Peptide neuromodulation in invertebrate model systems

Abstract

Neuropeptides modulate neural circuits controlling adaptive animal behaviors and physiological processes, such as feeding/metabolism, reproductive behaviors, circadian rhythms, central pattern generation, and sensorimotor integration. Invertebrate model systems have enabled detailed experimental analysis using combined genetic, behavioral, and physiological approaches. Here we review selected examples of neuropeptide modulation in crustaceans, mollusks, insects, and nematodes, with a particular emphasis on the genetic model organisms Drosophila melanogaster and Caenorhabditis elegans, where remarkable progress has been made. On the basis of this survey, we provide several integrating conceptual principles for understanding how neuropeptides modulate circuit function, and also propose that continued progress in this area requires increased emphasis on the development of richer, more sophisticated behavioral paradigms.

INTRODUCTION

Neuropeptides offer useful entry points to study how the brain controls behavior. For example, activation of neurons releasing Agouti-Gene Related Peptide promotes robust feeding behavior (Aponte et al., 2011). In addition, the Orexin/Hypocretin peptides closely regulate behaviors that coordinately involve feeding, metabolism and the alternation between sleep and wakefulness (Sakurai and Mieda, 2011). Likewise and strikingly, vasopressin is a critical factor in decision-making that underlies social affiliation (Pitkow et al., 2001). These three illustrations are not exclusive as important examples for the modulation of behavior by neuropeptides, and still they reveal the broad scope of behavioral biology that neuropeptides address. Hence there is increased interest in studying how peptide neuromodulators affect behavior at systems, cellular and molecular levels. Much recent work pursuing these questions has focused on behavior in invertebrates, due to several technical advantages such model systems offer. This article offers a selective overview of that work.

The literature we review is recent, but in general these efforts date back to pioneering experiments of the 1960's and 1970's, when two broad sets of observations laid the framework for modern studies. The first set derives from experiments in which extracts of brain, when injected back into animals, elicited complete behavioral routines. The possibility that endogenous brain chemicals could release and coordinate complex, fixed action patterns of behavior captured the imagination of many neuroscientists. For example, in the gastropod mollusk Aplysia, injection of an extract of the Abdominal Ganglion into mature animals engendered locomotor and feeding cessation, followed by stereotyped head-waving that facilitated the extrusion and deposition of an egg mass (Kupfermann, 1967, 1970) (Strumwasser et al., 1969; Toevs and Brackenbury, 1969). The active principle that triggered this ~hour-long fixed action pattern was a 36-amino acid peptide, the Egg Laying Hormone (ELH - (Heller et al., 1980); with the advent of molecular cloning, elh was one of the first identified genes directly implicated in the control of behavior (Scheller et al., 1982). Likewise in silkmoths, Truman and Sokolove (Truman and Sokolove, 1972) discovered a hormone in brain extracts called Eclosion Hormone (EH) that triggers the stereotyped behavioral sequence needed to shed the old cuticle and complete a molting cycle ((Truman, 1992). EH is a 70 AA peptide (Kataoka et al., 1987; Marti et al., 1987) that modulates behavior by working peripherally and centrally, as described below.

The second set of observations that inspired much current work on peptide modulation of behavior derives from studies of a tiny ganglion that innervates the gut of crustacea. From this work was born the concept that neural circuits are hard-wired, but may generate multiple outputs due to modulation. The stomatogastric ganglion contains roughly 30 identified neurons (the exact number depends on the species and even on the individual studied) and it supports rhythmic contractions of more than 40 pairs of gut muscles via two basic Central Pattern Generators (CPGs) (Marder and Bucher, 2007). The initial work sought to implicate the role each individual neuron plays and in doing so, reveal how specific neural properties and neural connections contrive to form a CPG. That body of work succeeded in producing useful models for how such neural networks are organized (e.g., (Maynard, 1972; Miller and Selverston, 1982). It was all the more remarkable when several groups subsequently discovered that the diminutive ganglion of ~30 neurons and its stable neural network produced much more than just the two basic rhythmic motor patterns (e.g., (Hooper and Marder, 1984, 1987) (Flamm and Harris-Warrick, 1986) (Nusbaum and Marder, 1989; Turrigiano and Selverston, 1990). It was soon apparent that several stable circuit configurations were latent within the system, elicited by the application of diverse modulatory substances. As now observed in many different neural circuits, modulatory inputs can change essentially all the functional components of a network (Marder and Bucher, 2007) (Bargmann, 2012; Brezina, 2010; Kristan et al., 2005; Kupfermann and Weiss, 2001).

An especially important class of modulators are neuropeptides – Neuropeptides refer to small peptides and peptide hormones derived from nerve cells whose molecular lengths range from as short as three amino acids (e.g., TRH – (Nillni et al., 1996) to as long as 70 or more (e.g., EH - (Truman, 1992). Neuropeptide receptors are primarily found among the large family of G protein-coupled receptors (GPCRs), however, there are notable exceptions - some neuropeptides directly gate ion channels (Cottrell, 1997); insulin, which is a neuropeptide in some invertebrates (e.g., (Brogiolo et al., 2001) signals through its traditional tyrosine kinase Insulin Receptor; finally, neurons secrete a multitude of other proteinaceaous factors (e.g., growth factors) that signal through diverse receptor types. To focus our efforts, we primarily restrict this review to a discussion of neuropeptides that activate GPCRs, because they belong to the broadest and most widely used neuropeptide receptor family. Even in spite of this restriction, we do not attempt a comprehensive review of the literature describing peptides and invertebrate behavior. Instead we overview selected studies of modulation of three different categories of behavior (feeding, ecdysis and locomotion) to illustrate what we consider some fundamental lessons learned so far. We pay special (though not exclusive) attention to studies in genetic model systems, as these have recently come to the fore in studies of neuropeptide modulation. Finally, we summarize by distilling what may be an initial list of principals for neuropeptide modulation of behavior, and indicate where future progress may lie. We apologize to colleagues for important work not here covered and refer interested readers to excellent, contemporary reviews of this literature (Holden-Dye and Walker, 2012; Nassel and Winther, 2010; Nusbaum and Blitz, 2012)

ILLUSTRATIVE EXAMPLES OF NEUROPEPTIDE MODULATION OF BEHAVIOR

Feeding Behavior

Behaviors associated with food seeking, recognition and ingestion can be categorized as appetitive versus consummatory, according to traditional analyses of animal behavior (Kupfermann, 1974a; Lorenz, 1950). The first category corresponds to exploratory behavior that is subject to environmental influence and that may displays variability. The consummatory category refers to the release and execution of innate behavioral sequences and its display is more or less invariant. Single neuropeptides (like Neuropeptide Y: NPY) contribute to both appetitive and consummatory feeding behaviors in mammals (Dailey and Bartness, 2009), and their roles in the fundamental neuronal circuits underlying feeding behaviors have been intensively studied. For example, detailed models are now emerging the explain how peptidergic neurons in the arcuate nucleus that secrete AgRP (Agouti-Related Peptide), GABA, and NPY promote feeding by inhibiting other neurons of the parabrachial, arcuate, and paraventricular nuclei of the hypothalamus (Aponte et al., 2011; Atasoy et al., 2012; Wu et al., 2009). That action depends on GABA and NPY more than AgRP, and is age-dependent and subject to hormonal modulation (Yang et al., 2011) (Luquet et al., 2005). To complement such advancing mammalian studies, invertebrate model systems offer sophisticated genetic manipulation and/or favorable cellular resolution: these features can help address the basis for the profound effects peptide modulators have on feeding behavior. In this section we overview several different invertebrate studies (in insects, in C elegans, in Aplysia) to illustrate potential cellular mechanisms of how peptide modulation may contribute to shape both appetitive and consummatory feeding behaviors.

