Authors: Jacqueline Y. Lo (1Buck Institute for Research on Aging; 8001 Redwood Blvd, Novato, CA 94945, USA.), Katelyn M. Adam (1Buck Institute for Research on Aging; 8001 Redwood Blvd, Novato, CA 94945, USA.; 2Leonard Davis School of Gerontology, University of Southern California; 3715 McClintock Ave, Los Angeles, CA 90089, USA.), Jennifer L. Garrison (1Buck Institute for Research on Aging; 8001 Redwood Blvd, Novato, CA 94945, USA.; 2Leonard Davis School of Gerontology, University of Southern California; 3715 McClintock Ave, Los Angeles, CA 90089, USA.; th; 4Center for Healthy Aging in Women; 8001 Redwood Blvd, Novato, CA 94945, USA.; 5Productive Health Global Consortium; 8001 Redwood Blvd, Novato, CA 94945, USA.; 6Lead Contact)
Categories: Article, neuropeptide, neuropeptidase, healthspan, aging, behavior, C. elegans, neural circuit
Source: Current biology : CB
Authors: Jacqueline Y. Lo, Katelyn M. Adam, Jennifer L. Garrison
Neural communication requires both fast-acting neurotransmitters and neuromodulators that function on slower timescales to communicate. Endogenous bioactive peptides, often called “neuropeptides,” comprise the largest and most diverse class of neuromodulators that mediate crosstalk between the brain and peripheral tissues to regulate physiology and behaviors conserved across the animal kingdom. Neuropeptide signaling can be terminated through receptor binding and internalization or degradation by extracellular enzymes called neuropeptidases. Inactivation by neuropeptidases can shape the dynamics of signaling in vivo by specifying both the duration of signaling and the anatomic path neuropeptides can travel before they are degraded. For most neuropeptides, the identity of the relevant inactivating peptidase(s) is unknown. Here, we established a screening platform in C. elegans utilizing mass spectrometry-based peptidomics to discover neuropeptidases and simultaneously profile the in vivo specificity of these enzymes against each of more than 250 endogenous peptides. Here we identified NEP-2, a worm ortholog of the mammalian peptidase neprilysin-2, and demonstrate that it regulates specific neuropeptides, including those in the egg-laying circuit. We found that NEP-2 is required in muscle cells to regulate signals from neurons to modulate both behavior and health in the reproductive system. Taken together, our results demonstrate that peptidases, which are an important node of regulation in neuropeptide signaling, affect the dynamics of signaling to impact behavior, physiology, and aging.
Neuropeptides are endogenous bioactive signaling molecules that function across considerable spatial and temporal scales as transmitters, modulators, and hormones.^1–3^ They can signal locally or between distant, unconnected cells over long timescales and facilitate distinct outputs from the same neuronal circuit in a context-dependent manner.^4,5^ This complexity allows flexible, reversible rewiring of brain circuits and coordination between tissues, but also makes functional analysis challenging.^6^ Mature neuropeptides are processed from larger proprotein precursors and are sometimes heavily post-translationally modified, often by enzymes localized within the vesicles from which the peptides are eventually released. Once secreted, they bind to specific G-protein coupled receptors (GPCRs) on target cells, where they activate intracellular signaling pathways to mediate diverse aspects of physiology and behavior, and their dysregulation has been implicated in the pathogenesis of many human diseases, including neurodegenerative and metabolic disorders.^7–9^ Considerable effort has gone into deorphanizing neuropeptide receptors and pairing them with their peptide ligands in vitro.^10^ However, information about the accompanying dynamics and signal termination of many peptidergic signaling systems is limited.
Unlike small molecule neurotransmitters, which are inactivated through either enzymatic degradation^11,12^ or cellular reuptake mechanisms that were elucidated in molecular detail decades ago^13^, neuropeptides are inactivated by receptor binding and internalization or via extracellular peptidases that degrade neuropeptides en-route to, or at, their sites of action.^14–17^ However, information about the identity of these peptidases, their target specificity, and their role as regulators of physiology and behavior is limited.^18–20^ Foundational studies have examined specific peptide-peptidase pairs using in vitro biochemical and electrophysiological experiments,^22–25^ but have not profiled peptide substrates for a given peptidase at a global in vivo organismal scale. This knowledge gap is a critical barrier to understanding the dynamics of neuropeptide signaling in vivo because peptidases represent a key regulatory node that controls the duration of signaling and the anatomic path by which neuropeptides can travel before being inactivated. In some instances, extracellular peptidase-mediated cleavage can also activate or amplify the actions of neuropeptides.^22–25^
The roundworm Caenorhabditis elegans makes rich use of neuropeptide signaling to regulate its behavior and physiology and shares with mammals a similar number of neuropeptide genes, many of which are highly conserved, and a conserved set of enzymatic pathways that regulate neuropeptide biogenesis, processing, transport, and exocytosis.^26^ Worm neuropeptides are broadly categorized into three gene families based on loose sequence flp genes (FMRFamide-like peptides) encode peptides containing the C-terminal sequence “FMRFamide,”^27^ ins genes (insulin peptide-like) encode insulin-like peptides^28^, and nlp genes (neuropeptide-like proteins) encode peptides distinct from the other two families^29^. The C. elegans genome also contains hundreds of putative peptidases, including more than 20 genes with homology to the mammalian neprilysin family.^30^ Mammalian neprilysins are type-II membrane bound proteins with active sites in the extracellular domain that have been implicated in the degradation of enkephalin and tachykinin peptides.^19^ However, most experiments on peptidase inactivation of neuropeptide signaling have been performed in vitro, with few in vivo studies that examine mechanisms by which any neuropeptidase regulates neural circuits that control behavior.
