Engineering stable peptide toxins by means of backbone cyclization: Stabilization of the α-conotoxin MII

Abstract

Conotoxins (CTXs), with their exquisite specificity and potency, have recently created much excitement as drug leads. However, like most peptides, their beneficial activities may potentially be undermined by susceptibility to proteolysis in vivo. By cyclizing the α-CTX MII by using a range of linkers, we have engineered peptides that preserve their full activity but have greatly improved resistance to proteolytic degradation. The cyclic MII analogue containing a seven-residue linker joining the N and C termini was as active and selective as the native peptide for native and recombinant neuronal nicotinic acetylcholine receptor subtypes present in bovine chromaffin cells and expressed in Xenopus oocytes, respectively. Furthermore, its resistance to proteolysis against a specific protease and in human plasma was significantly improved. More generally, to our knowledge, this report is the first on the cyclization of disulfide-rich toxins. Cyclization strategies represent an approach for stabilizing bioactive peptides while keeping their full potencies and should boost applications of peptide-based drugs in human medicine.

Venoms from marine snails of the Conus genus comprise a myriad of peptides called conotoxins (CTXs) for the rapid immobilization of prey (1, 2). These 12- to 30-aa peptides target membrane receptors with exquisite selectivity and potency and have become invaluable neurophysiological probes and drug leads. Recently, the CTX ziconotide (MVIIA) was approved for use in the treatment of severe chronic pain by the FDA, and other CTXs have entered clinical trials as treatments for pain (3, 4). In addition, CTXs have played a critical role in dissecting the molecular mechanisms of ion channel and transporter functions in the nervous system (2). One family of CTXs, the α-CTXs, consists of members that antagonize the nicotinic acetylcholine receptors (nAChRs). Ranging in size from 12 to 19 residues, α-CTXs are the smallest of all of the CTXs, yet this family is the most widely distributed among Conus venoms (5).

Despite their exciting applications, many peptide toxins are susceptible to enzymatic degradation by proteases. This characteristic may limit the therapeutic applications of CTXs, and, hence, methods that provide improvements in biological half-life are valuable. Cyclization has been used in the past as a strategy in the pharmaceutical industry for stabilizing and locking the conformation of small peptides (6). Similarly, microorganisms are known to produce cyclized peptides, such as cyclosporin A, which is now in widespread use as an immunosuppressant. Such a strategy has not been applied in the past to disulfide-rich proteins, but with the recent discovery of the cyclotide family of macrocyclic miniproteins (7), it is clear that the approach can be applied to disulfide-rich toxins to produce additional stabilization with the potential to dramatically increase the therapeutic potential of these molecules when limited by poor in vivo stability.

This study focuses on the cyclization of MII, a 16-residue α-CTX isolated from Conus magus (8). The 3D structure of MII consists of a central segment of α-helix with β-turns at the N and C termini (9, 10) and is stabilized by two disulfide bonds in a CysI-CysIII and CysII-CysIV configuration that is common to most members of the α-CTX family. In addition, the N and C termini of the peptide are in close proximity to each other, making MII a good candidate for studying the principles of backbone cyclization. MII is a potent inhibitor of the nAChR that is specific for the α3β2 subtype (8) and is also implicated in binding to the α6 nAChR, ligands of which are potentially important for Parkinson's disease therapy (11). There are currently a number of patents describing the use of MII in therapeutic applications.

To illustrate the advantage of cyclization of linear proteins, we designed and synthesized three cyclic MII analogues by adding a linker segment between the N and C termini. Structural studies of the analogues were undertaken, and activity and stability assays were performed. To our knowledge, this is the first study on the cyclization of CTXs. We also discuss the potential for backbone cyclization to enhance the therapeutic potential of peptide toxins.

Materials and Methods

Peptide design was based on an analysis of homology models generated by using the structural coordinate file of MII (Protein Data Bank ID Code 1MII), available from the PDB (www.rcsb.org/pdb), and the modeler module within insight ii (Accelrys, Inc., San Diego). Energy-minimized linkers of varying sizes were built into the linear MII molecule, and the resulting cyclic analogue models were evaluated.

