Metabotropic GABA signalling modulates longevity in C. elegans

The nervous system plays an important but poorly understood role in modulating longevity. GABA, a prominent inhibitory neurotransmitter, is best known to regulate nervous system function and behaviour in diverse organisms. Whether GABA signalling affects aging, however, has not been explored. Here we examined mutants lacking each of the major neurotransmitters in C. elegans, and find that deficiency in GABA signalling extends lifespan. This pro-longevity effect is mediated by the metabotropic GABAB receptor GBB-1, but not ionotropic GABAA receptors. GBB-1 regulates lifespan through G protein-PLCβ signalling, which transmits longevity signals to the transcription factor DAF-16/FOXO, a key regulator of lifespan. Mammalian GABAB receptors can functionally substitute for GBB-1 in lifespan control in C. elegans. Our results uncover a new role of GABA signalling in lifespan regulation in C. elegans, raising the possibility that a similar process may occur in other organisms.

A ging is a complex physiological process modulated by a multitude of genetic and environmental factors 1,2 . Recent work has revealed an important role of the nervous system in modulating aging 1,3,4 . For example, in insulin/IGF-1 signalling, one of the best characterized longevity-regulatory pathways, the DAF-2 insulin/IGF-1 receptor primarily acts in the nervous system to regulate lifespan in Caenorhabditis elegans 5 . Interestingly, work in mice revealed a similar mechanism 6 . Environmental factors also affect aging by impinging on the nervous system 1,3,4,7,8 . Apparently, neurons can act cell nonautonomously to modulate aging, and neuroendocrine signalling plays an important role in this process 1,3,4,7,8 . However, whether and how neurotransmitter signalling modulates aging is poorly understood.
GABA is the primary inhibitory neurotransmitter in the mammalian brain 9,10 . It is critical for the function and development of the nervous system and plays a key role in behavioural control [11][12][13][14] . GABA acts through both ionotropic and metabotropic receptors 15,16 . As ion channels, ionotropic GABA A receptors mediate the acute, fast actions of GABA 17,18 . By contrast, metabotropic GABA B receptors are G protein coupled receptors (GPCRs) that execute the slow, long-lasting effects of GABA [19][20][21][22] .
C. elegans is a popular model organism widely utilized to study the biology of aging, featuring a short generation time and lifespan, as well as conserved genetic mechanisms that regulate longevity 1,2 . C. elegans has also been widely used as a genetic model for neurobiology 23 . Although C. elegans possesses a relatively small nervous system, it produces all the major neurotransmitters (for example, ACh, glutamate, GABA and biogenic amines) and their cognate receptors, all of which are encoded by evolutionarily conserved gene families 23 . While the role of neurotransmitters in controlling behaviour and nervous system development and function has been extensively characterized, little is known about their role in aging.
Here we interrogated the role of neurotransmitters in lifespan control in C. elegans and found that GABA regulates lifespan through GABA B receptor signalling. We further identified a genetic pathway that cell nonautonomously transmits longevity signals from the GABA B receptor in motor neurons to the transcription factor FOXO/DAF-16 in the intestine. Our results identify a novel function of GABA beyond its conventional role in modulating animal behaviour and nervous system function and development.

Results
Deficiency in GABA signalling extends lifespan. The nervous system of C. elegans hermaphrodites is relatively simple 23,24 , making it convenient to characterize the role of neurotransmission in lifespan control. We first assayed the lifespan of unc-13 mutant worms that are defective in synaptic transmission 25,26 . These mutant animals are long lived ( Fig. 1a and Supplementary Table 1), consistent with previous results 27 . The fact that unc-13 mutant animals are long-lived indicates that deficiency in neurotransmitter signalling extends lifespan. To ascertain which neurotransmitter(s) may contribute to this effect, we examined mutant animals defective in each of the major neurotransmitters, including unc-17 (ACh), eat-4 (glutamate), cat-2 (dopamine), tph-1 (serotonin), tdc-1 (tyramine), tbh-1 (octopamine) and unc-25 (GABA) [28][29][30][31][32][33] (Fig. 1b-h). Among these  mutants, unc-25 worms lived the longest life, showing the most pronounced phenotype (Fig. 1h), while others only had modest or no effect ( Fig. 1b-g). While these data by no means exclude a role of other neurotransmitters, it reveals an important function of GABA signalling in lifespan control.
