|Home | About | Journals | Submit | Contact Us | Français|
The E200K mutation of the human prion protein (PrP) is known to cause familial Creutzfeldt-Jakob disease. In order to elucidate the effects of the mutation on the local structural stability of PrP, we performed ab initio fragment molecular orbital calculations for the wild-type human PrP and the E200K variant modeled under neutral and mild acidic conditions. The calculations revealed that this substitution markedly altered the intramolecular interactions in the PrP, suggesting that the local structural instabilities induced by the E200K mutation might cause initial denaturation of the PrP and its subsequent conversion to a pathogenic form. This work presents a new approach for quantitatively elucidating structural instabilities in proteins that cause misfolding diseases.
The key event leading to the development of Creutzfeldt-Jakob disease (CJD) is the conversion of the prion protein (PrP) from its cellular isoform to an abnormal isoform PrPSc.1 However, the mechanism underlying the conversion of PrP to PrPSc and the detailed characteristics of PrPSc remain obscure. Familial CJD (fCJD) is inherited as an autosomal dominant trait that has been linked to several point mutations in the PrP. The E200K mutation, which is associated with fCJD, is of particular interest because it can be transmitted to mammals,2,3 and its clinical and neuropathogenic phenotypes are indistinguishable from those of sporadic CJD. The E200K variant may mimic the PrPSc structure; therefore, it is necessary to analyze this variant for elucidating the mechanism underlying the initial conversion of PrP to PrPSc.
Nuclear magnetic resonance (NMR) analysis revealed that the C-terminal globular structure of the E200K variant resembles that of the wild-type PrP (Fig. 1A). However, it has been established that there are minor differences in the structural conformations of amino acid residues in the wild-type PrP and those in the E200K variant, and that these differences are responsible for certain differences in the electrostatic surface potentials.4,5 Nevertheless, the influence of these differences on the properties of the PrP remains to be resolved. It has been argued that the identification of such conformational changes may require a novel approach. Recently, it was reported that an atypical bovine spongiform encephalopathy (BSE) case harbored a mutation that was similar to the human E200K substitution, and this raised concerns whether such a mutation might cause sporadic BSE.6,7 This case also emphasized the importance of analyzing the E200K variant.
Intramolecular interactions between the amino acid residues in a protein are highly dependent on the structural configurations of residue pairs, and these interactions may play a key role in the determination of the intrinsic conformations of not only the residues but also the local or global protein structures, such as secondary, tertiary or quaternary structures.8,9 The ab initio fragment molecular orbital (FMO) method is a promising all-electron quantum chemical procedure for estimating the molecular interactions within biomolecules.10,11 This method has been used to quantify the inter- and intramolecular interactions in various proteins and nucleic acids—interactions that contribute to the binding affinities and structural stabilities of these molecules.12–15 Very recently, this method has been successfully used to reveal the intermolecular interactions between PrP and an anti-prion drug.15
In this study, we performed the FMO calculations on the globular domain of wild-type human PrP and the E200K variant in order to elucidate the differences in their intramolecular interactions and thus their local structural stabilities. This is the first study on the application of FMO calculations for investigating the intramolecular interactions in the PrP; the method represents a new approach for the structural analysis of proteins.
The mutation from glutamic acid to lysine markedly changed the attractive and/or repulsive interactions between adjacent amino acid residues. The selected top 10 interfragment interaction energies (IFIEs) are shown in Table 1. The negative values of IFIEs imply an attractive interaction (stable) and positive values imply a repulsive (unstable) interaction. E200, which has a negatively charged carboxylate acid side chain group, has strong attractive interaction with positively charged residues—lysine and arginine, and further, histidine in mild acidic pH—and has strong repulsive interaction with negatively charged residues—glutamic acid and aspartic acid. In contrast, K200 has attractive interaction with the negatively charged residues—glutamic acid and aspartic acid—but has repulsive interaction with the positively charged residues—lysine and arginine, and further, histidine in mild acidic pH. However, the total IFIE value between residue 200 and other residues ranged from −55 to −97 kcal/mol, and no significant difference was observed with respect to the E200K mutation. Thus, the total IFIEs of residue 200 could not reveal the difference in the structural stability between wild-type PrP and the E200K variant.
