Calcitonin Forms Oligomeric Pore-Like Structures in Lipid Membranes

INTRODUCTION

Calcitonin (CT) is a 32-amino acid peptide involved in hone calcium metabolism and clinically used in the treatment of ostcoporosis and other metabolic bone diseases. Its ability to lower the plasma calcium concentration has been widely studied and principally attributed to a strong inhibitory action on osleoclast-mediated bone resorption (1). It has been demonstrated that CT acts via a specific receptor found in many cell types and tissues, suggesting diverse biological roles for CT (2). In a recent work, Bums et al. demonstrated that CT gene-related peptide (CGRP) elevates calcium and polarizes membrane potential in MG-63 cells, an osteoblast cell line (3). They suggested that the peptide stimulates Ca^sup 2+^ influx in part through L-type voltage-dependent channels.

In the study of CT physiological activity, its ability of forming aggregates has seldom been considered. However, CT is generally considered an amyloid protein due to its aggregation behavior, which starts with the formation of oligomers and eventually results in the deposition of fibrils and plaques. Some authors consider annular aggregates as "off-pathway" intermediates in the process of fibrilization (4). Unexpected results were obtained when it was shown that CT is neurotoxic in vitro in the same manner and more so than other amyloid proteins involved in neurodegenerative pathologies (5,6). Many studies have been carried out to investigate the relationship between structural features and toxicity of amyloid proteins involved in diseases, such as ß-amyloid (ßA) in Alzheimer's disease (AD), a-Synuclein (aS) in Parkinson's disease (PD), prion in Creutzeldt-Jacob's disease, inslet amyloid polypeptide in type II diabetes, and others (7). Noticeably, Bucciantini et al. proved that oligomers of proteins not associated with any disease are also neurotoxic. suggesting that a shared structural feature of the amyloid oligomers is at the basis of neurotoxicity (8). This result is in good agreement with the general hypothesis that protein misfolding, leading to the exposure of hydrophobic regions of proteins, renders them potentially toxic, independent of their primary sequences (9). Misfolded proteins should have the ability to aggregate and interact with lipid membranes inducing cell damage and malfunction, mainly mediated via the formation of ion channels (7). Interestingly, this kind of toxicity named "hydrophobicity-based toxicity" is similar to those described for antibacterial toxins and viral proteins. This hypothesis is strongly supported by the observation that a single antibody, specifically designed for ßA, is able to recognize soluble oligomers of widely varying primary sequences of amyloid proteins (10).

In this framework, Lashuel et al. found a common structural feature for the A30P and A53T aS mutations associated with PD and ßA^sub ARC^ (arctic mutation) associated with AD (11). These proteins were shown to form oligomers rich in ß-sheet secondary structure, conformed as annular prototibrils with a diameter ranging 7-12 nm. Lin et al. (12) showed that ßA forms ion channels in lipid bilayers and imaged them for the first time by atomic force microscopy, suggesting that they were composed of hexamers or tetramers. Moreover, they measured the channel conductivity and tested the channel biological activity on neurons, demonstrating that ßA pore formation induces rapid neuritic degeneration and death by a Ca^sup 2+^-dependent toxicity mechanism (12). The pore formation and channel-like activity have been recently described for other amyloid proteins, suggesting a possible common mechanism for protein-misfolding diseases (13). A Ca^sup 2+^-dependent toxicity, based on pore-like prototibrils that cause membrane permeabilization, has been proposed for aS by Voiles et al. in PD (14,15).

Theoretical models have been proposed to predict ßA peptide ability to insert in lipid bilayer (16). Molecular dynamic simulation has been developed for ßA ion-channel formation in lipid bilayer on the basis of the amino acid sequence and experimental evidence of multilevel ion-channel conductance. Several pore arrangements can be obtained, all starting from a basic structure composed of a ß-hairpin followed by a helix-turn-helix motif (17).

Finally, it has been widely demonstrated that the secondary structure of membrane-bound ßA and its ability to insert into the lipid membrane are strongly affected by the lipid composition. In particular, the importance of "raft-like" lipid domains in the membrane has been recently pointed out, with particular attention to the presence of gangliosides (18). In particular. Kakio et al. showed that low ganglioside content favors the ß-structure conformation (19,20).