Loss of the GABA B receptor gene gbb-1 extends lifespan. GABA acts through both ionotropic and metabotropic receptors. The C. elegans genome encodes two metabotropic GABA B receptor genes, gbb-1 and gbb-2, which are highly homologous to their mammalian counterparts 34 . We found that gbb-1 but not gbb-2 mutants were long-lived (Fig. 2a). gbb-1;gbb-2 double mutant was indistinguishable from gbb-1 single mutant (Fig. 2a), indicating a specific role of gbb-1 in regulating lifespan. By contrast, mutant worms lacking unc-49, which encodes the sole evolutionarily conserved GABA A receptor family member 35 , were normally lived (Fig. 2b). These results identify a role for the GABA B receptor gene gbb-1 in lifespan control. gbb-1 is enriched in the nervous system but is also found to be expressed in other tissues such as muscle and intestine 34 . We therefore asked where gbb-1 acts to regulate lifespan. Using tissue-specific promoters, we found that expression of gbb-1 cDNA in neurons, but not in muscle or intestine, rescued gbb-1 lifespan phenotype (Fig. 3a). Thus, gbb-1 appears to act in neurons to regulate lifespan.
We crossed the gbb-1 neuronal transgene into wild-type background and found that it rendered worms short-lived (Fig. 3b). Thus, overexpression of gbb-1 shortens lifespan. This is consistent with our mutant data that loss of gbb-1 extends lifespan, providing further evidence supporting that gbb-1 regulates longevity.
GBB-1-dependent lifespan regulation requires DAF-16/FOXO. We wondered how gbb-1 controls lifespan. Various genetic and environmental factors regulate lifespan by converging on a subset of transcription factors 1 . We therefore examined the major transcription factors known to regulate lifespan, and found that RNAi of daf-16 completely suppressed the lifespanextending phenotype of gbb-1 mutant worms (Fig. 4a), while RNAi of other transcription factors such as hsf-1, skn-1 and pha-4 did not ( Fig. 4b-d). Thus, daf-16 is required for the function of gbb-1, suggesting that gbb-1 signals to daf-16 to regulate lifespan.
Gi/o-PLCb transmits longevity signals from GBB-1 to DAF-16. DAF-16 is best known to act downstream of insulin/IGF-1 signalling to regulate lifespan. Does GBB-1 genetically interact with insulin/IGF-1 signalling? To test this, we assayed the lifespan of gbb-1;daf-2 double mutant. daf-2 encodes the sole worm insulin/ IGF-1 receptor homologue 36 . The double mutant exhibited a lifespan significantly longer than both gbb-1 and daf-2 single mutants ( Supplementary Fig. 1), suggesting that gbb-1 and daf-2 may act in different pathways. However, as a reduction-offunction allele of daf-2 was used here (daf-2 nulls are lethal), we did not further test this model. Thus, we do not rule out the possibility that gbb-1 and daf-2 may act in the same or overlapping pathways.
Then how does GBB-1 signal to DAF-16? As a GPCR, GBB-1 is unlikely to communicate with DAF-16 directly. We then sought to identify genes that transduce longevity signals from GBB-1 to DAF-16. As the GABA B receptor is known to be coupled to Gi/o-PLCb signalling 10,37 , we examined this pathway. The worm genome encodes one Go orthologue, goa-1 (refs 38,39), and at least half a dozen Gi-related genes 40 . However, unlike gbb-1, goa-1 mutant worms are not long-lived ( Supplementary Fig. 2), suggesting the presence of functional redundancy from other Gi/o genes. To overcome this difficulty, we attempted to block Gi/o signalling using PTX (Pertussis toxin) by expressing it as a transgene in worm neurons. We found that inactivation of Gi/o by PTX extended lifespan (Fig. 5a). This PTX-dependent lifespan extension also required daf-16, as it can be blocked by daf-16 RNAi (Fig. 5a), suggesting that Gi/o may act upstream of DAF-16 to regulate lifespan. Further, the PTX transgene also suppressed the short-lived phenotype caused by overexpression of gbb-1 (Fig. 5b), suggesting that Gi/o acts downstream of gbb-1. These data support our model that Gi/o acts downstream of GBB-1 but upstream of DAF-16.
These results together support the model that G protein-PLCb signalling transduces longevity signals from GBB-1 to DAF-16.