Insect feeding: sNPF and TK peptides gate critical sensory input by pre-synaptic modulation

The motivational state profoundly influences the specific responses animals display in response to identical, food-associated stimuli (e.g., (Kupfermann, 1974b). In Drosophila, neuropeptides implicated in regulating feeding-associated behaviors include the NPY homolog, NPF (Krashes et al., 2009; Wu et al., 2005a; Wu et al., 2005b), which, like NPY, appears to be specifically dedicated to modulation of circuits involved in metabolism, stress, and energy homeostasis (Nassel and Wegener, 2011). Other feeding-associated peptides include hugin (Melcher and Pankratz, 2005), leukokinin (Al-Anzi et al., 2010) and Allatostatin A (Hergarden et al., 2012). For example, starvation increases food-searching behavior by the fly and increases physiological responsiveness in an identified olfactory glomerulus called DM1. DM1 normally responds to cider vinegar and its state-dependent responsiveness is increased due to the actions of a neuropeptide called small NPF (sNPF). sNPF is genetically distinct from NPF, is found at all levels of the neuraxis, and is probably involved in many diverse modulatory functions (Nassel and Wegener, 2011). Within the olfactory system, sNPF is expressed by the primary afferents (the olfactory sensory neurons – OSNs) and positively regulates OSN pre-synaptic release innervating DM1. Selective expression of an RNAi transgenic construct for either sNPF or sNPF-Receptor (sNPF-R) in OSNs abolishes both the behavioral and the physiological effects of starvation (Root et al., 2011). Evidence suggests sNPF signaling is diminshed by insulin signaling: the latter signals the satiety state, such that high insulin signals block sNPF receptor expression and thus diminishes DM1 glomerular responses to food (Root et al., 2011).

Neuropeptide modulation of olfactory sensory map in Drosophila includes negative regulation: Drosophila tachykinin (dTK) peptides are co-released from a subset of olfactory GABAergic local interneurons. The glomeruli contain very high levels of dTK peptide and OSN terminals contain dTK Receptor, dTKR (Ignell et al., 2009; Winther et al., 2006). RNAi knockdown of dTK in these brain regions led to deficits in the display of innate odor preference (Winther et al., 2006); likewise knockdown of dTKR, or its overexpression in OSNs, led to increased and decreased (respectively) responsiveness to specific odorants. GABA and dTK both reduce calcium levels in ORN terminals and thus reduce the likelihood of OSN transmitter release. These studies clearly support the hypothesis that peptide modulation shapes food-seeking behaviors by affecting pre-synaptic release properties of OSNs and thus modifying the map of odor preferences (Wang, 2012).

Insect feeding: NPF permits motivation to modulate appetitive behavior by gating the display of memory

Memory has been linked to these hunger signals and models of motivation include learned representations of cues associated with food, such as smell and taste, that provide additional incentive and direction to locate a particular food source (Toates, 1986). Through conditioning, Drosophila can be trained to associate odorants with sucrose reward (Tempel et al., 1983) and this appetitive memory performance is best displayed by flies that are hungry (Krashes and Waddell, 2008). That experimental situation permitted Waddell and colleagues to investigate how signals for hunger and satiety may interact with memory circuitry to regulate the behavioral expression of learned food-seeking behavior (Krashes et al., 2009). The authors directly implicated NPY as a critical element of a motivational switch that signals the hunger state and controls the output of appetitive memory.

Thus Krashes et al. (Krashes et al., 2009) used elegant genetic manipulations to focus attention on a serial, two-stage inhibitory neural circuit controlling appetitive memory performance in the Mushroom Bodies (MB) in the fly brain. The MB is a specialized neuropil comprising 1000s of neurons that integrate multi-modal inputs, with a special regard for the receipt of olfactory inputs. They are thought to play important roles in insect learning and memory (Strausfeld et al., 1998). Satiety leads to poor appetitive memory performance due to dopamine (DA) inhibition of the MB from six identified MB-MP neurons. Relieving that tonic inhibition, by precise genetic mosaic methods, unmasked the performance of an intact appetitive memory in a fully satiated fly. Hence MB-MPs normally block the expression of the memory at the level of MB neurons via specific DA inhibition. How does NPF fit into this feeding circuit? NPF is expressed in only a small set of neurons in the fly brain and stimulating those specific neurons by genetic manipulation revealed they operate upstream of the MB-MP inhibitory neurons: NPF neurons transmit the hunger state to unmask appetitive memory. Importantly, that neuropeptide NPF action was localized to MB-MP neurons by knocking down NPF Receptor selectively in MB-MP neurons – such a manipulation lead to a loss of appetitive memory display. Thus, NPF provides critical modulation of appetitive feeding behavior in the fly by directly inhibiting dopaminergic MB-MP cells which has the effect of dis-inhibiting MB neurons and therefore permitting the propagation of appetitive memory information. The likelihood that appetitive behavior is triggered by the conditioned odorant is determined by the competition between inhibitory systems in the brain (Krashes et al., 2009).

Aplysia feeding: neuropeptide modulation of behavior features feed forward signaling

In Aplysia, a central pattern generator produces two competing feeding motor programs – one supporting ingestion and the other supporting egestion. Neuropeptides operate in consummatory phases of feeding behaviors to promote a phase switching from the ingestive to egestive programs. How they produce this effect provides remarkable cellular detail to the mechanisms of peptide modulation. Two critical components in this CPG system are (i) the B20 interneuron, which promotes the egestive rhythm, and (ii) the B40 interneuron, which promotes the ingestive rhythm (Jing and Weiss, 2001, 2002). This form of circuit organization ensures that it is the balance of B40 and B20 activity that determines whether feeding responses to food stimuli are ingestive, intermediate, or egestive. As the animal ate and became sated, the subsequent change in feeding behavior was not simply an inhibition of ingestive responses, but instead a replacement of those responses with nonfunctional (intermediate) and/or egestive motor responses (Jing et al., 2007). B40 and B20 do not inhibit each other directly – instead the switch from ingestive to egestive behaviors as satiety increases represents the selection by external modulation. The Aplysia ortholog of the NPY neuropeptide, aNPY, contributes to this important modulatory control by acting as a critical trigger for reconfiguration of the multifunctional CPG network (Jing et al., 2007). aNPY released from gut afferents within the CNS acts on the B20 interneuron to promote the switch to egestion. Separate gut afferents activate the identified neuron B18. B18 in turn releases aNPY to act on B20 and help effect the switch from ingestive to egestive modes. The authors argue that the gut afferents effectively use “dual outputs” to more effectively promote the egestive motor program: (i) the first output is the direct biasing of the CPG by aNPY and (ii) the second is the enhancing action through a feedforward pathway via B18 (Figure 1A).

Figure 1. Peptidergic modulation of behavior features feedforward pathways.

A. In the mollusk Aplysia, feedforward mechanisms coordinate neuropeptide modulation of feeding circuitry. Satiation is signaled by sensory afferents from the gut (here symbolized by EN1 and EN2, blue). EN1 signaling is mediated in part by the NPY homologue apNPY that modulates many different elements including the B20 interneuron to switch from ingestion to egestion modes. A parallel feedforward pathway is mediated by EN2 afferents to activate the B18 interneuron (red) which also releases apNPY to augment the ingestion-to-egestion switch. Adapted from Jing et al., (2007).

B. In the mollusk Aplysia, feedforward mechanisms provide neuropeptide-mediated compensation in feed circuitry. The allatotropin-related peptide (ATRP) is released centrally by the CBI-4 interneuron (blue) shortening protraction duration, but it also increases the rate of protraction motorneurons (MN, red). By this feedforward mechanism, MNs release ATRP (and another neuropeptide, MM) to increase muscle contraction amplitude and so compensate for the effect of a decreased duration.