We calculated the internal interaction energies (ΔEInt) of each secondary structure element in the wild-type and variant PrPs at neutral pH in order to assess their structural stabilities (Table 2). Because ΔEInt is obtained by summing the IFIEs between the residues in the secondary structure element, in which negative IFIE values imply an attractive interaction (stable conformation) and positive values imply a repulsive (unstable) interaction, it represents the net structural stabilities of the elements. Thus, a negative ΔEInt value indicates structural stability, whereas a positive value indicates instability. The ΔEInt values were slightly lower in the variant PrP, implying that each element tends to become unstable as a consequence of the mutation. However, all the ΔEInt values in the wild-type and variant PrPs were negative, indicating that each element retained its attractive properties and was structurally stable. Although the ΔEInt for the α3 helix was relatively higher in the E200K variant (+118 kcal/mol), due to an inverse electrostatic interaction between glutamine and lysine, there was no significant difference in the structural stability of this element in the wild-type PrP and E200K variant under either neutral (Table 2) or mild acidic (data not shown) pH conditions.
The intramolecular interaction energies (ΔEPair) of the secondary structure element pairs in the wild-type PrP and E200K variant under neutral and mild acidic conditions are shown in Table 3, and the pair interactions are illustrated in Figure 2. The ΔEPair values signify the structural stability of the element pairs: a negative value indicates an attractive force between the elements and denotes a structurally stable conformation, whereas a positive value indicates repulsion and instability. It should be noted that the total ΔEPair value in the wild-type PrP was similar to that in the mutant at both neutral and acidic pH, indicating that the total structural stability of the wild-type PrP was similar to that of the mutant PrP. However, at the level of individual element pairs, certain differences in the ΔEPair values of the wild-type and E200K variant were observed. These interaction differences can be attributed to both the replacement of Glu with Lys at position 200 and the structural differences in backbone coordinates and the orientations of side chains between the wild-type PrP and the E200K variant. The strong repulsive interaction between α1 and α3 in the wild-type PrP changed to a very strong attractive interaction in the E200K variant. Under both neutral and mild acidic conditions, the wild-type and mutant PrPs exhibited very strong attractive interactions at the following sub-regions: L1-α1-L2 and L3-α2-L4. It can be suggested that these skeletal components play an important role in the maintenance of the stability of the whole PrP. Furthermore, the interactions between L1 and L2 and L2 and α3 became weak in the E200K variant (Fig. 2A and B). There were very strong repulsive interactions between the L2-α2 and L1-α2 pairs in both the wild-type and variant PrPs under mild acidic conditions and very strong repulsive interaction between L2 and L1 under acidic conditions in the E200K variant (Fig. 2C and D).
In proteins, there exist diverse intramolecular interactions that are highly dependent on the structural conformations of not only the residues but also the secondary or tertiary structures. Such interactions are strongly associated with the structural stability of proteins.8,9 Thus, elucidating the intramolecular interactions in proteins would be necessary to understand the structural stability of proteins. Although the FMO calculations at the MP2 level can provide quantitative information on the molecular interactions between the residues in a protein,12–15 such an analysis based on individual interactions involves enormous computational complexity; for example, PrP129–224, which contains ~100 residues, could have more than 5,000 molecular interaction energies for residue pairs. Conversion of the intramolecular interactions between residues into those for secondary structure elements or pairs dramatically reduces the computational complexity and eases the interpretation of the calculation results.
In the human PrP, the substitution of glutamine (which strongly attracts positively charged residues) at position 200 with lysine (which is functionally unlike glutamine) might cause a drastic change in the intramolecular interactions, especially for the residue at position 200 (Table 1), leading to the structural rearrangement of the PrP. However, with the exception of minor differences, the secondary and tertiary structures of the wild-type PrP and E200K variant are very similar (Fig. 1A).5 Although the FMO calculations indicated that the structural stabilities of the secondary structure elements of the wild-type PrP and E200K variant were similar (Table 2), there were considerable differences between the wild-type and variant PrP with respect to the intramolecular interactions between the pairs of the secondary elements (Table 3). Notably, the pair interactions between α1 and α3, L1 and L2, L1 and α3, and L2 and α3 in the E200K variant were different than those in the wild-type PrP (Table 3 and Fig. 2). These results indicate that the E200K mutation markedly alters the local structural stability of PrP.
