GABA Signalling Inhibits the Intestinal UPRmt
Neurotransmission is a core communication mechanism of the nervous system. To identify neurotransmitters that mediate systemic mitochondrial health, we screened neurotransmitter synthesis and transport mutants for changes to intestinal expression of the hsp-6p::gfp reporter, a well-established readout for the mitochondrial unfolded protein response (UPRmt) (8–12). We screened loss-of-function mutations affecting dopamine (cat-2), octopamine and tyramine (tdc-1), serotonin (tph-1), glutamate (eat-4), gamma-aminobutyric acid (GABA) (unc-25) and acetylcholine (ACh) (cha-1) signalling (Fig. 1a-b) (13–18). We found that loss of UNC-25/GAD, a glutamic acid decarboxylase required for GABA synthesis (Fig. 1c), induced intestinal hsp-6p::gfp expression (Fig. 1a-b). UNC-25 is expressed in 26 GABAergic neurons: 19 D-type motor neurons, 4 RME motor neurons, and 3 other neurons (AVL, DVB and RIS) (Fig. 1d) (19). Resupplying unc-25 cDNA to all GABA expressing neurons using the unc-25 promoter rescued intestinal hsp-6p::gfp expression levels in unc-25(-) animals (Fig. 1e), confirming a role for UNC-25 in non-autonomous regulation of the intestinal UPRmt. Furthermore, induced hsp-6p::gfp expression in unc-25(-) mutant animals was rescued by unc-25 cDNA expression under the D-type motor neuron unc-30 promoter, but not under the truncated unc-47 promoter that drives expression in the RMEs, AVL, RIS and DVB neurons (Fig. 1e) (20).
Animals lacking UNC-25 also showed increased intestinal expression of another mitochondrial stress marker; hsp-60p::gfp (Supplementary Fig. 1a-b). Similarly, sod-3p::gfp expression – a DAF-16/FOXO transcription factor regulated gene with roles in combatting oxidative stress in mitochondria – was also induced in unc-25(-) animals (Supplementary Fig. 1c-d), consistent with previously published data (21, 22). We found that UNC-25 loss did not affect hsp-16.2p::gfp, a reporter for cytosolic heat stress, suggesting that disrupted GABA signalling does not induce a general stress response (Supplementary Fig. 1e-f). UPRmt induction is regulated by ATFS-1, a mitochondrial unfolded protein response transcription factor, and the chromatin remodelling proteins DVE-1 and UBL-5 (10, 23). Using RNA-mediated interference (RNAi), we silenced atfs-1, ubl-5 and dve-1 in unc-25(-) animals and found that intestinal hsp-6p::gfp induction was suppressed (Supplementary Fig. 1g-h). Therefore, GABA signalling requires canonical transcriptional regulators to activate the intestinal UPRmt.
To corroborate the role of GABA signalling in controlling the intestinal UPRmt, we analysed other components of the GABA signalling pathway (Fig. 1c, e and Supplementary Fig. 1i-k). The vesicular GABA transporter UNC-47/VGAT is required for packaging GABA into pre-synaptic vesicles for synaptic release (24, 25). We found that loss of unc-47 induces hsp-6p::gfp expression (Fig. 1e and Supplementary Fig. 1i). The UNC-30/Pitx2 transcription factor is a terminal selector of D-type motor neuron identity, where it directly promotes unc-25 and unc-47 expression (Supplementary Fig. 1j) (26, 27). We found that unc-30 loss also induced hsp-6p::gfp expression and that expressing unc-30 DNA under the D-type motor neuron specific unc-30 promoter rescued this phenotype (Supplementary Fig. 1k). In contrast, expressing unc-30 in the intestine using the ges-1 promoter did not affect hsp-6p::gfp expression (Supplementary Fig. 1k). Taken together, these data reveal that GABA signalling specifically from the D-type motor neurons represses the UPRmt in the C. elegans intestine.
We next sought to determine how GABA influences mitochondria in distal tissues. Several GABA related processes regulate life and health-span, which are concepts closely related to mitochondrial health (28, 29). In C. elegans, GABA signalling transmits longevity signals through DAF-16/FOXO (21). An intermediate component in this pathway is the PLCβ homologue, EGL-8 (30). We found that loss of egl-8 did not induce intestinal hsp-6p::gfp expression (Supplementary Fig. 2a,c), suggesting that GABA signalling regulates the intestinal UPRmt and longevity through different mechanisms. We wondered if GABA was acting as a metabolite to influence mitochondrial health through the GABA shunt, where GABA is degraded to succinic semialdehyde (SSA) by GABA-transaminase (GTA-1) and then joins the TCA cycle (31). However, loss of the C. elegans GABA transaminase, GTA-1, did not influence intestinal hsp-6p::gfp expression (Supplementary Fig. 2b-c).
