Miquel Adrover*ab,
Pilar Sanchisab,
Bartolomé Vilanovaab,
Kris Pauwelscd,
Gabriel Martorelle and
Juan Jesús Pérezf
aInstitut Universitari d'Investigació en Ciències de la Salut (IUNICS), Departament de Química, Universitat de les Illes Balears (UIB), Ctra. Valldemossa km 7.5, E-07122, Palma de Mallorca, Spain. E-mail: miquel.adrover@uib.es
bInstituto de Investigación Sanitaria de Palma (IDISPA), Ctra. Valldemossa, 79, E-07010, Palma de Mallorca, Spain
cStructural Biology Brussels, Vrije Universiteit Brussels (VUB), Pleinlaan 2, 1050 Brussels, Belgium
dStructural Biology Research Centre, Vlaams Instituut voor Biotechnologie (VIB), Pleinlaan 2, 1050 Brussels, Belgium
eServeis Científico-Tècnics, Universitat de les Illes Balears (UIB), Ctra. Valldemossa km 7.5, E-07122, Palma de Mallorca, Spain
fDepartament d'Enginyeria Química, Universitat Politecnica de Catalunya (UPC), ETSEIB, Av. Diagonal, 647, E-08028, Barcelona, Spain
First published on 15th September 2015
Neuromedin C (NMC) is a peptide that regulates various processes in the central nervous system and gastrointestinal tract through its interaction with the bombesin receptor subtype-2 (BB2R). Hence, BB2R antagonists hold the potential to treat disorders that occur as a result of NMC dysfunction or misregulation. However, their efficient design requires a detailed understanding of the structural features of NMC, which hitherto are unknown. Herein, we describe the conformational ensembles of NMC in an aqueous solution, at five different TFE concentrations to decode its folding pathway and under its SDS micelle bound state. NMC displays a disordered but well defined backbone architecture that undergoes a progressive coil-helix transition with increasing TFE concentration, first at the C-terminus and then at the N-terminus. NMC also adopts a C-terminal α-helical conformation upon binding to SDS micelles. This micelle binding is directed by hydrophobic interactions that concur with the unfavourable deprotonation of His8 and its further insertion into the micelle. Moreover, NMR relaxation data reveal that the acquisition of the micelle bound α-helical conformation constrains the NMC flexibility more than the confinement itself. This comprehensive study of the structural behaviour of NMC provides essential mechanistic information that could be useful for the development of new therapeutics to treat neurological, cancer-related or eating disorders.
NMC exerts its physiological function mainly by its interaction with the subtype-2 bombesin receptor (BB2R), which is a member of the G-protein coupled receptor superfamily4 that is located in the gut and in the CNS.5 For instance, NMC mediates neurotransmission and neuromodulation,6 and it is able to excite specific neurons by decreasing the resting potassium conductance and increasing the non-specific conductance.7 NMC can also reduce the appetite, and therefore can act as a anorexia inducer,8 likely through its interaction with bombesin receptors in the central amygdala.9 In addition, its intravenous administration increases growth hormone levels in calves,10 while it can also act as an autocrine growth factor in human small-cell lung cancer.11 Moreover, NMC has been shown to regulate growth and/or differentiation of human tumors in a wide range of tissues, including carcinomas of pancreas, stomach, breast, prostate and colon.12 Accordingly, a novel protein vaccine comprising six covalently linked repeats of NMC was successful in suppressing the proliferation of breast tumors cells.13
As a result of NMC's pharmacological profile, BB2R antagonists are considered as prospective anticancer therapeutics14 and for the treatment of other illnesses.15 RC-3095, a peptidomimetic of NMC, was shown to produce long-lasting tumor regressions in different human models,16 as well as to show beneficial effects during the treatment of tumor necrosis factor-dependent chronic inflammatory conditions.17 More recently, a N-terminal modified NMC with acyclic tetraamines for binding of 99mTc ([99mTc]-demomedin C) was successfully targeted in BB2R expressing tumor cells as a potent agonist inducing selective intracellular calcium release and triggering GRP receptor mediated internalization of the radioligand.18 However, its tolerability, background radioactivity and retention in tumor lesions warrant future studies as these pharmacological aspects have led to the rejection of other 99mTc-bombesin analogs.19
The efficient design of more potent antagonists of BB2R requires a detailed understanding of the structure–activity relationships of NMC. However, very limited structural information on NMC is currently available. In contrast to bombesin or NMB, NMC was predicted not to adopt a α-helical conformation upon binding to non-polar sites, due to reduced hydrophobic interactions that arise from the replacement of Leu3 by His3 in NMB.20 Polverini et al. used CD spectroscopy to suggest that NMC could adopt a helical-like conformation upon binding to lipids.21 Furthermore, NMR spectroscopy was applied to solve the solution structure of the NMC–Ni2+ complex that consists of two connected turns,22 also likely adopted in the NMC–Cu2+ complex that could be physiologically involved in metal transport along the CNS.23 More recently, we used replica exchange molecular dynamics (REMD) to demonstrate that NMC, in a simulated aqueous environment, adopts different conformations resembling β-turns that are stabilized by different hydrogen bonds formed and broken along the trajectory.24
Although computer simulations can reveal the intrinsic conformational features of a peptide as encrypted in its sequence, caution should be taken about the thoroughness of the sampling. Therefore, we aim here to complement these preliminary computational results with further structural evidences. We combined different biophysical techniques to study the conformational ensemble of NMC in an aqueous solution, and at five different 2,2,2-trifluoroethanol (TFE)/water concentrations (i.e. at 10%, 25%, 40%, 60% and 90% TFE). As TFE can typically induce α-helicity in polypeptides,25 we complemented these data by analyzing the α-helical structure of NMC bound to SDS micelles, characterizing the NMC–SDS micelle complex and evaluating independently the structuring and the binding effects on the peptide flexibility. These data constitute a comprehensive overall picture about the conformational preferences of NMC under different environments and represents a new structural platform for the future development of BB2R antagonists.
