FLP-1 neuropeptides modulate sensory and motor circuits in the nematode Caenorhabditis elegans

Parasitic nematodes infect over one quarter of the population worldwide, causing morbidity in over one billion people. Current anthelmintic drugs are beginning to lose effectiveness due to the presence of resistant strains. We are interested in the role of neuropeptides, which regulate behaviors in all organisms, as another possible target for anthelmintic drugs. FMRFamide-related peptides (FaRPs) are a family of neuropeptides that are conserved throughout the animal kingdom. In particular, nematodes contain the largest family of FaRPs identified thus far and many of these FaRPs are identical among different nematode species; FaRPs in nematodes are collectively referred to as FLPs (FMRFamide-like peptides). However, little is known about the function of these FLPs. We are using the non-parasitic nematode Caenorhabditis elegans as a model for examining FLPs in nematodes. C. elegans contains at least 31 flp genes that encode 72 potential FLPs. Among the flp genes, flp-1 is one of the few that is universally found in nematodes. FLP-1 neuropeptides were previously reported to be involved in sensory and motor functions. However, previous alleles of flp-1 also disrupted a neighboring gene, daf-10. To understand the phenotypes of flp-1, new alleles that specifically disrupt flp-1 were characterized. The previously reported locomotory and egg-laying defects were found to be due to loss of flp-1, while the osmolarity defect is due to loss of daf-10. In addition, loss of flp-1 and daf-10 both cause several phenotypes that increase in severity in the double mutants by disrupting different neurons in the neural circuits.


PCR and generation of transgenic animals
To determine how the different deletions in the flp-1 alleles affected the transcripts, we isolated RNA from the different flp-1 alleles, reverse transcribed the RNA with an oligo dT primer, and amplified the flp-1 transcript with primers MR28 (5'-GAATTCACTTTATCATGACTCTGC T CTACCAGGTAGGGT-3') and 33P2 (5'-GGATTCACGATCGAAGTTGTC-3'), which encompasses the region between the initiator and stop codons. Amplified products were excised and the DNA sequence determined commercially (Genewiz).

Behavioral assays
All assays were done on day-old adult hermaphrodites unless otherwise specified. Animals were picked at the fourth larval stage and grown at 20˚C for approximately 24 hours prior to performing the assay.

Locomotion assays
Movement on a solid surface. To prepare plates for scoring movement on a solid surface, 100 microliters of E. coli (OP50/1) was spotted onto a MYOB plate [32] and grown overnight. The next day, a stamp (made by cutting a 1 cm square from the center of a 30 mm Sylgard circle) was applied to the bacterial lawn, leaving an even 1 cm square bacterial lawn. Ten to twenty day-1 adult worms were transferred to the bacterial lawn and left on an Nikon SMZ1000 stereomicroscope stage for 5 minutes to allow the worms to acclimatize to ambient oxygen levels and temperature. The movement of the animals was subsequently recorded at 8X magnification with a Nikon DS-F11 camera using Nikon Instruments Software-Elements Advanced Research (NIS-Elements AR) for 5-10 minutes. The number of body bends was counted for 1 minute starting at the 2-minute interval in the video.
Photographs of day-1 adult worms moving on bacterial food sources were also taken on a Nikon SMZ1000 stereomicroscope with a Nikon DS-F11 camera at 40X magnification. The wavelength and wave height (2x the amplitude) of the tracks in the bacterial lawn were calculated with the NIS-Elements AR imaging program. Five measurements were taken for each animal, averaged, and a mean value for each animal was determined.
Movement in liquid. To assess the effects of serotonin on movement [33], individual worms were transferred to a microtiter well containing serotonin solution (5 mg serotonin/ml M9 buffer; Sigma). After approximately 15 minutes in the serotonin solution, the number of body bends completed in 15 seconds was counted as modified from Miller et al. [34]. Each individual animal was counted three times and the three values were averaged to give a mean number for each animal.