It has been suggested that PrPSc accumulates in the endosomes of scrapie-infected cells,19 which are characterized by mild acidic conditions (pH 4.0–6.0); furthermore, it has been suggested that acidic pH might trigger the conformational transition of PrP to PrPSc.20,21 The calculations revealed prominent alterations in the intramolecular interactions between certain element pairs under mild acidic conditions. We focused on the element pairs with repulsive interactions (unstable conformations) specifically observed in the E200K variant because unstable pairs might be inherently related to protein denaturation. Our results suggest that the structural instabilities in the local regions around L1–L2, which are specifically observed in the E200K variant, might trigger PrP denaturation, thereby initiating its conversion to PrPSc. Therefore, by stabilizing these sub-regions, it might be possible to prevent the formation of PrPSc.
It has been proposed that the anti-prion drug, 2-pyrrolidin-1-yl-N-[4-(4-[2-pyrrolidin-1-ylacetylamino]-benzyl)-phenyl]-acetamide (GN8), is bridged to four amino acid residues of PrP [N159 and Q160 (L2), K194 (α2), and E196 (L4)], and it inhibits the conversion of PrP to PrPSc.15,22 Our calculations showed that the L2-α2 region with weak interactions in the neutral pH was considerably destabilized under mild acidic pH condition, although the structural stability of the L2–L4 region with strong attractive interaction in the neutral pH model did not alter considerably in the mild acidic pH (Table 3 and Fig. 2). This result may suggest that the binding of GN8, especially to L2-α2 region in the endosomes, was effective in structurally stabilizing PrP, partially supporting the proposed mechanism of GN8.15,22
The structural stability of the entire PrP can be observed by protein denaturation measurements.23 The observed thermodynamic stability of the wild-type human PrP is similar to that of the E200K variant;23 this is consistent with the calculation results of the total ΔEPair values (Table 3). It was believed that the structural stability of the E200K variant was irrelevant to the conversion of PrP to PrPSc,23 and that the changes in the electrostatic surface properties of PrP caused by the E200K mutation, which alters the binding characteristics of PrP, is a dominant cause of fCJD.5 On the contrary, our theoretical calculations show that the intramolecular interactions in the local regions of the E200K variant were remarkably different from those in the wild-type PrP (Table 3 and Fig. 2), providing a new insight into the structural instability of the E200K variant. We suggest that E200K mutation-induced alteration in the local structural stability may also play a role in the etiology of fCJD. Furthermore, electron microscopic studies have shown that PrPSc has a β-helical motif (residues 89–175), and the α2 and α3 conformation of cellular PrP is retained in PrPSc.24 It can be suggested that the denaturation of a specific PrP region triggers the conversion of PrP to PrPSc. This emphasizes that the local structural instability in PrP and not global stability is important for elucidating the PrPSc conversion mechanism. Although partial unfolding in unspecified region(s) of PrP has already been monitored,25 present technique could not be used to analyze the local denaturation status of PrP. It should be noted that, presently, there are no experimental data for verifying the obtained FMO calculations.
The present study demonstrates that FMO calculation is a powerful tool for quantitatively elucidating the local structural stability of proteins and has the potential to provide new insights into protein misfolding diseases. To the best of our knowledge, this is the first attempt to evaluate the intramolecular interactions in the PrP using FMO calculations.