The CD spectrum of NMC in an aqueous solution indicates that the peptide is disordered, as evidenced by the minimum located at 197 nm. A slight increase in the TFE percentage, from 0 to 10%, scarcely modifies the spectrum profile as well as the secondary structure content. However, the addition of TFE at percentages higher than 10% markedly enhances the intensity of the region between 190 and 202 nm and reduces that of the band located between 206 and 240 nm, which results in a notable increase of the α-helical content, while decreasing the percentages of β-strands, turns and random coil regions (Fig. 1A and Table S1†).
These results indicate that NMC adopts a predominant α-helix conformation upon addition of TFE, but only when the percentages are higher than 10%.
Chemical shifts corresponding to HN, N, Hα, Hβ, Cα and Cβ were used to determine the secondary structural propensity (SSP) scores (see Section 4.5) at different TFE/water ratios (Fig. 1B). The SSP plots obtained suggest that the structures of NMC at 0% and 10% TFE are similar, except for the residues located at the C-terminus, which seem to slightly enhance their helicity at 10% TFE. While both the structures seem to be randomly coiled, the increase in the TFE percentage above 10% clearly induces an enlargement of the α-helix content, which is more prominent in the central region of the peptide (Trp4–Gly7). On the other hand, most of the SSP values obtained at 90% TFE are higher than 0.5, which indicates that the corresponding residues display a well-defined α-helical structure.
These results are in agreement with the CD data and prove that the peptide central region (Trp4–Gly7) holds the higher α-helical tendency.
0% TFE | 10% TFE | 25% TFE | 40% TFE | 60% TFE | 90% TFE | SDS | |
---|---|---|---|---|---|---|---|
a Restraint statistics reported for unique, unambiguous assigned NOEs.b Violations are only reported when present in six or more structures.c Coordinate precision is given as the average pair-wise Cartesian coordinate root mean square deviations over the ensemble.d Values obtained from the PROCHECK-NMR analysis68 by using the Protein Structure Validation Server (PSV).69 | |||||||
Structural computed conformers | 18 | 17 | 20 | 20 | 19 | 20 | 20 |
![]() |
|||||||
Restrainsa | |||||||
Short-range (|i − j| ≤ 1) | 47 | 71 | 91 | 121 | 80 | 79 | 99 |
Medium-range (1 < |i − j| < 5) | 3 | 2 | 23 | 49 | 24 | 35 | 39 |
Long-range (|i − j| ≥ 5) | 0 | 0 | 2 | 7 | 3 | 0 | 0 |
NOE constrains per restrained residue | 7.1 | 9.0 | 12.8 | 19.3 | 13.1 | 12.1 | 13.8 |
Torsion angles restraints | 0.0 | 0.0 | 0.0 | 0.0 | 0.0 | 0.0 | 0.0 |
![]() |
|||||||
Restraints statisticsb | |||||||
Distance violations > 0.0 Å | 5 | 4 | 3 | 4 | 2 | 3 | 2 |
Torsion angle violations > 0° | 0 | 0 | 0 | 0 | 0 | 0 | 0 |
![]() |
|||||||
Target function value (Å2) | |||||||
Average/best | 0.0 | 0.0 | 0.02 | 0.0 | 0.02 | 0.0 | 0.0 |
![]() |
|||||||
Pairwise RMSD of residues 3–8 in Åc | |||||||
Backbone N, CA, C′ | 0.73 ± 0.26 | 0.63 ± 0.29 | 0.08 ± 0.07 | 0.01 ± 0.00 | 0.06 ± 0.04 | 0.25 ± 0.1 | 0.17 ± 0.7 |
Heavy atoms | 1.87 ± 0.51 | 1.40 ± 0.38 | 0.43 ± 0.30 | 0.05 ± 0.01 | 0.41 ± 0.20 | 1.05 ± 0.32 | 0.57 ± 0.35 |
![]() |
|||||||
Ramachandran plotd | |||||||
Most favoured regions (%) | 54.8 | 47.1 | 47.9 | 55.0 | 75.7 | 82.9 | 60.7 |
Additional allowed regions (%) | 45.2 | 52.9 | 52.1 | 45.0 | 23.4 | 17.1 | 39.7 |
Generously allowed regions (%) | 0.0 | 0.0 | 0.0 | 0.0 | 0.0 | 0.0 | 0.0 |
Disallowed regions (%) | 0.0 | 0.0 | 0.0 | 0.0 | 0.0 | 0.0 | 0.0 |
The NMC ensemble calculated in an aqueous solution displays a native random coil conformation. Nevertheless, the corresponding backbone RMSD is lower than that expected for a fully unstructured ensemble, mostly as a result of the short range NOEs detected between central amino acids, which create a backbone architecture resembling a distorted S (Fig. 1C).
