NMR Analysis of C. elegans FLP-18 Neuropeptides: Implications for NPR-1 Activation

Abstract

FMRFamide-Like-Peptides (FLPs) are the largest neuropeptide family in animals, particularly invertebrates. FLPs are characterized by a C-to-N-terminal gradient of decreasing amino acid conservation. NPR-1 is a GPCR (G Protein Coupled Receptor) which has been shown to be a strong regulator of foraging behavior and aggregation responses in Caenorhabditis elegans. Recently, ligands for NPR-1 were identified as neuropeptides coded by the precursor genes flp-18 and flp-21 in C. elegans. The flp-18 gene encodes eight FLPs including DFDGAMPGVLRF-NH₂ and EMPGVLRF-NH₂. These peptides exhibit considerably different activities on NPR-1, the longer showing lower potency. We have used NMR and biological activity to investigate structural features that may explain these activity differences. Our data demonstrate that long range electrostatic interactions exist between N-terminal aspartates and the C-terminal penultimate arginine as well as N-terminal H-bonding interactions that form transient loops within DFDGAMPGVLRF-NH₂. We hypothesize that these loops, along with peptide charge, diminish this peptide's activity on NPR-1 relative to that of EMPGVLRF-NH₂. These results provide some insight into the large amino acid diversity in FLPs.

FMRFamide was first discovered in 1977 by Price and Greenberg as a cardioexcitatory peptide from the clam Macrocallista nimbosa (1). FMRFamide-Like-Peptides (FLPs) are the largest family of neuropeptides found in invertebrates (2-5), but mammalian (even human) FLPs have also been identified (6-9). These peptides are characterized by a C- to N-terminal gradient of decreasing sequence conservation, and most end in RF-NH₂. This is true when FLP peptide sequences as a whole, from within taxa, within species, or even on specific precursor proteins are compared (2, 4, 5, 10). The peptides, like most neuropeptides and hormones, are synthesized as part of larger precursor proteins and processed in the secretory pathway (11). Peptides on a particular precursor have conserved regions in the mature peptides that are often associated with receptor binding and make up a subfamily (4). For many neuropeptides, including FLPs, this region is the C-terminus. However, other peptide families have different patterns of conservation; for example, in insect orcokinins the N-terminus is the conserved region (12) and in insulin, the cystine framework and other central residues portions are conserved (13). Two examples of FLP precursor proteins, flp-18 from the nematode Caenorhabditis elegans (4) and afp-1 from the nematode Ascaris suum (14) are shown in Table 1.

Table 1.

Sequence comparison of flp-18 and afp-1 precursor proteins^a

^aPeptide sequences red, processing sites blue, * denotes amino acid gap in peptide, _ denotes gap. Note that this analysis is not an alignment (4, 14).

^bThe longer peptides for each precursor are underlined.

The first nematode FLP, AF1, was isolated from Ascaris suum (15), and most subsequent early nematode FLP work was done on this species (14, 16-21). FLPs are highly expressed in nematodes, and thus are likely important chemical components of their anatomically simple nervous systems (3, 5, 18). In C. elegans, 28 different genes encoding well over 60 possible FLPs have been identified using bioinformatic approaches (4, 5, 22) and 28 of the putative processed peptides have been detected biochemically (23-27).

FLPs are involved in a wide range of biological processes that have been reviewed previously (10, 28-31). Some of the more prominent functional studies have focused on their role in cardioexcitation (1), muscle contraction (19), modulation of the action of morphine (32), egg laying (33) and feeding behavior (31) in nematodes. Also, disruption of the flp-1 gene in C. elegans resulted in a number of phenotypes (34). Though much work has been done to elucidate the activities of FLPs, the definitive biological functions of the vast majority of FLPs are still unknown.

Two types of receptors for FLPs have been identified: GPCRs (G-Protein Coupled Receptors) (35-40) and a sodium channel gated by FMRF-NH₂ (41-43). Other human/mammalian ion channel receptors have been identified whose activities are modulated by FLPs, including the Acid Sensing Ion Channels (ASICs) and Epithelial Na⁺ Ion Channels (ENaCs) (8, 44). NPR-1, a GPCR that modulates feeding behavior in C. elegans, is activated by two subfamilies of FLPs in C. elegans, including the FLP-18 peptides and FLP-21 peptide (35). All of the FLP-18 peptides occur on the same precursor and are presumably processed and released simultaneously (4). The most active of these peptides is EMPGVLRF-NH₂; by comparison DFDGAMPGVLRF-NH₂ is significantly less active (35). These observations motivated us to further investigate the structural properties of these peptides relative to their biological activities.

In previous studies, we have suggested that N-terminal hydrogen bonding can influence FLP activity (45). Structural interactions in small peptides such as FLPs are generally invisible to techniques such as X-ray crystallography, because small peptides are dynamic in solution. Also, some NMR parameters such as NOE correlations for distance measurements are of limited value on small peptides due to their dynamic properties in solution (46). However, these structural properties can modulate their activities (45). Although a high resolution X-ray crystallographic structure for a FLP-18 peptide bound to NPR-1 would be extremely useful, it would not provide any information on the unbound state of the peptides. As shown below, we have identified significant structural differences between different unbound FLP-18 peptides, and this work seeks to illuminate the relationship between free ligand conformations and their activities on the NPR-1 receptor.