All peptides were assembled on phenylacetamidomethyl resin by manual solid-phase peptide synthesis using the in situ neutralization/HBTU [2-(1H-benzotriazol-1-yl)-1,1,3,3-tetramethyluronium hexaf luorophosphate] protocol for Boc (tert-butoxycarbonyl) chemistry (12). The peptide was attached to the resin by means of a linker that generates a C-terminal thioester on hydrogen fluoride (HF) cleavage, which facilitates the intramolecular native chemical ligation reaction used to cyclize the peptide (13). Cleavage of the peptide from the resin was achieved by using HF with p-cresol and p-thiocresol as scavengers [9:0.8:0.2 (vol/vol) HF:p-cresol:p-thiocresol] at –5 to 0°C for 1.5 h. After cleavage, the peptide was precipitated with ether and then dissolved in 50% acetonitrile containing 0.05% trifluoroacetic acid (TFA) and lyophilized. The crude peptide was purified by RP-HPLC on a C₁₈ column using a gradient of 0–80% B (A, H₂O/0.05% TFA; B, 90% CH₃CN/10% H₂0/0.045% TFA) in 80 min. Analytical RP-HPLC and electrospray MS confirmed the purity and molecular mass of the synthesized peptide.

The linear reduced peptides were cyclized and oxidized by incubating in 50/50 0.1 M NH₄HCO₃, pH 8.2/isopropanol (0.3 mg/ml) overnight at room temperature. The reaction mixture was purified by RP-HPLC to yield the cyclic/oxidized peptides. Analytical RP-HPLC and electrospray MS confirmed the purity of the final products, and ¹H NMR was used to determine whether the cyclic peptides were folded.

The disulfide connectivity of the cyclic analogues was determined by using a reduction/alkylation procedure followed by MS/MS analysis. Partial reduction was achieved by incubating the peptide (50 μg) in 0.2 M citrate buffer (pH 3, 50 μl) with 50 equivalents of Tris(2-carboxyethyl)phosphine hydrochloride (TCEP) at 37°C. The reaction was left for 3 min, after which it was quenched by injection onto the RP-HPLC. The collected HPLC fractions were analyzed by MS to identify the one-disulfide (1SS) species. An equal volume of 60 mM N-ethylmaleimide in 0.2 M citrate buffer (pH 3) was then added to the HPLC fraction containing the 1SS peptide and left at 37°C for 1 h to alkylate the free Cys residues. The peptide was purified by RP-HPLC and then fully reduced by incubating with 10 equivalents of DTT in 0.1 M NH₄HCO₃ (pH 8.2) for 3 h at 37°C. The reduced alkylated peptide was isolated by RP-HPLC, lyophilized, and cleaved with EndoGluC by incubating for 3 h in 0.1 M NH₄HCO₃ with 2 equivalents of TCEP at 37°C. The peptide was purified by RP-HPLC, and the linear product, containing alkyl groups on one pair of cysteines, was analyzed by MS/MS, which enabled the identification of the two alkylated cysteines and hence the disulfide connectivity.

The structures of the cyclic MII analogues were derived by using NMR spectroscopy for samples dissolved in 90% H₂O and 10% D₂O. Avance 600 and DMX 750 MHz spectrometers (Bruker, Billerica, MA) were used in the acquisition of data. 2D NMR experiments included double quantum filtered (DQF)-COSY, exclusive (E)-COSY, TOCSY (total correlation spectroscopy), and NOESY, with mixing times of 150 and 300 ms. All spectra were recorded at 298 K. ¹H spectra acquired immediately after dissolution of the protonated peptide in D₂O at pH 4.7 were used to detect slow-exchanging amide protons. The peak areas in these spectra were measured and normalized to a nonexchangeable proton, and the exchange rates were calculated by fitting the volume of the decaying signals over time to the equation I = I_o exp(–k_ex× t) + I(∞) (14).