Here we report for the first time, to our knowledge, transmission electron microscopy (TEM) evidence of salmon calcitonin (sCT) annular oligomers very similar to those described for other amyloid proteins. Moreover, sCT oligomers form Ca^sup 2+^-permeable pores when inserted into rafts containing liposomes. Circular dichroism (CD) data demonstrate that the protein ß-structure content increases during the interaction with lipids.

MATERIALS AND METHODS

Preparation of sCT and sCT-liposome solutions

Lyophilized sCT (molecular mass = 3432 Da) was purchased from European Pharmacopoeia (EDQM. Strasbourg, France) and stored at -18

Transmission electron microscopy

Negative slain was obtained by a phosphotungstic acid (PTA) 2% w/v solution buffered at pH = 7.3 with NaOH. To avoid salt precipitation from PTA and/or NaOH, which can be misinterpreted as actual structures, the staining solution was liltered before each preparation through polycarbonate 0.2-µm pore filters. A droplet of the suspension, containing liposomes and proteins, was deposited onto 300 mesh copper grids for electron microscopy and covered with a very thin amorphous carbon film (~20 nm), and the excess of liquid was removed by placing the grid on filter work. When the grid was dried, a droplet of the staining solution was deposited and dried following the same procedure. The heavy metal (W) surrounding the specimen to he imaged scatters electrons more efficiently than the specimen itself, providing high image contrast and allowing a detailed examination of the structure (22). Moreover, heavy metal salts give good radiation protection and maintain the structural integrity under the electron beam bombardment. The samples were studied in a Zeiss 902 transmission electron microscope (Zeiss, Jena, Germany) operating at 80 kV and equipped with an electron energy loss filter. To enhance the contrast, the microscope was used in the electron spectroscopy imaging (ESI) mode filtering at ?E = 0 eV. The image acquisition was performed by a digital charge-coupled device camera model HSC2, 1k for 1k pixels, (Proscan, Lagerlechfeld, Germany), lhermoslated by a Peltier cooler model WKL 230 (LAUDA, Lauda-Königshofen, Germany). Image analysis and quantification was performed by a digital image analyzer analy SIS 3.0 (SiS, Klausdorf, Germany). This software allows us to enhance contrast and sharpness of the acquired images and to perform morphological quantification and statistics. The dimensional measurements were performed after a careful magnification calibration of the whole imaging system based on reference standards (cross grating and catalase crystal). For statistical analysis, ~260 objects were considered from at least five fields.

sCT immunogold labeling

sCT liposomes were deposited onto thin carbon film-coated grids for TEM observation and air dried. For immunolocalization of sCT, samples were preincubated for 5 min with 1% w/v bovine serum albumin (BSA) (Sigma Chemical, St. Louis, MO) in phosphate buffer (0.1 M, pH 7.0) and then incubated for 30 min at room temperature with mouse monoclonal anti-sCT (Abcam, Cambridge, UK) at a concentration of 40 µg/ml. After washing for 10 min by floating the grids on phosphate buffer drops containing 1% w/v BSA, samples were labeled with anti-mouse IgG 5-nm gold conjugate diluted 1:10 (Sigma Chemical) for 30 min at room temperature. Control samples were obtained as described above both by omission of the primary antibody and on liposomes without sCT. After washing with phosphate buffer, samples were negatively stained with PTA as described above.

Circular dichroism measurements

CD measurements were performed on a Jasco J-710 spectropolarimetcr (Jasco, Tokyo, Japan) in the far-ultraviolet region (260-195 nm). All spectra recorded were blank subtracted and were the mean of 8-16 different scans. Quartz cells of 0.1-cm path length were employed. The spectral step resolution was 0.1 nm and the speed scan 20 nm/min. All CD spectra were reported as ?e -= [T]/3300, where [T] is equal to (? × 100)/(1 × C) (? is the measured elliplicity, C is the aminnacid concentration (= 1.2 × 10^sup -5^ × 32), and 1 is the path length in centimeter). Conformational analyses of CD spectra were performed using the CDSSTR routine in a software package for analyzing protein CD spectra via the Internet (23).

