- Email: [email protected]
Identification of the importance of the secondary structure of Alzheimer's disease amyloid
Journal of
MOLECULAR STRUCTURE Journal of Molecular
Structure 408/409
(I 997) 18% 189
Identification of the importance of the secondary structure of Alzheimer ’s disease amyloid F. Cavillonajb”, A. Elhaddaouia,
A.J.P. AlixbTc, S. Turrella3*, M. Dauchezb’”
“Laboratoire de Spectrochimie Infrarouge et Raman (CNRS, UPR A 2631 L), UniversiiP des Sciences et Technologies de Lille, F-59655 Villeneuve d’Ascq, France bLaboratoire de Spectroscopies et Structures Biom~lkulaires, URCA, Facult6 des Sciences, B.P. 1039, F-51687 Reims Cedex 2, France ‘CHRU, INSERM U314, 45 rue Cognacq Jay, F-5/092 Reims Cedex, France Received 26 August 1996; revised 23 October 1996; accepted 28 October 1996
Abstract Studies on the nature of the interaction between the marker dye, Congo red and Alzheimer’s amyloid protein have been continued. Band-decomposition methods used on the infrared spectra clarify the changes in the SO; spectral region which result
from dye-protein interactions and confirm the implication of this group during fixation. In order to complete a model for the mechanism of fixation, macromolecular modelling techniques have been used to obtain optimal tertiary structures of amyloid peptides l-28 and l-43. Finally, docking calculations indicate that arginine-5 and lysine- I6 might be key residues involved in an interaction between Congo red and amyloid proteins. 0 1997 Elsevier Science B.V. Keywords:
Alzheimer’s
disease amytoid; Congo red: Infrared spectroscopy;
1. Introduction Alzheimer’s disease is a progressive and irreversible disease characterized by an intra- and extra-neuronal accumulation of beta-amyloid @A) proteins in the cerebral tissue forming senile plaques [l]. These deposits of PA peptide (composed of 39 to 43 aminoacid residues) are identified during post mortem examinations by their reactions with certain specific dyes, Congo red (CR) for example. In order to develop a dye molecule with the properties of CR but which can be used in in vivo examinations (i.e. detecting the presence of amyloid in living patients), it is necessary * Corresponding [email protected].
author. Fax: +33 3 2043 6755: e-mail: sylvia.tur-
Molecular modelling
to determine the characteristics essential for assuring the dye-protein interaction. Several studies have been undertaken to elucidate the mechanism of the interaction between CR and /3A [2-51. Most of the previous work concerned model proteins. However, NMR studies [6] suggested that the C-terminal 29-43 segment of PA is embedded in the plaque as a result of the deposition process, whereas the N-terminal l-28 segment should remain exposed, and hence be more available to interact with a staining dye. For this reason, our subsequent studies have concerned dye interactions with the /3( l-28) fragment of Alzheimer’s amyloid. However, the exact role and importance of each part of the fragment still remains a question. In the present work FT-IR and Raman studies have been undertaken on CR/amyloid
0022-2860/97/$17.00 0 1997 Elsevier Science B.V. All rights reserved PI/ SOO22-2860(96)09673-l
186
F. Cavillon et al./Joutnal of Molecular Structure 408/409 (1997) 185-189
systems. To complement the spectroscopic data, modelization techniques have been employed on the amyloid peptide to predict the probable sites of interaction with CR.