C. elegans provides a tractable system for dissecting the role of peptidergic signaling in behavior. The egg-laying circuit is regulated by both neuropeptidergic inputs and small molecule transmitters and modulators. The command neurons of this circuit, named HSN (Hermaphrodite-Specific Neuron), are responsible for initiating the active state of egg-laying, where eggs are released from the animal in short-bursts.^31,32^ HSNs utilize both serotonin and neuropeptides, and synapse directly onto both muscle cells and motor neurons to promote egg-laying behavior. Previous studies examining the neuronal inputs for egg-laying have long focused on serotonin and its multiple receptors, from identification of its release from HSNs^32,33^ to receptor binding on neurons and muscle cells in the circuit.35 More recently, specific neuropeptides, namely the peptides encoded by nlp-3 in HSNs, have also been shown to contribute to egg-release.^35,36^ The nlp-3 gene encodes a proprotein precursor that is enzymatically processed into five different bioactive peptides. However, it is not known which of the five peptides are necessary for egg-laying. This interplay of small molecule neuromodulator and neuropeptide signaling in a single circuit with a defined, measurable behavior enables close examination of how the three essential components of neuropeptide signaling – ligand, receptor, and peptidase – can influence circuit function. Here we identify a neuropeptidase in this circuit and explore its function to further understand how egg-laying behavior can be spatially and temporally regulated.
Defects in the egg-laying circuit can lead to egg-retention, which induces a phenomenon called matricidal hatching. The timing of embryo development in utero is precisely coordinated with egg-laying. Matricidal hatching occurs when embryos continue to develop and hatch inside the mother instead of being laid outside. This results in a phenotype called “bagging” (bag-of-worms^38^, Video S1). There are three primary reasons why C. elegans experience (1) genetic mutations that impair the egg-laying apparatus^39^, (2) environmental cues that indicate food scarcity or starvation^40^, and (3) age-related changes in the integrity of the egg-laying system^41^. While the first two causes are severe insults to the system, age-related decline is unrelated to use-dependent damage and appears to occur because of the natural course of aging. Intriguingly, this aspect of aging in the reproductive system can be uncoupled from aging of the organism. In several C. elegans models of longevity, including mutants that impact the insulin/IGF-1 signaling (IIS) pathway, there is an extension of both lifespan and reproductive span, but also a marked increase in the rate of matricidal hatching, implying that there are defects in egg-laying system function.^42,43^ For many C. elegans assays, including those that examine lifespan and reproductive biology, animals that experience matricide are censored from analysis. Thus, the impact of impaired reproductive health on lifespan and healthspan is not well explored.
Here, we establish a candidate-based screening approach in C. elegans to both identify neuropeptidases and link them to their endogenous neuropeptide target(s). Our experiments identified NEP-2, a worm ortholog of the mammalian peptidase neprilysin-2, that regulates specific neuropeptides, including those in the egg-laying circuit. We found NEP-2 is required in muscle cells to regulate signals from neurons to modulate both behavior and the health of the reproductive system. Taken together, our results demonstrate how a neuropeptidase can modulate the neuropeptide signaling component of the egg-laying circuit by altering the dynamics of signaling to impact behavior, physiology, and aging.
We established a screening platform in C. elegans using mass spectrometry-based peptidomics to identify neuropeptidases and simultaneously profile their in vivo specificity for more than 250 endogenous peptides. Our approach takes advantage of the fact that C. elegans expresses a complement of neuropeptides comparable to mammals that can be isolated and profiled by mass spectrometry^44^ and is genetically tractable. To screen for neuropeptidases, we compared neuropeptide levels in worm strains with candidate neuropeptidase genes knocked down (via RNAi) or knocked out (via genetic null mutation) with wild-type animals. We predicted that in a reduced or null neuropeptidase background, levels of mature neuropeptides that are the targets for that peptidase would increase. One of the hits from our screen, NEP-2, is a type-II membrane-bound ectoenzyme that is highly conserved and most similar to mammalian neprilysin-2 (NEP2/MMEL1, Figure S1A), with a small intracellular domain and a larger active-site containing extracellular domain (Figure 1A, Figure S1A–B). In C. elegans, there are 26 putative members of the neprilysin gene family and aside from very few exceptions^45,46^, most of these genes have not been studied.