C. In Drosophila, a feedforward mechanism is suggested by the arrangement of neuropeptide PDF signals to modulate circadian neural circuits. Genetic experiments suggest both the large LNv (blue) and the small LNv (red) contribute to coordination of daily locomotor rhythms under the influence of light and dark. Pharmacology and expression studies indicate the existence of a direct and an indirect (feedforward) PDF pathway to help synchronize other pacemakers in the circuit.

This dual output nature of aNPY actions represents an intriguing example of feedforward signaling. As elements of neural circuit design, feedforward pathways are instances in which the inputs and outputs of Neuron X are themselves directly connected. Feedforward pathways are termed coherent when both the indirect pathway to the output (via Neuron X) and the direct pathway (by-passing Neuron X) share the same sign. Coherent feedforward pathways may provide coordination among circuit elements that have divergent inputs and common outputs (e.g., (Jarrell et al., 2012). Modeling studies suggest that in transcriptional networks, they can provide subtle temporal variation in the control of target genes (Mangan and Alon, 2003). Additional studies of the Aplysia feeding CPGs support the hypothesis that neuropeptide modulation of behavior features extensive feedforward mechanisms (Jing et al., 2010; Wu et al., 2010). A novel neuropeptide (called ATRP) provides a striking additional example of a feedforward mechanism being used for compensation (Jing et al., 2010). ATRP acts centrally on the feeding CPG to accelerate the ingestion program, and does so by reducing the protraction phase. This action could conceivably compromise the ingestion program, because reducing the protraction phase would shorten the time available for protractor muscle contractions (and thus weaken them). However, there is feedforward aspect to ATRP actions: the same ATRP peptide is released directly onto the muscle by its motorneuron to act peripherally, and this second (local) action increases the rate of muscle contraction (Figure 1B). Thus, peptide modulation of behavior coordinates action at several synaptic levels, and is well-described by a feedforward design in the neuropeptide modulation of neuronal circuitry.

Worm feeding: NPR-1 and control of food-related behavior

The best characterized example of neuropeptide-modulated behavior in C. elegans nematode worms is food-related aggregation (or clumping). Some wild-type strains (including the commonly used N2 strain) forage on a lawn of bacteria in solitary fashion, while others aggregate into clumps of worms; this aggregation is termed “social” (de Bono and Bargmann, 1998). The genetic basis for this naturally occurring behavioral polymorphism has been identified as a single amino acid polymorphism in the npr-1 gene, which encodes a member of the neuropeptide Y receptor (NPYR family) (de Bono and Bargmann, 1998). Worm strains bearing null-mutant alleles of npr-1 are social, as are those bearing the partial loss-of-function allele encoding the 215Phe isoform (found in all the social strains), while strains bearing the allele encoding the 215Val isoform (including N2) are solitary (de Bono and Bargmann, 1998). The food-dependent aggregation of social npr-1 mutant worms relies on chemosensory responses in a number of different sensory neurons, including both external chemosensors and internal chemosensors (Coates and de Bono, 2002; de Bono et al., 2002). Cell-specific rescue of solitary feeding in social npr-1 mutants indicates that NPR-1 functions both in a subset of these sensory neurons “spokes”, as well as an interneuron “hub” to suppress chemosensory-driven aggregation (Coates and de Bono, 2002; Macosko et al., 2009). Genetic pathway analysis in sensory neurons suggests that NPR-1 suppresses their activation by chemosensory inputs, while its cellular function in the hub interneuron is likely more complex (Coates and de Bono, 2002; Macosko et al., 2009).

The ligands for NPR-1 turn out not to be NPY-related peptides, but rather FMRFamide-related peptides (FaRPs) encoded by the flp-18 and flp-21 genes (Rogers et al., 2003). FLP-18 and FLP-21 FaRPs activate NPR-1 expressed either heterologously in Xenopus laevis oocytes or ectopically in vivo in worm pharyngeal muscle (Rogers et al., 2003). Interestingly, the two peptides have a differential ability to activate the 215Phe and 215Val isoforms of NPR-1, with FLP-21 activating both isoforms, albeit with approximately ten-fold reduced affinity for the 215Phe isoform, while FLP-18 only activates 215Val (Rogers et al., 2003). FLP-21 appears to be the endogenous NPR-1 ligand required for its activation and consequent suppression of food-dependent aggregation, as flp-21 gene deletion increases food-dependent aggregation (Rogers et al., 2003). Conversely, FLP-21 overexpression from multiple transgenes with endogenous promoter sequences suppresses the social phenotype of npr-1 alleles encoding the 215Phe isoform, but not npr-1 null-mutant animals (Rogers et al., 2003). Taken together, these studies indicate that NPR-1 activation by FLP-21 plays a key role in altering the behavioral valence of chemosensory cues through neuronal modulation to encourage solitary feeding and discourage social feeding.

NPR-1 modulates behavioral responses not only to food and the presence or absence of other worms, but also to other key environmental parameters. One of the most important of these environmental parameters is ambient O ₂level: too little and cellular respiration fails, but too much is cytotoxic. NPR-1 plays key roles in modulating both direct responses of worms to O₂ gradients mediated by neuronal guanylate cyclases as well as the integration of sensory cues of food availability, internal metabolic state, and O₂ level (Chang et al., 2006; Macosko et al., 2009; Rogers et al., 2006). FLP-21/NPR-1 also modulate hypoxia-induced changes in NaCl preference, acute CO₂ avoidance, and acute responses and tolerance to ethanol, although the cellular loci and neural modulatory mechanisms for those effects have not been identified (Davies et al., 2004; Hallem and Sternberg, 2008; Pocock and Hobert, 2010). NPR-1 also functions in the hub neuron—described above in the context of social versus solitary feeding—to modulate Transient Receptor Potential ion channel (TRP)-mediated heat avoidance (Glauser et al., 2011), and in the hub neuron as well as external chemosensory neurons to repress food-leaving induced by food depletion (Milward et al., 2011). Finally, NPR-1 has also been demonstrated to influence the susceptibility of worms to infection by pathogenic bacteria, most likely through a combination of influences on animal behavior and innate immunity (Reddy et al., 2009; Styer et al., 2008).

Worm feeding: other neuropeptides modulating food-related behavior

The FLP-18 peptide that activates NPR-1 also activates NPR-4 and NPR-5, and this signaling pathway is important for modulating both foraging behavior and energy metabolism (Cohen et al., 2009). Worms with loss-of-function mutations in flp-18, npr-4, or npr-5 exhibit increased fat accumulation and a failure to appropriately switch from local search foraging to long-range dispersal upon severe food depletion (Cohen et al., 2009). Cell-specific rescue of flp-18, npr-4, or npr-5 mutants leads to a model in which FLP-18 peptides are secreted by a particular bilateral interneuron pair in response to sensory cues of food availability and then activate NPR-4 in other interneurons and the intestine to regulate foraging and fat storage, respectively (Cohen et al., 2009). Other neuropeptide systems besides NPY-related flp-21/npr-1 have been studied in the context of food-related sensorimotor integration. Unlike npr-1, which is expressed in numerous sensory neurons and interneurons, worm allatostatin/galanin-related receptor npr-9 is expressed solely in a single bilateral interneuron pair that has been previously shown to control local foraging search behavior (Bendena et al., 2008). npr-9 loss-of-function mutants exhibit increased local turning at the expense of long-range forward movements while on food, while overexpression of NPR-9 in this interneuron induces increased long-range forward movement at the expense of local turning (Bendena et al., 2008).