Ca^sup 2+^ influx measurements

Ca^sup 2+^ influx was measured using Fluo-4 as Ca^sup 2+^ indicator dye. Fluo-4 nonentrapped in liposomes was separated by filtration of 200 µl of liposomes solution on a 1-ml Biorad (Hercules, CA) AM-15 gel column, equilibrated in HEPES/EGTA buffer solution. Fractions containing liposomes were identified, with the first live fractions that were combined and diluted to 2 ml. Fluorescence measurements were performed at room temperature with a Shimadzu (Columbia, MD) RF5001PC spectrolluorimeler in a 1-cm cuvette. Excitation wavelength was 494 nm. whereas emission intensity was revealed in the range 500-540 nm with slits set to 5 mm. Ca^sup 2+^ influx was induced by addition of 20 µl of 17.2 mM CaCl^sub 2^ to the cuvette to obtain a final calcium concentration (0.172 mM) that does not cause fusion of liposomes and dispersion of the dye in the hulk (24).

RESULTS AND DISCUSSION

In previous articles, we investigated the early stages of sCT aggregation induced by ageing or oxidative conditions (25,26). In this work we report, for the first time, to our knowledge, images of annular sCT oligomers of shape and dimension very similar to ßA and aS protofibrils described in the literature.

Fig. 1 shows negatively stained sCT annular oligomers observed by TEM. The statistical distribution of their circle perimeters is also shown. As reported tor ßA, also in the case of sCT, the dominant symmetry was hexagonal even if small globules and incomplete hexagons and/or linear protolihrils were also observed. Considering sCT low molecular weight, the small globules could be oligomers themselves and the annular structures could be assemblies of oligomers. This structural feature has been described for ßA and aS and is considered to be at the basis of the "hydrophobicity-based neurotoxicity". However. sCT is not directly involved in any neurologic disease, even if, unexpectedly, its neurotoxicity was observed in in vitro experiments (5,6).

We reconstituted sCT in DPPC liposomes and in rafts containing liposomes (DPPC/cholesterol/monoganglioside-G^sub MI^) to investigate the differences in the interaction of the protein with these two model membranes, with particular attention to the effect induced by the presence of lipid rafts (27).

Immunogold labeling (Fig. 2) performed on sCT reconstituted in DPPC/eholesterol/G^sub MI^ liposomes allowed us to demonstrate that the protein strongly binds to rafts containing liposomes, whereas plain DPPC liposomes were seldom decorated, as we previously reported (26). This experiment clearly indicates that a specific stronger interaction of sCT with rafts occurred.

Fig. 3 shows TEM images of rafts containing liposomes opened onto amorphous carbon substrate in the presence of sCT. Gray islands surrounded by a dark background characterize the images. They represent liposomes opened and fused with each other, together with some unopened liposomes that appear white (28). As can be observed in Fig. 3, annular structures are easily recognizable on the dark background. Fig. 3, B and C, shows high magnification images of two different annular oligomers not interacting with liposomes, very similar in shape and dimension to those reported in Fig. 1 in the absence of liposomes. More interestingly, annular structures are also often detectable in liposomes, as shown in Fig. 3, D and E. They seem to be smaller than the isolated ones and characterized by a hexagonal symmetry. The diameter of the internal pores of these structures was ~3 mn. This is compatible with the pore size of most Ca^sup 2+^-permeable channels. In our opinion, they represent the direct visualization of the ion channels formed by sCT in the lipid bilayer. It is worth noting that these annular structures were never observed in the samples of both sCT-and raft-free liposomes.

These findings strongly suggest that sCT annular oligomers interact with the lipid bilayer-containing rafts, likely forming Ca^sup 2+^-permeable channels, as previously proposed by Stipani et al. (29) on the basis of electrophysiological measurements. This is the first direct evidence, to our knowledge, of sCT channel formation in a lipid environment and is in good agreement with the observation of pores formed by many other amyloid proteins reconstituted in liposomes and studied by atomic force microscopy (13). Moreover, the hexagonal symmetry has been proposed in the numerical simulation of the amyloid channels (17).

Fig. 4 shows CD spectra of sCT in the presence of liposomes, with or without rafts, together with the relative percentages of a-, ß-, and unordered structures obtained by deconvolution software (23). It is well known that the aggregation process of amyloid proteins is generally accompanied by an increase in ß-content (26.30). An evident rising of the ß-structure and a corresponding reduction of the random-coil content can be observed in rafts containing samples, whereas an essentially random-coil structure, very similar to the typical sCT conformation in water (26). is observed in plain DPPC liposomes. This conformational change can be clearly attributed to the interaction of sCT with the lipid rafts. Our result is very similar to that published for ganglioside-bound ßA (19) and is consistent with the secondary structure foreseen in the theoretical models (molecular dynamic) of ion channels (17).