2. Results and discussion Our previous results [3,4] indicated that the interaction between CR and Alzheimer-type proteins involved positively charged sites in the protein and the sulfonyl and ring groups of CR. The implication of the sulfonyl groups was based on changes in band shapes observed in the 1020 to 1270 cm-’ region of the FTIR spectra upon complexation. This region is principally assigned to vibrational modes of the S-O and C-S groups of the SO; part of CR. The FTIR spectra of CR with /3( l-28) present these same variations, as can be evidenced in Fig. 1, where a band decomposition is presented for this region. It is observed that this band appears to have three, not two components. Moreover, whether the reaction is with model peptides or with various amyloid fragments, it is the high-frequency component which disappears upon complexation. Ff and resonance Raman studies of these systems make it obvious that the binding between the dye and the protein involves not only interactions between the sulfonyl groups of CR and positively charged ionic sites in the protein, but also van-der-Waals-type attractions due to the extensive aromatic system. Thus the complexing site of the protein should possess a distance between positive entities which is comparable to that which separates the two SO; groups of CR, as well as an extended local conformation which can accommodate the flat aromatic part of the dye. In order to complete a model of CR fixing in such a hypothetical site, the tertiary structure of Alzheimer’s proteins needed to be understood. Table 1 Pairs of residues in the amyloid peptide separated Dielectric constant
1 (vacuum) 80 (solvent) 4r (solvent)
(E)
For the elucidation of the tertiary structures of the amyloid peptides, techniques of macromolecular modelling were used for obtaining (i) stable structures of the dye, (ii) stable 3D structures of the amyloid peptide with different secondary structure patterns and (iii) docking or interactions between the dye and the peptide. Various models of CR and /3( l-28) (a-helical and P-strand) were built and minimized using the BIOSYM package (INSIGHT II + HOMOLOGY + DISCOVER force field). In all cases, the models were optimized by first using 500 steps of steepest descent and then by using 2500 steps of conjugate gradient energy minimization methods. All the calculations were performed both in vacua (with a dielectric constant of E = 1) and in a non-explicit solvent (using dielectric constant values of E = 80 and E = 4r, the latter being distance-dependent). In the case of CR, with a torsion angle of 20’ between the two central phenyl rings, one obtains a molecule which is not quite planar but in which the sulfur atoms of the two SO; gro!ps are always separated by between 19.5 and 20.5 A. In the case of /3(1-28), the first approach was to construct an o-helical structure for the peptide fragment whose primary structure corresponds to that of the system in trifluoroethanol solution studied by NMR [5]. (The TFE promotes intramolecular hydrogen bonding, thus stabilizing the formation of a helical structure.) Secondly, using the primary structure (i.e. the sequence) of P( l-28) as input data, an (Yhelical structure (which could represent the soluble form of the proteins) and an all-P-strand conformation (which could represent the deposited form) were obtained by homology. The structures obtained using the NMR data, were then compared with the minimized structures obtained from the standard o-helical model built by homology. The only small differences observed between these
by a distance similar to that separating
Helical structure (using NMR data)
Helical structure (using homology)
Arg-5-Lys-16 Arg-5-Lys-16 Arg-SLys16
His-14-Lys-28 Arg-5-Lys-16
the sulfonyl groups in CR P-strand structure (using homology)
Arg-5-His- 13 His-6-His-14
F. Cavillon et al./Journal of Molecular Structure 408/409 (1997) 185-189
axn!qJosq~
187
188
F. Cavillon et ad/Journal of Molecular Structure 4OW409 (1997) 185-189
Fig. 2. Docking of the dye CR with the a-helical 28) amyloid peptide.
form of the fl( I-
structures are in the orientations of certain hydrophobic parts of the molecule [5]. In observing the nine final structures obtained for &l-28) (three models for each E value), very few sequences contain structures containing positively charged sites spaced by distances which can correspond to the separation between the two sulfonyl groups on CR. Appropriate structures are indicated in Table 1 and immediately one can note the repetition of the arginine-5-lysine-16 distance which is consistent with the NMR results. The highly hydrophilic residue lysine- 16 is always solvent-exposed and hence is a good candidate for an anchoring point. Further calculations were done with the full l-43 sequence in both of the previous conformations but
with additional constraints imposed on the different types of turns in the 28-32 region as proposed by standard predictive methods. It is found that while there is little difference for the starting N-terminal segment, the end C-terminal segment always adopts an extended P-strand conformation. In the final step of docking CR with the peptide models, initial structures were taken from the previously described minimized geometries. Calculations were performed using the DOCKING module in the BIPSYM package. In a first s,tep, a grid box with 5 A on each edge and a 0.5 A step was obtained for each structure. The hydrogen-bond option was then used and the non-bonded interactions and electrostatic contributions were evaluated for each point of the grid box. The partial charges used for the peptide were taken from the Amber libraries for proteins, while the charges for CR were derived from previous MNDO semi-empirical calculations. Numerous initial positions of both structures (peptide in alpha or beta forms plus CR) were then obtained. For each case, after the computation of non-bonded energies, the favorable assemblies were kept and minimized over all degrees of freedom. With the helical form of fl( l-289, the best interactions were obtained when the CR-sulfonyl groups were in interaction with arginine-5 (where a bifurcated hydrogen bond appeared) and lysine- 16 (see Fig. 2). In the case of the P-strand conformation for fl( l-28), the best association was obtained when interactions appeared between arginine-5 and histidine-13 (see Fig. 3). Regardless of the length of
Fig. 3. Docking of the dye CR with the P-strand form of the /3(I-28)
amyloid peptide.