Reproductive phenotypes
To determine an egg-laying rate, day-1 adult worms were singly plated and allowed to lay eggs for 2 to 8 hours. The number of eggs laid was counted and the egg-laying rate per hour was calculated. To determine the number of eggs retained in the uterus, day-2 adults were used. 50 microliters of a weak hypochlorite solution (1.2 ml of 6% NaOCl solution (Clorox), 0.5 ml 5 N KOH, and 8.3 ml water) was added to individual rings on the lid of a microtiter plate; individual day-2 worms were transferred to the rings. After the body of the worm was dissolved by the hypochlorite solution, the number of eggs was counted.

Nose touch
The nose touch response was performed as described [35]. Each day-1 adult was tested three times and the three values were averaged for each animal.

Osmotic avoidance
The ring assay for osmotic avoidance [28] was done with 10 worms in a 1.5 cm ring of 6 M glycerol (dyed with blue as described [22]). The number of animals within the ring was counted after 10 minutes.

Wandering
To assay whether animals stay on a food source, 10-20 day-1 adults were placed on a plate with a bacterial food source. The number of animals on the plate was counted at 12 hour intervals for a total of 48 hours.

Statistical analysis
All statistical analyses were carried out using Prism software (GraphPad). Values from wildtype, mutant, and/or transgenic animals were compared using a one-way ANOVA/Kruskal-Wallis and Tukey or Dunn's posthoc test, depending on whether the values passed a normality test.

Results
The flp-1 gene encodes eight peptides within exons 3 to 6 ( Table 1) and is alternatively spliced, giving rise to transcripts A, B, and C [36,37], which encode 7, 6, and 4 distinct FLP-1 peptides, respectively; all transcripts also encode a related, non-FLP-1 peptide [22]. The initial flp-1 alleles, yn2 and yn4, also deleted part of the daf-10 gene, making it difficult to determine which phenotypes correspond to loss of flp-1 and which to loss of daf-10 (Fig 1). The isolation of additional alleles that only disrupted flp-1 by the C. elegans Knockout Consortium allowed us to definitively categorize the phenotypes associated with loss of flp-1.
The flp-1 (ok2811, ok2781, and ok2505) alleles contain deletions that affect different splice sites or exons within the first three exons (Fig 1), making it difficult to predict a priori how transcripts would be affected. Hence, we performed multiple reverse transcription-polymerase chain reactions (RT-PCRs) for each allele, so that all possible transcripts would be isolated, and determined the DNA sequence of the products. For ok2811 mutants, the deletion affected the splice site at the end of exon 1 and extended past exon 2; the resultant transcript contained intronic sequences and disrupted the reading frame, suggesting that ok2811 is a null allele; no other transcripts were identified. The ok2781 deletion affected splicing at the 5' end of exon 2 and extended into intron 2; the resultant transcript included intronic sequence but maintained the reading frame, suggesting that ok2781 mutants will have phenotypes similar to wild type; no other transcripts were detected. The ok2505 deletion extended from intron 2 to intron 3, thereby removing exon 3; however, the resultant transcripts used cryptic splicing sites, resulting in two transcripts in which the coding sequences for two (KPNFMRFY-NH 2 and AGSDPNFLRF-NH 2 ) to four (KPNFMRFY-NH 2 , AGSDPNFLRF-NH 2 , SQPNFLRF-NH 2 , and ASGDPNFLRF-NH 2 ) FLP-1 peptides were deleted relative to transcript A. Hence, the three alleles were predicted to represent a presumptive null (ok2811), a presumptive wild type (ok2781), and a partial loss-of-function (ok2505). Transcription of daf-10 was not affected by the flp-1(ok2811, ok2781, ok2505) deletions by RT-PCR (S1 Fig).