In the present study we have used NMR for pH titrations, temperature titrations, and chemical shift analyses to identify transient long-range interactions within flp-18 peptides and designed analogues. The sequence and activity diversity among these peptides have motivated us to examine the structural properties of two extreme cases. We have focused on structural features of EMPGVLRF-NH₂ and DFDGAMPGVLRF-NH₂ that may influence the activity of each peptide on NPR-1. The material presented in this paper examines the hypothesis that local structure in the variable N-terminal regions of flp-18 peptides can modulate their binding to NPR-1.

Experimental Procedures

Peptide Synthesis

Peptides listed in Table 2 were synthesized using standard Fmoc solid phase methods, purified by HPLC, and verified by MALDI-TOF mass spectrometry at the University of Florida Interdisciplinary Center for Biotechnology Research (UF ICBR) protein core facility.

Table 2.

Peptides used in this study

Name	Sequence a	% Response (n)b
Peptide 1c	EMPGVLRF-NH2	100
Peptide 2d		29.1±5.7 (16)
Peptide 3		0 (3)
Peptide 4		19.0±2.6 (8)
Peptide 5	SGSGAMPGVLRF-NH2	118.7±11.0 (4)
Peptide 6	AAAAAMPGVLRF-NH2	62.4±3.2 (10)
Peptide 7e		49.3±16.5 (4)
Peptide 8		0 (3)
Peptide 9f		45.0±10.5 (6)
Peptide 10g		104.8±2.8 (8)
Peptide 11	PGVLRF-NH2	43.0±5.5 (14)
Peptide 12	PGVLRFPGVLRF-NH2	198.1±33.3 (10)

^aNaturally occurring sequences are underlined. The conserved PGVLRF-NH2 sequence is in bold. N-terminal “extension” sequences of native long flp-18 peptides are bold and: Red for C. elegans based sequences and Blue for A. suum based sequences.

^bPeptides (10^-6 M) were applied in 2 m pulses to Xenopus oocytes expressing NPR-1 215V. Results expressed as a % of response to 10^-6 M Peptide 1 (EMPGVLRF-NH₂) +/- SEM

^cMost active native C. elegans FLP-18 peptide

^dLongest and least active native FLP-18 peptide

^eLongest Ascaris AFP-1 peptide

^fChimera of long FLP-18 + long AFP-1

^gChimera of long AFP-1 + long FLP-1

Peptide Sample Preparation

Lyophilized peptides were weighed and dissolved to ∼1 mM in 95% H₂O and 5% D₂O, and the pH was adjusted to 5.5. These were aliquoted and frozen at -20 °C until being used for biological assays or NMR experiments. Small aliquots of each sample were submitted for amino acid analysis at the UF ICBR protein core to determine more accurate concentrations. For NMR spectroscopy, the pH-stable chemical shift standard DSS (2,2-dimethyl-2-silapentane-5-sulfonic acid) was added to 600 μL aliquots to a final concentration of 0.17 mM.

Biological Activity Assays - Expression in Xenopus laevis oocytes

Sense cRNA was prepared in vitro using the mCAPTM RNA Capping Kit (Stratagene, La Jolla, CA) from plasmid DNA containing full-length npr-1 215V cDNA cloned in pcDNA3 (Invitrogen Ltd., Paisley UK). RNA transcripts were synthesized using T7 RNA Polymerase (Stratagene, La Jolla, CA) after linearizing the plasmid with Apa I (Promega UK, Southampton, UK) and blunting the 3′ overhangs with T4 DNA Polymerase (Amersham Pharmacia Biotech, Little Chalfont, Bucks, UK). T7 RNA transcripts synthesized in vitro with the mCAPTM RNA Capping Kit are initiated with the 5′ 7MeGpppG 5′ cap analog. Sense cRNA was prepared in a similar manner from the GIRK1 and GIRK2 clones in pBS-MXT (47) (kindly donated by Drs. S.K. Silverman and H.A. Lester, California Institute of Technology, Pasedena, USA) after linearizing the plasmid with Sal I (Promega).

All experiments using Xenopus laevis were carried out under a Home Office (UK) Project License. Stage V and VI oocytes from virgin female adult X. laevis were prepared using standard procedures (35, 48-50). Oocytes were then injected with 50 ng of npr-1 receptor sense cRNA, either alone, or together with 0.5 ng each of GIRK 1 and GIRK 2 sense cRNA and incubated at 19oC for 2 - 5 days. Uninjected oocytes were used as controls. Electrophysiological recordings were made from oocytes using a two-microelectrode voltage-clamp technique, (35, 48-50).