We performed label-free quantitative mass spectrometry to compare neuropeptide levels in animals lacking nep-2 with wild-type animals (Figure 1B). To do this, we biochemically isolated endogenous neuropeptides from both wild-type animals and nep-2(ok2846) mutants and profiled them using tandem mass spectrometry. We then compared the relative abundance of each identified neuropeptide species and found levels of several neuropeptides increased, as predicted (Figure S1C). Neuropeptides whose relative abundance changed include members of both the nlp and flp gene families. However, consistent with previously published data^44^, we were unable to detect any peptides from the ins gene family. Interestingly, we observed increased levels of NLP-3 neuropeptides (Figure 1D–H) in animals lacking NEP-2. Mature neuropeptides are derived from a larger proprotein precursor via successive enzymatic processing steps that often occur inside the vesicle from which they are eventually released (Figure 1C). The NLP-3 proprotein precursor encodes five distinct peptides (Figure 1C), and all five NLP-3 neuropeptides are increased in nep-2(lof) mutants (Figure 1D–H). NLP-3 neuropeptides are known to promote egg-laying behavior. Other identified neuropeptides in this circuit include the inhibitory neuropeptides NLP-7 and FLP-11^47^; of the eight peptides detected from these genes, only one peptide from FLP-11 may also be regulated by NEP-2 (Figure S1D). NLP-3 neuropeptides are released from the command neuron, HSN, which also expresses nlp-8, nlp-15, flp-5, and flp-19.^48^ NEP-2 has no impact on the levels of other neuropeptides expressed in HSN (Figure S1E). These data suggest that NEP-2 may regulate the egg-laying circuit by degrading one or more NLP-3 neuropeptides. Based on the results of label-free mass spectrometry relative quantification, we identified NEP-2 as a neuropeptidase in C. elegans that regulates multiple endogenous neuropeptide targets, including all five NLP-3 neuropeptides.
The egg-laying circuit in C. elegans utilizes both neuropeptides and small molecule HSNs release both serotonin^33^ and NLP-3 neuropeptides^36^ to initiate egg-laying behavior (Figure 2A, left) via signaling to both motor neurons (VC neurons) and vulval muscle cells (vm). There are two types of vulval muscle cells, four vm1 cells and four vm2 cells, as well as the eight uterine muscle cells (um) which communicate via gap junctions. Vm2 cells can also receive input from VC motor neurons via the small molecule transmitter acetylcholine.^49^ Together, these muscle cells coordinate egg-release from the vulva, which is sensed by the uv1 endocrine gland cells that in turn release tyramine^50^ and express two neuropeptide genes, nlp-7 and flp-11^47^, that provide feedback to HSNs to turn-off the circuit. HSNs receive additional inhibitory signals from the VC neurons through acetylcholine^51^ and neuropeptides, FLP-10 and FLP-17^52^, which bind to the GPCR EGL-6 on HSN to downregulate its activity. Loss of nlp-3 leads to egg-retention (Egl, or egg-laying-defective, phenotype).^35^ In the nep-2(lof) animals, we observed elevated levels of NLP-3 peptides. Based on the circuit architecture, we predicted that elevated NLP-3 would lead to an increase in egg-laying behavior (Figure 2A, right).
To explore the role of NEP-2 in regulating this circuit, we asked whether elevated NLP-3 neuropeptide levels in animals lacking nep-2 impacted egg-laying behavior. We observed significantly fewer eggs in the uterus of nep-2(lof), (Figure 2B, Figure S2A), indicating a hyperactive egg-laying phenotype that is consistent with the established role of NLP-3 neuropeptides in egg-laying. Additionally, eggs in the uterus of nep-2(lof) animals consist of primarily early-stage embryos (Figure 2C, Figure S2B). In C. elegans, eggs are laid at specific embryo stages, namely after the 8+ cell stage. In nep-2(lof) animals, very early-staged embryos are laid prematurely (Figure 2D), further indicating hyperactive egg-laying behavior. We observed the same phenotype in mutant animals lacking the NEP-2 extracellular domain (Figure 1A, Figure 2B–D), as well as animals missing the entire gene, nep-2(pe379) (Figure 1A, Figure S2C–E), supporting NEP-2’s role as a neuropeptidase that degrades NLP-3 neuropeptides.
To examine the causal effect of NEP-2 on behavior, we performed gain- and loss-of-function experiments. When we restored nep-2 under regulation of its endogenous promoter, this rescued egg-laying deficiencies (Figure 2E–G, Figure S2F–G). Furthermore, when we overexpress nep-2, we can drive the phenotype in the opposite direction, resulting in egg-retention (Egl) (Figure 2H–J, Figure S2H–I). Animals that overexpress nep-2 retain significantly more eggs than wild-type, and both the eggs that are retained and the eggs that are laid outside of the animal are at a much later developmental stage than expected. In wild-type animals, the comma, plum, and pretzel stages of embryo development occur outside of the uterus. However, in nep-2 overexpressing animals, the eggs continue to develop through these stages inside the uterus before being laid.
One predicted phenotype of a hyperactive egg-laying mutant is increased output of eggs laid. Surprisingly, we did not observe increased frequency of egg-laying in animals lacking nep-2, and instead found that the egg-laying rate trends in the opposite direction (Figure S2J). One possible explanation for this observation is that there are sperm defects in these animals that interfere with the rate of ovulation.^53^ A key prediction of this hypothesis is that brood size from self-fertilization would be smaller than from mated fertilization. C. elegans has two sexes, hermaphrodites and males; hermaphrodites self-reproduce through internal fertilization with their own sperm and can also mate with males to produce cross-progeny.^54^ Indeed, the number of self-progeny from animals lacking nep-2 is smaller than wild-type (Figure S2K); this defect is restored when hermaphrodites are mated and provided with male sperm (Figure S2L), pointing to a potential defect in self-sperm. The hyperactive egg-laying behavior of animals lacking nep-2 is unaffected by mating (Figure S2M). Given that the egg-laying rate was unchanged in these animals, we hypothesized that there might be more eggs released per egg-laying event. We found that while wild-type animals will release one egg per lay over 90% of the time, animals lacking nep-2 have a greater frequency of releasing more than one egg (about 20% of the events) (Figure 2K). Taken together, NEP-2 plays a critical role in regulating egg-laying behavior, and the loss of nep-2 results in hyperactive egg-laying.
