Control of Neuropeptide Expression by Parallel Activity-dependent Pathways in Caenorhabditis elegans

Monitoring of neuronal activity within circuits facilitates integrated responses and rapid changes in behavior. We have identified a system in Caenorhabditis elegans where neuropeptide expression is dependent on the ability of the BAG neurons to sense carbon dioxide. In C. elegans, CO2 sensing is predominantly coordinated by the BAG-expressed receptor-type guanylate cyclase GCY-9. GCY-9 binding to CO2 causes accumulation of cyclic GMP and opening of the cGMP-gated TAX-2/TAX-4 cation channels; provoking an integrated downstream cascade that enables C. elegans to avoid high CO2. Here we show that cGMP regulation by GCY-9 and the PDE-1 phosphodiesterase controls BAG expression of a FMRFamide-related neuropeptide FLP-19 reporter (flp-19::GFP). This regulation is specific for CO2-sensing function of the BAG neurons, as loss of oxygen sensing function does not affect flp-19::GFP expression. We also found that expression of flp-19::GFP is controlled in parallel to GCY-9 by the activity-dependent transcription factor CREB (CRH-1) and the cAMP-dependent protein kinase (KIN-2) signaling pathway. We therefore show that two parallel pathways regulate neuropeptide gene expression in the BAG sensory neurons: the ability to sense changes in carbon dioxide and CREB transcription factor. Such regulation may be required in particular environmental conditions to enable sophisticated behavioral decisions to be performed.

and sophisticated roles for the BAG and URX neurons, prioritization and integration of information to guide behavioral responses may require modular regulation of neuropeptide expression.
Here we describe a system where the activity of the BAG neurons is crucial for the expression of a transcriptional reporter for the FLP-19 neuropeptide (flp-19::GFP). Furthermore, we show that the expression of flp-19::GFP is regulated via two parallel modules: 1) cGMP signaling through GCY-9 and PDE-1 and 2) the CREB transcription factor.

flp-19::GFP Expression in the BAG Neurons Requires Cilia Function.
It has previously been shown that neuropeptide expression is highly plastic, and can be controlled by neuronal activity 4,10,13 . The flp-19::GFP transcriptional reporter drives expression in the BAG, URX, AIN, AWA and HSN neurons in the hermaphrodite 25 .
Our previous studies showed that flp-19::GFP expression in the BAG neurons is exquisitely sensitive to perturbations in parallel transcriptional pathways that control BAG cell fate and function [26][27][28] . We hypothesized that neuronal activity may underpin the transcriptional regulation of flp-19::GFP in the BAG neurons. We therefore tested if the expression of flp-19::GFP is compromised when the activity of the BAG neurons is reduced or abolished.
As the BAG neurons are ciliated, we first examined if disruption of cilia structure, which is required for several behaviors in C. elegans 7,29,30 , would affect flp-19::GFP expression. We crossed the flp-19::GFP reporter (ynIs34) with a mutant of che-3, which encodes a dynein that is required for structural integrity of sensory cilia. We observed that che-3(e1379) mutant animals exhibit strong defects in the expression of flp-19::GFP in the ciliated BAG and AWA neurons but not in the non-ciliated neurons (Fig. 1A,B). To verify this regulation, we crossed che-3(e1379) mutant animals with an independent flp-19::GFP reporter (rpEx811) and observed a similar phenotype (Fig. 1C). We next asked if disturbance of cilia transport would produce a similar effect on flp-19::GFP expression. TUB-1, homolog of TUBBY in mammals, is required for correct localization of G protein coupled receptors to cilia [31][32][33] . We found that removal of TUB-1 function phenocopied the che-3 mutant loss of flp-19::GFP expression in the BAG neurons (Fig. 1A,B), indicating that correct cilia function is required for expression of flp-19::GFP.