NMR Spectroscopy

NMR data were collected at 600 MHz on a Bruker Advance (DRX)-600 console in a 14.1 Tesla magnet equipped with a 5 mm TXI Z-Gradient CryoProbe. Unless otherwise stated, all NMR experiments were collected at 288 K, and spectra were collected with a 6600 Hz spectral width and were referenced by setting the methyl proton resonance peak from DSS protons to 0.0 ppm. The ¹H carrier frequency was centered on water which was reduced using a WATERGATE sequence (51) or presaturation. Two-dimensional TOCSY (52) experiments were collected using a DIPSI-2 mixing sequence with a 60 ms mixing time. Two-dimensional ROESY (53) experiments were collected using a 2.27 kHz cw field spinlock applied for 250 ms.

Processing of 1D NMR spectra and creation of stack plots of pH and temperature titrations was done using Bruker XWINNMR and XWINPLOT ver 4.0 software. Two-dimensional NMR datasets were processed with NMRPipe (54) using standard methods: removing residual water by deconvolution, multiplying the data with a squared cosine function, zero-filling, Fourier transformation, and phase correction. Data were analyzed and assigned with NMRView (55) using standard ¹H-based methods (56).

One-dimensional pH titration experiments were performed for all peptides in Table 2 that contain aspartate and/or glutamate residue(s), as well as PGVLRF-NH₂ and SGSGAMPGVLRF-NH₂ as controls. One-dimensional NMR spectra were collected at increments of about 0.2 pH units from 5.5 to 1.9 by adding 1-3 μL of 0.01-0.1 M HCl for each pH value. pKa values and effective populations (c in Equation 1) of pH dependant resonance peaks were calculated using Origin 7.0 software and a modified version of the Henderson-Hasselbach equation below as previously described (57):

$$\begin{array}{c}\delta (pH)=\sum _{i=1}^j\frac{c_i({\delta }_a-{\delta }_b)\times {10}^{p{K}_{ai}-pH}}{1+{10}^{p{K}_{ai}-pH}};\sum _{i=1}^j{c}_i=1;\\ j=1\text{or}2\end{array}$$ Equation 1

where δ(pH) is the experimental chemical shift, δ_b is the chemical shift at the least acidic condition, δ_a is the chemical shift at the more acidic condition, pK_ai is the negative common log of the acid/base equilibrium constant for the i^th titration event, and c_i is the contribution of the i^th titration event to the total pH dependence of chemical shift.

One-dimensional NMR temperature titrations were collected on a standard TXI probe at 5 Kelvin (K) increments from 278 to 328 K, then ramped back to 278 K to check for sample integrity. The temperature for each experiment was calibrated using methanol (for 278.15 – 298.15 K) and ethylene glycol (for 308.15 – 328.15 K) and the corrected temperatures were used for the determination of all temperature coefficients (TC).

Results

Peptide Design Rationale and Physiological Responses

Three major considerations have motivated this study. First, we have been intrigued for some time by the amino acid diversity in FLPs (2, 10, 14-16, 45, 58, 59). In particular, as described above, FLPs display patterns of decreasing amino acid conservation from the C- to the N-termini, and the comparison of the C. elegans flp-18 (4) and A. suum afp-1 (14) genes suggests that the longer peptides produced by these genes are unique (see Table 1). Second, the activity at NPR-1 of the long FLP-18 peptide, DFDGAMPGVLRF-NH₂, is significantly lower than the shorter EMPGVLRF-NH₂ (35). Finally, in previous work on FLPs from mollusks, we found that different amino acid substitutions significantly changed the conformations of the peptides (45, 58) and that these conformational differences are correlated with their differences in activity (60).

In designing the peptides for this study, we considered several possibilities to explain the difference in NPR-1 activity between two native FLP-18 peptides, EMPGVLRF-NH₂ and DFDGAMPGVLRF-NH₂: First, the N-terminus could have intrinsic activity or act as a competitive inhibitor; second, a glutamic acid might be required in a position corresponding to the first residue of the more active EMPGVLRF-NH₂; third, the added bulk due to the extra amino acids could prevent the active portion of the peptide from efficiently binding to NPR-1; fourth, the N-terminal extension of DFDGAMPGVLRF-NH₂ could be involved in structural interactions that cause it to be less potent on NPR-1 than EMPGVLRF-NH₂. To address these possibilities, two native FLP-18 peptides, and a range of substituted and derived analogues (Table 2), were tested for their ability to activate the NPR-1 215V receptor expressed as described in the Experimental Methods. In the following, we use the term “activity” to indicate the magnitude of the potassium current evoked by a 10^-6 M pulse of peptide as a percentage of the response of the same oocyte to a 10^-6 M control pulse of peptide 1 (EMPGVLRF-NH₂). We will refer to peptides by their peptide number shown in Table 2.