Distance information was obtained from the NOESY spectrum with a mixing time of 300 ms. Backbone dihedral restraints were derived from ³J_HN-Hα coupling constants obtained from a 2D DQF-COSY spectrum or from a 1D ¹H NMR spectrum. The φ angle was restrained to 120 ± 30° for ³J_HN-Hα > 8 Hz and –60 ± 30° for ³J_HN-Hα < 5.8 Hz. Intraresidue nuclear Overhauser effect (NOE) and ³J_Hα-Hβ coupling patterns, determined from an E-COSY spectrum, were used in assigning the χ¹ angle conformations of side chains.

Initial structures were generated by using dyana software (15), and final structures were calculated in explicit water with cns (16) as described in ref. 17. Fifty structures were calculated, and the 20 structures with the lowest overall energies were retained for analysis. Structures were analyzed with promotif (18) and procheck_nmr (19).

The biological activity of the cyclic analogues was assessed by recording membrane currents using the dialyzed whole-cell patch clamp recording technique and by monitoring catecholamine secretion from adrenal chromaffin cells.

Electrophysiological Recordings. Chromaffin cells were prepared from bovine adrenal glands and maintained on glass coverslips as described in ref. 20. Glass electrodes were pulled, fire-polished (–2 to 3 MΩ) and filled with intracellular solution (in mM: 140 CsCl/2 CaCl₂/11 EGTA/2 MgATP/10 Hepes-KOH, pH 7.2). Agonists were diluted in bath solution (in mM: 140 NaCl/3 KCl/1.2 MgCl₂/2.5 CaCl₂/7.7 glucose/10 Hepes-NaOH, pH 7.35) and applied to cells by a 10-ms pressure ejection (15 psi, Picospritzer II, General Valve, Fairfield, NJ) from an extracellular pipette positioned ≈50 μm from the cell to evoke maximal responses to agonists (21). CTXs were bath-applied. Membrane currents evoked by agonist application were amplified and low-pass filtered (10 kHz) by using a MultiClamp 700B amplifier (Axon Instruments, Union City, CA), and voltage steps were generated by using pclamp 9.2 and a Digidata 1322A interface (Axon Instruments). All experiments were carried out at 22°C.

RNA preparation, oocyte preparation, and expression of nAChR subunits in Xenopus oocytes were performed as described in ref. 22. cDNA encoding the rat α2-7 and β2-4 nAChR subunits were provided by J. Patrick (Baylor College of Medicine, Houston). Oocytes were injected with 2.5 ng of cRNA and kept at 18°C in ND96 buffer (96 mM NaCl/2 mM KCl/1 mM CaCl₂/1 mM MgCl₂/5 mM Hepes, pH 7.4) supplemented with 50 mg/liter gentamycin and 5 mM pyruvic acid 2–5 days before recording.

Membrane currents were recorded from Xenopus oocytes by using an OpusXpress 6000A workstation (Axon Instruments). Electrodes were filled with 3 M KCl (–0.3 to 1.5 MΩ). During recordings, the oocytes were perfused with ND96 buffer at 22°C continuously at a rate of 1.5 ml/min, with 200-s incubation times. Acetylcholine (100 μM) was applied for 2 s at 5 ml/min, with 600-s washout periods. Cells were held at –80 mV with data sampled at 500 Hz and filtered at 200 Hz. Peak current amplitude was measured before and after incubation with cMII-7.

Catecholamine Secretion. Intact chromaffin cells were washed with buffer A (in mM: 145 NaCl/5 KCl/1.2 Na₂HPO₄/10 glucose/20 Hepes-NaOH, pH 7.4), incubated with native and cyclized CTXs for 20 min in the presence of 2 mM CaCl₂, and stimulated by 5 μM nicotine for 20 min. Supernatant aliquots were taken at the end of each experiment, and cells were lysed with 1% (vol/vol) Triton X-100 (Sigma). Both sets of samples were assayed fluorimetrically for catecholamines, and the amount released was expressed as a percentage of control (20).

