Introduction

Neuropeptides and peptide hormones have been discovered in animals ranging from cnidarians to mammals and are indispensable modulators of virtually all physiological processes in Metazoa [1]. By acting as neurohormones, neurotransmitters, or neuromodulators, neuropeptides fine-tune the communication between specialized cells and tissues in response to an altering environment. Most neuropeptides exert their effects via G protein-coupled receptors (GPCRs), which are prime pharmacological targets, making the study of neuropeptide signaling and their functions of great interest to both biological and medical research [2,3,4,5,6,7].

Neuropeptides are encoded by larger, inactive precursor proteins that are enzymatically processed by prohormone convertases and carboxypeptidases to yield mature, bioactive peptides [8]. In addition, post-translational modifications can be essential to acquire biological activity. These include N-terminal cyclization of glutamine/glutamic acid to pyroglutamate, N-terminal acetylation, C-terminal amidation, phosphorylation, or sulfation [9, 10]. Recognition sites for cleavage of neuropeptide precursors by proprotein convertases typically consist of a dibasic motif involving arginine and/or lysine residues (KR, RR, KK, and RK, in respective order of susceptibility to proprotein convertase cleavage) [11]. However, not all mature peptides are cleaved after dibasic cleavage sites. Frequently, basic residues that are interspersed with an even number of non-basic amino acids ([K/R]Xn[K/R], with n = 2, 4, 6, or 8) will serve as cleavage recognition sequences, or cleavage may take place following but a single basic residue (K, R) [12]. Less commonly so, certain bioactive peptides – such as NPFF – can be produced by processing via non-classic pathways, utilizing other recognition sites [13,14,15]. In mammals, endothelin-converting enzyme-2 (ECE-2) and carboxypeptidases A5 and A6 have been suggested as possible neuropeptide processing enzymes that act on sites other than the classic dibasic residues [13]. The cleavage sites recognized by these proposed peptidases are far less clear-cut than the simple K/R combination used by classic proprotein convertases, and remain to be investigated in detail [16].

Non-standard cleavage sites are an inherent drawback when in silico prediction methods are used for the prediction of bioactive neuropeptides in the proteome. While in silico analyses are invaluable and well-established tools, the algorithms available today are unsuitable to identify non-classically processed neuropeptides. Complementary in vivo techniques involving tissue extraction, chromatography, and Edman degradation can be applied, but are highly inefficient for large-scale identification of mature neuropeptide sequences [17,18,19].

Following developments in mass spectrometry (MS) and high-performance liquid chromatography (HPLC) in the early 21st century, it became possible to analyze integral tissue peptide extracts. The ability of studying the entire peptidome of a single sample introduced the term “peptidomics”, defined as the comprehensive characterization of peptides present in a biological sample [20]. Peptidomics approaches based on liquid chromatography coupled on-line to mass spectrometry (LC-MS) quickly expanded the repertoire of characterized peptides in a multitude of species [5,6,7]. MS and tandem mass spectrometry (MS/MS) enable de novo characterization of peptides and post-translational modifications: each experiment displays a snapshot of the full complexity of peptide profiles as they were present in vivo, without a strict requirement for prior in silico knowledge of the genome or transcriptome of the sampled organism [21]. Since 2001, the number of scientific publications dealing with peptidomics or the peptidome keeps increasing faster than the overall database, highlighting a continuously growing interest [22,23,24,25]. Indeed, peptidomics has become a highly invaluable tool for the analysis of (neuro)peptidomes in biological samples [7].

Recent years brought about major increases in sensitivity of mass spectrometers, which significantly improved the acquired data from mass spectrometric experiments. Recent Orbitrap and Q-TOF MS systems are specifically designed to provide fast scanning in combination with high mass accuracy, dynamic range, and resolving power [26]. When using high pressure liquid-chromatography (HPLC) in combination with MS/MS, not only the mass spectrometer determines identification rate but the ultra (U)HPLC column makes an important contribution to resolution as well. Continuous improvement of (U)HPLC columns leads to increasingly lower dispersion rates, resulting in a higher efficiency [27]. Comparison of platforms performed by parallelized high-resolution and normal-resolution peptidomics show a significant increase in peptide identifications from novel high-resolution set-ups. For instance, a comparative experiment using a nanobore high-resolution C18-LC-ESI-MS/MS and a low-resolution standard bore C18-LC-ESI-MS/MS system revealed a 12-fold increase in unique peptide identifications from gastrointestinal digestions of hemoglobin using the high-resolution system [28].