The NMC structure obtained at 10% TFE does not differ significantly to that found in water, which is in perfect agreement with previously reported CD and SSP data (Fig. 1A and B). However, while the depicted S-like conformation between His3–His8 is still preserved, the C-terminal region appears to adopt a more extended structure (Fig. 1C). This subtle but remarkable difference was already revealed by the SSP data, but it can also be observed from the overlapping of the 15N-HSQC spectra of NMC at 0% and 10% TFE, wherein the chemical shift perturbations of Val6, Gly7 and Leu9 are larger than those of Asn2, His3, Trp4 or Ala5 (Fig. 2A).
CD and SSP data already suggested that an increase in the TFE content from 10% to 25% implies a larger structural rearrangement than that occurring when the TFE percentage increase from 0% to 10%. Moreover, the overlapping of the 15N-HSQC spectra obtained at 10% and 25% TFE additionally shows that this rearrangement mainly occurs at the C-terminus, as is evidenced by the chemical shift perturbations displayed by the Val6–Met10 stretch (Fig. 2B). The NMC structure obtained at 25% TFE does not exhibit the S-like conformation observed at 0% and 10% TFE. In contrast, residues at the C-terminus (up to Trp4) roll up and adopt a helical-like turn, whereas the N-terminal region bends back towards the central residues (Fig. 1C).
The NMC solution structure obtained at 40% TFE shows the formation of a short but well defined α-helix at the C-terminus (Ala5–His8) as a result of an increased compactness of the structure displayed at 25% TFE (Fig. 1C). On the other hand, the N-terminal region remains unstructured, but adopts a newly extended conformation. Therefore, the increase in the TFE percentage from 25% to 40% implies an overall structural rearrangement that is also evident from the comparison of the corresponding 15N-HSQC spectra, wherein the entire resonances shift (Fig. 2C).
At 60% TFE, the chemical shifts of the residues between Val6 and Met10 do not show remarkable differences in comparison with those obtained at 40% TFE. However, more noticeable are the shifts of the cross peaks corresponding to His3 and Trp4 (Fig. 2D). These variations result from the preservation of the α-helical structure between Ala5 and His8 already formed at 40% TFE, whereas the residues at the N-terminus fold back also adopting a new α-helical conformation (Fig. 1C).
The increase in the TFE percentage mostly resulted in an overall chemical shift variation of the 1H–15N cross peaks towards high field, especially in the 1H dimension (Fig. 2), which can be attributed to the lower capacity of TFE to form hydrogen bonds relative to water.28 This was also the case when the TFE percentage increased from 60% to 90%, except for His3 and Trp4, whose cross peaks shifted towards low field, suggesting a further structural rearrangement at the N-terminus (Fig. 2E). The structure of NMC obtained at 90% TFE shows an enlargement of the Ala5–His8 α-helical stretch depicted at 60% TFE towards Leu9, but also towards Trp4 and His3 (Fig. 1C). Hence, the more compact conformation adopted by the N-terminus when going from 40% to 60% TFE becomes fully α-helical at 90% TFE.
Thus, NMC is a disordered peptide in an aqueous solution, although its central region exhibits a constrained backbone architecture resembling a distorted S. Upon increasing TFE/water ratios, the C-terminal region first stretches to fold back into an α-helical structure, which also occurs at the N-terminus but only at higher TFE concentrations.
In addition, we complemented these results by studying the biophysical characteristics of NMC upon interaction with SDS micelles, which were chosen as a model system.
We ran diffusion-oriented (DOSY) NMR experiments on a solution containing NMC alone or in the presence of SDS micelles. The resulting diffusion coefficients (D) were independent of the NMC concentration, indicating that no self-aggregation occurred in both the samples. Assuming that the slight viscosity change linked to the presence of SDS micelles equally affects the reference (i.e. acetate and DSS signals) and the NMC signals, it is clear that NMC reduces its overall mobility in the presence of SDS micelles (Fig. 3A), which potentially suggests their binding.
Isothermal titration calorimetry (ITC) was then used to thermodynamically characterize the binding process. The titration curve of SDS in acetate buffer resulted in the appearance of initial exothermic peaks accounting for the low-temperature energy-favoured demicellization.29 These peaks became less exothermic as they approached the midpoint of the inflection (critical micelle concentration; cmc),30 to finally become endothermic as a result of the micelle dilution effect29 (Fig. S2†).