It was observed that that the long native FLP-18 peptide 2 was much less effective than the shorter FLP-18 peptide 1 at activating the receptor (Table 2), confirming previous observations (35). To test if the N-terminus of peptide 2 could have intrinsic activity or act as a competitive inhibitor, we designed and analyzed the effect of peptide 3. This peptide had no intrinsic activity on the receptor (Table 2) and did not block the effects of 1 μM pulses of peptide 1 (n = 3) (data not shown). To test the possibility that a glutamic acid might be required in a position corresponding to the first residue peptide 1, we examined peptide 4, where a glutamic acid residue is substituted for the alanine at position 5 in peptide 2. However, this substitution did not improve, and in fact weakened, the effectiveness of the long peptide. It also seemed possible that the added bulk of peptide 2 due to the extra amino acids could be preventing access to the NPR-1 binding site. Thus, we analyzed peptides 5 and 6. Peptide 5 was designed to eliminate any potential structure in the N-terminus based on commonly used flexible linker sequences in fusion protein constructs (pET fusion constructs, Novagen Inc.). The N-terminus of peptide 6 was designed to induce a nascent helical structure in the same region (61, 62). It can be seen that peptide 5 completely restored activity compared to peptide 2, whilst peptide 6 only partially restored activity in comparison with the short native peptide.

As shown in Table 1 the longest native peptide from afp-1 in A. suum (19), peptide 7, is two amino acids longer than the corresponding longest peptide from the C. elegans flp-18 gene (4), peptide 2. Thus, we also synthesized and tested peptide 7, as well as its N-terminus, peptide 8. Peptide 7 was slightly more effective than the peptide 2. However, the short N-terminal peptide sequence was again inactive (Table 2) and did not block the effects of 1 μM pulses of peptide 1 (n = 3) (data not shown). In addition, we also made chimeras of the long C. elegans and A. suum sequences, peptides 9 and 10. Peptide 9, in which two extra amino acids (SM) are introduced into the center of peptide 4 to give it the same number of amino acids as peptide 7, showed similar activity to that of the long native Ascaris peptide itself, GFGDEMSMPGVLRF-NH₂. However, peptide 10 showed similar activity to that of peptide 1.

As shown later, our results indicate that the conserved C-terminal PGVLRF-NH₂ is largely unstructured in solution. Thus, we tested peptide 11 and also peptide 12, in which the conserved sequence was duplicated. The activity of the peptide 11 was less than that of peptide 1 and similar to that of peptides 7 and 9. When compared to peptides 1 and 5, the reduced activity of peptide 11 could indicate that methionine preceding proline is important for activity. However, the C-terminal duplicated peptide with no methionine was approximately twice as active as peptide 1.

To further investigate the effects of changing the structure of the N-terminal sequence of the C. elegans long FLP-18 peptide, we determined full dose response curves for the peptides 1, 2, and 5 (Fig. 1).

Figure 1.

Dose response curves of select FLP-18 peptides. Peptides were applied to Xenopus oocytes expressing NPR-1 215V, and the responses from inward rectifying K+ channels were recorded and normalized to the response of peptide 1 (EMPGVLRF-NH₂) at 10^-6 M. Filled circles are peptide 1 (EMPGVLRF-NH₂) (EC₅₀ = 10^-6.80 M), open squares are peptide 5 (SGSGAMPGVLRF-NH₂) (EC₅₀ = 10^-6.12 M), and open circles are peptide 2 (DFDGAMPGVLRF-NH₂) (EC₅₀ = 10^-5.28 M). Three measurements at each peptide concentration were obtained and results are shown +/- SEM.

From Figure 1, both peptides 2 and 5 are less potent compared to peptide 1. This suggests that elimination of structure at the N-terminus of peptide 2 can increase its potency on the receptor. Also, peptides 2 and 5 are more efficacious at higher concentrations than peptide 1, suggesting that longer peptides might be more efficacious on NPR-1 than shorter peptides.

NMR Chemical Shifts Reveal Regions of flp-18 Peptides with Significant Structure

Chemical shifts are extremely sensitive to molecular and electronic environments and thus provide unique atomic probes in molecules. Specifically, in peptides and proteins, chemical shifts of many nuclei along the polypeptide backbone have been shown to be dependent on secondary structure (46, 63-68). Thus, the first step in NMR analysis is the assignment of resonance peaks in spectra to atoms in the molecule. For all peptides in Table 2, nearly complete NMR resonance assignments were made using standard two-dimensional ¹H-based methods (56) (Supplementary Material S2).

Short, linear peptides are often very dynamic, lack a 3D hydrophobic core, and interconvert rapidly between many different conformations. Despite this inherent flexibility, numerous studies have demonstrated that regions of short peptides can be highly populated in specific types of secondary structure (69-72). A fundamental hypothesis of this study is that differences in local structure of variable N-termini of free FLPs could partially explain differences in their potencies on receptors.

In order to compare chemical shifts from one peptide to another and to identify regions that contain significant populations of secondary structure, it is useful to compare experimental chemical shifts to random-coil values (64-66, 68). Fig. 2 plots the difference between experimental and random-coil values for some of the peptides analyzed in this study. The white and black bars represent deviations from random-coil values for amide and alpha protons, respectively, and the magnitude of the deviations reflects the population of local structure along the backbone of the peptides (64, 68). All data were compared to random coil values at pH 2.3.

Figure 2.