F. Cavillon et al./Journal
of Molecular
the sequence used (calculations were also done for &l-43)), the docking of CR is optimal when the Nterminal fragment (l-28), which would normally be found at the surface of the plaque, adopts an o-helical secondary structure. In all cases, the most accessible anchoring points for the dye in Alzheimer’s disease amyloid proteins seem to be arginine-5 and a site located between histidine- 13 and lysine- 16.
Acknowledgements The authors would like to acknowledge financial from the France Alzheimer Association.
support
Structure
408/409 (1997) 185-189
189
References [l] G.G. Glenner and C.W. Wong, Biochem. Biophys. Res. Commun., 120 (1984) 885. [2] E. Klunk, J.W. Pettegrew and D.J. Abraham, J. Histochem. Cytochem., 20 (1989) 1273. [3] A. Elhaddaoui, E. Pigorsch, A. Delacourte and S. Turrell, J. Mol. Struct., 347 (1995) 363. [4] E. Pigorsch. A. Elhaddaoui and S. Turrell, Spectrochim. Acta, 12 (1994) 2145. t-1 F. Cavillon, A. Elhaddaoui, M. Dauchez, S. Turrell and A.J.P. Alix, in: Spectroscopy of Biological Molecules, Eds. J.C. Merlin, S. Turrell and J.P. Huvenne (Eds.), Kluwer, Dordrecht, 1995, p. 491. [6] M.G. Zagorski and C.J. Barrow, Biochemistry, 31 (1992) 5621.
MOLECULAR STRUCTURE Journal of Molecular
Structure 408/409
(I 997) 18% 189
Identification of the importance of the secondary structure of Alzheimer ’s disease amyloid F. Cavillonajb”, A. Elhaddaouia,
A.J.P. AlixbTc, S. Turrella3*, M. Dauchezb’”
“Laboratoire de Spectrochimie Infrarouge et Raman (CNRS, UPR A 2631 L), UniversiiP des Sciences et Technologies de Lille, F-59655 Villeneuve d’Ascq, France bLaboratoire de Spectroscopies et Structures Biom~lkulaires, URCA, Facult6 des Sciences, B.P. 1039, F-51687 Reims Cedex 2, France ‘CHRU, INSERM U314, 45 rue Cognacq Jay, F-5/092 Reims Cedex, France Received 26 August 1996; revised 23 October 1996; accepted 28 October 1996
Abstract Studies on the nature of the interaction between the marker dye, Congo red and Alzheimer’s amyloid protein have been continued. Band-decomposition methods used on the infrared spectra clarify the changes in the SO; spectral region which result
from dye-protein interactions and confirm the implication of this group during fixation. In order to complete a model for the mechanism of fixation, macromolecular modelling techniques have been used to obtain optimal tertiary structures of amyloid peptides l-28 and l-43. Finally, docking calculations indicate that arginine-5 and lysine- I6 might be key residues involved in an interaction between Congo red and amyloid proteins. 0 1997 Elsevier Science B.V. Keywords:
Alzheimer’s
disease amytoid; Congo red: Infrared spectroscopy;
1. Introduction Alzheimer’s disease is a progressive and irreversible disease characterized by an intra- and extra-neuronal accumulation of beta-amyloid @A) proteins in the cerebral tissue forming senile plaques [l]. These deposits of PA peptide (composed of 39 to 43 aminoacid residues) are identified during post mortem examinations by their reactions with certain specific dyes, Congo red (CR) for example. In order to develop a dye molecule with the properties of CR but which can be used in in vivo examinations (i.e. detecting the presence of amyloid in living patients), it is necessary * Corresponding [email protected].