A similar trend was observed when titrations were carried out on solutions containing NMC (Fig. S2†). However, several differences that could be ascribed to the SDS–NMC interaction were observed. Difference enthalpograms reveal an initial exothermic heat flow that rapidly levels off as the SDS concentration increases (Fig. 3B). This effect likely corresponds to specific and cooperative electrostatic interactions occurring between the SDS sulfate group and cationic His.31 The amplitude of this variation scales linearly with NMC concentration (Fig. 3C) and the slope obtained (αΔH ∼ −79 ± 5 kJ mol−1 mM−1) indicates that there is a large change in the ΔH of the system upon a small change in both protein and surfactant [αΔH is proportional to (d2H)/(dnproteindnSDS) and measures the enthalpy of NMC–SDS interaction32].
Next, the enthalpy difference became slightly endothermic, which could be attributed to a conformational rearrangement of NMC.33 Then, the curves began to deviate exothermically from the control curve at a SDS concentration of ∼3.3 mM, which corresponds to the onset for binding of NMC to SDS (critical aggregation concentration; cac).34 The subsequent exothermic variation is attributed to the association of SDS and NMC,33 whose saturation is at the inflection point and corresponds to cmc, which slightly increases with the NMC concentration (3.9–4.2 mM). SDS injected beyond this point remains in the micellar form, having fewer NMC molecules to interact and leading to the final asymptotic curve (Fig. 3D). The amplitude of this curve linearly scales with the NMC concentration and the obtained αΔH ∼ −0.92 ± 0.09 kJ mol−1 mM−1 (Fig. 3C) proves that the nature of the interaction between SDS micelles and NMC is much weaker than an electrostatic one.
The cac and cmc values were additionally used to calculate the Gibbs free energy change of aggregation (ΔGmic) and the Gibbs free energy change of aggregation in the presence of NMC (ΔGag) through the application of the charged phase separation and mass-action model.35 The obtained ΔGmic and ΔGag values were −24 ± 2 and −25 ± 1 kJ mol−1, respectively, which proves that the micellar behaviour of SDS and the formation of NMC/SDS mixed micellar junctions are both similarly thermodynamically favoured. Moreover, ΔHmic and ΔHag, both <−0.1 kJ mol−1 (Fig. 3D), are much smaller that the terms TΔSmic or TΔSag (∼24 kJ mol−1), revealing that both the aggregation of SDS in the absence and in the presence of NMC is entropy driven.
According to Lindman and Thalberg,35 the free energy to drive 1 mol of monomeric SDS into NMC-bound micelle (ΔGps = ΔGag − ΔGmic) is indicative of the binding strength of SDS onto NMC. The obtained ΔGps ≈ −1 kJ mol−1 reveals that the binding between NMC–SDS is only slightly thermodynamically favoured in comparison to that occurring between SDS–SDS molecules.
NMR was then used to obtain residue level insights on this helical rearrangement. The comparison of the 15N-HSQC spectra of NMC obtained in the absence and in the presence of d25-SDS show notable chemical shift perturbations as a result of the binding. This occurs for all residues except for the cross peaks of Asn2 (i.e. HN–N and HNδ–Nδ), which indicates that the chemical environment of Asn2 is only slightly modified upon micelle interaction (Fig. 4B). All 1H, 15N, and 13C resonances (except those for N and HN of Gly1) were unambiguously assigned, deposited to BMRB (25525), and used to calculate the SSP values, which were also comparable to those obtained at 60% TFE (Fig. 1B). Hence, CD and SSP data strongly suggest that the structure of NMC bound to SDS micelles must be similar to that obtained at 60% TFE.
The geometrical restrains automatically obtained from the NMR assignment and the 1H–1H-NOE intensities (Fig. S1†) were used to calculate the solution structure of NMC bound to d25-SDS micelles. The ensemble obtained, deposited to the PDB under the accession code 2n0h, has an excellent Procheck-NMR score satisfying all convergence criteria for structure calculations (Table 1). The structure reveals that the binding process induces the formation of an α-helical stretch between Ala5 and Leu9, whereas the N-terminus retains the native random coil conformation (Fig. 4C). In fact, this structure is very similar to that obtained in 60% TFE, which is ascertained by the low RMSD value arising from the alignment of the averaged structures of both ensembles (0.52 Å for the backbone atoms) (Fig. 4D).
Our results show that NMC binds to SDS micelles through a process that implies the formation of an α-helical stretch at the C-terminus, whereas the N-terminal segment remains disordered.
The extent to which Trp4 was buried into the micelles was determined using the acrylamide quenching experiments. Equal amounts of acrylamide were added to solutions containing either free NMC or NMC–SDS micelles complex. The presence of micelles decreased the Stern–Volmer constant (Ksv) of NMC threefold (∼13 M−1 for the free form vs. ∼4 M−1 for the complex) (Fig. S3†), indicating a high degree of protection of the Trp4 side chain against the solvent. Hence, the Trp4 side chain inserts into the SDS micelles during the binding process.