We next sought to determine if the hyperactive egg-laying behavior of nep-2(lof) is due to elevated levels of NLP-3 neuropeptides in the animal. We used epistasis analysis to compare double-mutant animals lacking both nep-2 and nlp-3 to each single mutant and observed that nep-2; nlp-3 behaved like nlp-3(lof) (Figure 3A–C, Figure S3A–B). This suggests that NEP-2 degrades at least one, if not multiple, NLP-3 neuropeptides to regulate egg-laying behavior. We also observed, in a separate experiment (Figure 2H–J, Figure S2H–I), that animals overexpressing nep-2 have a more severe Egl phenotype (eggs retained in uterus, stage of eggs laid by the animal). This suggests that NLP-3 peptides may not be the only targets regulated by NEP-2 in the egg-laying circuit.
To establish where NEP-2 is acting in the egg-laying circuit to regulate NLP-3 neuropeptide levels, we reporter-tagged NEP-2 at its endogenous locus via gene editing. Consistent with previous studies^46^, we observed NEP-2 in muscles and a subset of head and tail neurons (Figure S4A). In one mammalian NEP-2 isoform, the extracellular domain can be cleaved and shed into the extracellular space to act on target cells distant from the site of expression.^55–57^ Although the cleavage sequence of this mammalian isoform is not contained in the C. elegans protein sequence (Figure S1A), we wondered whether the worm NEP-2 protein might also be cleaved from the cell surface under some circumstances. In C. elegans, proteins that are released into the pseudocoelemic space are taken up by large scavenger cells called coelomocytes.^58^ Using a NEP-2 reporter tagged on the C-terminus (extracellular, Figure S1B), we observed accumulation of this protein in coelomocytes (Figure S4B). Accumulation in coelomocytes was not observed in the N-terminally tagged reporter strain (where the cleaved extracellular portion of NEP-2 would not contain a fluorescent label), suggesting that the fluorescent signal is not due to nep-2 gene expression in coelomocytes (Figure 4B). These data point to the possibility that NEP-2 is also cleaved and secreted in C. elegans, but perhaps via a different protease cleavage recognition site.
We found that NEP-2 is expressed in both the uterine and vulval muscles, as well as in body wall muscle (Figure 4A). This NEP-2 reporter expression pattern was consistent with a published C. elegans sc-RNA-seq cell atlas^59^ (Figure 4B). Notably, we did not observe any NEP-2 expression in the HSN neurons, where nlp-3 is expressed in the egg-laying circuit. Intriguingly, when we profiled nep-2 transcript expression across the life of the animal, we found that expression peaked during the latter half of the reproductive span (Figure 4C). This suggests manipulation of nep-2 can modulate occurrence of matricide, since this time point coincides with a higher probability that progeny will hatch internally, which can result in the death of the mother.
We next leveraged RNAi in C. elegans to determine where nep-2 expression is required for its role in egg-laying behavior. Feeding RNAi in C. elegans functions in all tissues except neurons.^60^ We performed feeding RNAi against nep-2 in wild-type animals and observed hyperactive egg-laying similar to the genetic null nep-2 mutants (Figure 4D, Figure S4C). However, nep-2 RNAi in a background where only neurons are sensitized to RNAi61 displayed normal egg-laying behavior (Figure 4E, Figure S4D). A recent study reported that the strain long-used for neuron-specific RNAi in C. elegans showed RNAi responses in non-neuronal tissue and generated a new, more robust neuron-specific RNAi strain.^62^ We repeated neuronal RNAi-mediated nep-2 knockdown in this strain and observed the same result. (Figure S4E). While we could not confirm nep-2 knockdown via qPCR in this background, we note that nep-2 is only expressed in a small subset of neurons (Figure 4B) that may obscure our ability to measure knockdown when assessed at the organismal-level. When nep-2 expression was restored selectively in the muscle, including the uterine and vulva muscle cells (Figure S4F), egg-laying behavior was rescued (Figure 4F, Figure S4G–J). Collectively, these data point to NEP-2 functioning in the vulval and/or uterine muscle to modulate the egg-laying circuit, where nep-2 is highly expressed (Figure 4G, data from Roux et. al.^59^). When we examined the frequency of egg-release, we found that NEP-2 was required specifically in the muscle to restrict the number of eggs released per egg-lay event (Figure 4H, Figure S4K). Interestingly, while both the developmental stage of eggs laid and the distribution of egg stage inside the uterus were rescued by nep-2 restoration in the muscle, the number of eggs inside the uterus was only partially restored (Figure S4H). This is consistent with the hypothesis that NEP-2 may also have a separate role in sperm function.