These studies on the various food-related organismic functions modulated by neuropeptides, their cellular loci, and their cellular and molecular mechanisms paint a picture of neuropeptide signaling pathways that regulate the key survival traits of the worm: obtaining things that are necessary for life and avoiding things that are dangerous to life. These receptors and ligands are expressed in multiple neurons, and act to both gate sensory inputs and alter the network state of central processing modules (such as the one defined by the described hub interneuron). The key issues left experimentally unaddressed by these studies are the physiological and/or environmental food-related stimuli (if any) that regulate ligand secretion and the regulated patterns of ligand secretion and consequent receptor activation that induce adaptive alterations of neuronal information processing.

Worm sensorimotor gain control: neuropeptide-mediated sensorimotor feedback

In addition to feedforward modulation of sensorimotor integration as embodied in the examples described above, there is also evidence for neuropeptide-mediate sensorimotor feedback loops in which sensory or motor activation induces neuropeptide secretion that then modulates the gain or temporal properties of that activation process. NLP-1, a buccalin-related peptide, is expressed in a chemosensory neuron and acts upon the NPR-11 receptor in an interneuron to modulate the dynamics of the odor-evoked response in that same chemosensory neuron, suggesting the existence of a feedback connection between the interneuron and the chemosensory neuron (Chalasani et al., 2010). This feedback connection is mediated by an insulin-related peptide (INS-1) secreted by the interneuron (Chalasani et al., 2010). The NLP-12 peptide is expressed specifically in a stretch receptor neuron, and loss-of-function mutants of nlp-12 or its receptor (ckr-2) eliminate pharmacologically induced presynaptic potentiation of ACh release at the neuromuscular junction and result in decreased locomotion rates (Hu et al., 2011). In addition, imaging analysis of fluorescently tagged NLP-12 suggests that its secretion is stimulated by the pharmacological agent that induces presynaptic potentiation and that stimulation is prevented by a TRP channel mutation that disrupts mechanosensation in the stretch receptor (Hu et al., 2011). These results support a model in which NLP-12 mediates a feedback loop that couples motor-induced activation of a stretch receptor to the strength of the neuromuscular junction, although future work is required to identify the cellular locus and molecular mechanisms by which CKR-2 receptor activation closes the loop.

Worm reproduction: neuropeptide modulation of egg laying and sexual behavior

Neuropeptides also modulate worm reproductive behaviors, including egg laying and copulation. Neuropeptides encoded by the flp-1 gene (the first worm neuropeptide gene whose mutation was shown to induce behavioral defects; Nelson et al., 1998) promote transition from the behavioral state of egg retention to active egg laying, as flp-1 loss-of-function mutant worms spend longer in the egg-retaining state than wild-type worms (Waggoner et al., 2000). FLP-1 peptide regulation of egg-laying is bidirectional, as flp-1 mutant worms also fail to suppress egg-laying in food-poor environments (Waggoner et al., 2000). Egg-laying behavior is also modulated by the EGL-6 neuropeptide receptor whose ligands are related FaRPs encoded by the flp-10 and flp-17 genes (Ringstad and Horvitz, 2008). These peptides are expressed in sensory neurons that inhibit egg-laying, as when they are ablated, egg-laying is increased, while egl-6 is expressed in motor neurons that innervate egg-laying muscles (Ringstad and Horvitz, 2008). This leads to a simple model in which sensory stimuli relevant to the suitability of the environment for egg-laying control FLP-10/FLP-17 secretion, which directly modulates the activity of the egg-laying motor neurons to promote egg-laying in suitable environments and suppress it when unsuitable. Mutation of genes encoding components of non-FaRP neuropeptide pathways also influence egg-laying—including Pigment Dispersing Factor (PDF) and adipokinetic/gonadotropin releasing hormone—although the cellular loci and molecular mechanisms underlying these effects have not been addressed (Lindemans et al., 2009; Meelkop et al., 2012).

While worms are generally hermaphoditic and internally self-fertilize, under certain environmental conditions, males develop and engage in copulation with hermaphrodites. Loss-of-function mutations in the flp-8, flp-10, flp-12, or flp-20 neuropeptide genes of males each induce the phenotype of repetitive turning, where instead of making a single turn around the hermaphrodite before initiating copulation, the male engages in repeated turning, thus delaying copulation (Liu et al., 2007). These particular flp genes are expressed in male-specific neurons, touch receptor neurons, and some interneurons, but touch receptor-specific rescue of flp-20 mutants completely restores single-turn male behavior (Liu et al., 2007). This suggests a model for flp-20 in which it conveys somatosensory information relevant to termination of turning and initiation of copulation to unknown target neurons.

Insect Ecdysis

Ecdysis describes behavior by which insects shed their old cuticle in favor of a newly-generated one that permits growth of the body or completion of a new body form (as occurs during metamorphosis). Ecdysis must coincide precisely with the internal physiology of the animal (its growth or new developmental stage): for example the older cuticle is loosened by internal digestion to permit its rapid and efficient removal; the new cuticle is transiently softened to permit rapid inflation, then subsequent hardening. In some cases, the old cuticle has a simple shape (like that of the caterpiller – essentially a tube). In many other cases however, the old cuticle is an elaborate costume that must be delicately and precisely removed – consider the ecdysial behaviors needed to remove old cuticle from the highly articulated legs of a locust (Fabre, 1917) or cricket (Carlson, 1977). Such an elaborate procedure requires a multi-step behavioral sequence wherein coordination must be balanced by efficiency, as the animal is naturally very vulnerable throughout this period.

Insect Ecdysis: ETH targets diverse peptidergic neurons

Ecdysis is controlled by a complex interplay of peptide factors derived from both central neurons and peripheral endocrine cells. Two specific peptides, Eclosion Hormone (EH) and Ecdysis Trigger Hormone (ETH), represent critical interacting factors: their actions and interactions illustrate aspects that are central to the peptide modulation of behavior. In the moth Manduca, ETH (and the co-synthesized P-ETH peptide) derives from endocrine cells associated with trachea and elicits coordinated behavior by directly activating diverse neural targets (Zitnan et al., 1996). To discover the cellular basis for this precise modulatory mechanisms, Adams and colleagues identified the receptors specifically tuned to ETH – these GPCRs are most closely related to receptors for mammalian Neuromedins and TRH (Hewes and Taghert, 2001a; Park et al., 2003). ETH-R is alternatively spliced into two variant protein isoforms and, in Drosophila, the two proteins display largely non-overlapping patterns of expression in the CNS. ETH-RA has been most extensively described: remarkably it is largely confined to diverse sets of peptidergic (DIMM-positive) neuroendocrine neurons (Hewes and Taghert, 2001a; Park et al., 2008). These ETH targets include identified neurons that (differentially) express large amounts of the peptides dFMRFa, leukokinins, CAPA peptides and EH (Kim et al., 2006a). Thus, rather than targeting the motor, or even the immediate premotor elements of the CPG that drives rhythmic ecdysial muscle activity, ETH modulation focuses on a collection of peptidergic elements as immediate targets. ETH modulation represents the temporal orchestration among these diverse peptidergic elements (Figure 2).

Figure 2. The ETH neuropeptide triggers a complex behavioral sequence by direct and sequential activation of intermediate peptidergic targets.

Ecdysis behavior in insects is a composite of several behavioral sub-sequences and is triggered by a complex set of neuropeptides. The ETH neuropeptide originating from peripheral endocrine cells acts high in the control hierarchy. It acts centrally to elicit a sequence of behaviors (timeline to the right) by directly activating a heterogeneous set of target neurons, most of which are peptidergic. Their activation is sequential, due to presumed parallel inhibitory pathways that are activated by ETH. Each peptidergic target has restricted and overlapping control of specific behavioral subsets. Arrowheads symbolize activating inputs; knobs symbolize inhibitory inputs. Adapted from Kim et al. (2006).