NMR chemical shift deviations from random coil values. Experimental chemical shifts at pH ∼2.3 were subtracted from sequence corrected random coil values (64, 65). Filled and open bars represent alpha and amide protons, respectively.

Several features in Figure 2 are worth noting. First, the chemical shifts of residues in the conserved PGVLRF-NH₂ regions of each peptide are close to random-coil values, and rather similar among all the peptides examined. This suggests that this conserved sequence is unstructured in solution and that flexibility is important for binding to NPR-1. This flexibility may help the ligand diffuse/maneuver more effectively into a binding pocket on the receptor. Second, the N-terminal extension of peptide 2 shows significant deviation from random-coil values. In particular, the G4 amide proton has a very large deviation, suggesting significant structure. Third, the chemical shift deviations of amide and alpha protons of peptides 3 and 8 are nearly identical to the corresponding regions of the full-length peptides 2 and 7, respectively. This indicates that these N-terminal extensions are behaving as independent structural units. Finally, peptide 5, designed to lack N-terminal structure, indeed shows a consistent very small deviation from random-coil values in its first five residues.

pH Dependence of Amide Proton Chemical Shifts Reveal Regions of flp-18 Peptides with Significant Structure

The sensitivity of NMR chemical shifts to electronic structure and hydrogen bonding make them ideal probes of longer-range interactions with titratable side-chains. NMR studies of peptides that utilize amide protons often need to be conducted below ∼pH 6 to prevent amide proton exchange (56, 73). By varying the pH from about 5.5 to 2, both aspartic and glutamic acid side-chains will be converted from negatively charged and deprotonated to neutral and protonated. These different charge states of the carboxylate groups will produce changes in the electronic environment in interacting atoms proportional to 1/R³, where R is the distance between the charged group and chemical shift probe. Backbone amide resonances are particularly sensitive to interactions such as H-bonding (57, 66) and thus provide ideal probes of long-range interactions with side-chain carboxylates. This phenomenon provides a powerful mechanism to study long-range hydrogen bonding and salt bridge interactions in small peptides (45, 73, 74).

Many of the FLP-18 peptides contain aspartic or glutamic acids, so we performed 1D NMR pH titration experiments on all peptides in Table 2 containing these residues and, as controls, on peptides 5 and 11, neither of which showed any pH dependence in the proton chemical shifts. Stackplots of the amide region for a representative set of peptides are shown in Fig. 3.

Figure 3.

Amide region of one-dimensional NMR data, collected as a function of pH from about 1.9 to 5.5. Peaks are labeled with their assigned amino acids and the panels correspond to the following peptides: (ADD peptide sequences here as well) A. Peptide 3 = DFDGAM-NH₂, B. Peptide 2 = DFDGAMPGVLRF-NH₂, C. Peptide 8 = GFGDEM-NH₂, D. Peptide 7 = GFGDEMSMPGVLRF-NH₂, E. Peptide 5 = SGSGAMPGVLRF-NH₂, F. Peptide 1 = EMPGVLRF-NH₂. pH dependent interactions are summarized in Figure 5, and complete pKa analyses are provided in supplementary material.

No chemical shifts in peptide 5 (Figs 3E and 4E) or peptide 11 (data not shown) have any pH dependence, demonstrating that backbone amide proton chemical shifts are not intrinsically pH-dependent in this pH range. Second, several resonances in other peptides have large pH-dependent shifts. To our knowledge no systematic study has been undertaken to identify the maximum change in chemical shift of backbone amide protons as a function of pH, but Wüthrich and coworkers showed that in a small protein with a well-defined (∼70-90% populated) hydrogen-bond between an aspartic acid side chain and a backbone amide proton led to a change of 1.45 ppm over the titratable range of the aspartic acid (75). Thus, in Figure 3, some amide proton resonances of non-titratable amino acids have pH-dependent shifts that are characteristic of significant H-bond interactions. Others have smaller pH-dependent changes, suggesting either more transient dynamic interactions or much longer and weaker H-bonds. In contrast, several resonances in peptides with titratable groups show little or no pH dependence, showing that these effects are relatively specific. Next, the spectra from the N-terminal truncated peptides 3 and 8 are highly dependent on pH and are nearly perfect subsets of the same regions in their full length counterparts. Consistent with chemical shift data in Figure 2, this demonstrates that the N-terminal extensions of peptides 2 and 7 behave as independent structural units. The extensions also do not interact significantly with the more C-terminal backbone atoms, which show relatively little pH dependence, indicating that the conserved C-termini are less structured than the N-terminal extensions.

Figure 4.