Finally, Fig. 5 shows fluorescence spectra of liposomes loaded by Fluo-4. The presence of sCT reconstituted in rafts containing liposomes allows Ca^sup 2+^ entry, whereas the addition of Ca^sup 2+^ in the solution of plain rafts containing liposomes does not alter the spectral shape. This result demonstrates that Ca^sup 2+^ enter liposomes only when sCT is present and strongly suggests that the pore-like structures observed in the lipid bilayer are Ca^sup 2+^-permeable ion channels.

In addition, we found that the presence of rafts is required for Ca^sup 2+^ entry, since fluorescence spectra of plain DPPC liposomes do not evidence any significant difference after the reconstitution of sCT (Fig. 5, C and D). This result a supports the strategic role played by lipid rafts in facilitating the membrane insertion of amyloid proteins.

All our results showed a dramatic similarity between ßA, aS, and sCT oligomers. In particular, we observed an annular morphology for sCT similar to the one observed for the A30P and A53T aS and ßA^sub ARC^ mutations (11), and we directly visualized that these structures actually interact with lipids. Moreover, in agreement with CD data relative to other amyloids (11,19), we found an increase in ß secondary structure when sCT interacts with model membranes containing rafts. Finally, we show that annular oligomers are permeable to Ca^sup 2+^, providing a demonstration of their functional activity as in the case of ßA. This activity is strictly dependent on the presence of rafts, substantiating their role in the possibly harmful pore formation.

All these structural and functional data are in good agreement with the general hypothesis (9) that protein misfolding leads to the exposure of hydrophobic regions and renders proteins potentially cytotoxic via the formation of ion channels. In addition, these results again corroborate the hypothesis that amyloid proteins, although lacking sequence homology, have a similar aggregation behavior. It is interesting to note that CT has been found to be toxic for neurons in vitro (5,6).

Most CT activities are mediated by a G protein-coupled transmembrane receptor in osteoclasts (2). Recently, it has been proposed that CT could form ion channels in osteoblast plasma membrane, leading to Ca^sup 2+^ influx (29). Here we provide, for the first time, experimental evidence that sCT oligomerization actually leads to the formation of annular structures able to interact with model membranes, in the presence of lipid rafts, with Ca^sup 2+^-permeable channel activity.

CONCLUSIONS

The structural investigation here reported clearly indicates that sCT, physiologically involved in the regulation of calcium in the bone tissue, is able to form annular oligomers during the early stages of fibrillar aggregation. These oligomers are very similar to those observed for proteins involved in important neurodegenerative diseases, such as AD and PD. Moreover, the presence of rafts in model membranes strongly favors the binding of sCT, with the consequent increase of its ß-secondary structure content. Conversely, sCT interacts slightly with plain DPPC liposomes, without any significant conformational modification (26). The annular sCT oligomers have been directly visualized in the interaction with raft-containing liposomes, forming amyloid ion channels. Finally, we demonstrate that these structures play a functional role, allowing Ca^sup 2+^ to cross the lipid membranes. It is worth noting that in the absence of rafts, sCT reconstituted in plain DPPC liposomes does not induce Ca^sup 2+^ entry.

Recently, a mechanism has been proposed (schematically represented in Fig. 6) based on the formation of pore-like oligomers that penetrate the lipid bilayer to explain the neurotoxicity of several amyloid proteins (15). However, we wonder if this mechanism also has physiological significance in the case of the role played by CT in calcium homeostasis.

We thank Dr. Giovanna Mancini of the National Research Council, Istituto di Metodologie Chimiche, and Dipartimento di Chimica, Università degli Stndi di Roma "La Sapienza" for the kind cooperation and helpful discussions, and Mr. Carmine De Sena of Dipartimento del Farmaco, Istituto Superiore di Sanità. for technical assistance.

This research was supported by a grant from the Italian National Health Fund for the project "Safety of Drugs Used in Diseases of the Elderly" (Progetto 1%-Ricerca Sanitaria Finalizzata 2000).

© 2006 Biophysical Society Provided by ProQuest LLC. All Rights Reserved.