To visualize such orchestration, Kim et al. (Kim et al., 2006b) monitored GCAmP fluorescence intensity as a proxy for cellular activity in ETH-RA expressing neurons following exposure to ETH in vitro. They described a remarkable and highly predictable order by which the different ETH peptidergic targets displayed transient activation over the course of the ~ 30 minutes subsequent to ETH exposure. Thus, despite the fact that each group visualized expresses ETH-RA, there must exist parallel inhibitory interactions established within the target network to assure an orderly temporal activation and thus a proper sequencing of target peptide actions (Figure 2). To support their hypothesis, Kim et al showed that RNAi knockdown of individual peptide targets (e.g., knockdown of dFMRFa or of leukokinin) produced partial deficits in ecdysial behavior, deficits entirely consistent with a presumption of their sequential recruitment by the command chemical ETH. Thus in this highly detailed model of peptide modulation underlying the release of a complex innate behavior, the following model is put forth. The command chemical (neuropeptide ETH) directly activates a series of secondary neuropeptide messengers, and does so reliant on circuit interactions among targets that assure their proper temporal ordering.

Insect Ecdysis: Positive feedback directs release of the ETH Command Peptide

With the strong evidence that numerous different neuropeptides act as command signals to trigger innate behaviors, an important emerging question becomes – under what conditions are such critical factors released? From the study of insect ecdysis, there is excellent cellular, genetic and endocrinological evidence to suggest that positive feedback loops are employed for a stepwise, incremental approach to reach a threshold for peptide release. The best evidence is found in the control of ecdysis behavior in Lepidoptera. As mentioned above, in Manduca, injection of either EH or ETH can elicit complete ecdysis routines (albeit with different latencies). In part this reflects the fact that both peptides are positive regulators of release of the other. ETH secretion begins at the onset of preecdysis behavior and increases by positive feedback from EH to maximal activation, driving the behavioral sequence to its conclusion.

The current model suggests that stage-specific ecdysis behavior is produced by small triggering steps (probably due to other releasing factors, including the neuropeptides corazonin (COR) (Kim et al., 2004) and a diuretic hormone (DH) related to mammalian corticotropin releasing factor (Kim et al., 2006a). COR and DH release first produce a small amount of ETH release from the peripheral Inka endocrine cells into the blood. ETH starts to act on its central targets, one of which is the pair of EH-producing VM neurons of the brain. ETH-triggered VM cell activity initiates EH release which acts back on the ETH-producing Inka cells to eventually cause a massive ETH release and finally this helps cause massive VM release of EH (Clark et al., 2004; Ewer and Truman, 1997; Kingan et al., 1997). Ecdysis is a ballistic behavior: it happens infrequently, but it must happen at the correct time; it only lasts for minutes to hours, but cannot be reversed. Its control must therefore be precisely in synchrony with the proper internal state. The precision is due in part to the positive feedback between the two peptide hormone anchors (ETH and EH): this system offers an incremental, processive and interlocked decision-making mechanism. The final massive release events (that causes release of most ETH and EH stores) are only achieved as the final stage of mutual positive interactions that ensure a timely and proper endocrine resolution and subsequent triggering of behavior.

Positive feedback has also been suggested to control ultimate release of neuropeptides that trigger other innate behaviors in insects. Specifically Luan et al. (Luan et al., 2012) have analyzed the decision-making network for wing-spreading behavior of newly-emerged adult Drosophila which is triggered by the protein hormone bursicon (BUR). BUR is released from a pair of command interneurons (called Bseg) to provoke release of the same hormone from other neurons (called Bag) to support hardening of wing cuticle. The authors infer a loop wherein Bseg activity feeds back positively to permit its own concerted release of BUR and release of the proper behavioral sequence (Luan et al., 2012). In endocrinology, the classic example of a positive feedback system is the control of ovulation in mammals, in which the hypothalamus and ovary interact positively to generate the luteinizing hormone surge that coordinates ovulatory events (Clarke, 1995). A threshold level of estrogen is reached in the follicular phase of the ovarian cycle and this signals changes from a negative feedback to a positive one: it now activates cells within the brain, probably dis-inhibiting inhibitory systems and activating positive inputs including kisspetin and noradrenergic afferents to GnRH neurons (Smith et al., 2011). Thus, positive feedback loops may be a more general organizing principal to control release of peptide modulators that trigger stage-specific behaviors.

Drosophila circadian circuits: peptide support of 24-hr rhythmic outputs from a dedicated circuit

This final section features an overview of the lessons learned from the study of neuropeptide actions in circadian physiology. In mammals, neuropeptides play important roles in the critical neuronal circuits of the hypothalamic Suprachiasmatic Nucleus (SCN) which is a principal center in the hypothalamus for control of daily rhythms. In particular, Vasoactive Intestinal Polypeptide (VIP), Pituitary Adenylate Cyclase Activating Peptide (PACAP), Vasopressin (VP) and Gastrin Releasing Peptide (GRP) are prevalent in the SCN and have demonstrable neurophysiologic and genetic actions (e.g.,(Aton et al., 2005; Hannibal et al., 2008; Irwin and Allen, 2010; Maywood et al., 2011). Of special note is the role of VIP in supporting rhythmicity to the SCN circuitry: it is essential to maintain normal rhythm generation at the molecular, cellular and behavioral levels (Aton and Herzog, 2005). Because it has a central role as a synchronizing agent in the SCN, understanding the mechanisms of VIP actions and the controls on its release represent a fundamental problem in circadian biology. In this regard, parallel studies of peptide modulation in circadian physiology in the numerically simpler neuronal circuits of invertebrates offer useful points for comparison.

The invertebrate peptide of particular relevance to circadian physiology is the Pigment Dispersing Factor (PDF). PDF is a member of the larger family of Pigment Dispersing Hormones (PDHs) that originates from earlier studies of crustacean endocrinology – PDH causes chromatophore dispersion in diverse extra-retinal and epithelial pigment-bearing cells (Rao and Riehm, 1993). As a function of time of day, the distribution of pigment granules within chromatophores is either constricted to permit greater light sensitivity, or extended for light shielding. In insects, there is a single Pdf gene, which encodes a ~100 amino acid precursor, the final portion of which contains the 18 amino acid PDF.

There are many reasons why this example represents one of the most advantageous contexts within which to study the complex mechanisms underlying the modulation of behavior by neuropeptides, and here we list three general ones. [Full disclosure – both authors of this review work on this system]. (i) PDF peptide modulation works in the context of the 24-hour daily rhythm generated by circadian clocks present in a network of interacting pacemakers. This issue raises important questions of how the pacemaker properties of PDF neurons influences PDF release, and how PDF may feed back and modulate the same clock properties. (ii) This system has access to several rhythmic behavioral outputs and is sensitive to the different environmental inputs that entrain clock rhythms. Thus, peptide release and actions can be categorized within a broader context of upstream and downstream circuit interactions, and interactions of the organism with its environment. (iii). There are a substantial number of laboratories currently studying PDF modulation in the fly brain using genetic analyses to address its mechanisms in unbiased ways. This concerted focus means there will be opportunities not easily possible in non-genetic systems to make novel connections between the details of PDF synthesis, release and signaling and other aspects of neuronal cell biology. In general we submit this peptide modulatory system has unique features because it combines the benefits of a genetic model system with the clarity of a neural network that displays cellular resolution.

PDF expression is restricted to the CNS (Helfrich-Forster, 1997; Nassel et al., 1993): there are approximately 16 neurons that also display strong circadian clock protein expression – called the large and small LNv (Lateral Neuron ventral). There are other PDF-expressing neurons in the CNS, but they are few in number and probably contribute little to the generation of rhythmic locomotor activity (Shafer and Taghert, 2009). In the circadian pacemaker network of the fly brain, approximately 10% of the pacemakers (16 of ~150) express PDF, while PDF-R is expressed by about 60% of all pacemakers. Interestingly, PDF receptivity is found in nearly all of the pacemaker cell groups (Shafer et al., 2008), but in most groups the PDF-R is only found in a subset (Im and Taghert, 2010) – for example, in the six-cell LNd pacemaker group, PDF-R is expressed by only three, and in the 15-cell DN1 group, PDF-R is expressed by only 6-7.