Peptidomics can provide valuable knowledge on peptidergic effectors involved in physiological and behavioral processes. Neuropeptidomics have revealed thousands of bioactive peptides, guiding functional studies on neuropeptide functions. Often, small invertebrate model organisms have been favored for such studies [5, 29,30,31,32]. Caenorhabditis elegans has emerged as a powerful model to study peptidergic modulation on a cellular and molecular level, facilitated by its completely mapped nervous system constituting 302 neurons [30, 33, 34]. In C. elegans, neuropeptides are classified into three major families based on their mature sequences: insulin-like peptides (INS), FMRFamide-related peptides (FLPs), and non-insulin/non-FMRFamide-related peptides (NLPs) [35]. Wormbase (release WS259) specifies 40 INS precursors, 31 FLP precursors, and 59 NLP precursors (NLP-1 to 57, PDF-1, and NTC-1) encoded by the C. elegans genome. Additional NLP precursors have been suggested in recent phylogenetic studies, expanding the amount of putative NLP precursors to 75 [36, 37]. Previous peptidomic studies in C. elegans reported up to 75 NLP- and FLP-type neuropeptides in a single experiment, processed from 29 distinct precursors [38].

Here, we prepared peptide-enriched extracts from whole-mount C. elegans samples [39] and relied on quadrupole-Orbitrap LC-MS/MS to biochemically identify mature neuropeptides. Using this pipeline, we identified 203 unique C. elegans peptides, including 35 peptide sequences that so far have not been predicted by in silico methods. We discovered 29 novel peptides encoded by known or in silico predicted precursors, as well as eight mature peptides originating from seven prepropeptides that were not yet annotated as neuropeptide precursors. By collecting robust data from mixed-stage cultures reared under standard conditions, this study represents the currently most comprehensive reference neuropeptidome of C. elegans.

Experimental

C. elegans Strains

The wild-type N2 Bristol strain was obtained from the Caenorhabditis Genetics Stock Center (University of Minnesota). Worms were cultivated in standard liquid cultures supplied with flash frozen E. coli K12 as food source as described previously [40, 41]. The concentration of bacteria was verified daily, and new K12 bacteria were added to maintain optimal food levels (OD600 = 1.68).

Extraction of Peptides

Wild-type mixed stage worms from liquid cultures were collected and pooled. This pool was aliquoted into 10 samples, each containing approximately 1 mL of biological material. Peptides were extracted using an acidified methanol extraction solvent as described previously [31]. This precipitates larger proteins, while smaller peptides remain in solution. A size exclusion column (Sephadex PD MiniTrap G-10, GE Healthcare) and a 10 kDa cut-off filter (Amicon Ultra-4, Merck Millipore) were used to enrich the samples for peptide content by isolating the 700–10,000 Da mass fraction. Samples were briefly stored at 4 °C prior to MS analysis.

Quadrupole-Orbitrap LC-MS/MS

Quadrupole-Orbitrap LC-MS/MS experiments were conducted using a Dionex UltiMate 3000 UHPLC coupled on-line to a Thermo Scientific Q Exactive mass spectrometer. The UHPLC is equipped with a guard pre-column (Acclaim PepMap 100, C18, 75 μm × 20 mm, 3 μm, 100 Å; Thermo Scientific) and an analytical column integrated in the nano-electrospray ion source (EASY-Spray, PepMap RSLC, C18, 50 μm × 150 mm, 2 μm, 100 Å; Thermo Scientific). The sample was separated at a flow rate of 300 nL/min, using a 45-min. linear gradient from 3% to 55% acetonitrile containing 0.1% formic acid. MS data were acquired using a data-dependent (dynamic exclusion settings at 15 s) Top10 method, choosing the most abundant precursor ions (charge selection ranging from 2 to 5) from a full MS survey scan for Higher-energy Collisional Dissociation fragmentation (HCD). Full MS scans were acquired at a resolution of 70,000 at m/z 200, with a maximum injection time of 256 ms and scan range of 400 to 1600 m/z. The resolution for MS/MS scans after HCD fragmentation was set at 17,500 at m/z 200, with a maximum injection time of 64 ms.