GCY-9
Regulates flp-19::GFP Expression Cell-autonomously in the BAG Neurons. Acute responses to CO 2 are regulated by a neuronal circuit that includes the BAG neurons. The BAG neurons sense carbon dioxide through the GCY-9 receptor-type guanylate cyclase, homolog of the human GC-D 26,34,35 . As such, gcy-9 mutants are unable to sense and respond to changes in CO 2 concentration 34 . We asked whether the CO 2 -sensing function of the BAG neurons is required for flp-19::GFP expression. To this end, we crossed two independent deletion alleles of gcy-9 (n4470, tm2816) with the flp-19::GFP reporter and observed a reduced number of animals expressing GFP in the BAG neurons ( Fig. 2A,B). When we resupplied gcy-9 cDNA driven by a BAG specific gcy-33 promoter into gcy-9(n4470) mutant animals, flp-19::GFP expression was restored (Fig. 2B). The BAG neurons are also involved in sensing downshifts in O 2 concentration, through expression of the soluble guanylate cyclases GCY-31 and GCY-33 22 . To ask whether O 2 -sensing function of the BAG neurons is also required to regulate flp-19::GFP expression, we crossed gcy-31(ok296) and gcy-33(ok232) mutants with the flp-19::GFP reporter. We found no detectable change in expression of flp-19::GFP when BAG O 2 -sensing function was ablated (Fig. 2C). As it has been previously shown that the URX and BAG communicate with each other 23,24 , we asked whether removal of O 2 -sensing from the URX neurons would affect the expression of flp-19::GFP. We therefore crossed the flp-19::GFP reporter into gcy-35(ok769) and gcy-36(db42) mutant strains, in which soluble guanylate cyclases required for URX O 2 -sensing function are mutated 3,4,22,36 . However, we observed no defect in the expression of flp-19::GFP in the BAG or URX neurons ( Fig. 2C and data not shown). Furthermore, animals which lack O 2 sensing function in both the BAG and URX neurons (gcy-31 gcy-36; gcy-33; gcy-35 quadruple mutant) exhibit wild type expression of flp-19::GFP (Fig. 2C). Taken together, these data show that CO 2 sensing function, and not O 2 sensing function, regulates the expression of the flp-19::GFP reporter in the BAG neurons.
To ask whether CO 2 sensing has a general effect on neuropeptide expression in the BAG neurons, we analyzed the expression of other neuropeptides in gcy-9(n4470) mutant animals. We crossed the gcy-9(n4470) mutant with reporters for flp-17(ynIs64) and flp-13(ynIs37) and observed no change of expression compared to wild type ( Figure S1A). Furthermore we tested other reporters expressed in the BAG neurons (soluble guanylate cyclases gcy-31(rpIs29) and gcy-33(rpIs7) and the transcription factor egl-13(rpIs32)) in the gcy-9 mutant background and we observed no change of expression ( Figure S1A). Therefore, the regulation of flp-19::GFP expression by GCY-9 is somewhat specific.
cGMP Levels Regulate flp-19::GFP Expression in the BAG Neurons. gcy-9 encodes a receptor-type guanylate cyclase. The role of these enzymes is to generate cGMP for gating downstream cyclic nucleotide-gated TAX-2/TAX-4 cation channels. As such GCY-9 regulates the activity of the BAG neurons through the control of cGMP levels. We speculated therefore that cGMP is a key regulator of flp-19::GFP expression in the BAG neurons. In order to test this hypothesis, we genetically manipulated cGMP levels. Phosphodiesterases are enzymes that catalyse the breakdown of cGMP to GMP and it has previously been shown that the phosphodiesterase PDE-1 is expressed in the BAG neurons 34 . We therefore asked whether the predicted increase in cGMP levels in pde-1 mutant animals would affect flp-19::GFP expression. We crossed the pde-1(nj57) mutant with the flp-19::GFP reporter and observed that BAG expression of flp-19::GFP was unaffected (Fig. 3A). Next we tested if loss of pde-1, and therefore increase of cGMP in the BAG neurons, was sufficient to derepress flp-19::GFP expression in the gcy-9 mutant. We therefore examined flp-19::GFP expression in the pde-1(nj57); gcy-9(n4470) double mutant and found that expression in the BAG neurons was fully restored (Fig. 3A). This suggests that reduced levels of cGMP in the BAG neurons causes transcriptional downregulation of flp-19::GFP.
Activation of GCY-9 by CO 2 normally triggers the conversion of cGMP from GTP 35 . Subsequently, cGMP opens the cyclic nucleotide-gated channels TAX-2/TAX-4, through which the neuron is activated by calcium influx 5,6 . We therefore hypothesized that GCY-9-mediated regulation of the flp-19::GFP reporter was through this canonical pathway. Indeed, we found that in tax-4(p678) mutant animals flp-19::GFP expression is undetectable in the BAG neurons (Fig. 3B). In addition, we observed loss of flp-19::GFP expression in the URX O 2 -sensing neurons of tax-4(p678) mutant animals, suggesting that similar mechanisms of regulation may exist in these   TAX-4 channels in the URX neurons. To confirm that the BAG neurons are present in tax-4 mutant animals, we crossed tax-4(p678) mutant animals with an independent reporter for BAG neurons (egl-13::GFP), and observed no loss of expression ( Figure S1B). Furthermore, we found that loss of pde-1 was not able to restore the tax-4(p678) mutant flp-19::GFP expression, indicating that tax-4 acts downstream of cGMP regulation, and that the channels are necessary for the expression of flp-19::GFP regardless of the levels of cGMP (Fig. 3B).
The data we have presented suggest that activity of the BAG neurons is important for the expression of flp-19::GFP. To reinforce this assertion, we inactivated the BAG neurons by expressing a constitutively-active EGL-2(GF) potassium channel using the gcy-33 promoter 4,37,38 . We found that animals carrying the Pgcy-33::egl-2(gf) transgene exhibit a decrease in the expression of flp-19::GFP in the BAGs (Fig. 3B).
Taken together, our data show that GCY-9 and the downstream cGMP-regulated TAX-2/TAX-4 channels are required for the expression of flp-19::GFP in the BAG neurons. These data suggest that the conversion of GTP to cGMP by GCY-9 triggers opening of the TAX-2/TAX-4 channels, resulting in calcium influx and that this change in activity regulates the transcription of the neuropeptide reporter flp-19::GFP (Fig. 3C).