Loss of flp-1 or daf-10 causes sensory defects
The daf-10 flp-1(yn2 and yn4) double mutants showed defects in two sensory modalities: mechano-and osmosensation. Wild-type animals recoil when touched on their nose (nose touch sensitivity) [35]; this nose touch response is mediated by the ASH multimodal sensory neuron [38]. While wild-type animals respond to nose touch 97% of the time (N = 371), all three flp-1 mutant alleles showed a significant decrease in the nose touch response (ok2811, ok2781, and ok2505 responded 72%, 68% and 55% of the time, respectively), although none of the decreases were as severe as the yn2 allele, which responded only 7% of the time (Fig 2; Table 2). The nose touch defect was rescued by wild-type flp-1 in the ok2781 and ok2505 mutants. Because daf-10 mutants have defective sensory cilia, we expected that the mutants Table 1  would show no nose touch sensitivity, similar to the yn2 double mutants. Accordingly, the daf-10 mutants were nose touch insensitive and had slightly, but not significantly higher (17-19%) response rates than yn2 mutants, and significantly lower response rates than the flp-1 mutants (Fig 2; Table 2; Dunn's posthoc test, p<0.0001). These results suggest that loss of flp-1 slightly exacerbates the sensory loss in the nose touch circuit in daf-10 mutants. Wild-type animals are highly sensitive to environmental conditions. For instance, C. elegans avoids areas of high osmolarity [28] and will avoid crossing a high osmolarity ring (97% within ring; N = 168; no. trials = 16; Table 2). yn2 double mutants are insensitive to high osmolarity (11% within ring; N = 140; no. trials = 14; [22]). Similarly, daf-10 single mutants showed no response to high osmolarity (19% and 17% within ring for e1387 and gk795, respectively; N = 150; no. of trials = 15; Table 2). By contrast, all flp-1 single mutants were sensitive to high osmolarity (81%, 84%, and 92% within ring for ok2811 (N = 150; no. of trials = 15), ok2781 (N = 161; no. of trials = 16), and ok2505 (N = 150; no. of trials = 15), respectively), indicating that the osmolarity defect in yn2 double mutants is due to loss of daf-10 alone.

Increased levels of FLP-1 peptides decrease locomotion
In its natural habitat, C. elegans moves between soil and liquid, depending on the moisture in the soil. To mimic these conditions in the laboratory, we assayed crawling and swimming movement of animals on an agar surface, which is analogous to soil, and in liquid, respectively. We previously reported that yn2 double mutants showed hyperactivity and a larger waveform when crawling compared to wild type, while overexpression of flp-1 inhibited movement and showed a smaller waveform [22]. Here, we further dissected the locomotory patterns of yn2 double and flp-1 single mutants. Movement of yn2 double mutants is bimodal. When yn2 double mutants are measured only in their locomotory phase, they demonstrate hyperactivity as reported in Nelson et al. [22] and showed a larger waveform than wild type (Fig 3A; Table 2). However, yn2 double mutants also showed long periods of inactivity interspersed with rapid movements; for instance, 59% of yn2 double mutants (n = 134) showed no forward movement during our 1 minute analysis, whereas only 7% of wild-type animals (n = 358) showed no forward movement. Therefore, if movement was measured as activity for a specific length of time, such as one minute, yn2 double mutants showed hypoactivity, similar to daf-10 mutants (Fig 3B; Table 2). The single flp-1 mutants, ok2811 and ok2781, showed increased crawling mobility and larger waveforms (Fig 3A and 3B; Table 2). Hence, the locomotion phenotype of yn2 double mutants is the combined result of the lack of sensory input due to the daf-10 mutation and the hyperactivity due to loss of FLP-1 peptides. In contrast to the hyperactive movements of ok2811 and ok2781 mutants, ok2505 mutants were very sluggish ( Fig 3B; Table 2) and showed decreased movement. Because ok2505 mutants are predicted to express only a subset of the FLP-1 peptides, these results suggest that different FLP-1 peptides have different effects on the locomotory circuit.
Exogenous serotonin inhibits locomotion in C. elegans [33]. Wild-type animals exposed to exogenous serotonin stop swimming within minutes (Fig 3C; Table 2). yn2 double mutants were insensitive to the serotonin-induced sluggishness [22]. To determine whether this serotonin-insensitivity is due to loss of flp-1 or daf-10, single flp-1 and daf-10 mutants were similarly exposed to exogenous serotonin. Like wild-type animals, daf-10 mutants stopped swimming in the presence of serotonin. By contrast, all flp-1 mutants continued swimming in the presence of serotonin (Fig 3C). The swimming phenotype was rescued by a wild-type flp-1 fragment. Hence, the serotonin-insensitivity is due to loss of FLP-1 peptides.
yn2 double mutants, have an additional sensory-motor phenotype, which we termed wandering. Wild-type animals stay on a bacterial food source; however, yn2 double mutants wander off the food source and up the side of the plate, where they die [22]. The wandering phenotype is not due to the loss of chemosensory input, as yn2 double mutants can chemotax to a bacterial food source [22]; furthermore, daf-10 single mutants, which have impaired sensory responses, and flp-1 mutants did not exhibit wandering (Fig 4). To determine whether the wandering phenotype was a result of loss of sensory input and flp-1 signaling, we mated flp-1 (ok2811) mutants with a second mutant that has decreased sensory responses, tax-4. tax-4 encodes a subunit of a cyclic nucleotide-gated channel that is homologous to the vertebrate rod photoreceptor cGMP-gated channel and that is found in a subset of sensory neurons; loss of tax-4 results in defects in several sensory behaviors [39]. The tax-4; flp-1 double mutants showed a wandering behavior, although not as robustly as the yn2 mutants. We suggest that wandering is the result of a synergistic interaction between loss of sensory input and FLP-1 signaling in downstream circuits.