Mass Spectrometry Data Analysis

To correct for inter-run variation causing retention time shifts between the 10 different runs, all data files were aligned using Progenesis LC-MS software (Nonlinear Dynamics). Peak picking was done in automatic mode, using default sensitivity settings. Results were then filtered on charge state, retaining all features with charges ranging from 1 to 7. All selected features were exported to a .csv file containing the m/z, charge, deconvoluted mass, abundance, and retention times. For peptide annotation, we developed a custom R script [42] that compares all the detected masses in the Progenesis .csv file to an in-house peptide library containing the masses of 354 C. elegans peptides and their post-translational modifications (supplementary Tables S1 and S2). Deconvoluted masses that match within an error margin of 5 ppm were interpreted as a positive hit.

MS/MS fragmentation data were analyzed using PEAKS software (Bioinformatics Solutions) with a custom-made library containing all known C. elegans proteins (Wormbase ver. WS256), E. coli proteins (SwissProt, acquired September 2016), and a list of common protein contaminants (common Repository of Adventitious Proteins, cRAP). Parent mass error was set at 10 ppm, and fragment mass error at 0.02 Da. The following variable modifications were taken into account: oxidation (+15.99 Da), glycine-loss in combination with amidation (–58.01 Da), pyroglutamation from glutamic acid (–18.01 Da), pyroglutamation from glutamine (–17.03 Da), phosphorylation of serine, threonine, or tyrosine (+79.97 Da).

Filtering Parameters for the Discovery of Novel Putative Neuropeptide Precursors

All peptide identifications were exported from PEAKS as a .csv file. A custom R script was devised to cross-reference peptides in this file with the known C. elegans neuropeptides in our in-house database (supplementary Tables S1 and S2), thereby creating a dataset containing peptides hitherto not present in the predicted C. elegans peptidome. In a next filtering step, the location of the peptide within its corresponding precursor was examined, and only peptides flanked by a basic amino acid (K, R, or a combination of both) were retained. Two variations to this rule were allowed: first, if no N-terminal basic residue is present, the peptide must be located 30 amino acids or less from the N-terminus of the precursor protein. This considers that neuropeptides may be located right after the signal peptide cleavage site, typically situated no more than 30 amino acids from the N-terminus of the precursor protein. Second, if no C-terminal basic amino acid is present, the peptide must be located at the C-terminus of the precursor protein. All precursor proteins containing peptides that comply with this set of rules were retained. The resulting list of precursor proteins was subjected to SignalP to look for the presence of a signal peptide sequence, reducing the number of candidate neuropeptide precursors from the initial 20 to 16 [43]. The location of the signal peptide cleavage site was compared with C-termini of peptides that were identified as being located right after the signal peptide sequence, to confirm consistent processing. After manual inspection of MS/MS spectra, seven candidates were withheld for further analysis. All of these candidate precursors were found to be shorter than 300 amino acids, supporting reported observations for neuropeptide precursors to typically not be longer than this length [44, 45]. Indeed, all known FLP and NLP precursors in C. elegans are shorter than 300 amino acids, with the single exception of an NLP-16 splice variant, which is 326 amino acids long. For the prediction of putative proprotein convertase cleavage sites, we used both the NeuroPred and ProP algorithms [46, 47].

Sequence Alignment of Putative Novel Neuropeptide Precursors

Protein BLAST was used to search for protein homologs in all Metazoa of the presumed neuropeptide precursors we found in C. elegans. Potential homologs were selected based on the conservation of the (di)basic cleavage sites as well as the presence of conserved motifs within the putative neuropeptide itself. These hits were used for sequence alignment with the PSI-Coffee algorithm, which is optimized for the alignment of distantly related proteins using homology extension [48]. Shading was added using ver. 3.21 of BOXSHADE [49].