The use of non-deuterated SDS resulted in the appearance of new NOEs that were unambiguously assigned to specific intermolecular NMC–SDS contacts. The side chain of Trp4 is fully embedded into the SDS micelles since its indole group exhibits strong NOEs with different SDS methylene groups. In addition, the HN of Val6, Leu9 and Met10, as well as the Hα of Trp4, Ala5, Val6 and Leu9 display different NOEs with the aliphatic tail of SDS, proving that the backbone at the C-terminus is also inserted into the SDS micelles. NOEs signals connecting the Val6 and Met10 side chains with protons of C1 in SDS were also found, indicating that these regions protrude from the hydrophobic core of the micelle (Fig. 5A and S5†) (Table 2).
![]() | ||
Fig. 5 The mapping of the interactions observed between NMC and SDS micelles. (A) Overlapping of the 1H–1H NOESY spectra of NMC obtained at 15 °C when it was bound to SDS micelles (red) and when it was bound to d25-SDS micelles (blue). The chemical structure of SDS is shown above the NMR spectra. Carbon atoms are numbered arbitrarily. (B) Model representation of the interaction between NMC and SDS micelles. Aliphatic chains of SDS are coloured in grey, while the corresponding sulfate group is coloured in yellow (sulphur atoms) and red (oxygen atoms). NMC residues are coloured based on the previously determined free energy transfer of each amino acid from water to lipid bilayers (ΔG = −1.8 kcal mol−1, red; ΔG = 1.8 kcal mol−1, blue).40 His3 is coloured taking the ΔG value determined for protonated His, while His8 is coloured based on the ΔG value determined for neutral His. |
NMC | |||
---|---|---|---|
CH2 (1)b | CH2 (2)b | CH2 (3–11)b | |
a The intensities of the NOE signals are divined as: “+++” high intensity; “++” medium intensity; “+” low intensity.b Atoms are numbered arbitrarily according the chemical structure of SDS. | |||
HN-V6 | ++ | ||
HN-L9 | + | ||
HN-M10 | + | ||
Hα-W4 | ++ | ||
Hα-A5 | ++ | ||
Hα-V6 | ++ | ||
Hα-L9 | + | ||
Hβ2-W4 | + | ||
Hβ3-W4 | ++ | ++ | |
Hβ3-H8 | + | ||
Hβ3-M10 | + | ||
Hγ2-M10 | + | ||
Hε1-W4 | + | + | ++ |
Hε3-W4 | + | ++ | |
Hζ2-W4 | + | ++ | |
Hδ1-W4 | + | + | ++ |
Hζ3-W4 | + | +++ | |
Hδ2-H8 | ++ | ||
Hγ1-V6 | ++ | ||
Hγ2-V6 | ++ |
When analyzing the structure of NMC upon binding to SDS micelles (Fig. 4C), it is difficult to understand how this interaction can occur since an amphipathic-like architecture is lacking. For instance, His8 points towards the same face of the helix than Trp4, while His3, Val6 and Met10 are in the opposite one.
At pH 4.0 His side chains must be protonated, the pKa of the imidazole protons ranges from 4.9 to 6.6 in micellar media,37 and therefore highly unlikely to penetrate into the micelles. This has been the case for His containing peptides, where micelle insertion was only observed when the pH increased above the pKa of His.38 However, we unexpectedly detected unambiguous NOEs between Hβ/Hδ2 of His8 and the aliphatic methylene groups of SDS (Fig. 5A and S5†) (Table 2), which proves that His8 inserts into the micelles even at pH 4.0. This could only occur if the insertion occurs together with His8 deprotonation. In fact, the Hδ2 and Hε2 chemical shift values (highly sensitive to imidazole protonation state) in His8 are shifted upfield upon micelle binding in a range comparable to that observed during the pH-induced deprotonation (i.e., ∼0.2 ppm for Hδ2 and ∼0.4 ppm for Hε2).38,39 This was not the case for His3, since its Hδ2 and Hε2 values only underwent a slight downfield shift upon micelle binding (Table 3).
Hδ2a | Hε1a | Na | H1a | H2a | |
---|---|---|---|---|---|
a Values are given in ppm. | |||||
His3 in free NMC | 7.08 | 8.42 | — | — | — |
His3 in micelle bound NMC | 7.28 | 8.61 | — | — | — |
His8 in free NMC | 7.14 | 8.43 | — | — | — |
His8 in micelle-bound NMC | 7.02 | 8.03 | — | — | — |
Ans2 side chain in free NMC | — | — | 113.3 | 7.53 | 6.93 |
Asn2 side chain in micelle-bound NMC | — | — | 112.8 | 7.52 | 6.88 |
C-terminal amide in free NMC | — | — | 108.2 | 7.54 | 7.15 |
C-terminal amide in micelle-bound NMC | — | — | 105.6 | 7.20 | 7.03 |
Our data indicate that NMC binds to SDS micelles only through the insertion of its C-terminal region, while the N-terminal tail remains out of the micelle (Fig. 5B). This is additionally supported by the upfield chemical shifts of the C-terminal amide, whereas the chemical shifts corresponding to the amide side chain of Asn2 remain unaltered (Table 3) (Fig. 4B). The binding process occurs through the energetically favoured insertion of Trp4, Val6, Gly7, Leu9 and Met10 have a negative ΔG for transfer form water to lipid bilayers,40 which must energetically compensate the deprotonation and the further insertion of His8 into the micelle (Fig. 5B).