author. Fax: +33 3 2043 6755: e-mail: sylvia.tur-
Molecular modelling
to determine the characteristics essential for assuring the dye-protein interaction. Several studies have been undertaken to elucidate the mechanism of the interaction between CR and /3A [2-51. Most of the previous work concerned model proteins. However, NMR studies [6] suggested that the C-terminal 29-43 segment of PA is embedded in the plaque as a result of the deposition process, whereas the N-terminal l-28 segment should remain exposed, and hence be more available to interact with a staining dye. For this reason, our subsequent studies have concerned dye interactions with the /3( l-28) fragment of Alzheimer’s amyloid. However, the exact role and importance of each part of the fragment still remains a question. In the present work FT-IR and Raman studies have been undertaken on CR/amyloid
0022-2860/97/$17.00 0 1997 Elsevier Science B.V. All rights reserved PI/ SOO22-2860(96)09673-l
186
F. Cavillon et al./Joutnal of Molecular Structure 408/409 (1997) 185-189
systems. To complement the spectroscopic data, modelization techniques have been employed on the amyloid peptide to predict the probable sites of interaction with CR.
2. Results and discussion Our previous results [3,4] indicated that the interaction between CR and Alzheimer-type proteins involved positively charged sites in the protein and the sulfonyl and ring groups of CR. The implication of the sulfonyl groups was based on changes in band shapes observed in the 1020 to 1270 cm-’ region of the FTIR spectra upon complexation. This region is principally assigned to vibrational modes of the S-O and C-S groups of the SO; part of CR. The FTIR spectra of CR with /3( l-28) present these same variations, as can be evidenced in Fig. 1, where a band decomposition is presented for this region. It is observed that this band appears to have three, not two components. Moreover, whether the reaction is with model peptides or with various amyloid fragments, it is the high-frequency component which disappears upon complexation. Ff and resonance Raman studies of these systems make it obvious that the binding between the dye and the protein involves not only interactions between the sulfonyl groups of CR and positively charged ionic sites in the protein, but also van-der-Waals-type attractions due to the extensive aromatic system. Thus the complexing site of the protein should possess a distance between positive entities which is comparable to that which separates the two SO; groups of CR, as well as an extended local conformation which can accommodate the flat aromatic part of the dye. In order to complete a model of CR fixing in such a hypothetical site, the tertiary structure of Alzheimer’s proteins needed to be understood. Table 1 Pairs of residues in the amyloid peptide separated Dielectric constant
1 (vacuum) 80 (solvent) 4r (solvent)
(E)
For the elucidation of the tertiary structures of the amyloid peptides, techniques of macromolecular modelling were used for obtaining (i) stable structures of the dye, (ii) stable 3D structures of the amyloid peptide with different secondary structure patterns and (iii) docking or interactions between the dye and the peptide. Various models of CR and /3( l-28) (a-helical and P-strand) were built and minimized using the BIOSYM package (INSIGHT II + HOMOLOGY + DISCOVER force field). In all cases, the models were optimized by first using 500 steps of steepest descent and then by using 2500 steps of conjugate gradient energy minimization methods. All the calculations were performed both in vacua (with a dielectric constant of E = 1) and in a non-explicit solvent (using dielectric constant values of E = 80 and E = 4r, the latter being distance-dependent). In the case of CR, with a torsion angle of 20’ between the two central phenyl rings, one obtains a molecule which is not quite planar but in which the sulfur atoms of the two SO; gro!ps are always separated by between 19.5 and 20.5 A. In the case of /3(1-28), the first approach was to construct an o-helical structure for the peptide fragment whose primary structure corresponds to that of the system in trifluoroethanol solution studied by NMR [5]. (The TFE promotes intramolecular hydrogen bonding, thus stabilizing the formation of a helical structure.) Secondly, using the primary structure (i.e. the sequence) of P( l-28) as input data, an (Yhelical structure (which could represent the soluble form of the proteins) and an all-P-strand conformation (which could represent the deposited form) were obtained by homology. The structures obtained using the NMR data, were then compared with the minimized structures obtained from the standard o-helical model built by homology. The only small differences observed between these
by a distance similar to that separating
Helical structure (using NMR data)
Helical structure (using homology)
Arg-5-Lys-16 Arg-5-Lys-16 Arg-SLys16
His-14-Lys-28 Arg-5-Lys-16
the sulfonyl groups in CR P-strand structure (using homology)
Arg-5-His- 13 His-6-His-14
F. Cavillon et al./Journal of Molecular Structure 408/409 (1997) 185-189
axn!qJosq~
187
188
F. Cavillon et ad/Journal of Molecular Structure 4OW409 (1997) 185-189
Fig. 2. Docking of the dye CR with the a-helical 28) amyloid peptide.