Arg H^ε region of one-dimensional NMR data, collected as a function of pH from 1.9 to 5.5. These long-range interactions on the penultimate C-terminal Arg result from the titratable carboxylate groups on the N-termini. pH dependent interactions are summarized in Figure 5, and complete pKa analyses are provided in supplementary material. Legend: A. Peptide 4 = DFDGEMPGVLRF-NH₂, B. Peptide 2 = DFDGAMPGVLRF-NH₂, C. Peptide 10 = GFGDAMPGVLRF-NH₂, D. Peptide 7 = GFGDEMSMPGVLRF-NH₂, E. Peptide 5 = SGSGAMPVLRF-NH2, F. Peptide 1 = EMPGVLRF-NH₂

pH Dependence of Arginine Side-Chains Reveal Long-Range Interactions

The penultimate arginine residue is highly conserved and found in the same position in all FLPs. This arginine is at least 7 residues away from any carboxyl groups, so we were surprised to find in several FLP-18 peptides that its epsilon proton (Arg H^ε) is pH dependant (Fig. 4).

Peptide 5 demonstrates that there is no intrinsic pH dependence Arg H^ε over the pH range investigated, and we conclude that in other peptides there are long-range interactions between the Arg and the N-terminal carboxylates. Such an interaction would indicate a non-covalent ring structure. These interactions show up in most of the FLP-18 analogues having N-terminal carboxyl sidechains and, at first glance, do not appear to relate to the activity of the peptides (Table 2). For example, peptide 1 (one of the more active peptides) has nearly the same Arg H^ε pH dependence as peptide 4 (the least active PGVLRF-NH₂ containing examined). Moreover, peptide 10, with similar activity on to peptide 1, has nearly no Arg H^ε pH dependence. Thus, Arg interaction with acidic residues alone is not sufficient to explain the difference in activity among the FLP-18 analogues tested.

Quantitative Determination of pKa Reveals Multiple Interactions

Several of the peptides in Table 2 have more than one carboxylate, so it is not always obvious which is responsible for the pH dependence of a particular resonance. If the titrating groups have distinct pKa values, then it should be possible to determine the contribution of each carboxylate on each titrating resonance using Equation 1. Every peak that exhibited pH-dependent chemical shifts was fitted using first one, then two, then three pKa values. In all cases we used the minimum number of interacting pKa values to get a good quality of fit and maximum linear regression coefficient (R²) to the experimental data. In the peptides with three titrating groups, inclusion of three interacting groups in the calculation did not improve the fits more so than including only two.

The complete table of relative pKa contributions (c from Eq 1) and pKa values are provided as supplementary material (Table S1), and the interactions are represented graphically in Fig. 5. As we discuss below, the interactions between titrating groups and resonances in these peptides is rather complicated and dynamic. The data presented here illustrate that, though there is a heterogeneous ensemble of H-bonding interactions between various backbone amide protons, certain ones are prominent.

Figure 5.

Proposed H-bonding Interactions Between Backbone Amide Protons and Carboxyl Side-chains: DFDGAM-NH₂ = Peptide 3, DFDGAMPGVLRF-NH₂ = Peptide 2, DFDGEMPGVLRF-NH₂ = Peptide 4, SGSGAMPGVLRF-NH₂ = Peptide 5, EMPGVLRF-NH₂ = Peptide 2. Each H-bond acceptor residue is color coded to match the arrows leading from it to its H-bond donors. The arrow widths are proportional to the relative extent to which that particular interaction affects the chemical shift of the amide proton at the point end of the arrow. The bar plots show the temperature coefficient of the backbone amide proton resonances

Using the pKas calculated for the pH dependent resonances for the peptides in this study we can assign most H-bonding interactions between titrating carboxyl side-chains and either backbone amide or Arg H^ε protons. Figure 5 also illustrates the relative strength of these interactions. The most significant interaction (the largest shift from a long range interaction) is from a hydrogen bond between the D1 carboxylate and G4 amide in all peptides containing the N-terminal DFDG sequence. It contributes 40% to the observed titration of the G4 amide in peptides 2 and 3, and 55% to that of peptide 4 (supplementary data). The calculated pKa of D1 (∼3.0) is significantly lower than that of D3 (∼4.0), indicating that D1 is likely interacting with the positively charged amino terminus and stabilizing its negative charge. This is also seen in peptide 1, as E1 also has an unusually low pKa (∼3.5). Additional support for this pKa assignment comes from the pKa of the alpha protons of D1 and D3 of peptide 3 and D1 of peptide 2, which are 3.23, 4.09, and 2.97, respectively (supplementary data).

The interactions observed from pH titrations of peptides beginning in GFGD are different from and less substantial than those beginning in DFDG. For example, the largest pH dependant chemical shift change of the amide proton of a non-acidic amino acid for peptide 7 is ∼0.12 ppm (G3), whereas G4 in peptide 2 is ∼0.28 ppm. Also, the arginine sidechain of peptides 7 and 9 show a rather small chemical shift change in the pH titration (Fig. 4D) compared the same resonance for peptides 2 and 4. Additionally, the D4 sidechain of GFGD containing peptides seems to interact primarily with backbone amides N-terminal to it (Supplemental Material S1). This is a different conformation entirely than that observed in DFDG containing peptides, where D1 has a substantial interaction with G4.