In an aqueous solution, R1 and R2 constants are lower than 1.3 s−1, whereas all HET-NOEs are negative, which indicates that both the features are typical of intrinsically disordered peptides (IDP).41 The addition of 60% TFE increases R1 and R2 constants by ∼2–3 times, while most of the HET-NOEs become positive as a result of the rigidity linked to the α-helix formation. These variations are even more pronounced when NMC binds to SDS micelles. In this case, R2 increases by ∼9 times and most of the HET-NOEs display values close to 0.5, thus proving that the binding additionally constrains the α-helical NMC structure (Fig. 6A–C).
The R1 and R2 constants obtained in an aqueous solution for the Trp4-indole group are similar to the backbone ones. However, HET-NOE is ∼0.9 units higher, likely a result of the rigidity linked to the contacts of Trp4-Hδ1, Hε3 with Ala5, which must reduce the side chain dynamics. The presence of 60% TFE slightly increases the R1 and R2 constants, revealing that the mobility of the Trp4 side chain is slightly reduced upon folding. Moreover, the confinement of the indole group into the SDS micelles enlarges by ∼4 times the R2, thus proving that the binding also additionally constrains the mobility of the Trp4 side chain (Fig. 6A–C).
The R1/R2 ratios within one standard deviation of the mean were then used to determine the correlation time (τc) of each structural ensemble. Calculations carried out with the r2r1_tm and TENSOR 2.0 software gave similar values, being 0.8 ± 0.1 ns in buffer, 2.1 ± 0.1 in 60% TFE, and 6.9 ± 0.1 ns under its micelle bound state. Although the NMC molecular size scarcely changed in the presence of 60% TFE (Fig. 1C), τc increased by ∼3 times as a result of the increased viscosity of the TFE/water mixture.42 Furthermore, τc of the NMC–SDS micelle complex is ∼8 times bigger than that of free NMC, which can be ascribed to the resulting high molecular weight complex. In addition, this τc value is also ∼1 ns bigger than that of free SDS micelles,43 which points to the formation of 1:
1 NMC–SDS micelle complex.
The backbone R1 and R2 values and the energy-minimized representative conformers of each NMR-derived solution structure were used to estimate the diffusion tensor (D∥/D⊥) using the isotropic, axially symmetric and fully anisotropic diffusion models in the software Quadratic-Diffusion.44 The D∥/D⊥ values obtained for NMC were 1.2 in buffer, 0.98 in 60% TFE, and 0.8 under its micelle bound state. Hence, the diffusion model that best describes the NMC rotational behaviour under these experimental conditions is the isotropic one (D∥/D⊥ < 1.3).
The 15N-relaxation parameters were then analyzed assuming an isotropic rotational diffusion model and according to the Lipari–Szabo model-free formalism45 (Tables S2–S4†). The order parameters (S2; indicative of the amplitude of internal ps–ns timescale motions) of NMC in water have an average value of 0.45 ± 0.12, being within the typical range found in IDPs (Sav2 ∼ 0.3–0.6).46,47 However, these values are not homogeneous along the NMC sequence. The Sav2 of the residues between Trp4 and His8 is notably higher than that arising from the three N-terminal and the two C-terminal residues (ΔSav2 ∼ 0.18) (Fig. 6D). This striking variation clearly proves that central amino acids, although integrated within the fully disordered NMC structure, display much slower motions than the terminal ones, likely as result of their inter-residual interactions.
The acquisition of the α-helical structure in the presence of 60% TFE reduced by ∼46% the flexibility of the central residues (His3–His8), as evidenced by ΔSav2 of ∼0.23. This was not the case for the still unstructured two C-terminal residues, whose mobility was nearly unaltered upon folding of the Ala5–His8 stretch (ΔSav2 ∼ 0.03). The insertion of the folded NMC into SDS micelles additionally reduced the mobility of the central amino acids (His3–His8) by ∼20% (ΔSav2 ∼ 0.1), whereas that corresponding to the C-terminal Leu9 and Met10 was reduced by ∼90% (ΔSav2 ∼ 0.36) as a result of their confinement in the SDS micelle (Fig. 6D).
Our results reveal that NMC in an aqueous solution displays highly different conformational motions along its sequence. Moreover, the NMC α-helical folding markedly reduces the mobility of the central residues, affecting the peptide flexibility to a larger extent than the subsequent structural confinement into SDS micelles.