form of the fl( I-
structures are in the orientations of certain hydrophobic parts of the molecule [5]. In observing the nine final structures obtained for &l-28) (three models for each E value), very few sequences contain structures containing positively charged sites spaced by distances which can correspond to the separation between the two sulfonyl groups on CR. Appropriate structures are indicated in Table 1 and immediately one can note the repetition of the arginine-5-lysine-16 distance which is consistent with the NMR results. The highly hydrophilic residue lysine- 16 is always solvent-exposed and hence is a good candidate for an anchoring point. Further calculations were done with the full l-43 sequence in both of the previous conformations but
with additional constraints imposed on the different types of turns in the 28-32 region as proposed by standard predictive methods. It is found that while there is little difference for the starting N-terminal segment, the end C-terminal segment always adopts an extended P-strand conformation. In the final step of docking CR with the peptide models, initial structures were taken from the previously described minimized geometries. Calculations were performed using the DOCKING module in the BIPSYM package. In a first s,tep, a grid box with 5 A on each edge and a 0.5 A step was obtained for each structure. The hydrogen-bond option was then used and the non-bonded interactions and electrostatic contributions were evaluated for each point of the grid box. The partial charges used for the peptide were taken from the Amber libraries for proteins, while the charges for CR were derived from previous MNDO semi-empirical calculations. Numerous initial positions of both structures (peptide in alpha or beta forms plus CR) were then obtained. For each case, after the computation of non-bonded energies, the favorable assemblies were kept and minimized over all degrees of freedom. With the helical form of fl( l-289, the best interactions were obtained when the CR-sulfonyl groups were in interaction with arginine-5 (where a bifurcated hydrogen bond appeared) and lysine- 16 (see Fig. 2). In the case of the P-strand conformation for fl( l-28), the best association was obtained when interactions appeared between arginine-5 and histidine-13 (see Fig. 3). Regardless of the length of
Fig. 3. Docking of the dye CR with the P-strand form of the /3(I-28)
amyloid peptide.
F. Cavillon et al./Journal
of Molecular
the sequence used (calculations were also done for &l-43)), the docking of CR is optimal when the Nterminal fragment (l-28), which would normally be found at the surface of the plaque, adopts an o-helical secondary structure. In all cases, the most accessible anchoring points for the dye in Alzheimer’s disease amyloid proteins seem to be arginine-5 and a site located between histidine- 13 and lysine- 16.
Acknowledgements The authors would like to acknowledge financial from the France Alzheimer Association.
support
Structure
408/409 (1997) 185-189
189
References [l] G.G. Glenner and C.W. Wong, Biochem. Biophys. Res. Commun., 120 (1984) 885. [2] E. Klunk, J.W. Pettegrew and D.J. Abraham, J. Histochem. Cytochem., 20 (1989) 1273. [3] A. Elhaddaoui, E. Pigorsch, A. Delacourte and S. Turrell, J. Mol. Struct., 347 (1995) 363. [4] E. Pigorsch. A. Elhaddaoui and S. Turrell, Spectrochim. Acta, 12 (1994) 2145. t-1 F. Cavillon, A. Elhaddaoui, M. Dauchez, S. Turrell and A.J.P. Alix, in: Spectroscopy of Biological Molecules, Eds. J.C. Merlin, S. Turrell and J.P. Huvenne (Eds.), Kluwer, Dordrecht, 1995, p. 491. [6] M.G. Zagorski and C.J. Barrow, Biochemistry, 31 (1992) 5621.