Temperature Dependence of Amide Chemical Shifts Corroborates Regions with H-Bonding

Although complicated and often over-interpreted, the temperature dependence of amide proton chemical shifts in polypeptides can be associated with hydrogen bonding (46, 76). Additionally, some peptides analyzed here lacked carboxyl side-chains, so pH titration results were not valid in determining possible structural interactions in these peptides. We therefore measured temperature coefficients (TCs) for several relevant peptides (Fig. 5). A rough guideline to interpreting TCs is that an absolute value less than 4 indicates an internal hydrogen bond, values between 4 and 6 indicate weak hydrogen bonding, and values greater than 6 are not involved in hydrogen bonding (46, 76, 77). The magnitudes of the temperature coefficients for all amide protons in this study are inversely correlated with the magnitude of chemical shift pH dependence for those resonances (Fig. 6), which is consistent with H-bonding interactions as described above.

Figure 6.

Relationship between Temperature Coefficients and pH dependence of chemical shift among Backbone Amide Protons: Plotted here is the chemical shift change with pH vs with Temperature for backbone amide resonances. R² for the linear fit is 0.58. All data from peptides 1-4 are represented.

Overall Peptide Charge is Correlated With Activity on NPR-1

The experimental data presented above demonstrate that acidic residues in the N-terminal regions of FLP-18 peptides can interact with numerous amide protons and the conserved penultimate Arg. Although there are many additional factors influencing activity as addressed below, there appears to be a qualitative relationship between their charge properties (particularly of the N-terminus) and activities on NPR-1. This relationship is demonstrated in Figure 7 where the overall net charge at pH 7 of the entire peptide is plotted against its activity on NPR-1.

Figure 7.

Relationship between overall peptide charge and activity on the NPR-1 receptor. The overall peptide charge at neutral pH is plotted against activity for all of the peptides analyzed in this study. For the linear fit, R² = 0.67. Legend: Filled circle = 4, Open Circle = 2, Filled Square = 9, Open Square = 5, Filled Triangle = 6, Open Triangle = 7, Filled Diamond = 1, Open Diamond = 10, Filled Hexagon = 11, Open Hexagon = 12 (numbers correspond to peptide numbers in Table 2)

Discussion

The goal of this work has been to determine the conformational properties of unbound FLP-18 neuropeptides from C. elegans and how these may affect their potencies on NPR-1. The starting point for this study was the knowledge that two of the peptides encoded by the flp-18 gene have significantly different potencies on NPR-1 (35). The major findings reported above can be summarized as follows:

• The backbone of the conserved PGVLRF-NH₂ is predominantly unstructured.

• DFDG forms a structural loop stabilized by H-bonding.

• Another loop forms when N-terminal acidic residue(s) interact with the conserved C-terminal penultimate arginine side-chain.

• The DFDG loop interacts with the second loop to form a dynamic bicyclic structure that might influence binding to NPR-1.

• Charge also affects the activity of FLP-18 peptides on NPR-1.

The backbone structure of the conserved PGVLRF-NH₂ is predominantly unstructured

All NMR structural parameters measured in this study for the PGVLRF-NH₂ region of FLP-18 peptides indicate that the peptide backbone of this conserved sequence is predominantly unstructured. The only significant evidence for any kind of structural motif is the interaction between the conserved penultimate arginine side-chain and acidic residues in the N-termini. These results suggest that the primary receptor-binding region of FLP-18 peptides is highly flexible before interacting with NPR-1.

DFDG forms a structural loop stabilized by H-bonding

We observe transient H-bonding and ionic interactions within FLP-18 peptides beginning in the sequence DFDG. Specifically, acidic residues in the variable N-termini form substantial H-bonds to backbone amides N-terminal to the conserved proline (Figures 5 and 8).

Figure 8.

Model of interactions thought occurring within native FLP-18 peptides. This figure shows the most significant H-bonding interactions supported by NMR data. A: For DFDGAMPGVLRF-NH₂, H-bonds between the D1 side-chain carboxylate and G4/A5 backbone amide protons as well as H-bonding/ionic interaction between the D3 side-chain by red dashed lines. The N-terminal loop structure implicated in inhibiting binding to NPR-1 is circled in black. B: EMPGVLRF-NH₂ is shown with the most significant H-bonding and ionic interactions for which we have evidence. Notice that it has no N-terminal loop, in contrast to DFDGAMPGVLRF-NH₂. Also, the same unstructured region for both peptides is shown in ribbon view. The N- and C- termini are also labeled on both peptides.

In the DFDG containing peptides, G4 has the smallest temperature coefficient of all amides in the study; this is characteristic of involvement in a significant H-bonding interaction (46, 76, 78). This phenomenon is particularly prominent in peptide 4, where the D1 pKa rather than that of D3 is the most significant contributor to the G4 amide proton titration. It is also the least active PGVLRF-NH₂ containing peptide tested. Also, weak ROESY peaks were observed between D1 beta protons and G4 alpha protons in both peptides 2 and 3 (data not shown). This further corroborates the pH titration results that indicate significant long-range H-bonding between the D1 sidechain and G4 backbone amide proton of peptides beginning with DFDG. In contrast, the N-terminal SGSG region of peptide 5 is unstructured based on our NMR results, and is one of the most active peptides analyzed.