The above results describe a neuropeptidase node in the peptidergic arm of the egg-laying circuit. We next asked which neuropeptide receptor(s) signals in this circuit. One way to add a layer of regulation in a peptidergic circuit would be to selectively express the relevant peptidase in specific cells that also express the receptor. The GPCR NPR-36 is known to interact with NLP-3-1 and NLP-3-2 peptides to regulate egg-laying (co-submitted manuscript, Michael Koelle). RNAi-mediated knockdown of npr-36 in nep-2(lof) animals rescued hyperactive egg-laying behavior (Figure 4I–K, Figure S4L). Additionally, previous studies59 showed that npr-36 is also expressed in the uterine and vulval muscles (Figure 4L). Taken together, these data suggest that NEP-2 acts in the vulva and/or uterine muscle to regulate egg-laying behavior (Figure 4M). We note that while nep-2 expression in the muscle cells is required, NEP-2 may be either membrane-bound or cleaved and shed from the membrane to regulate NLP-3 signaling.
Improper egg-retention in C. elegans can lead to a phenomenon called matricide – when embryos hatch inside the uterus and progeny eat their way out, resulting in maternal death (Video S1).^38^ Intriguingly, the long-lived insulin receptor mutant (daf-2) experiences extraordinarily high rates of matricide.^42^ Several common C. elegans assays require censoring these animals at the day of matricide and thus, are not considered in the final data analysis. We wondered whether enhancing egg-laying behavior by removing nep-2 in the daf-2(lof) background could alleviate the high matricide rate. Consistent with previous studies, we measured the rates of matricide and vulval defects in daf-2(lof) at ~ 40%, and this phenotype is completely rescued in the double mutant (Figure 5A, Figure S5A–B). We then measured nep-2 expression in daf-2(lof) animals and found that nep-2 levels are elevated in daf-2(lof) compared to wild-type at both day 2 and day 5 of adulthood (Figure S5C), when reproduction peaks in daf-2(lof) animals, suggesting dysregulation in NLP-3 signaling that could influence egg-laying. Consistent with this hypothesis, the double mutant also rescued the tendency towards late-stage embryos observed in daf-2(lof) mothers (Figure 5B–C, Figure S5D, F). Additionally, in animals aged to the mid/late reproductive period, when this phenotype is more extreme in daf-2(lof), the nep-2; daf-2 double mutant is still capable of rescuing the egg-retention defect (Figure 5D–E, Figure S5E).
We next investigated whether there were any benefits to reducing matricide in the insulin receptor mutant animals. Surprisingly, double mutants lacking nep-2 and daf-2 lived significantly longer than the already long-lived daf-2(lof) animals (Figure 5F). Standard protocols for C. elegans lifespan analysis censor animals that experience matricide (bagged animals) from the lifespan curve. When bagged animals are included in daf-2(lof) lifespan experiment analysis, the maximal lifespan does not change, but the median lifespan is reduced.^63^ Since the nep-2;daf-2 double mutant lacks this sub-population of bagged animals, this could be one reason we observed a further extension of lifespan. We confirmed that the increased lifespan in the double mutant is a direct result of improved egg-laying in daf-2 by measuring lifespan in the presence of a drug that inhibits larval production (5'-fluorodeoxyuridine, FUDR). Without larval production, animals will never experience matricide. This paradigm completely abolished the double mutant lifespan extension (Figure 5G). Finally, we tested whether the extended lifespan in worms lacking both nep-2 and daf-2 is also accompanied by improved healthspan. We challenged animals with chronic paraquat exposure, an oxidative stress paradigm, and found that double mutant animals perform better than daf-2(lof) (Figure 5H). These animals also are more active at older ages (day 10, data not shown). Although longevity in daf-2(lof) mutants is driven in part by elevated DAF-16/FoxO activity,^64–66^ we found transcriptional targets of DAF-16 are not further elevated and instead are downregulated in both nep-2(lof) and nep-2; daf-2 (Figure S5G–H). This suggests that the increased lifespan of nep-2; daf-2 occurs through a DAF-16-independent mechanism.
An alternate paradigm that also causes matricidal hatching in C. elegans is mating hermaphrodites to male animals.^67,68^ We wondered whether loss of nep-2 can also improve matricidal health in this context. Consistent with previous studies, we observed that mated wild-type mothers experience high matricide rates and early death, and that this is significantly delayed in animals lacking nep-2 (Figure 5I–J). These data suggest that NEP-2 can improve lifespan and healthspan in animals experiencing high rates of matricide by reducing egg-retention.
Neuropeptides are unique signaling molecules that coordinate physiology with behavior, yet the physiological and behavioral consequences of peptidase-mediated termination of neuropeptide signaling have not been elucidated in most systems,^69–72^ particularly when compared to classical neurotransmitter systems. Neuropeptidases, by inactivating neuropeptides, ensure the spatial and temporal fidelity of signaling and represent a key under-explored regulatory node in peptidergic circuits.^73^ We found that NEP-2, a mammalian neprilysin homolog, is a neuropeptidase that regulates several neuropeptide targets in C. elegans, including NLP-3 peptides. While studies on mammalian NEP2 have been largely focused on identification, localization, and characterization of substrate targets in vitro,^55–57,74,75,76^ here we delineate, for the first time, an extensive set of endogenous targets for an identified peptidase, NEP-2, and interrogate how NEP-2 regulates an in vivo behavioral circuit.