Most of the small peptides do not behave as pure random coils because their residues usually do not sample all sterically accessible regions, but rather exhibit local structural preferences.49 Hence, we used NMR to gain residue insights on the structural preferences of NMC in an aqueous solution. The 15N-HSQC spectrum only displays nine signals proving that either there is a main conformational state or that the dynamic equilibrium between different conformers is a fast exchange regime (Fig. 2A). This agrees with our REMD prediction of a low energy structure among the different NMC conformers.24 The NMR ensemble possesses a well-defined backbone architecture resembling a distorted S (Fig. 1C), which is built through short range contacts between the central residues. We already predicted these turns using REMD, since ∼20% of the sampling displayed the segment Trp4–Gly7 stabilized by a hydrogen bond. In addition, others turns, such as His3–Trp4, Trp4–Ala5 or Gly7–His8, were only ∼10% sampled.24 This S-like architecture is also adopted in the NMC–Ni2+ complex since two turns are formed, one involving the three first residues coordinating the metal and the other linking Ala5 to His8.22
NMC folds into a helical structure upon increasing the TFE percentage as result of the reduction of the dielectric constant, which favours the formation of intramolecular hydrogen bonds. However, the ease of adopting this α-helical structure is not the same along the entire sequence. The presence of only 10% TFE induces the stretching of the C-terminal region that folds back into a helical-like structure only when the TFE percentage increases to 25%. The C-terminal helical structuring is finally strengthened at 40% TFE and it does not further change at 60 or 90% TFE. The N-terminal segment is more resistant to undergo the coil-helix transition, since it only acquires a helical-like structure when the TFE content is 60%, becoming fully α-helical at 90% TFE.
In addition, we studied the molecular complex formed between NMC and SDS micelles. ITC was used to characterize the binding process. Initially, it appeared as a highly exothermic event proportional to the NMC concentration and was attributed to electrostatic interactions.31,33 This showed that NMC is the limiting reactant in this region of the plot and that the binding is notably weak. NMC started to bind SDS at a cac of ∼3.3 mM, consistent with what was observed when using hydrophobically alkali-soluble emulsion polymers (HASE) (cac ∼ 4 mM).50 The cac value was independent of the NMC or HASE concentration, although it is concentration-dependent in folded proteins.33 The binding saturation of NMC occurred at [SDS] ∼ 4 mM, which agrees with the cmc of SDS at 15 °C,29 and it did not change with the NMC concentration, thus differing from what occurs in folded proteins.33 SDS added beyond cmc remains in micellar form and leads to an asymptotic curve that is related to the NMC hydrophobicity.34,50 Its αΔH value is ∼ 86 kJ mol−1 mM−1 lower than that determined for the electrostatic interactions between NMC and SDS monomers, which proves that the diving force leading to the NMC/SDS micelle binding is weaker than an electrostatic one and likely to be hydrophobic. This αΔH is also ∼15 times lower than that determined for folded proteins,33 which must account for the lower hydrophobicity of NMC in comparison to larger polypeptides. The formation of micelles alone or in the presence of NMC was always thermodynamically favoured through an entropy-driven process. Moreover, the calculation of ΔGps (which compares the stability of the interactions between NMC–SDS and SDS–SDS) demonstrated that NMC–SDS was only ∼−1 kJ mol−1 more favoured than the SDS–SDS interaction, being weaker than that determined for HASE–SDS interactions (∼−4 kJ mol−1).50
Next, we calculated the solution structure of NMC under its micelle bound state and characterized the architecture of the complex. Although it was predicted that the replacement of Leu3 in NMB by His3 in NMC would hinder the acquisition of a helical membrane-bound structure,20 we have shown that the C-terminal region of NMC folds into an α-helix upon micelle insertion, whereas the N-terminal segment remains unstructured. This folding process must be directly related to the energetically favoured insertion of Trp4, Val6, Gly7, Leu9 and Met10 (residues with a ΔG < 0 of transfer form water to lipid bilayers40) into the non-polar micelle, which was experimentally observed through intermolecular NOEs (Fig. 5B). However, this hydrophobic insertion cannot occur without the enclosure of His8 into the micelle, which is expected to be fully protonated at pH 4.0, and therefore highly unfavourable. Nevertheless, NMR data reveal that His8, but not His3, deprotonates during the NMC insertion. These observations enabled us to hypothesize that the favourable hydrophobic interactions of the residues near His8 would energetically compensate for its unfavourable deprotonation and insertion. This idea is also supported by the small ΔGps value of the NMC–SDS complex in comparison to other peptide–SDS complexes.50
The micelle-bound NMC structure displays Trp4, His8 and Leu9 oriented toward the same face. These residues correlate with Trp8, His12 and Leu13 in bombesin, which are essential for the binding to the bombesin family of receptors.51 Hence, it is likely that Trp4, His8 and Leu9 also reorient as NMC approaches its receptor, which further validates the micelle-bound NMC structure, also within a biologically relevant context.
The ability of NMC to form a short α-helix during its micelle insertion may result in conformation and/or orientation selective interactions. Hence, the wide spectrum of similar but not identical biological activities of bombesin-related peptides raises the possibility that fluctuations of secondary structure can modulate their accessibility to different receptors, a mechanism already proven for neurokins and opioid peptides.52 Hence, we completed our structural data by analyzing the dynamics of NMC in water, in the presence of 60% TFE and under its micelle bound state. The fact that NMC at 60% TFE displays a structure similar to that adopted when it is embedded into micelles has allowed us to discriminate the folding and the binding effects on the molecular tumbling and dynamics.