Stability Assays. Initial stability assays were performed by using the protease EndoGluC, which cleaves on the C-terminal side of glutamic acid. Peptides were incubated with enzyme at a ratio of 25:1 [peptide:enzyme (wt/wt)] in 0.1 M NH₄HCO₃ at 37°C for 10 h. Aliquots (3 μl) were taken from 0 to 10 h at hourly intervals, quenched with 5% formic acid (57 μl), and stored at –20°C until analyzed. The amount of intact peptide remaining was determined by integration of the corresponding peak in the liquid chromatography/MS trace. The cleavage product was also monitored.

The stability of the peptides in human serum (Sigma-Aldrich) was measured as a guide to in vivo stability by using methods described in refs. 23–25. Triplicate samples of native and cyclic peptides were assayed simultaneously at a concentration of 20 μM. At 0, 4, 8, 12, 16, and 24 h, 200 μl of the mixture was removed and added to 100 μl of 15% aqueous trichloroacetic acid to precipitate serum proteins. After 15 min of incubation at 4°C, the precipitate was spun down, and the supernatant (200 μl) was stored at –20°C until analyzed by RP-HPLC. Aliquots of the samples were injected, and the amount of intact peptide remaining was determined by integration at 215 nm. Peptide recovery after precipitation was found to be >90% for all peptides.

Results

Peptide Design and Synthesis and Disulfide Determination. Examination of the 3D structure of MII revealed that the distance between the N and C termini was 11.2 ± 0.3 Å. Using molecular modeling techniques, the number of residues needed to span this distance was estimated to be five or more. Any fewer residues was likely to cause strain on the molecule and perturb the structure and hence activity of the peptide. We therefore designed three cyclic MII analogues, with five-residue (GGAAG, cMII-5), six-residue (GGAAGG, cMII-6), and seven-residue (GAGAAGG, cMII-7) linker sequences (Fig. 1). Gly and Ala residues were used in the linkers because their side chains are small and less likely to interact with the rest of the molecule. Ala residues were interspersed in the linker sequences to assist in NMR spectral assignment.

Fig. 1.

Sequences of MII, cMII-5, cMII-6, and cMII-7. Disulfide bonds are shown as thin lines. The thick line joining the N and C termini of cMII-5, cMII-6, and cMII-7 indicates backbone cyclization.

The three cyclic analogues were synthesized by using Boc (tert-butoxycarbonyl)/HBTU [2-(1H-benzotriazol-1-yl)-1,1,3,3-tetramethyluronium hexafluorophosphate] in situ neutralization chemistry and an intramolecular native chemical ligation strategy to facilitate backbone cyclization. The peptides were cyclized and oxidized in one pot by using 50/50 0.1 M NH₄HCO₃/isopropanol at room temperature overnight. The RP-HPLC traces of the oxidative profile for the three cyclic analogues are shown in Fig. 2. The oxidative folding of the five-residue analogue resulted in a series of closely eluting peaks. In contrast, the oxidative profile of the six-residue analogue consisted of two distinct peaks in a ratio of ≈1:1, whereas the oxidation of the seven-residue analogue resulted in only one product. The oxidation products were separated by RP-HPLC and analyzed by ¹H NMR. The purified peptides, marked in Fig. 2 for each cyclic analogue, gave well dispersed ¹H NMR spectra and were further analyzed by using 2D NMR spectroscopy. The other isomers seen in the RP-HPLC traces (Fig. 2) gave poorly dispersed ¹H NMR spectra, indicating lack of a well ordered 3D structure.

Fig. 2.

Cyclization/oxidation RP-HPLC profiles of the cyclic MII analogues. Profiles are shown for the analogues containing a five-residue linker (GGAAG) (a), a six-residue linker (GGAAGG) (b), and a seven-residue linker (GAGAAG) (c). The peaks corresponding to the peptides subsequently named cMII-5, cMII-6, and cMII-7 are indicated by asterisks.