Drosophila circadian circuits: PDF modulation derives from distinct peptidergic cell types

An interesting aspect of the PDF cell population is the stark heterogeneity of its cellular properties. PDF expressing pacemakers are comprised of two groups – the 4-5 large LNv and the four small LNv (Helfrich-Forster, 1995). Both cell types contribute (non-redundantly) to the generation of rhythmic locomotor activity (Cusumano et al., 2009; Helfrich-Forster, 1998; Shafer and Taghert, 2009; Sheeba et al., 2010; Yang and Sehgal, 2001). Both large and small LNv express the molecular clockworks, but they differ in many other important ways. (i) The large cells are neuromodulatory and form a large projection tangential to the retinotopic projections of axons from the eye, within a distal layer of the medulla (Helfrich-Forster, 1997; Taghert et al., 2000); In contrast, the small LNv make a precise topographic projection to dorsal protocerebrum, for which incorrect targeting by even a few microns is enough to abrogate their informational functions (Helfrich-Forster, 1998). (ii) Large cells express the bHLH transcription factor DIMM and give no evidence of utilizing a small classical co-transmitter (Taghert et al., 2001). DIMM-expressing neurons are dedicated and diverse neurosecretory cells which are generally large and which produce and episodically release large amounts of neuropeptides (Park and Taghert, 2009). Small LNv do not express DIMM and also co-secrete small conventional transmitters (Choi et al., 2012; Johard et al., 2009; Taghert et al., 2001; Yasuyama and Meinertzhagen, 2010). (iii) Large cells are directly light-sensitive and play important roles in setting the arousal state and the phase of the evening activity peak (Chung et al., 2009; Parisky et al., 2008; Sheeba et al., 2008); (iv) Large LNv act more like hourglasses than circadian oscillators: when placed in constant darkness, large LNv lose their PER molecular rhythm within a single cycle; in contrast, small LNv display durable molecular oscillations in constant darkness and contribute critical PDF signaling under those conditions (Lin et al., 2004; Peng et al., 2003; Yang and Sehgal, 2001); (v) Large cells express no or low amounts of PDF-R, while small LNv are PDF-sensitive and relay light information from the large LNv (Im and Taghert, 2010; Kula-Eversole et al., 2010; Shafer et al., 2008) (Helfrich-Forster et al., 2007). It is therefore an interesting though unexplained feature of this critical modulatory system that it displays such a degree of cellular heterogeneity, consisting of large peptidergic modulators (l-LNv) working with small, more conventional neurons that employ peptides along with classical small transmitter(s). Whether this particular cellular profile represents an essential element of a modulatory system remains to be determined.

Drosophila circadian circuits: PDF receptors define where modulation takes place

Of the two broad classes of neuropeptide GPCR families, PDF-R is a member of the smaller one – called the Secretin Receptor or B1 family Receptors of this group traditionally signal via Gs alpha and calcium (Harmar, 2001). Experiments in vitro and in vivo indicate PDF-R probably signals though cAMP (Mertens et al., 2005; Shafer et al., 2008) (Hyun et al., 2005) (Choi et al., 2009) and perhaps also via Ca2+ (Mertens et al., 2005). When small LNv express a gene encoding a “tethered PDF”, their resting membrane potential is depolarized even when they are decoupled from neuronal signaling networks by bath application of tetrodotoxin, which block Na⁺-dependent action potentials (Choi et al., 2012), suggesting that PDF generates electrogenic responses in PDF-R-expressing neurons. In the tethered peptide design, the PDF peptide sequence is fused by a linker region to a membrane-integral GPI anchor; the PDF moiety is located extracellularly and is able to interact with and activate cognate receptors expressed by the same cell (Choi et al., 2009) (Fortin et al., 2009; Ibanez-Tallon and Nitabach, 2012).

PDF-R present on PDF neurons (autoreceptors) may have different functions from those found in non-PDF pacemakers. While PDF signals received by non-PDF pacemakers are both necessary and sufficient for circadian rhythm generation, PDF signals received by the PDF-secreting LNvs themselves are largely dispensible (Im and Taghert, 2010; Lear et al., 2009). However, PDF signaling to autoreceptors on the PDF-secreting LNvs plays a key role in the circadian allocation of daily rest and activity between morning and evening (Choi et al., 2012). Consistent with the observation that cell-specific rescue of PDF-R expression solely in non-PDF pacemakers of PDF-R-null flies rescues free-running rhythms (Im and Taghert, 2010; Lear et al., 2009), cell autonomous activation of PDF-R solely in PDF-negative pacemaker neurons with a membrane-tethered PDF construct promotes strong rhythmicity in Pdf-null flies, which would otherwise be poorly rhythmic (Choi et al., 2012).

Flies deficient in PDF or PDF-R display severe deficits in circadian rhythms and alterations in PER molecular rhythms during constant dark (DD) conditions. Among the 4 small LNvs, rhythms are maintained but become de-synchronized (Lear et al., 2005; Lin et al., 2004). Among PDF target pacemaker groups like the LNd, the amplitude and period of the PER rhythm decrease but cells remain synchronized (Lin et al., 2004; Yoshii et al., 2009). Thus PDF neuropeptide acts over many daily cycles to promote the amplitude and pace of PER cycling – it has access to the molecular clockworks in diverse pacemakers and affects them differently. Recent observations have begun to shed light on the signaling pathways by which PDF affects synchronization and how these may differ according to cell type.

Drosophila circadian circuits: defining the importance of circadian signalosomes

Because PDF modulation system profoundly affects the circadian molecular oscillator within individual pacemaker neurons, the molecular details of the signaling pathway downstream of PDF-R gains in significance. Among the identified neurons in the pacemaker network, the PDF-expressing subset are termed M cells based on their abilities to influence “Morning” activity levels; several non-PDF pacemakers are termed E cells based on their abilities to influence “Evening” time activity levels (Grima et al., 2004; Stoleru et al., 2004; Yoshii et al., 2004) reviewed by (Helfrich-Forster, 2009). Duvall and Taghert (Duvall and Taghert, 2012) recently used an RNAi-mediated genetic approach to report that adenylate cyclase 3 (AC3) underlies PDF signaling in M cells. Surprisingly, disruption of AC3 does not alter PDF-R mediated responses in non-PDF pacemakers (specifically, in the PDF-R(+) LNd). Moreover, AC3 disruptions in small LNv did not alter GPCR signaling by other ligands that elevate cAMP levels in these neurons (dopamine and the neuropeptide DH31). Hence, within small LNv, PDF-R signaling occurs via discrete molecular pathways that are distinct from those controlled by other cAMP-elevating ligands. This provides a molecular mechanism underlying the observation that PDF-R activation in small LNv has potent effects on daily allocation of rest and activity, while DH31-R activation does not (Choi et al., 2012). Furthermore, PDF-R association with a different AC(s) supports PDF signaling in the other circadian pacemakers. Thus critical pathways of circadian synchronization are mediated by highly specific second messenger components. These findings support a hypothesis that PDF signaling components within target cells are sequestered into “circadian signalosomes”, whose compositions differ between different pacemaker cell types (Duvall and Taghert, 2012). The molecular identities of components within the signalosomes, and the functional significance of two different PDF modulatory pathways, are now issues for future consideration.

Drosophila circadian circuits: when does PDF modulation occur during the day?