Regulation of Acetylcholine Release by GABAB Receptors Controls the Intestinal UPRmt
We speculated that GABA signalling mediates the UPRmt from the D-type motor neurons by signalling to downstream GABA receptors. We therefore screened available GABA receptor mutants for hsp-6p::gfp induction, focusing on those receptors previously implicated in life and health-span (22). We found that loss of the inhibitory UNC-49 and excitatory EXP-1 ionotropic GABAA receptors did not influence the intestinal UPRmt (Fig. 2a-c) (32, 33). However, loss of both components of the metabotropic GABAB receptor – GBB-1 and GBB-2 – induced intestinal hsp-6p::gfp expression (Fig. 2a-c). Intestinal hsp-6p::gfp expression in the gbb-2(-); gbb-1(-) compound mutant was not significantly different to either single mutant, indicating that both components of the GABAB receptor are required in the same pathway to control the UPRmt (Fig. 2a-c). Likewise, intestinal hsp-6p::gfp expression in the unc-25(-); gbb-2(-); gbb-1(-) triple mutant was not additive compared the unc-25(-) single mutant, showing that GABA synthesis and the metabotropic GABA receptors act within the same genetic pathway to control the systemic UPRmt (Fig. 2d). Mutant strains containing the unc-25(e156) mutation, however, exhibited slightly stronger hsp-6p::gfp expression than gbb-1(-) and gbb-2(-) mutants, suggesting that GABA signalling may also have roles outside of metabotropic signalling in controlling systemic UPRmt activation. Canonically, GBB-1 and GBB-2 act in concert to reduce neuronal excitability, however, previous studies found that GBB-1 can act independently to influence longevity through DAF-16/FOXO (21). This is supported by our data, where sod-3p::gfp expression – a readout for DAF-16 activity – is regulated by GBB-1, and not GBB-2 (Supplementary Fig. 3). These data support a role for GABA in UPRmt activation independent to its role in longevity. Single cell sequencing data shows that gbb-1 and gbb-2 are expressed primarily in neurons (34). Therefore, we resupplied gbb-1 cDNA under the pan-neuronal rgef-1 promoter in gbb-1(-) animals and found that this rescued the increased intestinal hsp-6p::gfp levels (Fig. 2e), confirming that the GABAB receptor complex acts in neurons to influence intestinal UPRmt activation.
Within the C. elegans ventral nerve cord, D-type GABAergic motor neurons synapse with body wall muscle and adjacent cholinergic neurons, which are the only motor neurons that express gbb-1 and gbb-2 (Fig. 3a) (35, 36). To determine whether the metabotropic receptor complex acts in cholinergic motor neurons to mediate the intestinal UPRmt, we resupplied gbb-1 cDNA specifically to cholinergic ventral nerve cord motor neurons using a 1882bp acr-2(s) promoter (37). We found that gbb-1 expression in these cholinergic motor neurons rescued the increased intestinal hsp-6p::gfp levels of gbb-1(-) animals (Fig. 2e), confirming that the GABAB receptor complex acts in cholinergic motor neurons to influence intestinal UPRmt activation. As GABA is generally an inhibitory neurotransmitter, and ACh is the primary excitatory neurotransmitter, these neurons work in a negative feedback loop to regulate each other and the muscles they innervate (38). When the inhibitory GABA signal is lost, cholinergic neurons are overactive, leading to increased ACh release (36). We therefore speculated that GABA regulates the intestinal UPRmt through downstream ACh signalling. Our initial screening data showed that loss of the choline acetyltransferase CHA-1/ChAT, which is required for ACh production, significantly reduced intestinal hsp-6p::gfp expression (Fig. 1a). Likewise, loss of the ACh vesicular transporter UNC-17/VAChT also reduced intestinal hsp-6p::gfp expression (Fig. 3b-c). Importantly, loss of unc-17 rescued the increased hsp-6::gfp observed in unc-25(-) and gbb-1(-) animals (Fig. 3b-c and Supplementary Fig. 4). These data support the concept that ACh levels can influence the systemic UPRmt, and that increased ACh release in animals lacking GABA signalling would increase the intestinal UPRmt. To test this hypothesis, we examined animals lacking two acetylcholinesterases, ACE-1/2, which have an approximately two-fold increase in systemic ACh (39). We found that, as with loss of GABA signalling, increased ACh in ace-2(-); ace-1(-) animals induced intestinal hsp-6p::gfp expression (Fig. 3d-e).