DKF-2/PKD acts upstream of DAF-16 to regulate lifespan. DAF-16 is best known to be regulated by kinases 1 . We next set out to identify kinases that act downstream of PLCb but upstream of DAF-16. Protein kinase C (PKC), a group of kinases activated by DAG and/or Ca 2 þ , is a primary downstream target of PLCb. The worm genome encodes four PKC homologues: tpa-1, pkc-1, pkc-2 and pkc-3. However, none of these genes, when mutated or knocked down, gave rise to a long-lived phenotype as did egl-8 and gbb-1 ( Supplementary Fig. 3a-d). These data suggest that PKC is unlikely to mediate the effect of EGL-8/PLCb in longevity signalling. (b) Overexpression of gbb-1 cDNA in neurons shortens lifespan (log-rank test, P ¼ 0.001, n ¼ 53-72 for different genotypes). Each plasmid DNA listed above was injected at a concentration of 50 ng ml À 1 . All lifespan assays were carried out at 20°C and were repeated at least three times. Please see Supplementary Table 1 for detailed statistical analysis of lifespan data.  For RNAi experiments, NGM plates included carbenicillin (25 mg ml À 1 ) and IPTG (1 mM). HT115 bacteria-carrying vector or RNAi plasmid were seeded on RNAi plates 2 days before the experiment. Worms were fed RNAi bacteria from the egg stage. All RNAi lifespan assays were carried out at 20°C, and were repeated at least three times. Please see Supplementary Table 1 for detailed statistical analysis of lifespan data.
In addition to PKC, DAG can also activate another type of kinases, protein kinase D (PKD), in both a PKC-dependent andindependent manner 44 . Two PKD homologues are found in C. elegans: dkf-1 and dkf-2 (ref. 45). We thus considered a potential role of these two PKD genes. dkf-1 mutant worms were short-lived ( Supplementary Fig. 3e-f), suggesting that it is probably not involved. By contrast, worms lacking dkf-2 were long-lived (Fig. 6a), a phenotype similar to gbb-1 and egl-8 mutants. A previous study also showed that loss of dkf-2 extends lifespan and does so by regulating DAF- 16 (ref. 45). Indeed, we found that the long-lived phenotype of dkf-2 worms was completely suppressed by loss of daf-16 (Fig. 6b), consistent with the notion that dkf-2 acts upstream of daf-16 to regulate lifespan.
To obtain further evidence that dkf-2 regulates daf-16, we examined whether loss of dkf-2 can promote daf-16 function. We found that the mRNA levels of DAF-16 target genes, such as mtl-1, sod-3 and dod-3, were upregulated in dkf-2 mutant background (Fig. 6c). We also assayed the protein level of SOD-3::GFP fusion, a commonly used reporter for DAF-16 activity, and found that it was upregulated in dkf-2 worms (Fig. 6d). Thus, dkf-2 appears to regulate daf-16 function, providing additional evidence supporting that DKF-2 acts upstream of DAF-16 to control lifespan.
DKF-2 acts downstream of EGL-8 to regulate lifespan. We then asked whether dkf-2 functions in the gbb-1 pathway to regulate lifespan. dkf-2;gbb-1 double mutant exhibited a lifespan similar to single mutants (Fig. 6a), suggesting that they act in the same pathway. To gather additional evidence, we first rescued dkf-2 long-lived phenotype by expressing wild-type dkf-2 cDNA as a transgene in neurons in dkf-2 mutant worms (Fig. 6e). In addition, this dkf-2 transgene modestly shortened lifespan in wildtype background (Fig. 6f and Supplementary Table 1). Importantly, this transgene suppressed the long-lived phenotype of gbb-1, as well as egl-8 mutant worms (Fig. 6g-h), providing additional evidence supporting that DKF-2 acts downstream of GBB-1 and EGL-8/PLCb to regulate lifespan.
Neuron-to-intestine signalling transmits longevity signals. Apparently, the aforementioned genetic pathway, which transmits longevity signals from GBB-1 to DAF-16, does not inform where the pathway operates. Our data showed that GBB-1 acts in neurons to regulate lifespan (Fig. 3a), yet DAF-16 is best known to function in the intestine 46 . This raises the question whether GBB-1 signals DAF-16 through neuron-to-intestine signalling. However, DAF-16 also functions in neurons 46 , which confounds our data interpretation. To clarify this issue, we interrogated the site of action of DAF-16 in the GBB-1 pathway. We found that transgenic expression of daf-16 cDNA in the intestine but not in neurons rescued the lifespan phenotype of gbb-1;daf-16 double mutant (Fig. 7a), indicating that DAF-16 acts in the intestine in the GBB-1 pathway. GBB-1 is broadly expressed in the nervous system 34 . We further asked in which groups of neurons GBB-1 regulates lifespan. Transgenic expression of gbb-1 cDNA in sensory neurons under the osm-6 promoter did not rescue gbb-1 mutant phenotype (Fig. 7c), nor did expression of gbb-1 in subsets of interneurons using the glr-1 promoter (Fig. 7b). By contrast, restoring gbb-1 in ventral cord motor neurons through the acr-2 promoter fully rescued the gbb-1 longevity phenotype (Fig. 7d). Thus, GBB-1 can act in ventral cord motor neurons to regulate lifespan. This suggests a motor neuron-tointestine signalling axis that transmits longevity signals from GBB-1 to DAF-16.