The DFDG loop may interact with the second loop to form a dynamic bicyclic structure which reduces binding to NPR-1

There is no direct or simple correlation between the activity data and any one set of NMR data. However, the two carboxylate residues in peptides 2 and 4 allow both the N-terminal loops as well as the ionic interaction between the conserved arginine and the aspartates (Fig. 8A). The increased activities of peptides 7 and 9, along with their apparent weaker interaction between the penultimate arginine and acidic residues relative to peptides 2 and 4, illustrate that the residues SM inserted in the middle of these peptides can interfere with loop formation between the N- and C- termini. FLP-18 peptides are short and flexible, and both loop interactions are likely dynamic. However, there is a possibility that the bulk of the N-terminal loop in DFDG containing peptides is brought into proximity of the conserved receptor-binding region by the action of the second loop involving the penultimate arginine. We propose that this bicyclic structure reduces binding to NPR-1.

Charge is also important in determining the activity of flp-18 peptides on NPR-1

There is a significant correlation between charge and activity such that more positively charged peptides tend to activate NPR-1 better than more negatively charged ones. Interestingly, the vast majority of predicted FLPs in C. elegans tend to be positively charged (4), including the peptide encoded by flp-21, which has an overall charge of +3 and is active on both naturally occurring isoforms of NPR-1 (215F and 215 V). However, peptides 4 and 9 have the same charge but different activities on NPR-1. The acidic residues of peptides 4 and 9 differ substantially in their interaction with the C-terminal arginine. This is likely due to the insertion of the residues SM in the middle of peptide 9. Thus, the N-terminal DFDG loop in peptide 9 does not interact well with the penultimate arginine, whereas that of peptide 4 does. This further supports the bicyclic model and the affect of a two-loop conformation on the activities of DFDG containing peptides.

Peptides 6 and 12 were often outliers in our attempts to correlate specific NMR data parameters to activity results. Peptide 6 was designed to possess a helix in the N-terminus, and we predicted reduced binding to NPR-1 resulting in activity similar to that of peptide 2. This prediction was incorrect, and peptide 6 had more activity than peptide 2. However, with no carboxylates, peptide 6 lacks the ability to form sidechain mediated H-bonding loops, which our model suggests should give it an activity more like that of peptides 1 and 5. Thus, the activity of peptide 6 (intermediate between peptides 1 and 2) suggests that other properties of its structure modulate its potency.

Peptide 12 unexpectedly had nearly exactly twice the activity of peptide 1. It is composed of two copies of the conserved PGVLRF sequence that is responsible for FLP-18 activity on NPR-1. Previous studies on FLP receptors show that the C-terminal amide group is necessary for activity (40), so it is extremely unlikely that the C-terminal PGVLRF in peptide 12 can interact with the active site of NPR-1. However, this peptide is also the most positively charged of all among those tested. This is consistent with our observation that a peptide's charge influences its activity on NPR-1.

Both native FLP-18 peptides in this study, DFDGAMPGVLRF-NH₂ and EMPGVLRF-NH₂, differ in both potency and efficacy. We have shown that N-terminal structure, peptide charge, loop formation and backbone flexibility in PGVLRF-NH₂ all modulate the activity of FLP-18 peptides on NPR-1. One interesting feature of the dose response curves in Figure 1 is that the two longer peptides have a larger maximal response and a steeper linear portion than the shorter peptide. This suggests that the native peptides could induce different configurations of the NPR-1 receptor with different abilities to couple the G-protein pathway under study and perhaps to other additional second messenger pathways as yet untested (79-82). Both the A. suum and C. elegans long peptides have been isolated (19, 23), demonstrating that these exist in vivo. However, other studies (83, 84) have shown that many peptide degradation products can also be found in cells. Perhaps multiple forms of FLP-18 peptides could shape the behavioral response to NPR-1 activity in a way that could not be achieved by any one peptide alone. It is possible that the ensemble of peptides functions as a bouquet to achieve a unique, beneficial, fine tuned response (2, 45).

Supplementary Material

si20060225_113. Supporting Information Available.

S1: Calculated pKas for all pH dependent resonances in all peptides examined by NMR pH titration and calculated proportion contributions for each pKa to each pH dependant resonance.

S2: NMR chemical shift assignments for all peptides examined by NMR

The abbreviations used are

FMRFamide

Phe-Met-Arg-Phe-NH₂

FLP

FMRFamide-Like-Peptide

NMR

nuclear magnetic resonance

AAA

Amino Acid Analysis

TOCSY

total correlation spectroscopy

NOESY

nuclear Overhauser effect spectroscopy

ROESY

rotating frame Overhauser effect spectroscopy

D₂O

deuterium oxide

GPCR

G-Protein Coupled Receptor

C. elegans

Caenorhabditis elegans

TC

Temperature Coefficient

K

Kelvin

NPR-1

neuropeptide resemblance receptor-1

flp-n

gene name

flp-n

precursor protein

FLP-N

peptide name