Initial atomic coordinates for the wild-type human PrP and E200K variant were obtained from the Protein Data Bank (PDB): codes 1QM3 and 1FO7, respectively. The wild-type PrP and E200K variant contain five major secondary structure elements: three α-helices (α1: 144–154, α2: 172–194 and α3: 200–224) and two β-strands (β1: 129–131 and β2: 161–163), as described previously.5 The loop regions (L1: 132–143, L2: 155–160, L3: 164–171 and L4: 195–199) were defined as those located between the major elements, as indicated in Figure 1B and andC.C. The C-terminal oxygen atom was included in the wild-type models owing to the lack of relevant coordinate information. The following assumptions were made for the neutral pH model: the lysines, arginines and N terminus were considered to be positively charged; the glutamic acids, aspartic acids and C terminus were considered to be negatively charged; and histidines were considered to be neutrally protonated at the Nτ atom. All the other amino acid residues were considered to be neutral. In the mild acidic pH model, as described previously,5 the protonation states were the same as in the neutral pH model, with the exception of histidines that had imidazolium side chains and thus were positively charged. The proteins were modeled in explicit water by arranging water molecules in a sphere at a distance of 10 Å from the surface of the protonated protein. After the models were optimized using an AMBER99 force field with constraints of a fixed force field for the whole protein, the water molecules were restricted to within 4.5 Å from the surface (typically, ~250 molecules); subsequently, the protein models in water were optimized under constraints of fixed force fields for heavy atoms, with the exception of the C-terminal carboxyl group in the wild-type PrP. The modeling procedures were performed by using the molecular operating environment (MOE).16
All-electron quantum chemical calculations using the ab initio FMO method were carried out using the commercial version17 of ABINIT-MP11 at the MP2/6-31G level.18 The IFIE, ΔEIJ, between fragments I and J was evaluated as ΔEIJ = EIJ − EI − EJ − Tr(ΔPIJVIJ), where EIJ is the dimer electronic energy for a fragment comprising the fragments I and J; EI and EJ are monomer electronic energies for fragments I and J, respectively; and ΔPIJ and VIJ are the difference density matrix and environmental electrostatic potential11 for the IJ dimer, respectively. During the calculation, the protein structure was divided into fragments per residue unit, except for the covalently bridged disulphide residues, Cys179 and Cys214, which were treated as a single S-S bond fragment. All the interaction energies were averaged for the top three conformers in the bundle of 20 conformers deposited in the PDB. The N-terminal region 125–128 for both the wild-type and variant proteins and the C-terminal region 225–228 for the wild-type PrP and region 225–231 for the E200K variant were excluded from the molecular interaction evaluations because of their relatively high structural flexibility.
The internal interaction energy ΔEPInt for the secondary structure element P was evaluated on the basis of the sum of all IFIEs (ΔEIJ) between the non-covalently bound residues, except for the S-S bond fragment:
The pair interaction energy between the elementsΔEPQPair between the elements P and Q was evaluated on the basis of the sum of three energy components:
where ΔEnon is the sum of the IFIEs (ΔEIJ) between the non-covalently bound residues I and J that belong to the elements P and Q, respectively. However, this equation excludes the S-S bond fragment; the molecular interaction energy contributions of this bond are taken into account in ΔESS as follows:
For the pair interaction, in which one of the element pairs P and Q is either α2 or α3, an interaction energy contribution ΔESS for the S-S bond fragment was added to ΔEPair and simply evaluated as half the interaction energy obtained from the sum of the IFIEs between the S-S bond fragment and each residue in the secondary structure element, except for α2 and α3. For the pair interaction between the covalently bridged α2 and α3, the S-S bond fragment contribution was excluded. In the case of covalently adjoining secondary structure element pairs, molecular interaction energy ΔEcov between the N-terminal and C-terminal residues in the adjoining elements was added to ΔEPair. This was simply evaluated from the IFIE between further cleaved side chain fragments or between a side chain group and a conventional fragment of corresponding pairs, in which the residues, except for glycine and threonine, were further divided into two fragments—main and side chain portions—by cutting the bond-detached atoms10,11 at the Cβ atom of serine, asparagine, aspartic acid, histidine, tyrosine and valine; at Cγ for glutamine and glutamic acid; and at Cδ for arginine and lysine. For valine, the interaction energy was evaluated as the sum of the IFIEs for two fragmented methyl groups.
We are very grateful to Dr. T. Nakano, National Institute of Health Science, for his kind support. We also thank Drs. O. Hino and M. Kobayashi, AdvanceSoft Corporation, for their helpful advice. The numerical calculations were supported by the Ministry of Agriculture, Forestry and Fisheries Research Network (MAFFIN). This study was supported by a Grant-in-Aid from the BSE and other Prion Disease Control Projects of the Ministry of Agriculture, Forestry and Fisheries, Japan.
Previously published online: www.landesbioscience.com/journals/prion/article/10890