Increased levels of FLP-1 peptides decrease egg-laying rates
Wild-type animals lay between 5-8 eggs per hour [22], but the rate of egg laying is highly affected by environmental stresses and the animal's metabolic rate. For example, under harsh environmental conditions, such as low food availability, high osmolarity, and high vibration frequency, egg-laying rates are decreased [40]. yn2 double mutants had a severely decreased egg laying rate of roughly one egg per hour (Fig 5A; Table 2). All of the single flp-1 mutants also showed a significantly decreased egg-laying rate, although none showed phenotypes as severe as that in the yn2 double mutants. Rather than being rescued by microinjection of a wild-type flp-1 fragment, the transgenic flp-1 mutants as well as the yn2 double mutants showed an even more severe egg-laying defect. The daf-10(e1387) allele also showed a significantly decreased egg laying rate, which was not seen in the daf-10(gk793) allele.
The decreased egg-laying rate in flp-1 mutants can result from several variables: mutants are generating fewer eggs, a sensor controlling the number of eggs in the uterus is defective, or muscles are not responsive to the HSN and VC motor neuron output [40]. To determine whether the decreased egg-laying rate was due to a decreased number of available eggs for release, we counted the number of eggs in the uterus. Wild-type animals have 10-15 eggs in their uterus at any one time [40,41], although we had slightly higher numbers in our current assays (2 day adults: 15.91±0.55 (N = 113); Fig 5B; Table 2). The yn2 double and ok2811 null mutants did not show any significant differences compared to wild type. By contrast, the ok2781 and ok2505 mutants showed significantly fewer and significantly greater numbers of retained eggs, respectively ( Fig 5B; Table 2). Microinjection of the flp-1 fragment significantly increased egg retention in all flp-1 mutants (Fig 5B; Table 2). Loss of daf-10 significantly decreased egg retention but only slightly decreased egg-laying rates. Overexpression of flp-1 in the daf-10 mutants, however, increased egg retention. p 0.001, ÃÃ , p 0.01, Tukey's post hoc test; transgenic line significantly different from mutant allele: +++ , p 0.001, ++ , p 0.01, Tukey's post hoc test, each strain was tested in at least three independent trials. https://doi.org/10.1371/journal.pone.0189320.g003