To confirm the disulfide connectivity, oxidized cyclic peptides were partially reduced with Tris(2-carboxyethyl)phosphine hydrochloride under acidic conditions, and the resulting one-disulfide species were alkylated with N-ethyl-maleimide. After fully reducing each peptide, the cyclic backbone was cleaved by using the protease EndoGluC, and the resulting linear sequence was analyzed by MS/MS. The MS/MS data indicated that for both the cMII-6 and cMII-7 analogues, C2 and C8 were alkylated with the N-ethyl-maleimide group. Therefore, both cMII-6 and cMII-7 possessed a I-III, II-IV disulfide connectivity, which is the same as native MII. The disulfide connectivity of cMII-5 was not determined because it was not possible to partially reduce the peptide.

NMR Structure Determination and Analysis. NH-NH_i+1, Hα-NH_i+1, and Hβ-NH_i+1 connectivities obtained from the NOESY spectrum were used in the sequential assignment of individual spin systems determined from the TOCSY spectrum. Sequential Hα-NH_i+1 connectivities were seen for the entire peptide chain except for at P6. The observations of Hα-Hδ_i+1 NOE correlations indicated that P6 was in a trans configuration. Chemical shift data are listed in Tables 2–4, which are published as supporting information on the PNAS web site.

Deuterium exchange experiments on MII, cMII-6, and cMII-7 revealed the presence of several slowly exchanging amide protons, including, N5, V7, C8, and L10. Table 1 lists the rate constants for disappearance of nonoverlapped amide protons in MII, cMII-6, and cMII-7. The cyclic analogues have significantly reduced rate constants relative to the native peptide, showing that cyclization produces increased protection of the intramolecular hydrogen-bond network.

Table 1. Exchange rates (k_ex, min^–1) for MII, cMII-6, and cMII-7 at pH 4.7 ± 0.2 and 298 K.

Residue	MII	cMII-6	cMII-7
G1	*	14	>500
C2	>500	>500	>500
C3	>500	>500	>500
S4	>500	>500	100
N5	177	61	98
V7	54	15	39
C8	178	55	90
H9	>500	>500	>500
L10	284	46	99
E11	†	†	103
H12	†	>500	>500
S13	†	†	>500
N14	>500	>500	>500
L15	>500	†	†
C16	220	†	105

^*N-terminal residue not detected in linear peptide.

^†Overlapped in ¹H NMR spectrum.

A comparison of the chemical shifts for the cyclic analogues to those of MII provided preliminary information on the structural similarities of these molecules. Fig. 3 compares the Hα secondary shifts of native MII (9), cMII-5, cMII-6, and cMII-7. The Hα chemical shifts of MII, cMII-6, and cMII-7 are closely correlated, whereas those of cMII-5 differ substantially and tend toward random coil values. This finding indicates that the structures of cMII-6 and cMII-7 are similar to MII, whereas cMII-5 is significantly different and is likely to be more flexible. In addition, the negative secondary shifts for residues P6 to H12 indicate helicity in this region of cMII-6 and cMII-7, which is consistent with the structure of native MII. The magnitude of these negative shifts in cMII-5 is reduced, reflecting a decrease in helical character.

Fig. 3.

A comparison of the αH secondary shift values of residues 2–15 for MII (solid line, circles), cMII-5 (dashed line, triangles), cMII-6 (solid line, squares), and cMII-7 (solid line, diamonds). The values were calculated by subtracting the observed values from the random coil values (33).

A large difference in chemical shift between two geminal β-protons is generally indicative of a well ordered structure. The chemical shift differences of geminal β-protons for residues C2, C3, N5, P6, C8, E11, and C16 in MII, cMII-6, and cMII-7 are all >0.2 ppm (Fig. 7, which is published as supporting information on the PNAS web site). In contrast, only the β-protons of N5 are separated by >0.2 ppm in cMII-5. In addition, MII, cMII-6, and cMII-7 exhibit very similar βH shift differences for most residues, whereas they are substantially different in cMII-5.