Rat GABA B receptor can functionally substitute for GBB-1. We expressed the rat GABA B receptor, GB1/GB2 (refs 47-49), as a transgene in gbb-1 mutant background using a neuron-specific promoter, and found that the transgene rescued the long-lived phenotype of gbb-1 mutant worms (Fig. 8a). This suggests that mammalian GABA B receptor can functionally substitute for worm GBB-1 in regulating lifespan.
Unlike worm GBB-1, the mammalian GABA B receptor has been extensively characterized pharmacologically 15,48,50,51 . We thus took this advantage by testing some known antagonists of the GABA B receptor such as CGP36216 (ref. in wild-type animals (Fig. 8b,c). Remarkably, when these chemicals were applied to transgenic worms expressing the rat GABA B receptor GB1/GB2, they extended lifespan and did so as efficiently as gbb-1 mutation (Fig. 8d,e). These data present further evidence supporting that the mammalian GABA B receptor can functionally substitute for worm GBB-1 in lifespan regulation, raising the possibility that it may play a similar role in mammals.

Discussion
As the primary inhibitory neurotransmitter, GABA is best known to regulate the function and development of the brain and it plays a key role in controlling animal behaviour [11][12][13][14] . In the current study, we identified a new role of GABA signalling in aging. Interestingly, the effect of GABA on longevity is mediated by GABA B rather than GABA A receptor. As a GPCR, GABA B receptor generally meditates the slow, long-lasting actions of GABA, which is consistent with its role in regulating longevity. A previous study reported that knocking down Drosophila GABA B receptor in insulin-producing cells shortens lifespan 54 , an effect opposite to that observed in C. elegans. However, unlike extended lifespan, it is difficult to interpret a short-lived phenotype. As such, it remains unclear whether GABA and GABA B receptor modulate aging in this organism. On the other hand, mammalian GABA B receptor can functionally substitute for its worm homologue in lifespan control; furthermore, antagonists of mammalian GABA B receptor can extend the lifespan of transgenic worms expressing this receptor, raising the possibility that GABA and GABA B receptor may play a role in aging in other organisms. It will be interesting to test this hypothesis in future studies.
Recent studies have uncovered an increasingly important role of the nervous system in longevity 1,3,4,7,8 . Neurons regulate lifespan in a cell-nonautonomous manner presumably through neurotransmission 1,3,4,7,8 . Aside from neuroendocrine signalling, neurotransmitters are also believed to play a key role in this process. For example, endoplasmic reticulum (ER) stress in neurons affects lifespan through neurotransmitter rather than neuroendocrine signalling 55 . The identity of this neurotransmitter(s), however, remains elusive. By characterizing mutant strains defective in each of the major neurotransmitters in C. elegans, we found that GABA signalling plays an important role in regulating lifespan. To the best of our knowledge, this represents the first comprehensive analysis examining the role of all neurotransmitters in lifespan control in any organism.
Our results uncovered a genetic pathway that transmits longevity signals from GABA via G protein-PLCb signalling to    ARTICLE the FOXO transcription factor DAF-16, which is a key regulator of lifespan (Fig. 8f). Specifically, this genetic pathway consists of GBB-1, G protein, EGL-8/PLCb, DKF-2/PKD and DAF-16/ FOXO (Fig. 8f). It is interesting that PKD rather than PKC transmits signals from PLCb in our case, although PKC plays an important role in temperature-dependent lifespan regulation in C. elegans 56 . One remaining question concerns how DKF-2 regulates DAF-16. The simplest model would be that DKF-2 phosphorylates DAF-16. However, we did not detect such phosphorylation in in vitro kinase assays (J.G., X.Z.S.X. and J.L., unpublished data). This suggests that DKF-2 likely regulates DAF-16 indirectly. Consistent with this model, we found that GBB-1 acts in motor neurons while DAF-16 functions in the intestine. Expression of DKF-2 in neurons is also sufficient to rescue dkf-2 lifespan phenotype. Other components in the GBB-1 genetic pathway are also best known to act in the nervous system 41,42 . This reveals a motor neuron-to-intestine signalling axis that transmits longevity signals from GBB-1 to DAF-16.