flp-19::GFP Expression Does not Require Neuropeptide or Neurotransmitter Signaling.
We have shown that the expression of flp-19::GFP is sensitive to GCY-9 and TAX-4-controlled activity. We next asked whether the expression of flp-19::GFP was regulated by neuropeptide and neurotransmitter signaling, through either an autocrine or paracrine fashion. We first tested whether flp-19::GFP expression was affected by abolishing neuropeptide signaling. To this end, we crossed the flp-19::GFP reporter with two mutants that do not have proper neuropeptide signaling: egl-3(nr2090), responsible for maturation of neuropeptides, and unc-31(e169), involved in dense core vesicle function [39][40][41] . The expression of flp-19::GFP in the BAG neurons in egl-3 and unc-31 mutants was unchanged when compared to wild type ( Table 1). The BAG neurons are glutamatergic as they express the vesicular glutamate transporter EAT-4 42,43 . We found that the glutamatergic function of the BAG neurons is not required for the regulation of flp-19::GFP as expression is unchanged in an eat-4(ky5) mutant (Table 1). Furthermore, we asked if neurotransmitter release through synaptic vesicle exocytosis was required for flp-19::GFP expression. We crossed flp-19::GFP with the unc-13(e1091) mutant, defective in neurotransmission due to compromised synaptic vesicle fusion and the snb-1(md247) mutant, defective in synaptic transmission [44][45][46] . We found that unc-13 and snb-1 mutant animals exhibit wild type expression of flp-19::GFP in the BAG neurons (Table 1). These data suggest that the regulation of flp-19::GFP in the BAG neurons occurs through a BAG-intrinsic mechanism.

CRH-1/CREB, an Activity-dependent Transcription Factor Acts in Parallel to GCY-9 to Regulate flp-19::GFP Expression in the BAG Neurons.
In order to better understand the mechanism through which flp-19::GFP is regulated, we studied mutants of various genes involved in activity-dependent expression in the nervous system (Table 1). Our screen found that both crh-1 and kin-2 mutants display defects in the expression of the flp-19::GFP reporter in the BAG neurons. CRH-1 is the C. elegans homolog of CREB and functions in neurons to regulate various behaviors. CRH-1 regulates lifespan, tap habituation and it has been linked with the control of thermotaxis behavior from the AFD neurons [47][48][49] . KIN-2 is the homolog of the regulatory subunit of cAMP-dependent protein kinase (PKA) and it can regulate CREB activity through a cascade of phosphorylation events [50][51][52][53][54] . We found that two independent null alleles of crh-1 (tz2 and n3315) exhibit reduced expression of flp-19::GFP in the BAG neurons (Fig. 4A). In addition, kin-2(ce179) mutants show a similar phenotype to crh-1 mutants, and the crh-1(tz2); kin-2(ce179) double mutant is not significantly different from either single mutant, suggesting that they function in the same genetic pathway (Fig. 4A). We speculated that CRH-1 might be the downstream effector in the GCY-9 cascade. To ask whether gcy-9 and crh-1 act in the same genetic pathway to control flp-19::GFP expression we constructed a gcy-9(n4470); crh-1(tz2) double mutant. Surprisingly, we found that these double mutant animals exhibited complete abrogation of flp-19::GFP expression (Fig. 4B). We confirmed that the BAG neurons are present in the gcy-9(n4470); crh-1(tz2) double mutant by examining a flp-13::GFP transgene ( Figure S1C). Together, these data suggest that gcy-9 and crh-1 act in two separate genetic pathways to regulate flp-19::GFP expression (Fig. 5). In summary, we propose a model where flp-19::GFP expression is regulated by two parallel pathways: 1) GCY-9 regulation of the TAX-2/TAX-4 cation channels through the control of cGMP levels, signaling to an unknown transcription factor and 2) CRH-1/CREB regulation by the kinase KIN-2. likely through a CO 2 -independent signaling pathway (Fig. 5).