Because the chemical shift analysis indicated that cMII-6 and cMII-7 were structurally similar to native MII, a full 3D structure of these two cyclic analogues was determined. Although the structure of cMII-5 is clearly different from that of MII, cMII-6, and cMII-7, based on the chemical shift data, a full 3D structure could not be determined because of a relative lack of interresidue NOEs, a characteristic that is indicative of a flexible molecule. The structure of cMII-6 was calculated by using 18 dihedral angles and 72 distance restraints, which consisted of 47 sequential, 19 medium-range, and 6 long-range NOEs. The φ angles were restrained to –120 ± 30° for N5 and H12 and –65 ± 15° for residues C2, C3, C8, H9, L10, S13, and H12. The χ¹ angle restraints and corresponding stereospecific assignments were applied to residues C2, C8, H9, and E11 (all 180 ± 30°) and C3, N5, H12, L15, and C16 (–60 ± 30°) based on coupling constants and NOE patterns. The structure of cMII-7 was calculated by using 18 dihedral angles and 91 distance restraints, which consisted of 61 sequential, 25 medium-range, and 5 long-range NOEs. The φ angles were restrained to –120 ± 30° for N5 and H12 and –65 ± 15° for residues C2, C3, C8, H9, L10, H11, and H12. The χ¹ angle restraints and corresponding stereospecific assignments were applied to residues C2, C8, H9, and E11 (all 180 ± 30°) and C3, N5, H12, L15, and C16 (–60 ± 30°) based on coupling constants and NOE patterns. Fifty structures were calculated for each peptide, and the 20 lowest-energy structures were chosen as representatives of the solution structures. Fig. 4 a and b shows the ensemble of 20 structures for cMII-6 and cMII-7, respectively, superimposed over the backbone atoms of residues G1 to C16. All structures satisfied the experimental restraints with only minor deviations from idealized covalent geometry (Table 5, which is published as supporting information on the PNAS web site).

Fig. 4.

The 3D structures of cMII-6 and cMII-7. (a) Stereroview of the superposition of backbone heavy atoms of 20 NMR-derived structures for cMII-6. (b) Stereroview of the superposition of backbone heavy atoms of 20 NMR-derived structures for cMII-7. (c) Superposition over the backbone heavy atoms of native MII (red), cMII-6 (blue), and cMII-7 (green), with disulfide bonds shown in yellow. (d) Ribbon representation of cMII-6. (e) Ribbon representation of cMII-7. (f) Ribbon representation of native MII (PDB ID code 1MII).

The structures of both cyclic analogues are well defined. The mean rms deviation (residues 1–16) from the average structure for the cMII-6 ensemble (Fig. 4a) is 0.41 ± 0.17 Å for the backbone atoms and 0.91 ± 0.19 Å for all heavy atoms; for cMII-7 (Fig. 4b), the value is 0.43 ± 0.13 Å for the backbone atoms and 0.98 ± 0.17 Å for all heavy atoms. Analysis of the structures with promotif indicates an α-helical region from residues P6 to G11 for both molecules, as shown in the ribbon representations of cMII-6 and cMII-7 in Fig. 4 d and e, respectively. The linker region of both peptides is disordered, which indicates either flexibility in the peptide backbone or a lack of medium- and long-range NOEs in this region of the molecule.

The structures of both cyclic analogues are consistent with that of MII (Fig. 4f). Fig. 4c illustrates an overlay of the peptide backbone of the mean structures of MII and its cyclic analogues. The three mean structures closely overlay, with a backbone rms deviation of 0.5 Å over residues G1 to C16.

Biological Activity. Bath application of 1 μM α-CTX MII reversibly inhibited nicotine (100 μM)-evoked depolarizing membrane currents in dissociated bovine chromaffin cells voltage-clamped at –60 mV to 59 ± 5% (n = 6) of control peak amplitude. cMII-5 failed to inhibit nicotine-evoked membrane currents (Fig. 5a). The maximum concentration of cMII-5 tested was 30 μM. The cyclic analogue cMII-6 was less active than the native α-CTX MII (75 ± 9%; n = 6), but increasing the linker by one residue (cMII-7) restored the potency of the toxin to inhibit nicotine-evoked currents (60 ± 4%; n = 6). MII inhibited nicotine (5 μM)-evoked catecholamine secretion with a half-maximal inhibitory concentration (IC₅₀) of 1.0 ± 0.2 μM (n = 12). cMII-5 failed to inhibit nicotine-evoked currents and secretion at tested concentrations. cMII-6 was slightly less active than the native toxin (IC₅₀ = 2.1 ± 0.2 μM; n = 12). Importantly, cMII-7 retained the full activity of the native peptide (IC₅₀ = 1.3 ± 0.4 μM; n = 12; Fig. 5b).