Results

Knowledge of the C. elegans neuropeptidergic repertoire provides invaluable information for further functional and behavioral studies. We ventured to maximize detection of the integral C. elegans peptidome using an in-house developed peptidomics pipeline. In order to include low-abundant and stage-specific neuropeptides as much as possible, we increased sample sizes and made use of mixed-stage cultures. This study focuses on the FLP and NLP families of neuropeptides for which our extraction protocol is optimized.

Current predictions estimate that the 31 FLP and 75 NLP precursor proteins in the C. elegans genome could generate up to 354 mature neuropeptides (supplementary Tables S1 and S2). The effectiveness of our peptidomics pipeline was evaluated by assessing how much of the predicted C. elegans peptidome could be observed in each individual sample, compared with the overall success of peptide identification. Peptide mass matching analysis of MS1 data of 10 mixed-stage C. elegans samples identified 203 individual peptides, corresponding to nearly 58% of the predicted peptidome (Figure 1, Supplementary Tables S1–S2). Furthermore, 131 neuropeptide sequences were successfully detected by tandem MS, confirming 37% of the known and predicted C. elegans peptide sequences (Figure 1, Supplementary Tables S1, S2). Robustness of the method was examined by counting the occurrence of individual peptides in the 10 replicates (Supplementary Figure S1). Approximately 80% of the 203 detected peptides were measured in seven or more replicates, indicating a high reproducibility of the method.

Figure 1
figure 1

Percentage of the predicted C. elegans peptidome (FLP and NLP) observed with LC-MS. In total, 203 neuropeptides were detected in MS mode, together almost 58% of the C. elegans peptidome (containing 354 known and predicted NLP/FLP-type neuropeptide sequences prior to this study). We confirmed 131 neuropeptides – or 37% of this peptidome – with MS/MS

Full size image

Detection of Novel Peptides Derived from Known C. elegans FLP and NLP Precursors

Next to the identification of already predicted neuropeptides, our peptidomics dataset revealed several peptides that are additionally processed from known C. elegans FLP or NLP precursors. These neuropeptides were identified with MS/MS and retained as targets of interest based on the presence of flanking basic residues. Our analysis identified 29 hitherto unstudied peptide sequences, present in 22 precursor proteins: FLP-1, FLP-3, FLP-5, FLP-6, FLP-9, FLP-10, FLP-15, FLP-18, NLP-3, NLP-6, NLP-8, NLP-9, NLP-10, NLP-12, NLP-17, NLP-42, NLP-43, NLP-49, NLP-50, NLP-51, and NLP-52 (supplementary Table S3). Often, these peptides are located in the precursor between the already known neuropeptides sequences, implying that they may be remnants following proprotein cleavage. However, as we do not detect other remnants of these precursors, nor similar remnants for others, and since all these detected sequences are conserved in Caenorhabditis species, they may be biologically active peptides.

Mass Spectrometric Evidence for the Recently Predicted C. elegans Neuropeptide Precursor Proteins NLP-53, NLP-55, NLP-56, NLP-57, NLP-58, and NLP-65

The C. elegans peptidome has recently been extended with the genes nlp-53 (Y12A6A.2), nlp-54 (C30H6.10), nlp-55 (F14F11.2), nlp-56 (Y57G11C.45), and nlp-57 (F08G12.8). Additionally, we assigned gene names to the novel peptide precursors (supplementary Table 1) suggested by Mirabeau and Joly and Koziol et al., including nlp-58 (T07C12.15) and nlp-65 (R06F6.7) [36, 37]. Based on predictions, these genes are presumed to encode neuropeptide precursor proteins (supplementary Table S2). We here present mass spectrometric evidence that these genes indeed give rise to mature neuropeptides. Using MS/MS, we successfully identified NQGAGSVSLDSLASLPMLRYamide from NLP-53 (supplementary Figure S2), MYINPDYYYVEQLPTM from NLP-55 (supplementary Figure S3), and SSIMTDDVEPPQLLTRQL from NLP-56 (supplementary Figure S4). Three mature neuropeptides were identified for NLP-57: SPIHGIWNNLPAPPQ, VYGFYNYLPKEEDDRD, and NTILLLTPNEDYVE (supplementary Figure S5). In NLP-58 we detected ARIFDGQEEQ as a mature neuropeptide (supplementary Figure S6), but not VPMMSLKGLRamide (present in MS data only, supplementary Table S2), which was predicted by Mirabeau and Joly (2013). Finally, in C. elegans protein NLP-65, we identified the previously in silico predicted neuropeptides DGLPSFYDIR and GLPSAYDIR (supplementary Figure S7) [37].

