental conditions, while the metabolic switch was executed by fine-tuning its directionality.
To understand the metabolic switch mechanism, we reconstructed the pathway that converts lipid-derived acetyl-CoA to carbohydrates via glyoxylate shunt and gluconeogenesis (Fig. 4b, Additional files 7 and 8: Fig. S7 and S8a). We focused on several reactions that act as branch points or “metabolic turnouts.” The first set of reactions determines whether acetyl-CoA, via isocitrate, will enter the TCA or the glyoxylate cycle (Fig. 4b and Additional file 7: Fig. S7). We observed 3.5-fold upregulation of the glyoxylate cycle enzyme ICL-1 in dauers compared to L3 larvae (Fig. 4b and Additional file 7: Fig. S7). This was consistent with previous studies [29, 46] and suggested that dauers have higher glyoxylate pathway activity. Thus, they convert isocitrate to malate and succinate without losing carbon atoms in the TCA cycle via decarboxylation (Fig. 4b and Additional file 7: Fig. S7). Whether these carbon atoms are used for gluconeogenesis depends on the reactions at the second branch point. This point will determine whether the oxaloacetate produced downstream of the glyoxylate pathway is recycled by the citrate synthase or converted to phosphoenolpyruvate (PEP) by phosphoenolpyruvate carboxykinase (PEPCK) (Fig. 4b and Additional file 8: Fig. S8a). The production of PEP by PEPCK is the first and pathway-specific step of gluconeogenesis. Similar to ICL-1, the two isoforms of PEPCK, PCK-1 and PCK-2, were 2-fold elevated in daf-2 dauers compared to L3 larvae (Fig. 4b and Additional file 8: Fig. S8a). Thus, dauers have a higher ability to use acetyl-CoA for gluconeogenesis. This ability is further supported by ~ 2-fold increase of oxaloacetate-producing malate dehydrogenase (MDH-1) and enzymes shared between gluconeogenesis and glycolysis such as enolase (ENOL-1) and aldolase (ALDO-1) (Fig. 4a, Additional files 7 and 8: Fig. S7 and S8a).
As shown above, simultaneous inactivation of DAF-12 and AAK-2 in daf-2 at 25 °C prevents the gluconeogenic mode and the developmental arrest (Fig. 3d–f). Thus, we asked whether DAF-12 or AAK-2 is responsible for the altered expression of the enzymes. To test this, we quantified all 43 enzymes in daf-2 on DA and daf-2;aak-2 with or without DA at 25 °C. We first analyzed the TCA/glyoxylate cycle branch point. As expected, larvae with high gluconeogenic mode (daf-2 on DA and daf-2;aak-2 without DA) showed similar upregulation of ICL-1 as in daf-2 dauers (Fig. 4b and Additional file 7: Fig. S7). Interestingly, daf-2;aak-2 on DA also had higher amounts of this enzyme despite the low gluconeogenic mode (Fig. 4b and Additional file 7: Fig. S7). We reasoned that not only the absolute levels of ICL-1 but its molar ratio with respect to competing TCA cycle enzymes (isocitrate dehydrogenases) controls the isocitrate flow into the glyoxylate shunt (Fig. 5a). Indeed, ICL-1 displayed the lowest molar ratio to all isocitrate dehydrogenase subunits and isoforms in daf-2 L3 larvae at 15 °C (Fig. 5b, c). This was consistent with a more intensive TCA cycle. The ratios were overall higher in daf-2 and daf-2;aak-2 at 25 °C with or without DA (Fig. 5b, c). However, in stages with pronounced gluconeogenic mode (daf-2 without or with DA at 25 °C, daf-2;aak-2 at 25 °C), this elevation was much more pronounced (3.5-, 3.8-, and 4.2-fold) compared to daf-2;aak-2 at 25 °C with DA (2-fold, Fig. 5b). In DA-treated daf-2;aak-2, ICL-1 was much less dominant in respect to the IDHG-1 subunit of the NAD+-dependent isocitrate dehydrogenase and, importantly, to the NADP+-dependent IDH-1, which has been implicated in the regulation of DA signaling [14] (Fig. 5c, Additional file 8: Fig. S8e and S8f). Thus, simultaneous inactivation of DAF-12 and AAK-2 lowers the capacity of the glyoxylate shunt.
We next asked whether DAF-12 or AAK-2 controls the production of PEP by PEPCK at the second branch point. DA-treated daf-2 at 25 °C had similarly elevated PCK-1 and PCK-2 as daf-2 dauers (Fig. 4b and Additional file 8: Fig. S8a). The AAK-2-deficient strain, however, showed low PCK-1 and PCK-2 levels regardless of the presence or absence of DA (Fig. 4b and Additional file 8: Fig. S8a). This indicated that AAK-2 might promote the conversion of oxaloacetate to PEP (Fig. 5d). Accordingly, daf-2 at 25 °C with or without DA showed lower molar ratios of citrate synthase to PEPCK, whereas in daf-2;aak-2 at 25 °C with or without DA, they were much higher (Fig. 5e, f). It is, therefore, possible that the gluconeogenic mode in daf-2;aak-2 at 25 °C without DA is maintained through higher recycling of oxaloacetate through the glyoxylate shunt. Future analysis will determine the precise metabolic fluxes in these animals.
Enzyme stoichiometry may also influence the flux through bidirectional reactions that are catalyzed by separate enzymes in the opposite directions if the molar ratios between these enzymes change. One such step, the rate-limiting interconversion of fructose 6-phosphate and fructose 1,6-bisphosphate in glycolysis and gluconeogenesis, largely determines the balance between the two pathways and is under the control of multiple metabolic regulators including AMPK [49]. However, when we compared L3 and dauer larvae, we did not detect significant changes in the molar ratio between the glycolytic enzyme PFK-1 and the gluconeogenesis-specific enzyme FBP-1 which are responsible for these reactions (Additional file 8: Fig. S8g). A limitation of our study is the lack of information on the post-translational modifications of the enzymes that would complement the absolute molar quantification. However, we can conclude that in such reactions, most probably, conserved regulation mechanisms via post-translational and allosteric control take place.
Together, our results suggest that, via control of molar ratios of enzymes, both DAF-12 and AAK-2 could induce the switch from TCA to glyoxylate cycle, while AAK-2 promotes the entry of carbon from these cycles into gluconeogenesis. Thus, the stoichiometry of metabolic enzymes is unequivocally associated with the transition from growth to quiescence.