Two key questions for the field to determine are - when during the 24-hr cycle is PDF released and when does it act? There are strong suggestions that PDF is rhythmically released, with peak accumulation in the sLNv dorsomedial terminals and activity-mediated release occurring in the morning (Cao and Nitabach, 2008; Park et al., 2000). In addition, manipulation of the biophysical membrane properties of PDF-secreting pacemakers with a membrane-tethered spider toxin that cell-autonomously inhibits voltage-gated Na⁺ channel inactivation induces phase advance both of daily morning activity and of rhythms of staining for PDF in sLNv terminals (Wu et al., 2008). These studies suggest that the phase of rhythmic PDF secretion determines the phase of morning activity. Further suggestion that PDF release occurs primarily in the early daytime comes from the genetic evidence that PDF signaling and CRY photoreception interact (Cusumano et al., 2009; Im et al., 2011; Zhang et al., 2009), as described below. All these lines of evidence are indirect measures however; direct observation of normal PDF release events in vivo remains a useful goal for the field.

Drosophila circadian circuits: PDF modulation converges with environmental inputs

An unexpected aspect to PDF modulatory actions in the Drosophila circadian neural circuit is its interaction with CRY signaling. In flies, CRY is a blue light photosensor (Panda et al., 2003) expressed in diverse circadian pacemaker neurons (Yoshii et al., 2008) and at many levels of circadian neural circuit, there is precise co-expression with PDF-R (Im et al., 2011)(Im and Taghert, 2010). These anatomical data complement genetic evidence indicating that PDF and CRY signaling interact in specific pacemaker subsets to support the phase and amplitude of the circadian molecular oscillator in pacemaker-specific fashion (Cusamano et al., 2009; Zhang et al., 2009; Im et al., 2011). The locomotor behavior of flies that are doubly mutant for Pdf and cry (or for Pdf–r and cry) is much more disrupted that for any single mutant; in the critical LNd neurons, the amplitude of the PER rhythm is greatly attenuated, the phase delay between the peaks of PER cytoplasmic and nuclear sub-cellular localization is lost, and the daily clearance of PER protein from the nucleus is no longer apparent (Im et al., 2011). Remarkably, the PER oscillator rhythm is normal in small LNvs. This is another indication that PDF signaling via autoreceptors has different signaling cosequences from PDF signaling in non-PDF pacemakers.

Exactly how to interpret the interactions between PDF and CRY signaling remains a point for study. (Cusumano et al., 2009) and (Zhang et al., 2009) concluded that PDF normally gates CRY signaling and in so doing delays the phase of an otherwise robust evening peak. An alternative, or perhaps complementary, interpretation suggested that PDF and CRY signals converge as a requisite input to sustain robust molecular oscillations in non-PDF pacemakers (Im et al., 2011) (Figure 3A). A similar convergence of sensory input and neuropeptide (EH) action explains the enhanced emergence of adult flies immediately following a lights-on signal (McNabb and Truman, 2008) (Figure 3B). Likewise, as described above, the avoidance of noxious ambient temperatures by C. elegans depends on convergent signaling by both a molecular thermoreceptor (a TRPV family member) and the FLP-21/NPR-1 neuropeptide signaling pathway (Glauser et al., 2011) (Figure 3C). Similar convergence of an environmental with an intrinsic signal may also help explain the switch from perch selection to wing expansion behaviors following eclosion (Peabody et al., 2009). These examples suggest that the coincidence of peptide release (signifying an internal state) with a specific environmental signal (the external state) is a generalizable concept. The extent to which it may support the neuropeptide modulation of behavior more generally remains to be determined.

Figure 3. Peptidergic modulation of behavior features convergence with specific environmental signals.

Orange - environmental signal; Blue - sensory receptor; Red - Peptidergic neuron

A. Light activates CRY in non-PDF (“E”) pacemakers in Drosophila in parallel to PDF signaling: CRY and PDF-R are co-expressed in individual E pacemakers. The convergence supports rhythmic oscillations and locomotor activity. Adapted from Im et al. (2011).

B. Light triggers rapid adult eclosion in Drosophila during a narrow temporal gate, provided there was prior neuropeptide EH release. EH activates an excitatory pathway for behavior and a parallel inhibitory one; light dis-inhibits the inhibitory pathway and the convergence promotes eclosion behavior. Adapted from McNabb and Truman (2009).

C. Temperature triggers rapid avoidance behavior in C. elegans; heat sensation mediated in part by OSM9 (~TRPV)-expressing sensory neuron(s); proper responses requires converging inputs from FLP21/NPR-1 peptide signaling that originates in RMG interneurons. Adapted from Glauser et al. (2011).

Drosophila circadian circuits: a role for feed forward signaling in PDF modulation?

The configuration of PDF neurons and PDF receptors in the Drosophila brain suggests the involvement of a feedforward effect, perhaps akin to that proposed for the role of modulatory peptides in the Aplysia feeding CPG (Jing et al., 2007; Wu et al., 2010). In this case, the connection from large LNv to non-Pdf pacemakers is the direct pathway and the large LNv to small LNv to non-Pdf pacemakers is the indirect one. The evidence for these different signaling pathways is varied and comes from different studies (Helfrich-Forster et al., 2007; Im and Taghert, 2010; Shafer et al., 2008; Shafer and Taghert, 2009) (Kula-Eversole et al., 2010) (Blanchardon et al., 2001; Cusumano et al., 2009; Renn et al., 1999b; Sheeba et al., 2010). These combined data suggest PDF in the circadian circuit acts at each of two levels and may thus be used in a feedforward fashion (Figure 1C). There is considerable genetic evidence to suggest that large and small LNv have different functional roles. A feedforward hypothesis for modulatory PDF actions may help design future experiments to better understand the logic of this cellular configuration.

EVOLUTIONARY CONSERVATION OF PEPTIDE MODULATORY SYSTEMS

It is possible that some of the neuropeptide modulators illustrated by these studies in invertebrates derive from ancestors that produced similarly-acting modulators in mammals. That is to say, a modulator affecting a specific behavior in an invertebrate may (in the simplest hypothesis) have a close sequence ortholog that acts in similar fashion in a vertebrate. This hypothesis is undermined by the many examples of modulatory peptides that appear not present in vertebrates (e.g., proctolin) or not present in Drosophila (e.g., Gonadotropin releasing Hormone, GnRH). It is also true however that, while peptide sequences are often short and hence difficult to use as bioinformatic probes, conserved features within peptide receptors are often more easily detected. Hence, while there are no Drosophila peptides clearly related to mammalian GnRH, the fly genome encodes a GPCR that is clearly a member of an ancestral GnRH receptor family (Lindemans et al., 2011). The presumption that vertebrates and invertebrates share orthologous modulatory pathways is therefore strengthened by genomic analyses showing that genes encoding many of the principal mammalian peptide GPCRs have orthologs in insect genomes (Fan et al., 2010; Hewes and Taghert, 2001b) (Fredriksson and Schioth, 2005; Hauser et al., 2008). This suggests that, comparable to the conservation of developmental signaling pathways like the Notch and hedgehog pathways, many of the key neuropeptide signaling pathways have been conserved over hundreds of millions of years, and that functional lessons learned in invertebrate model systems will continue to be instructive for the studies of vertebrates as well. However, this general observation leaves open the important questions - how and when are such “conserved” modulatory pathways deployed in unrelated animals?

We can first return to the simplest question and ask whether, in different animals, highly orthologous neuropeptides, and their receptors, are used in similar behavioral contexts, for apparently similar purposes. Some examples do support such a model of evolutionary constancy for the use of specific modulatory mechanisms. For example, the NPY/NPF family of peptides in mammals and in invertebrates (and their related receptors) are involved in feeding, stress responses, metabolism and reproduction (Nassel and Wegener, 2011). In the context of feeding, they affect both appetitive and consummatory phases of feeding behavior, as reviewed above. Likewise, the hugin family of peptides that negatively regulates Drosophila feeding (Melcher and Pankratz, 2005) activates receptors that are orthologous to the mammalian Neuromedin U family of GPCRs. Neuromedins have also been implicated as anorexigenic peptide modulators (Hanada et al., 2004; Howard et al., 2000). Hence the modulatory actions of the hugin/NeuromedinU (Melcher et al., 2006) and NPY/NPF families of peptides (among others) exhibit evolutionary constancy in regulation of neural circuits related to feeding behaviors and serve as clear examples of neuropeptide modulators whose functions may be relatable across broad evolutionary distances.