The Intestinal ACR-11 Nicotinic α7 Receptor Regulates Mitochondrial Health
We theorized that excess ACh acts directly on intestinal cells to induce the UPRmt. Therefore, we used RNAi to knock down intestinally-expressed ACh receptors (ACR-6, ACR-7 and ACR-11) in the ace-2(-); ace-1(-) compound mutant, screening for inhibition of hsp-6p::gfp induction (Fig. 3f). This analysis revealed that only knockdown of the nicotinic α7 receptor (α7 nAChR) ACR-11, an ortholog of human CHRNA7, abrogated intestinal hsp-6p::gfp induction in ace-2(-); ace-1(-) animals (Fig. 3f). Importantly, ACR-11 knockdown also prevented hsp-6p::gfp induction in unc-25(-) animals (Fig. 3g). To corroborate our RNAi results, we used CRISPR/Cas9 to generate independent acr-11 deletions in wild type and ace-2(-); ace-1(-) mutant animals (Fig. 4a). As acr-11 and ace-2 are tightly linked on chromosome I, the acr-11 deletions were generated separately. Confirming our RNAi data, the acr-11(rp192) deletion prevented hsp-6p::gfp induction in ace-2(-); ace-1(-) animals (Fig. 4b). However, loss of acr-11 alone was not sufficient to lower hsp-6p::gfp expression, as observed in unc-17 mutant animals (Fig. 4b,c,e). We wondered whether other intestinal α7 nAChR may support optimal UPRmt activation in conditions of physiological ACh levels. We therefore used CRISPR/Cas9 to generate acr-6 and acr-7 deletions (Fig. 4a) and measured hsp-6p::gfp levels in single, double and triple mutant combinations with acr-11(rp191) animals. We found that hsp-6p::gfp was only reduced when all three intestinally expressed α7 nAChR were deleted (Fig. 4c). This implies a small but significant role for ACR-6 and ACR-7 in UPRmt activation under basal ACh conditions, with ACR-11 the main responder to elevated ACh.
To examine the expression pattern of acr-11, we generated transgenic animals expressing green fluorescent protein (GFP) under control of the 2024bp acr-11 promoter and detected expression in the intestine and pharynx, and not in neurons or body wall muscle (Supplementary Fig. 5). To determine whether ACR-11 is required in the intestine, we resupplied acr-11 cDNA in the intestine (ges-1 promoter) to the ace-2(-) acr-11(-); ace-1(-) triple mutant, which restored the hsp-6p::gfp expression levels to that of the ace-2(-); ace-1(-) double mutant (Fig. 4d-e). Furthermore, we found that acr-11 overexpression in the intestine greatly induced hsp-6p::gfp expression in wild-type animals, and that this induction is rescued by preventing ACh release using an unc-17 mutation (Fig. 4f-g). These data reveal that ACh acts through ACR-11 in the intestine to regulate the UPRmt.
Calcium plays key roles in mitochondrial signalling, function and health, including in stress response activation (40, 41). As ACR-11 is an α7 nAChR, which are highly permeable to calcium ions (42), we investigated intracellular calcium storage using an intestine-specific fluorescence resonance energy transfer (FRET)-based calcium indicator (43, 44) in animals with perturbed GABA and ACh signalling (Fig. 5a-b). We found that both unc-25(-) and ace-2(-); ace-1(-) animals had increased intracellular calcium levels (Fig. 5a-b). Furthermore, animals lacking ACR-11, either alone or when combined with the ace-2(-); ace-1(-) compound mutant, had wild-type calcium levels (Fig. 5a-b). These data mirror hsp-6p::gfp reporter levels in these mutants, supporting our hypothesis that ACR-11 responds to ACh release by increasing intracellular calcium levels in the intestine, which likely activates the UPRmt.