Discussion
In identifying new drug targets for anthelmintic drugs, peptide signaling systems that inhibit locomotion or reproduction are attractive targets. FLP-1 neuropeptides were previously found to have roles in locomotion and egg laying [22,24]. Because the flp-1 gene lies within the first intron of the daf-10 gene and the initial flp-1 alleles, yn2 and yn4, affected the activity of both flp-1 and daf-10, we examined more recently isolated single flp-1 alleles for sensory, locomotory, and reproductive defects. Surprisingly, although the ok2781 mutants were predicted to have wild-type behavior because no peptide coding regions are lost, the mutants acted more as loss-of-function mutants, more often showing phenotypes similar to or less severe than ok2811 mutants, which are the presumed flp-1 null mutants. These results suggest that some FLP-1 peptides are still expressed, albeit at low levels, in ok2781 mutants. The phenotypes of ok2505 mutants varied greatly compared to ok2811 mutants, sometimes showing more severe phenotypes (e.g., nose touch response, crawling rate, resistance to serotonin, and number of eggs retained in the uterus) and sometimes acting similar to wild-type (e.g., waveform), suggesting that the different FLP-1 peptides affect neural circuits differently. Because ok2505 mutants expressed only a subset of FLP-1 peptides, we cannot distinguish at this time which FLPs promote and which inhibit different behaviors.
DAF-10 IFT-A is required for proper cilia formation in amphid and phasmid neurons in C. elegans [26,42]. The cilia, which are located at the distal endings of the dendrites, contain the molecular machinery for sensory transduction and translating information from the environment [26]. The loss of DAF-10 IFT-A causes the animals to have defects in osmo-and chemosensitivity [42], whereas flp-1 single mutants have no osmolarity defect. Hence, the osmolarity defect previously reported in the yn2 double mutants as caused by flp-1 [22] was actually due to the loss of daf-10.
The ASH sensory neuron, whose sensory cilia are functionally dependent on DAF-10 activity, mediates two distinct sensory stimuli: nose touch and avoidance of noxious chemicals and hyperosmolarity; activation of ASH results in signaling to AVA interneurons, the driver cells for backward locomotion, causing the animal to initiate a reversal [43]. However, nose touch and avoidance to noxious chemicals activate distinct cellular signaling pathways, so that loss of one ASH sensory modality does not imply loss of other ASH sensory modalities [43]. yn2 double mutants were not responsive to nose touch or hyperosmolarity [22], whereas flp-1 mutants were nose touch insensitive, but hyperosmolarity sensitive. flp-1 is expressed in the AVA interneuron [22]. Because the yn2 double mutant has a more severe nose touch insensitivity than the single mutants alone, we propose that the enhanced nose touch defect in the yn2 double mutants is due to the reduction of the ASH signal due to the faulty cilia in daf-10 mutants, compounded by the reduction of downstream AVA interneuron signaling caused by the loss of flp-1. Alternative signaling pathways mediate the hyperosmolarity response in flp-1 mutants.
Disrupted sensory input and exogenous serotonin decrease locomotion [33,44]. The locomotory activity of yn2 double mutants was further characterized. The yn2 double mutants displayed extended periods of inactivity, which we attribute to loss of daf-10, as the daf-10 single mutants show a similar phenotype. When the double mutants were active, the speed with which they moved [22] and the sinusoidal waveform they exhibited were much greater than that of wild type. flp-1 single mutants, ok2811 and ok2781, were also hyperactive and had exaggerated waveforms (Fig 3A and 3B; Table 2). Hence, the speed and waveform defects are due to loss of flp-1. The slowed speed and wild-type waveform of ok2505 mutants suggests that different FLP-1 peptides have different effects on locomotory circuits. Our results do not differentiate between whether the lost FLP-1 peptides in ok2505 mutants activate locomotion or the remaining peptides inhibit locomotion. Both the yn2 double mutants and flp-1 single mutants were insensitive to the serotonin-induced decreased locomotion, whereas the daf-10 single mutants stop swimming in the presence of exogenous serotonin. Collectively, our results indicate that the overall activity of FLP-1 peptides, like serotonin, is to inhibit locomotory circuits. These flp-1 locomotory circuits can over-ride the locomotory circuits in which serotonin is active, as flp-1 mutants continue to swim in the presence of serotonin while wild-type and daf-10 single mutants stop swimming.
When C. elegans are starved, they become more active as they constantly forage for food [45]. Often times, starved animals wander off the agar medium and climb onto the side of plates where they die, a phenotype we refer to as wandering. flp-1 and daf-10 single mutants do not wander. By contrast, the yn2 double mutants, which can sense food [22], climb onto the side of plates even in the presence of food. We suggest that wandering in yn2 double mutants is a synergistic effect caused by the loss of sensory input due to the knockout of daf-10 and the loss of downstream interneuron signaling by flp-1. Similarly, decreasing sensory input by a different mutation, such as tax-4, in flp-1 mutants also caused a wandering phenotype. Because tax-4 is only expressed in a subset of amphidial neurons [39] and DAF-10 is expressed in all amphidial neurons, we predicted and found that the wandering phenotype was not as severe in tax-4; flp-1 double mutants.
The yn2 double mutant has a severe egg-laying defect [22]. Because only one daf-10 allele showed a decreased egg laying rate, we conclude that loss of amphidial and phasmid sensory input affects egg-laying rates only slightly. By contrast, flp-1 single mutants exhibited a significantly decreased egg-laying rate, albeit not as severe as the double mutant. The egg-laying neural circuit consists of two motor neurons, HSN and VC, which synapse onto the vulval muscles to cause muscle contraction and egg release [46]. Both neurons receive little synaptic input from other cells: VC neurons receive input from HSN and other VC neurons, while HSN neurons receive input from the VCs, two interneurons, and PLM, a body touch mechanosensory neuron [46]. Because flp-1 is not expressed in the egg-laying neural circuit, these results suggest that FLP-1 peptides act extra-synaptically to modulate the egg-laying neural circuit. Furthermore, because both loss and overexpression of FLP-1 peptides decrease egg-laying rates, either different FLP-1 peptides have different effects on the egg-laying circuit or high levels of FLP-1 peptides, as in the transgenic animals, cause activation of additional circuits to decrease egg laying.