Fig. 5.

Retained biological activity for the cyclic analogues cMII-6 and cMII-7. (a) Superimposed nicotine (100 μM)-evoked currents recorded from isolated bovine chromaffin cells voltage-clamped at –70 mV in the absence (control) and presence of 1 μM native and cyclic derivatives of MII. (b) Concentration–response curves for inhibition of nicotine (5 μM)-evoked catecholamine release from isolated bovine chromaffin cells by increasing concentrations of MII, cMII-5, cMII-6, and cMII-7.

To examine the selectivity of cMII-7 for different nAChR subtypes, rat neuronal nAChR subunit combinations were expressed in Xenopus oocytes. Bath application of cMII-7 (100 nM) reversibly inhibited acetylcholine-mediated currents in the following order of efficacy: α3β2 > α7 ≫ α4β2 > α3β4 = α4β4 (n = 4–10) (Fig. 8, which is published as supporting information on the PNAS web site). This observation is consistent with the nAChR subtype selectivity observed for MII (8), demonstrating that cyclization does not perturb the receptor selectivity of the peptide.

Biological Stability. The stability of MII, cMII-5, cMII-6, and cMII-7 against cleavage by proteases was determined by incubating with EndoGluC, which cuts the peptide backbone on the C-terminal side of glutamic acid residues. In the case of MII and its cyclic analogues, the cleavage site is between E11 and H12. Incubation of the peptides for 10 h at 37°C followed by liquid chromatography/MS analysis of aliquots taken at various time points revealed that the stability of cMII-6 and cMII-7 was significantly increased relative to MII (Fig. 6a).

Fig. 6.

The stability of MII and its cyclic analogues against biological degradation. (a) The stability of MII (red), cMII-6 (blue), and cMII-7 (green) against proteolytic attack by EndoGluC. (b) The stability of MII, cMII-6, and cMII-7 in human blood plasma demonstrating that the stability of cMII-6 and cMII-7 is improved by 15–20% over native MII over 24 h.

To provide a biologically relevant assessment of stability, MII, cMII-6, and cMII-7 were incubated in human plasma at 37°C over 24 h, and the amount of intact peptide remaining was determined by RP-HPLC analysis. Fig. 6b shows that, as with the EndoGluC trial, a significant enhancement in stability was observed for cMII-6 and cMII-7 over MII. The combined data show that cyclization with appropriate linkers results in peptides that have similar 3D structures and activities to the native peptide but have reduced flexibility and improved stability to isolated enzymes and in human plasma.

Discussion

In this study, we have described the design, synthesis, and characterization of three cyclic analogues of α-CTX MII that contain a five-residue (cMII-5), six-residue (cMII-6), or seven-residue (cMII-7) linker joining the N and C termini. Of these three analogues, both cMII-6 and cMII-7 have 3D structures highly analogous to the native peptide, whereas the structure of cMII-5 is perturbed. The biological activity data reflect the structural data; cMII-6 and cMII-7 both have similar activities to MII, whereas cMII-5 is inactive. When incubated with the protease EndoGluC and in human plasma, the two active cyclic analogues are significantly more stable than the native peptide. These findings demonstrate that backbone cyclization improves the stability of peptide toxins while maintaining biological activity.

The 3D structures of MII and its two cyclic analogues, cMII-6 and cMII-7, are strikingly similar. Both cyclic analogues retain the same disulfide connectivity and possess the major secondary structural element, the α-helix between residues 6 and 11. Recently, it has been reported that mutation of residues N5, P6, and H12 to alanine causes a dramatic decrease in activity of MII at the α3β2 nAChR, whereas mutation of residues H9 and L15 causes a moderate reduction (26). The side-chain orientations of all of these residues in the two cyclic analogues are similar to those in MII, and the surface topologies of all three peptides are also almost identical. We therefore anticipated that the potency and selectivity of the cyclic molecules would be similar to the native peptide, which proved to be the case.