Discovery of Seven Putative Novel Neuropeptide Precursors in C. elegans 

Besides providing evidence for the actual in vivo presence of peptides from the predicted C. elegans peptidome, the MS/MS dataset was mined further for possible neuropeptides encoded by yet to be annotated neuropeptide precursor genes. The dataset was filtered using several neuropeptide characteristics as parameters (Material and Methods: § “Filtering Parameters for the Discovery of Novel Putative Neuropeptide Precursors”). Seven proteins, which we suggest to name NLP-76 to NLP-82, were previously overlooked in bioinformatic surveys and complied with these assumptions. The peptide sequences that we identified for these proteins are: SPIVELYPVVDSGNVEPEAFPAYFRF in C02B4.4 (Figure 2), pQPAGGQDVPPFL in C06A8.3 (Figure 3) GDVDSVFFSPFRIIamide and ALLAGPHDYDLGDFISNPNV in C16D2.2 (Figure 4, LSADFRNEIPPPDYI in F09E8.8 (Figure 5), HSAGSTYPESL in T04C12.3 (Figure 6), NDFFLRSA in Y67D8B.4 (Figure 7), and FILQDLPLERFE in H32K21.1 (Figure 8).

Figure 2
figure 2

C. elegans protein C02B4.4 encodes the putative neuropeptide SPIVELYPVVDSGNVEPEAFPAYFRF. (a) Precursor sequence with the predicted signal peptide marked in yellow and putative proprotein convertase cleavage sites in green (coloring is based on results from SignalP, NeuroPred, and ProP prediction algorithms). The peptide detected with MS/MS is shown in red. (b) Fragmentation spectrum of the detected neuropeptide SPIVELYPVVDSGNVEPEAFPAYFRF

Full size image
Figure 3
figure 3

C. elegans protein C06A8.3 encodes putative neuropeptide pQPAGGQDVPPFL. (a) Precursor sequence with the predicted signal peptide marked in yellow and putative proprotein convertase cleavage sites in green (coloring is based on results from SignalP, NeuroPred, and ProP prediction algorithms). The peptide detected with MS/MS is shown in red. (b) Fragmentation spectrum of the detected neuropeptide pQPAGGQDVPPFL. The N-terminal glutamine has undergone a post-translational modification to pyroglutamic acid

Full size image
Figure 4
figure 4

C. elegans protein C16D2.2 encodes putative neuropeptides GDVDSVFFSPFRIIamide and ALLAGPHDYDLGDFISNPNV. (a) Precursor sequence with the predicted signal peptide marked in yellow and putative proprotein convertase cleavage sites in green (coloring is based on results from SignalP, NeuroPred, and ProP prediction algorithms). The peptides detected with MS/MS are shown in red. (b) Fragmentation spectrum of the detected neuropeptide GDVDSVFFSPFRIIamide, with transformation of the C-terminal glycine into an amide as post-translational modification. (c) Fragmentation spectrum of the detected neuropeptide ALLAGPHDYDLGDFISNPNV

Full size image
Figure 5
figure 5

C. elegans protein F09E8.8 encodes putative neuropeptide LSADFRNEIPPPDYI. (a) Precursor sequence with the predicted signal peptide marked in yellow and putative proprotein convertase cleavage sites in green (coloring is based on results from SignalP, NeuroPred, and ProP prediction algorithms). The peptide detected with MS/MS is shown in red. (b) Fragmentation spectrum of the detected neuropeptide LSADFRNEIPPPDYI