Conclusions
Sensory neurons send signals to interneurons, which in turn signal to motor neurons, resulting in an animal responding to a change in the environment [47]. DAF-10 IFT-A activity in amphidial and phasmid sensory neurons is necessary for ciliogenesis and, therefore, sensory function. flp-1 is expressed in many interneurons [22], which serve to transmit information from the sensory neurons to the motor neurons. Hence, decreased DAF-10 IFT-A activity affects behaviors that are dependent on sensory information, such as osmolarity and nose touch responses, locomotion, and egg laying; with the exception of osmolarity, these sensory deficits are exacerbated when flp-1 signaling is lost in the interneurons. As FLP-1 peptides are not expressed in the egglaying neural circuit, the FLP-1 peptides must be acting extra-synaptically to influence egg-laying rate. Furthermore, different FLP-1 peptides must be affecting egg-laying circuits differentially as both increased and decreased FLP-1 peptide levels result in decreased egg-laying rate. Determining the function of specific FLP-1 peptides awaits future studies.
The wandering behavior is the result of a synergistic effect of loss of both sensory function and downstream signaling through FLP-1 peptides. Wandering animals are able to chemotax to food, but wander off food within a few hours. Chemoattraction to food is dependent on amphidial neurons; however, loss of amphidial function in daf-10 mutants does not result in wandering animals, suggesting that a different pathway that utilizes FLP-1 peptides to sense food is operating. Loss of both pathways results in wandering.