Native and cyclic derivatives of MII were tested for functional activity by means of their effects on nicotine-evoked currents and the release of catecholamines from bovine chromaffin cells. These cells are a valuable model for understanding the structure and function of nAChRs, and there is also a tight relationship between nAChR activation and catecholamine secretion. Native MII is a highly selective antagonist for α3β2 nAChR combinations (8). Nicotine-evoked catecholamine release was inhibited with an IC₅₀ of 1.0 μM, which is consistent with previous observations (27).

The three cyclic analogues of MII were tested for retained activity after molecular engineering. The striking structural similarity of cMII-6 and cMII-7 is reflected in their sustained ability to inhibit the activity of nAChR channels. The analogue cMII-7 retained the full activity of MII in both the inhibition of nicotine-evoked membrane currents and nicotine-induced catecholamine secretion from bovine chromaffin cells. Investigation of cMII-7 inhibition of cloned rat neuronal nAChR subunit combinations expressed in Xenopus oocytes demonstrated that the selectivity for α3β2 over α7, α3β4, α4β2, and α4β4 was unchanged from that reported previously for MII (8, 28). As a result of reducing the linker length by one residue, cMII-6 was marginally less active than MII and cMII-7, presumably due to subtle perturbations in the structure. Increasing strain further by reducing the linker size to five residues (cMII-5) results in the complete loss of structural integrity, as indicated by the NMR chemical shift analysis, and hence biological activity.

The dramatic enhancement in stability against proteolytic attack of the two cyclic analogues possessing a native-like structure can likely be attributed to a reduction in backbone flexibility of these two peptides. In general, potential cleavage sites within a protein are more susceptible to attack by proteases if they are located in flexible regions of the backbone (29). The deuterium exchange rates measured for MII and its cyclic analogues provide an indication of the flexibility of the peptide backbone. The deuterium exchange rates for both cMII-6 and cMII-7 were significantly decreased compared with those for MII, suggesting that the backbones of the cyclic peptides are more rigid. This increased rigidity appears to be subsequently reflected in the improved resistance of cMII-6 and cMII-7 to proteolytic attack.

Although cyclization of the peptide backbone provides a CTX with excellent protection against attack from proteases, it is clear that careful consideration is needed in selecting a correctly sized linker to span the distance between the N and C termini. As seen in the case of cMII-5, having a linker sequence that is too short introduces strain and perturbs the folding relative to the native structure. Yet by increasing the linker size by one residue, as in the case of cMII-6, this strain is sufficiently reduced to permit the formation of the correctly folded, and hence biologically active, peptide. When the linker size is increased to seven residues, the peptide forms almost exclusively the correctly folded form. The importance of linker size has also been illustrated in a study on the relative stability of a series of cyclic analogues of the PIN WW domain (30). By joining the termini of a linear molecule, it was possible to increase its thermodynamic stability, but the adverse effects of removing stabilizing charge–charge interactions between the termini were only overcome if a linker of correct length was added.

CTXs make excellent drug leads because they are small, easily synthesized, and selective for specific membrane receptors and ion channels. However, one drawback is their potential susceptibility to enzymatic degradation. The results of this study clearly demonstrate that cyclization is a successful method for overcoming this. Although MII is only one of thousands of conopeptides, the conserved features of the α-CTX family (i.e., α-helix and disulfide connectivity) indicate that cyclization would likely be successful for other α-CTXs.

The approach used for cyclizing CTXs is potentially applicable to a wide range of toxins and bioactive peptides and indeed to larger proteins because many have their termini within close proximity. An analysis of proteins in the Protein Data Bank (31) showed that ≈20% of all entries had their termini within 15 Å of one another (32). Because the termini of proteins are typically disordered, cyclization can be expected to substantially improve binding efficiency by means of entropic contributions, as well as confer protease resistance. Cyclization is potentially a powerful and widely applicable approach to improving the suitability of disulfide-rich peptides as drugs.