C. elegans egg-laying behavior provides a compelling model to explore the intricate interplay between neuropeptide signaling and physiologic outcomes. Governed by a network of neurons, endocrine, and muscle cells, this behavior relies on the coordinated signaling of both neurotransmitters and neuropeptides. Our results uncouple the physical process of egg-release from ovulation rate. We found that the regulation of NLP-3 neuropeptides by NEP-2 specifically regulates egg-release by the uterine and vulval muscles. This is likely to be driven by NLP-3-1 and NLP-3-2, the ligands which bind the GPCR receptor NPR-36 in the circuit (co-submitted manuscript, Michael Koelle). Rescuing nep-2 in muscle does not fully restore the number of eggs retained in the uterus, which suggests that ovulation rate may be altered in animals lacking nep-2. In C. elegans hermaphrodites, sperm in the germline secrete protein signals (MSP, major sperm proteins) that promote ovulation. We observed potential sperm defects in nep-2(lof) that could result in reduced ovulation, possibly through an interaction between NEP-2 and the other NLP-3 neuropeptides or one of the other NEP-2 neuropeptides targets identified in the mass spectrometry screen.
NEP-2 has been previously implicated as a regulator of olfactory plasticity,^46^ but this is the first study to delineate NEP-2’s in vivo targets and explore its role in regulating spatial and temporal aspects of peptidergic signaling. In addition to NLP-3 peptides, we found that NEP-2 targets many distinct neuropeptides that do not share any obvious similarities at the level of amino acid sequence, expression patterns, or circuits (Figure S1C). Given that most neuropeptidase gene families recognize specific regions of a peptide and, at that region, target amino acid pairings rather than defined sequences, specificity for neuropeptide targets may be encoded spatially through co-localization. While we have identified an extensive set of changes in the neuropeptidome of nep-2(lof) animals, it is possible that some of these changes are due to indirect effects of NEP-2. We observed that levels of some neuropeptides went down significantly in nep-2 null mutants, suggesting more complex regulation and through-network effects via other NEP-2 peptide targets. Follow-up experiments to identify specific cleavage sites and empirically determine degradation products for each neuropeptide target identified in this mass spectrometry screen would add to our understanding of NEP-2 substrate recognition.
One driver of matricide in C. elegans is the age-related decline of the egg-laying system, which is uncoupled from overall lifespan. The long-lived insulin receptor mutant, daf-2,^77,78^ has a reproductive span that is three times as long as wild-type animals. But these benefits are paired with the pleiotropic effect of increased matricide. We were able to eliminate the age-related matricide associated with daf-2(lof) by also removing nep-2. This suggests that the health of the reproductive apparatus contributes significantly to overall organismal health. Since these tissues are also among the first to show signs of aging, future studies will focus on addressing axes of inter-tissue communication that regulate overall lifespan.
Mammalian NEP2/MMEL1 is highly expressed in reproductive tissues, including the testes, ovary, and the female reproductive tract.^78^ However, NEP2/MMEL1 activity in these tissues has not been investigated in detail. To corroborate existing expression data, we analyzed additional published RNA-sequencing datasets to evaluate specifically where and when NEP2/MMEL1 could be acting in murine and human uteri. Based on our data in C. elegans, we predicted that NEP2/MMEL1 would be most highly expressed in the myometrial layer of the uterus, either at the onset of menses in humans or at the onset of labor in either model. Surprisingly, we found that NEP2/MMEL1 is primarily expressed in human endometrium during the secretory phase of the menstrual cycle.^79^ Examination of the protein localization via Human Protein Atlas indicates that NEP2/MMEL1 expression is specific to the epithelial cells that line the uterus.^80^ Additionally, NEP2/MMEL1 was not detected in non-pregnant mouse uteri,^81^ but rather, potentially downregulated at the onset of labor in a murine model of preterm labor,^82^ mirroring the role of NEP-2 we observed in maintaining uterine quiescence in C. elegans. The expression and localization of NEP2/MMEL1 suggests that it plays a role in uterine physiology; whether it mediates smooth muscle contraction in mammals is an interesting open question. A recent study identified a juxtacrine signaling pathway between epithelial and myometrial cells in the uterus that promotes labor onset by modulating prostaglandin production.^83^ One possible mechanism by which NEP2/MMEL1 could modulate contractions is by regulating neuropeptides essential to prostaglandin production in the epithelium during menstruation or pregnancy. Pregnancy-related complications increase with age,^84,85^ often due to irregular uterine muscle activity. Inappropriately timed contractions can result in both pre- or post-term birth. Some mechanisms governing the regulation of uterine smooth muscle function have been identified,^86^ and neuropeptide signaling has been shown to modulate uterine contractions.^87,88^ Delineating the temporal and spatial expression of NEP2/MMEL1 in the mammalian uterus may point to specific nodes of regulation that are required for uterine homeostasis. Additionally, pharmacologic modulation of neuropeptidases can alter the levels of their neuropeptide substrates, implicating these enzymes as important targets and control points for pharmacological intervention.
Given that there are hundreds of neuropeptides of varying size, sequence, and function in both worm and mammalian genomes, there are also peptidases of varying specificity, regulation, and localization. Future studies systematically identifying neuropeptidases responsible for inactivating specific neuropeptides within particular brain regions or circuits will offer new insights into how peptidergic signaling dynamics are controlled in vivo. Understanding the regulatory mechanisms and neuropeptide substrates of these peptidases could also yield novel pharmacological targets for modulating peptidergic signaling across tissues. Indeed, neuropeptidases represent particularly attractive drug targets, since they may degrade multiple peptides that share specific features (e.g. sequence, function, or location), that might function together in a circuit or pathway. Modulating peptidase function provides a powerful and unique strategy to potentially target them simultaneously.