Discussion
Activity-dependent regulation is common occurrence in sensory neurons and can contribute in changes in behavior. Chemoreceptor genes can be regulated by different mechanisms such as developmental changes, neuronal activation, or in a paracrine fashion through pheromones. Transcriptional changes of chemoreceptors might be a strategy to modulate external responses. For example, the TAX-2/TAX-4 calcium channel regulates the expression of chemoreceptors such as STR-2 and SRD-1 in the AWC and ASI neurons respectively 11,55 . In addition, TRPV channels control the biosynthesis of serotonin through regulation of tryptophan hydroxylase expression in the ADF neurons 56 .
We have presented data to show that flp-19::GFP expression in the BAG neurons is regulated by two parallel pathways. The GCY-9 pathway controls flp-19::GFP expression by modulating the levels of cGMP, counterbalanced by the phosphodiesterase PDE-1. PDEs and GCYs are known to function together in other neurons to regulate activity. For example, gcy-12 and pde-2 control cGMP levels to determine body size in C. elegans 57 . Similarly, in the AFD neurons, the opposing roles of GCY-8 and PDE-2, control C. elegans thermotaxis behavior 58 . It might be possible that PDE-1 is involved in sensing the levels of CO 2 in the BAG neurons; potentially to set a quantitative threshold or temporal window of TAX-2/TAX-4 channel opening. We have not identified the effector downstream the TAX-2/TAX-4 signaling, but we speculate that it may be a transcription factor regulated by calcium or calmodulin dependent activation. In parallel to GCY-9, CRH-1/CREB also controls the level of flp-19::GFP through the activity of the PKA kinase. As PKA is a cAMP regulated kinase, this may suggest that a cAMP-regulated pathway controls flp-19::GFP expression through CRH-1.
It is necessary to further study the implications of the two parallel pathways we have identified to understand the advantages they may provide in the natural habitat of C. elegans. Interestingly worms are attracted to CO 2 as dauers while L4 larvae avoid CO 2 5,6,34 . Furthermore it has been shown how in juvenile infective stages of parasitic worms (related to C. elegans) BAG neurons are involved in the attraction to CO 2 59 . It might be possible that this change in the attraction/repulsion to CO 2 is regulated by the expression of neuropeptides such as FLP-19. When worms are grown in low O 2 levels, they become attracted to lower levels of O 2 , instead of being repelled. This change of behavior is due to changes in the expression level of guanylate cyclases that detect O 2 . It might be possible that similar adaptation occurs when worms are grown in hypercapnic conditions, and the transcriptional regulation of flp-19 could be involved in such adaptation. However, the role of flp-19 may not be directly related to CO 2 sensing, but to other broader functions. Indeed, the BAG neurons are involved in lifespan regulation, as is the CRH-1 transcription factor, therefore it would be interesting to examine whether flp-19 mutant animals display defects in longevity.
flp-19 is expressed in a subset of neurons that are distinct in class and function: from oxygen (URX/BAG) and carbon dioxide sensing (BAG), chemotaxis (AWA), egg laying (HSN) or pheromone sensing (CEM). Similar means of flp-19 activity-dependent regulation maybe present in these other neurons through different molecular pathways, providing this neuropeptide with multiple layers of control that may be required in particular ephemeral habitats.

Methods
Strains used in this study. Strains were grown using standard growth conditions on NGM agar at 20 °C on Escherichia coli OP50 60 . Transgenic animals were created as previously described 61 . Strain information is detailed in Table S1.
Microscopy. Worms were anesthetized in 20 mM NaN 3 on 5% agarose on glass slides and images were taken using an upright fluorescence microscope (Zeiss, AXIO Imager M2) and ZEN software (version 2.0). Neuronal scoring: Neurons were given a numerical value according to their expression levels. Wild-type expression scored 1, decreased expression scored 0.5 and abolished expression scored 0. Percentage of GFP expressing animals was then correlated to the theoretical maximum score using the equation below. Statistical analysis. Statistical analysis was performed in GraphPad Prism 6 using one-way ANOVA with Newman-Keuls Multiple Comparison Test. Values are expressed as mean + /− s.d. Differences with a P value < 0.05 were considered significant.