UPRmt activation can have a positive or negative effect on mitochondrial health (45). An overactive stress response in the absence of stressors may prime mitochondria to manage subsequent stress exposure. Conversely, UPRmt induction due to poor mitochondrial health may cause sensitivity to mitochondrial stressors. To assess whether perturbed GABA/ACh signalling influences mitochondrial stress resistance, we examined acute paraquat sensitivity. Paraquat exposure induces mitochondrial stress by disrupting complex I of the electron transport chain and increases superoxide levels (46). We found that loss of GABA biosynthesis (unc-25 mutant) or metabotropic GABA receptors (gbb-1 and gbb-2 mutants) were sensitive to paraquat (Fig. 5c and Supplementary Fig. 6a). Paraquat sensitivity was rescued in the gbb-1(tm1406) mutant by resupplying gbb-1 cDNA in neurons (Fig. 5c). Further, we found that ace-2(-); ace-1(-) animals were sensitive to paraquat exposure (Fig. 5d). Thus, two conditions where ACh signalling is amplified causes sensitivity to a mitochondrial stressor. In contrast, acr-11(-) single and ace-2(-) acr-11(-); ace-1(-) triple mutant animals exhibited increased paraquat resistance (Fig. 5d). Neither acr-6(-) nor acr-7(-) caused changes to paraquat survival (Supplementary Fig. 6b), highlighting the specific importance of ACR-11. Furthermore, reduced ACh signalling in unc-17(-) animals conferred paraquat resistance, and was sufficient to rescue paraquat sensitivity induced by loss of unc-25 (Fig. 5e). We found that intestinal acr-11 cDNA overexpression severely impeded paraquat survival (Fig. 5f), mirroring the dramatic increase in hsp-6p::gfp expression observed when acr-11 is overexpressed in the intestine. This sensitivity to paraquat was rescued by reducing ACh signalling via loss of unc-17 (Fig. 5f). Therefore, increased UPRmt activation via increased ACh signalling to ACR-11 is associated with mitochondrial damage and paraquat sensitivity, whilst decreased ACh signalling to ACR-11 is associated with increased mitochondrial fitness, leading to increased paraquat survival.
Another indicator of mitochondrial health is mitochondrial morphology. In general, more mitochondrial fusion indicates increased OXPHOS functionality, whereas more mitochondrial fragmentation indicates increased autophagy and/or biogenesis (47). Based on our paraquat data, we hypothesized that unc-25(-) and ace-2(-); ace-1(-) animals would have more fragmented mitochondria, and that acr-11(-) single and ace-2(-) acr-11(-); ace-1(-) triple mutants would have increased mitochondrial fusion. Using a mitochondrial-targeted GFP reporter expressed in the intestine (ges-1p::gfpmt) (48), we found that unc-25(-) and the ace-2(-); ace-1(-) animals had increased mitochondrial fragmentation in the intestine compared to wild type animals, as predicted (Supplementary Fig. 7a). This phenotype was intestine-specific, as we did not detect mitochondrial morphology changes in body wall muscle using a ubiquitously expressed reporter (cox-4p::gfpmt) in unc-25(-) animals (Supplementary Fig. 7c-d) (49). Unexpectedly, the acr-11(-) single mutant also displayed a fragmented mitochondria phenotype (Supplementary Fig. 7a). Further, the ace-2(-) acr-11(-); ace-1(-) triple mutant exhibited more mitochondrial fragmentation than the acr-11(-) single and ace-2(-); ace-1(-) double mutants (Supplementary Fig. 7a). These data imply that, in the context of mitochondrial morphology, ACh signalling and the ACR-11 receptor act in distinct pathways.
We wondered how similar mitochondrial fragmentation phenotypes observed in unc-25(-), ace-2(-); ace-1(-) and acr-11(-) animals could lead to opposing mitochondrial fitness in terms of paraquat resistance. We posited that, in a pathway separate from GABA/ACh signalling, ACR-11 may repress autophagy and thus mitochondrial turnover, perhaps as an energy conservation mechanism. This means that ACR-11 loss would induce fragmentation as mitochondria are excessively cleared from the system. Using a mCherry::gfp::lgg-1 reporter to measure autophagy (50), we indeed found that the number of intestinal autophagosomes and autolysosomes in acr-11(-) L4 larvae is higher compared to control animals (Fig. 5g-i). This suggests that animals lacking ACR-11 exhibit elevated mitochondrial turn-over in unstressed conditions that conveys a survival advantage when exposed to oxidative stress.