Full size image
Figure 6
figure 6

C. elegans protein T04C12.3 encodes putative neuropeptide HSAGSTYPESL. (a) Precursor sequence with the predicted signal peptide marked in yellow and putative proprotein convertase cleavage sites in green (coloring is based on results from SignalP, NeuroPred, and ProP prediction algorithms). The peptide detected with MS/MS is shown in red. (b) Fragmentation spectrum of the detected neuropeptide HSAGSTYPESL

Full size image
Figure 7
figure 7

C. elegans protein Y67D8B.4 encodes putative neuropeptide NDFFLRSA. (a) Precursor sequence with the predicted signal peptide marked in yellow and putative proprotein convertase cleavage sites in green (coloring is based on results from SignalP, NeuroPred, and ProP prediction algorithms). The peptide detected with MS/MS is shown in red. (b) Fragmentation spectrum of the detected neuropeptide NDFFLRSA

Full size image
Figure 8
figure 8

C. elegans protein H32K21.1 encodes putative neuropeptide FILQDLPLERFE. (a) Precursor sequence with the predicted signal peptide marked in yellow and putative proprotein convertase cleavage sites in green (coloring is based on results from SignalP, NeuroPred, and ProP prediction algorithms). The peptide detected with MS/MS is shown in red. (b) Fragmentation spectrum of the detected neuropeptide FILQDLPLERFE

Full size image

We subjected these precursor protein sequences to a BLAST search to look for possible orthologs. All seven precursors – including the peptide sequence and position of the basic residues – are well-conserved in other Caenorhabditis species such as C. brenneri, C. briggsae, and C. remanei (supplementary Figures S7–S14), suggesting a possible biological role. Precursor proteins C02B4.4, C06A8.3, C16D2.2, F09E8.8, and Y67D8B.4 were found to be conserved in several parasitic nematode species as well, including Ancylostoma ceylanicum, Necator americanus, Dictyocaulus viviparus, and Oesophagostomum dentatum (supplementary Figures S7, S8, S9, S10, and S13). Overall, alignments of the seven precursors with their closest homologs in other nematode species display the evolutionary conservation of the detected neuropeptides and the flanked basic cleavage sites. Our BLAST search did not detect any homologs for these new precursor proteins in species other than nematodes.

Discussion

When studying animal physiology and in particular neuronal signaling underlying behavior, aging, and learning and memory formation, an extensive knowledge of the organism’s complete peptidome is highly valuable. Various efforts in creating an integral peptidome library through peptidomics have been made in a wide range of animals, including nematodes (C. elegans, Ascaris suum), arthropods (Tribolium castaneum, Aedes aegypti, Glossina morsitans, Drosophila melanogaster, Apis mellifera, Schistocerca gregaria, Triatoma infestans), molluscs (Aplysia californica, Lymnaea stagnalis), and vertebrates (Rattus norvegicus, Danio rerio) [5, 50,51,52]. Some of these species are considered pests (S. gregaria, A. suum), are known vectors of disease (A. aegypti, G. morsitans, T. infestans) or are invaluable to our ecosystem and to human survival (A. mellifera). Therefore, knowledge of their neuropeptide signaling serves diverse interests, e.g., as an interesting target for pest control, or to aid in ways to promote the survival of biologically valuable species. Since several of these organisms lack a well-annotated reference genome, peptidomics is often the best option to screen for bioactive neuropeptides. On the other hand, knowledge of the integral peptidome of well-established (in)vertebrate model organisms such as C. elegans, D. melanogaster, D. rerio, R. norvegicus, and Mus musculus can provide indispensable information that may be extended to human disease studies. While both the rat and mouse peptidomes have been studied extensively [53,54,55], scientists continuously aim to optimize peptidomics work flows in order to unveil new peptides and advance insights into neuropeptidomes, as exemplified by recent work on dissected rat brains [52].