Recent studies have pointed to a critical role of such neuron-tointestine signalling in lifespan control 55,57 . However, exactly how longevity signals are transmitted between neurons and intestinal cells remains a difficult question to address. Future efforts are needed to unravel the underlying mechanisms. Despite the observation that loss of GABA signalling seems to elicit the most pronounced effect on lifespan among all the major neurotransmitters, this does not necessarily indicate that other neurotransmitters do not have a role in lifespan control. Previous studies have demonstrated that worms experience a decline in dopamine and serotonin levels with age 58 . Octopamine has also been reported to play a key role in mediating CREB-regulated transcription coactivator (CRTC)-dependent lifespan regulation in neurons 57 . These neurotransmitters all bind to multiple types of receptors, including both GPCRs and ion channels. It is possible that their receptors have opposing effects on longevity, with some promoting lifespan and others inhibiting it, which may account for the modest effect resulting from a complete loss of  Table 1 for detailed statistical analysis of lifespan data. these neurotransmitters. It is also possible that these neurotransmitters may play a more prominent role in controlling lifespan under certain specific physiological conditions. Our studies will encourage others to investigate how the nervous system regulates lifespan through neurotransmitter signalling, an interesting but poorly understood question in the biology of aging. dkf-1(ok2695) Â 4 outcrossed. TQ5447: dkf-1(pr2) Â 6 outcrossed. The pkc-3 RNA interference (RNAi) clone was generated in the laboratory, while other RNAi clones were from the Ahringer library and were confirmed by sequencing. All mutants in the GBB-1 pathway are null alleles, including gbb-1, egl-8, dkf-2 and daf-16. These mutants have been outcrossed to our N2 strain for four to eight times before lifespan assay.

Methods
Microinjections were performed using standard protocols. Each plasmid DNA listed above in the transgenic line was injected at a concentration of 50 ng ml À 1 . For those experiments involving transgenes, three independent transgenic lines were tested for lifespan to confirm the results. For simplicity and clarity, only the data from one transgenic line were shown.
Lifespan assay. Lifespan studies were performed on 60-mm nematode growth medium (NGM) plates at 20°C as previously described 56,59,60 . For each lifespan assay, 70-110 worms were included and transferred every other day to fresh NGM plates with 14 worms per plate. The first day of adulthood was considered day 1. Survival was scored every 1-2 days, and worms were censored if they crawled off the plate, hatched inside or lost the vulva integrity during reproduction. 5-Fluoro-2 0 -deoxyuridine was included in assays involving unc-13, unc-17 and egl-8 mutant worms, which show egg-laying defects. For RNAi experiments, NGM plates included carbenicillin (25 mg ml À l ) and isopropyl-b-D-thiogalactoside (1 mM). HT115 bacteria with vector or RNAi plasmid were seeded on RNAi plates 2 days before experiment. Worms were fed RNAi bacteria, beginning at the egg stage.
For GABA B receptor antagonist experiments, 1 mM CGP36216 or SCH50911 (Tocris) was first spread out on NGM plates to let it diffuse for 1 day. L4 hermaphrodites were then transferred to these plates to assay lifespan. Worms were transferred to fresh drug plates every 2-3 days until day 10, after which animals remained on the same drug plates until death.
Graphpad Prism 5 (GraphPad Software Inc.) and IBM SPSS Statistics 19 (IBM Inc.) were used to analyse lifespan data. P values were calculated with the log-rank (Kaplan-Meier) method.
Overexpression of rat GABA B receptor in C. elegans. cDNA encoding the rat GABA B receptor subunits, GB1 or GB2, was driven by the neuron-specific promoter rgef-1. The two plasmids were co-injected (50 ng ml À 1 each) into gbb-1 mutant worms to generate the transgenic animals overexpressing rat GABA B receptor.
Quantification of SOD-3::GFP fluorescence intensity was performed on an Olympus BX51 upright microscope as described previously 56 . Images were acquired with a Roper CoolSnap charge couple device camera controlled with MetaMorph (Molecular Devices Inc.) and analysed with ImageJ (NIH). SOD-3::GFP is encoded by the transgene muIs84 (ref. 46).