Further information and requests for resources and reagents should be directed to and will be fulfilled by the lead contact, Jennifer L. Garrison (jgarrison@buckinstitute.org).
All plasmids and strains generated in this study will be deposited with Addgene and the CGC, respectively, and are also available from the lead contact upon request.
C. elegans were cultured using standard techniques.^98^ Worm strains were maintained at 20°C on NGM plates seeded with 300 μl of OP50 bacteria grown overnight without agitation at 37°C. Unless indicated otherwise, animals were assayed at day two of adulthood (four days post timed egg lay). For RNAi experiments, NGM plates were supplemented with 1 mM IPTG (Zymo cat. No. 11–230B) and 100 μg/ml carbenicillin and seeded with 300 μl of HT115 bacteria expressing the RNAi clone, grown overnight at 37°C shaking at 220 rpm. For FUDR experiments, NGM plates were supplemented with 40 μM FUDR (Sigma cat. no. F0503, dissolved in water). All strains were un-starved and maintained at stable temperature (20°C) for at least three generations before being used in any experiments. For worm populations used in label-free mass spectrometry experiments, strains were grown from L1 to adulthood in liquid culture, according to standard protocol,^99^ with OP50 in the liquid culture at OD600 1.68, at 20°C, shaking at 200 rpm. Worms were collected at day 2 of adulthood for mass spectrometry analysis. Ten biological replicates of 50,000 worms were used per strain. Strains used are listed in the key resources table. Animals were backcrossed 2–6x to wild-type N2. Double mutants were generated by standard genetic techniques. Some strains were provided by the CGC as indicated, which is funded by the NIH Office of Research Infrastructure Programs (P40 OD010440).
Biochemical isolation of endogenous neuropeptides from C. elegans was performed as previously described.^44^ Briefly, Day 2 adult hermaphrodite worms were collected from culture and flash-frozen in liquid nitrogen. During worm collection, animals were washed 3–5 times with M9 to remove bacteria, and all liquid was removed before flash-freezing and storing at - 80°C. Worm pellets were dounce homogenized as a solid frozen pellet in 10 ml of 9:1 methanol/water/acetic acid, followed by probe sonication (15 seconds on, 45 seconds off; 4 cycles total at 80% amplitude with 3 mm ultrasonic probe tip, Cole-Parmer cat. no. SI-04712-12) after transferring lysates to a 15 ml conical tube. Lysates were then centrifuged at max speed at 4°C (Eppendorf 5810R centrifuge), and the supernatant was saved. Methanol was removed using a speedvac (volume reduced from 10 ml to 1 ml). The 1 ml of sample was transferred to low-binding 1.5 ml tubes (Thermo Fisher cat. no. 3451) then centrifuged again at max speed at 4°C (Eppendorf 5430R centrifuge) and the supernatant was transferred to a new glass culture tube. The sample was then delipidated with equal volume hexanes (Sigma cat. no. 139386) in the glass culture tube, twice, recovering the aqueous phase each time and combining. The aqueous fraction containing neuropeptides was passed through a size-exclusion filter (GE Healthcare sephadex PD MidiTrap G-10 cat. no. 28918011) according to manufacturer’s protocol to remove molecules < 700 Da. Finally, samples were desalted using a C18 spin column (Thermo Fisher cat. no. 89873) according to manufacturer’s protocol and stored in the stated elution buffer at 4°C before mass spectrometry analysis. Low-binding 1.5 ml tubes were used throughout the protocol after lysis unless otherwise stated to mitigate loss of sample.
Neuropeptide samples were run on a Waters nanoACQUITY coupled online to a Thermo Scientific Q Exactive Plus mass spectrometer using an EASY-Spray analytical column, 15 cm x 75 μm ID, PepMap C18, 3 μm; Thermo Scientific ES800. The gradient used was as flow rate of 300 nL/min, 3–8% acetonitrile with 0.1% formic acid over 5 minutes, 8–20% acetonitrile with 0.1% formic acid over 50 minutes, 20–36% acetonitrile with 0.1% formic acid over 10 minutes. Mass spectrometer settings were as data-dependent (dynamic exclusion settings at 25 s) Top10 method (charge selection excluding unassigned, 1); HCD (Higher-energy collisional dissociation); full MS scan resolution of 70,000, with a maximum injection time of 100ms and scan range of 400 to 1600 m/z; MS/MS scan resolution of 17,500, with a maximum injection time of 120 ms.
LC-MS/MS data were analyzed using PEAKS Studio X and the Q module (Bioinformatics Solutions, Inc.). The data was searched against a database containing all C. elegans proteins (Wormbase), E. coli proteins (SwissProt), and common protein contaminants (CRAPome)^100^. Parent mass error tolerance was set at 20 ppm and fragment mass error tolerance was set at 0.03 Da. Enzyme was set at “none” and digest mode to “unspecific.” Variable modifications amidation (-0.98), enzymatic glycine removal leaving an amidated C-terminus (-58.01), half of a disulfide bridge (-1.01), oxidation on M (+15.99), phosphorylation on STY (+79.97), pyro-glu from E (-18.01), and pyro-glu from Q (-17.03). Label-free quantification analysis was performed using the Q module of the PEAKS software, using TIC intensity selected as normalization.