Increasing the knowledge on the C. elegans peptidome comes with the clear benefits of it being a seasoned model organism, in the sense that an as-complete-as-possible peptidome can quickly integrate with knowledge from diverse biochemical, cellular circuit, and functional studies; hence, it represents a valuable tool for researchers. It should be noted that while mature neuropeptides from both FLP and NLP precursors have been detected in vivo, the complete INS family remains fully based on in silico predictions [56]. Knowledge on how INS protein precursors are expressed and processed into mature INS peptides is thus lacking so far. Previous peptidomics studies in C. elegans have detected up to 75 neuropeptides in a single experiment [57]. Per replicate, we identified on average 170 neuropeptides, with a minimum of 160 and a maximum of 174. When pooling the data from all 10 replicates, we effectively identified 203 out of the 354 known and predicted C. elegans neuropeptides, and confirmed 131 neuropeptides with MS/MS. All of this was done with an optimized peptidomics pipeline, vastly increasing the amount of identifications compared with previous studies. While the identification of 72 neuropeptides, originating from 51 precursors, is based on MS1 data only (supplementary Tables S1 and S2), 10 of these precursors were found to contain multiple MS1 detected neuropeptides (24 in total), which makes it unlikely they were found by random chance.

MS profiling experiments do not allow subjecting the entire predicted C. elegans peptidome to LC-MS/MS analysis. Analyzing whole-mount C. elegans extracts implicates the presence of various metabolites, phospholipids, and proteins in the sample. Their ubiquity may push the lower abundant neuropeptides into the background, resulting in problematic ionization and low-quality fragmentation spectra. To overcome this, we enriched the sample for the 0.7–10 kDa mass range, as the bulk of C. elegans neuropeptides lie within this mass range. There are 33 predicted C. elegans neuropeptides that have a mass lower than 700 Da, which implies they are lost during size exclusion chromatography. Essentially, for discovery (profiling) experiments this is a necessary trade-off with the removal of metabolites from the samples. The maximum number of identifications that could theoretically be achieved with our peptidomics pipeline would therefore encompass 321 (354 minus 33) peptides of the predicted C. elegans peptidome, of which we were able to identify 64%. Altogether, we present an extension of the existing C. elegans peptidome with 35 new mature neuropeptide sequences, including eight neuropeptides originating from seven novel precursors. Thereby, we have expanded the number of C. elegans FLP/NLP neuropeptide precursors to 113, coding for a total of 391 mature neuropeptides. This number may still rise, as the presence of multiple (di)basic cleavage sites and sequence architecture in the seven novel precursors (supplementary Figures S8–S13) suggests the existence of yet to be detected neuropeptides.

Several reasons may underlie the fact that 118 predicted neuropeptides remain undetected in this study. First, the knowledge we have on C. elegans neuropeptides that have not been confirmed by peptidomics consists completely of in silico predictions. All amino acid sequences enclosed by (di)basic cleavage sites in a neuropeptide precursor are considered putative mature neuropeptides. It is plausible that not all sequences encoded in the precursor are processed to mature neuropeptides. In addition, when three adjacent basic amino acids are present, there are two potential cleavage products from proteolytic processing of the precursor that can be predicted. These peptides only differ by one N-terminal basic amino acid residue (K or R) that can be part of the mature peptide if the first two basic residues are used as a cleavage signal. Our peptidomics data reveals a preference for these peptides where the third basic amino acid of the string is part of the mature peptide (e.g., FLP-1, FLP-9, FLP-12, FLP-15, FLP-17), if only one mature peptide is observed. While this may reflect endogenous processing preference, this may also be a technical artefact due to the expectedly more effective ionization of peptides containing basic residues. Individual peptides may deviate from this general trend, and we also observed mature neuropeptides with and without the additional basic residue originating from the tribasic cleavage sites (e.g., FLP-8, FLP-14). Second, as they are modulators of behavior in response to changing environments, peptides may be expressed under specific conditions only. If the cue for their expression – be it internal or external – is absent under standard conditions, these peptides are likely not present in our reference samples. For instance, all NLPs of the nlp-29 gene cluster (nlp-27, nlp-28, nlp-29, nlp-30, nlp-31, nlp-34) are considered antimicrobial peptides that are upregulated after infection [58]. Therefore, it is no surprise that we could detect almost none of those peptides by sampling healthy worms. Finally, it is possible that certain neuropeptides did not reach the detection limit due to highly localized expression and/or low abundance: some peptides are only expressed in one cell and/or at very low levels.