The R1/R2 ratios were used to determine the τc values, which represent the time of the molecule to tumble as a function of size, shape and viscosity. The τc of NMC in water is similar to that found for other peptides of similar size.53 The addition of 60% TFE raised τc ∼ 3 times, which is attributed to an increased solvent viscosity typical of the TFE/water mixtures42 and not to changes in the peptide size.53,54 The τc of the NMC–SDS complex was ∼1 ns bigger than that of free SDS micelles proving that NMC slightly reduces the micellar tumbling rate as a result of the formation of a 1:
1 complex; the stoichiometry mainly observed in peptide–micelle complexes.46,55
15N relaxation data were used to determine the diffusion tensor of NMC in water, at 60% TFE and under its micelle-bound state. All the D∥/D⊥ values were <1.3, suggesting an isotropic rotational behaviour. This model has already been adopted to study the dynamics of other small peptides, either in their free form56 or under their micelle-bound states.46 NMC dynamics were studied applying the model-free approach,45 which fits the relaxation data to one of the five models characteristic of the complexity of the residue level dynamics. The nine residues of NMC in water were described by model 2, indicating internal motions (τe) on ps–ns timescales. This was also the case for most of the residues of NMC at 60% TFE, except that His3 was fitted to model 4, and Leu9/Met10 that were fitted to model 5, hence suggesting complex internal motions. Five out of the nine residues in the NMC–SDS complex were fitted to model 1, proving their lack of flexibility (Tables S2–S4†).
Residue level mobility was qualitatively compared within the same NMC structure and across the three different NMC structures through the analysis of S2. Folded proteins exhibit Sav2 of ∼0.8, whereas mobile terminal residues display a Sav2 of ∼0.6 (ref. 57) similar to IDPs (Sav2 ∼ 0.3–0.6).46,47 The Sav2 of NMC in water was within the typical range of IDPs. However, the S2 values notably changed between the central and the terminal residues, pointing towards a constrained mobility of the central residues, a trend also observed at 60% TFE. S2 values also revealed that the formation of intramolecular hydrogen bonds linked to the α-helical folding reduces much more the backbone flexibility than the intermolecular hydrophobic interactions associated to the insertion of NMC into the SDS micelles. Hence, the coil-helix transition has a higher impact on the dynamics than the NMC confinement.
The 15N-relaxation data acquired at 60% TFE could be affected by the increase in the viscosity (Δη ∼ 1 cp at 25 °C (ref. 42)). However, data comparing the NMR dynamics of the Escherichia coli orthologue of frataxin (CyaY) in water (η ∼ 0.89 cp (ref. 42)) and in hen egg white (η ∼ 4 cp (ref. 58)) prove that viscosity does not affect the CyaY fold nor the HET-NOE values, but notably decreased and increased the R1 and R2 values, respectively.59 The viscosity change scarcely affected the S2 values of CyaY, calculated assuming an axially symmetric diffusion model (ΔSav2 ∼ 0.06) (Fig. S6†). In contrast, R1, R2, HET-NOE and S2 values of NMC were notably enhanced when going from pure water to 60% TFE, thus proving that these changes are associated with structural alterations rather than viscosity modifications.
TENSOR 2.0 was also used to calculate the generalized order parameters describing the amplitudes of internal motions (S2). The 15N relaxation constants and the energy-minimized solution structures were analyzed according to the molecular diffusion derived by Woessner in combination with the Lipari–Szabo model-free analysis of local flexibility.45 The amide bond length was fixed at 1.02 Å. Five different models were tested to characterize the internal dynamics of the NH groups:75 model 1 (S2), model 2 (S2, τe), model 3 (S2, Rex), model 4 (S2, τe, Rex) and model 5 (Sf2, Ss2, τe). τe is the effective internal correlation time (describes motions on a timescale > 20 ps), Rex is a chemical exchange term (describes slow timescale motions on the order of μs–ms), and Sf2 and Ss2 are terms that result from splitting the generalized order parameter into two order parameters reflecting slower and faster motions, respectively. The confidence levels were estimated using 100 Monte Carlo simulations per run in combination with c2 and F-test criteria.
F0/F = 1 + Ksv[Q] | (1) |
The Gibbs free energy changes of aggregation (ΔGmic) and aggregation in the presence of NMC (ΔGag) were calculated from the ITC data through the application of the charged phase separation and mass-action models (eqn (2)).35
ΔGmic/ag = (1 + K)RT![]() | (2) |
A factor of (1 + K) is needed to calculate the free energy of ionic SDS, where K is the micellar charge fraction with a value of 0.85.76 The enthalpy change and free energy changes were also used to calculate the entropy changes of micellization (ΔSmic) and aggregation in the presence of NMC (ΔSag).
Footnote |
† Electronic supplementary information (ESI) available. See DOI: 10.1039/c5ra12753j |
This journal is © The Royal Society of Chemistry 2015 |