Single day 2 adult hermaphrodite animals were dissolved in 20% bleach in water to release the eggs in the uterus without damage. Images of the eggs were taken within 15 minutes of bleaching and the number of eggs per animal were counted and the stage of each egg was scored.
20–30 day 2 adult hermaphrodite animals were placed on a 3 cm seeded plate (50 μl of OP50, seeded the previous day) and allowed to lay eggs for 30 minutes. Animals were removed after the egg-laying period. Eggs on the plate were scored for the stage of embryo development. Scoring was performed for a 5 minute period after animals were removed.
Five day 2 adult hermaphrodite animals were placed on a 3 cm seeded plate (3 μl of OP50 seeded the previous day) and recorded for two hours with a dissecting scope connected to a Pixelink camera (PL-D721MU) using the Pixelink Capture software. Multiple videos were taken per strain. Egg-laying behavior was analyzed manually for frequency of egg-laying events and number of eggs released per event per animal.
For self-progeny counts, L4 animals were transferred to new plates every day (3 cm plates seeded with 50 μl of OP50), and the number of progeny was counted for each day and summed for total self-brood size. Animals were censored from the total self-brood size count if they bagged or left the plate and a full reproductive span could not be assessed. For mated progeny, L4 animals were mated with day 1 adult wild-type male animals (1 hermaphrodite with 3 males on a single plate) for 24 hours and the hermaphrodite was subsequently transferred daily (mating males were removed). The number of progeny was counted for each day (and assessed as being male or hermaphrodite) and summed for total mated-brood size. Animals that did not meet the threshold of 40% males in total progeny were censored, as they were deemed to be insufficiently mated. Additionally, mated animals were scored for both lifespan and matricide (bagging). In this case, since animals typically died with matricide, bagged animals were not censored in the lifespan analysis.
Animals were paralyzed with 0.167 mM levamisole for live imaging on 15-well multitest slides (VWR cat. no. IC096041505). Images of the uterus, as well as Figure S4C, used a Zeiss Axio Imager M2 microscope with a Zeiss Axiocam 503 monochrome camera. Figures 4A and S4A utilized a Zeiss Axio Examiner LSM980 microscope.
30 day 1 adult animals were transferred to a 6 cm seeded plate (300 μl of OP50) at the beginning of the assay, and four plates were set-up per strain. Animals were transferred every-other-day during the reproductive period to move them away from progeny, except if in the presence of FUDR (5'-fluorodeoxyuridine, Sigma cat. no. F0503, dissolved in water, added to NGM media at 40 μM). Animals were scored for dead or alive by gentle probing, and animals that bagged (matricidal hatching), had vulva deformities (burst), escaped (crawled off the plate), or were on contaminated plates were censored. OASIS 2 was used for analysis. See Table S2 for details and statistics.
Animals also were scored for both the presence of hatched larvae inside the mother’s body cavity (bagging) and/or deformities of the vulva (bursting) during each day of adulthood. The cumulative sum of total animals subjected to matricidal hatching and vulval defects was graphed.
Paraquat (methyl viologen dichloride hydrate, Sigma cat. no. 856177, dissolved in water) was added on top of 3 cm seeded plates (100 μl of OP50) to a final concentration of 10 mM, as calculated by volume of NGM in the plate. Day 1 adult animals were transferred to assay plates, and animals were scored daily for dead or alive. Animals that escaped (crawled off the plate) were censored.
Animals were collected at the specified ages by washing off the culture plates with M9 into 1.5 ml tubes and washing in the tubes three times with M9 to remove bacteria and progeny. Liquid was removed from the tubes, Trizol was added, and the worms were stored in - 80°C until RNA extraction. RNA was extracted with the Direct-zol RNA Miniprep kit (Zymo cat. no. R2050) according to the manufacturer’s protocol. cDNA was synthesized with the iScript cDNA Synthesis kit (Bio-Rad cat. no. 1708890) according to the manufacturer’s protocol. qPCR was performed using iTaq Universal SYBR Green Supermix (Bio-Rad cat. no. 1725120) using a Bio-Rad CFX96 Touch Real-Time PCR Detection System. Data was analyzed using the 2^–∆∆Ct^ method and normalized to act-1. Primers used are listed in Table S1.
Descriptions of statistical tests utilized and sample sizes for each experiment are indicated in the figure legends. Statistical analyses were performed with GraphPad Prism 10 software. Unpaired t-test was used for two strains with one variable, and one-way ANOVA with Tukey’s multiple comparison test was used when there were more than two strains with one variable. 2-way ANOVA was used with Šídák's multiple comparisons test when comparing two strains with two variables and with Tukey’s multiple comparisons test for more than two strains with two variables. Fisher’s exact test was used for categorical data comparing the stage of the eggs laid by different strains and the number of eggs released per egg-lay. Log-rank (Mantel-Cox) test was used for analysis of lifespan curves. Data are presented as mean ± s.e.m. In all figures, ns indicates not significant, and p-values indicated by * p < 0.05, ^^ p < 0.01, ^^ p < 0.001, ^***^ p < 0.0001.