This study presents the first evidence on the in vivo presence of NLP-53-, NLP-55-, NLP-56-, NLP-57-, NLP-58, and NLP-65-derived peptides. We added these putative NLP neuropeptide precursors to Wormbase. NLP-65 was predicted in silico by Koziol and coworkers, although without the C-terminal arginine in both peptides, which they assumed to be part of the cleavage site [37]. Up to now, there was no evidence from peptidomics studies to support their processing to yield mature bioactive peptides. The only recently added NLP-precursor for which we were unable to provide supporting MS-evidence would be NLP-54. Both C. elegans TRH-like peptides that are contained within NLP-54 have, however, a low molecular weight, thereby passing below the 700 Da mass cut-off that is used in our implemented size-exclusion chromatography. Due to their small size they most likely occur as singly charged ions, also escaping MS/MS detection [59, 60]. In addition, no mature peptides processed from NLP-59, -61, -62, -63, -66, -67, -68, -70, and -71 could be detected. The present mass spectrometric detection of the eight peptides contained within the NLP-53, 55-58, and NLP-65 precursors confirms the predicted coding potential of these corresponding peptide encoding genes.

In addition to identifying known or predicted peptides within FLP and NLP precursors, our MS/MS data also revealed 29 novel peptides which are processed from known FLP and NLP precursor proteins (supplementary Table S3). All these peptides are flanked by (di)basic cleavage sites and display conservation in other Caenorhabditis and/or nematode species, with the exception of the newly found NLP-42- and NLP-58-derived peptides. This conservation at least suggests that the peptides could be functionally relevant themselves. We confidently detected eight peptides that are processed from seven distinct proteins previously unannotated as neuropeptide precursors (Figures 2, 3, 4, 5, 6 and 7, supplementary Table S4). All encoded peptides are processed adjacent to (di)basic cleavage sites and the precursor proteins contain a signal peptide, suggesting that they are indeed putative preproproteins harboring mature neuropeptides. Most of these precursors are also enriched in neurons according to Wormbase, which further strengthens a putative function as neuropeptide precursors. The discovery of these novel precursor proteins by our peptidomics study underscores the continued need for wet-lab discovery in peptidomics.

The increased number of predicted and biochemically characterized neuropeptides in C. elegans presents an increasingly complete picture of the reference peptidome of this model organism. The accuracy and completeness of this reference are essential for follow-up (differential) peptidomic studies in C. elegans [61]. This model organism is a powerful model for neurobiological research, furthered even more by the complete mapping of its nervous system [33, 62]. Many neuropeptides are involved in modulating neural circuits, and their functional output often evokes interesting phenotypical changes in C. elegans [34, 63,64,65]. Given a complete picture of the C. elegans peptidome, it will also become possible to map in which neurons specific neuropeptides are present. On top of accumulating expression-based information, recent advances in mass spectrometry suggest that knowledge of peptide content of single neurons in C. elegans will become feasible. Since this may provide additional insights into the regulation of neuronal signaling in this model, we and others are working towards a future in which peptidomics can contribute to the study of neuropeptidergic neuromodulation on a single neuron level, using C. elegans as a research model.

Conclusion

Aiming to expand the C. elegans reference neuropeptidome, we are able to identify 203 neuropeptides based on mass matching, of which 131 were confirmed at the sequence level using LC-MS/MS. This includes neuropeptides in the newly annotated precursors NLP-53, NLP-55, NLP-56, NLP-57, NLP-58 as well as the previously unannotated protein R06F6.7 (now NLP-65), hitherto known by in silico predictions only. Furthermore, we found evidence for 29 putative additional neuropeptides present in known FLP and NLP precursors, as well as eight neuropeptides encoded by seven novel neuropeptide precursors, which were not yet annotated as such. We present both our in-house database of known and predicted peptides, as well as our most complete evidence for in vivo occurrence of these peptides under standard conditions to the scientific community. We hope this resource may be of valuable help in the biochemical and functional studies of C. elegans neurobiology.