An evolutionary parallel is also suggested in the case of neuropeptides that modulate circadian control circuits in mammals (Vasoactive Intestinal Peptide: VIP) and insects (PDF), respectively. However, this case has a clear and important distinction. The contributions of these two peptide signaling systems to circadian physiology in the two sets of animals are highly similar (reviewed by (Vosko et al., 2007). In the Drosophila brain, PDF supports rhythmicity through distributed actions across the pacemaker network that affects both electrical properties and molecular cycling; VIP acts similarly in the mouse brain. In the SCN, VIP is expressed by approximately 10% of pacemakers, similar to the PDF situation in the fly brain. Like the VIP-receptor system in mouse, the Drosophila PDF receptor is broadly but heterogeneously expressed throughout the pacemaker network, with a significant display of autoreceptors (An et al., 2012; Im and Taghert, 2010; Shafer et al., 2008). Knockout mice that are deficient for VIP or for its receptor (VPAC2) display altered behavioral, cellular and molecular rhythms (Aton et al., 2005; Colwell et al., 2003; Harmar et al., 2002); a very similar profile of rhythmic phenotypes is observed in Pdf and Pdf-R deficient flies (Hyun et al., 2005; Lear et al., 2005; Mertens et al., 2005; Renn et al., 1999a). It is interesting therefore to consider that neither PDF and VIP - nor the PDF-R and VIP receptors - are strict sequence orthologs. It is probably significant however that PDF-R and VPAC2 are related, in that both are members of the Family B1 GPCR group (Harmar, 2001), PDF-R is more related to the receptors for CGRP and calcitonin (Hewes and Taghert, 2001b; Hyun et al., 2005; Lear et al., 2005; Mertens et al., 2005). Hence, in highly divergent animals, the modulation of 24-hour activity cycles generated by circadian neural circuits features a prominent role for Family B1 GPCR signaling pathways. These results suggest a lesson when considering possible conservation of modulatory systems: evolution may sometimes select functionally-related, though not precisely orthologous, signaling mechanisms.

GENERAL PRINCIPLES GLEANED FROM STUDY OF INVERTEBRATE PEPTIDES

Neuropeptide modulation is typically exerted extrinsically to motor networks

Neuropeptides frequently modulate motor outputs generated by central pattern generators—such as the switching of the crab STG network between distinct gastric mill rhythms—or initiate complex fixed action patterns—such as ecdysis and eclosion. This suggests a general principle that neuropeptides act from outside motor networks to modulate their intrinsic functional properties or outputs.

Neuropeptides control the gain of sensory inputs

Combined genetic and physiological studies have shown both in Drosophila and C. elegans that neuropeptides control the gain—and hence behavioral salience—of various sensory inputs. This can be a result of direct activation of peptide receptors in sensory neurons themselves—as seen in both fly and worm olfactory neurons—but also in interneurons that relay sensory information for further processing—such as the hub interneuron of the worm.

Neuropeptides operate in feedforward circuits

There are several examples of neuropeptides that operate in homotypic feedforward circuits, where a particular peptide acts not only at downstream effector sites, but also to increase secretion of that same peptide by intervening neurons to then act downstream. This is seen in the fly circadian control network, where PDF secreted by lLNv neurons acts both directly on dorsal clock neurons as well as to increase PDF secretion by sLNv neurons to also act on dorsal clock neurons. Similarly, the ATRP peptide acts in Aplysia both on the STG pattern generator to accelerate ingestion, but also is released by motor neurons onto muscle fibers to encourage that same end.

Neuropeptide release is controlled by peptidergic feedback loops

Neuropeptides can participate in feedback loops where one neuron secretes a peptide that acts on its receptor in a second neuron to increase secretion of a different peptide by that second neuron that then acts on its receptor in the first neuron. In some cases, such a positive feedback loop implements a bistable switch to ensure that once a behavioral sequence is initiated, it proceeds inexorably to its conclusion, such as in the EH/ETH positive feedback loop controlling insect ecdysis (Figure 4A). In other cases, feedback loops modulate sensory inputs, such as in the worm sensory feedback loop wherein a peptide secreted by a sensory neuron acts in an interneuron which, in turn, secretes a peptide that acts on the sensory neuron (Figure 4B).

Figure 4. Control of neuropeptide release by peptidergic feedback loops.

A. Positive peptidergic feedback initiates the all-or-nothing insect eclosion behavior. Corazonin (COR) and Diuretic Hormone (DH) induce modest ETH secretion by the peripheral Inka cells. This ETH propagates through the hemolymph and acts on central EH neurons to induce modest EH secretion, which in turns propagates through the hemolymph and-closing the feedback loop-acts on the Inka cells to induce massive ETH secretion. These high levels of ETH then act on the central EH neurons to induce massive EH secretion. These high levels of EH finally act on the eclosion CPG to initiate eclosion.

B. Peptidergic feedback modulates sensory responses in C. elegans. The AWC sensory neuron releases NLP-1 neuropeptide in response to sensory stimuli. NLP-1 acts on the AIA interneuron to modulate INS-1 peptide secretion, which-closing the feedback loop-acts on the AWC sensory neuron to modulate its responsiveness to sensory stimuli.

OUTLOOK FOR FUTURE STUDIES

Increasingly sophisticated genetic methods promise the availability of tools (genetic toolkits) to systematically categorize neuropeptide and neuropeptide receptor content for individual cell types. It is now possible to assay the functional contributions of such modulatory signaling: the contributions of single peptides in cells that secrete multiple peptides versus the aggregate signaling from that cell type. This is particularly important in C. elegans, in which the entire nervous system contains only approximately 300 neurons, but expressed over one hundred distinct neuropeptides. Such technical facility will increase even more the value of genetic model organisms flies and worms for studies of the neural basis of behavior. But it is important to recognize that C. elegans and Drosophila melanogaster are highly derived species whose genomic signatures and behavioral profiles are highly specific to their evolutionary history. They thus offer views of genetic machinery and behavioral repertoires that must be interpreted in light of species-specific evolution in the application of lessons learned there to mammals.

This brings us to our final point—that the regulation of behavior by neuropeptides in invertebrates relies on three types of studies—but only two of these are currently given the attention they deserve. The first type involves, genetics, genomics, and endocrinology. What are neuropeptide sequences, and which are their receptors? Where are these proteins expressed and how do they signal? How unique or redundant are their actions? The second type involves neurophysiology, functional imaging, and neuroanatomy. When are neuropeptide signals sent and how quickly do they act? At what system levels do they work? What is their relation to sensory inputs, to CPGs, and to motor outputs? The third type is behavioral biology. What are the details of animal behavior that are modulated by neuropeptides and what are the behavioral consequences of such modulation.

Almost all current behavioral paradigms rely on placing animals in intentionally impoverished environments so as to isolate a specific feature of a single behavior for experimental isolation and manipulation. However, this divorces neuropeptide modulation from what we consider to be one of its most important roles: the integration of internal physiological state variables—such as metabolic state, circadian time, reproductive drive, etc—with external sensory cues—such as food or mate availability, light and temperature, suitability for egg deposition, etc—in the choice of what behaviors to perform at any given moment in time. This requires the development and use of richer, more sophisticated naturalistic behavioral paradigms that will permit this behavioral choice function of neuropeptide modulation to be directly experimentally addressed.