Skip to main content
NCBI home page
Protein Science : A Publication of the Protein Society logoLink to Protein Science : A Publication of the Protein Society
. 2014 Aug 23;23(11):1559–1571. doi: 10.1002/pro.2534

Conformational and aggregation properties of the 1–93 fragment of apolipoprotein A-I

Jitka Petrlova 1, Arnab Bhattacherjee 2, Wouter Boomsma 2, Stefan Wallin 2, Jens O Lagerstedt 1,*, Anders Irbäck 2,*
PMCID: PMC4241107  PMID: 25131953

Abstract

Several disease-linked mutations of apolipoprotein A-I, the major protein in high-density lipoprotein (HDL), are known to be amyloidogenic, and the fibrils often contain N-terminal fragments of the protein. Here, we present a combined computational and experimental study of the fibril-associated disordered 1–93 fragment of this protein, in wild-type and mutated (G26R, S36A, K40L, W50R) forms. In atomic-level Monte Carlo simulations of the free monomer, validated by circular dichroism spectroscopy, we observe changes in the position-dependent β-strand probability induced by mutations. We find that these conformational shifts match well with the effects of these mutations in thioflavin T fluorescence and transmission electron microscopy experiments. Together, our results point to molecular mechanisms that may have a key role in disease-linked aggregation of apolipoprotein A-I.

Keywords: disordered protein, amyloidogenic mutations, secondary-structure probability, fibril formation, molecular simulation

Introduction

Apolipoprotein A-I (apoA-I) is the major protein in high-density lipoprotein (HDL), and active in the reverse cholesterol transport pathway. A great deal has been learned in recent years about the structural properties of apoA-I in both lipid-free1,2 and lipid-bound3,4 form. For C-terminally truncated lipid-free apoA-I, a crystal structure was derived, which revealed a helical half-circle dimer organization.5 The response of apoA-I when exposed to nanoparticles with different surface charges was also investigated.6 Finally, computational studies based on coarse-grained models have provided insights into the formation of HDL particles.7–9

Variants of apoA-I carrying amino acid substitutions or deletions have been linked to certain diseases, including familial systemic amyloidosis.10,11 Disease-associated fibrils of apoA-I often contain N-terminal fragments of the protein, such as the 1–93 portion, apoA-I(1–93).12 The cleavage points cluster in the 83–96 region,13 which forms a loop in the 243 residues long apoA-I structure.14 A recent study investigated the impact of disease-linked, amyloidogenic mutations on properties of the apoA-I(1–93) fragment.15 All variants were shown to be disordered in monomeric form, with a weak propensity to form transient helical structure. The aggregation rate was increased by some of the mutations, but reduced by others. Predicted protection factors for hydrogen exchange, by contrast, pointed to a common trend for the mutations to enhance conformational fluctuations in the 88–110 part of the native full-length protein.15 This finding suggests that one effect of disease-linked mutations is to promote the generation of N-terminal fragments, by facilitating cleavage in the aforementioned loop region. However, the molecular mechanisms by which the fragments aggregate remain largely unknown.

In this article, we present a combined computational and experimental study of lipid-free apoA-I(1–93) in wild-type (WT) and mutated forms, aimed at providing a more detailed picture of the conformational ensemble sampled by the monomers and at identifying potential aggregation mechanisms. To characterize the monomer, we use atomic-level Monte Carlo (MC) simulations and circular dichroism (CD) spectroscopy. In addition, we study fibril formation by means of thioflavin T (ThT) fluorescence as well as aggregation by transmission electron microscopy (TEM).

To identify relevant conformational features, we compare five different apoA-I(1–93) variants, namely WT, G26R, S36A, K40L, and W50R. The G26R (“Iowa”) and W50R mutations are disease-linked and known to accelerate apoA-I(1–93) aggregation.15 The other two, S36A and K40L, are in part selected based on our WT simulations. S36 is conserved in non-fish (acidic in fish)16 and the S36A mutation has been linked to hypoalphalipoproteinemia.17,18 K40 is highly conserved in mammals and partly in fish16 but no disease-related K40 variants have to our knowledge been described.

One important property monitored in our simulations is the propensity of individual residues to form different types of secondary structure. We find that the β-structure probability profile of WT exhibits minima near the sites of the amyloidogenic mutations G26R and W50R. This observation leads us to examine how this profile is altered by these mutations as well as by two other mutations located near such minima, namely S36A and K40L. We find the mutational effects to be strong for G26R and W50R, less strong for K40L, and weak only for S36A. This conclusion, pertaining to the monomer, matches well with the effects that these mutations have on apoA-I(1–93) aggregation in our ThT fluorescence and TEM experiments, suggesting that the simulation approach can help to identify novel amyloidogenic variants of apoA-I.

Results

Choice of mutants

The apoA-I(1–93) monomer is known to be disordered.15 Here, we use atomic-level MC simulations to gain further insight into the conformational properties of this fibril-associated apoA-I fragment. As described in Materials and Methods, in these simulations, we observe two distinct forms of the monomer, one disordered and highly dynamic, and the other with a significant β-strand content. Of these two, the disordered form seems to dominate under the conditions studied experimentally. Therefore, in this and the next subsection, our analysis focuses entirely on the disordered form. At the end of the section, we briefly discuss the observed β-sheets.

We begin our discussion of the disordered form by presenting results obtained for WT, which will guide us below in selecting other variants to study. Using DSSP19 to classify secondary structure, we find overall α-helix and β-strand contents of 23.7% ± 0.2% and 4.5% ±0.1%, respectively, for WT. Figure 1 shows the corresponding profiles of residue-specific α- and β-structure probabilities, along with a prediction produced by PSIPRED.20 The PSIPRED prediction contains three long helical stretches. Our secondary-structure profiles are in good overall agreement with this prediction, although the simulated α-helix content is modest in parts of the regions predicted helical by PSIPRED. The simulated α-helix probability is >10% for most residues and as high as 40–50% for residues 50–60 and 72–74.

Figure 1.

Figure 1

Open in a new tab

Simulated residue-specific α-helix and β-strand probabilities for WT apoA-I(1–93) in disordered form, based on DSSP secondary-structure assignments.19 The maximum statistical uncertainties, at the one-sigma level, are about 0.01, or similar to the symbol size. Vertical lines indicate the positions of the mutations studied. The amino acid sequence is displayed below the figure. PSIPRED20 predicts three helical regions, whose locations are indicated by the blue bars above the figure.

The β-strand probability is lower than the α-helix probability for almost all residues. However, somewhat elevated β-strand probabilities occur in the 11–70 region. This region contains two stretches predicted to be particularly aggregation-prone parts of apoA-I (residues 15–20 and 50–57),15 and two segments (1–43 and 46–59) known to form amyloid on their own.21,22 Our simulated β-strand profile exhibits several local minima in the 11–70 region (at residues 25–26, 35, 39, 53–54, 61–62, 65–66, and 69), which represent potential turn locations. There is a line of evidence to suggest that in the aggregation of Aβ, which is the perhaps best studied amyloid-forming peptide and known to adopt a β-turn-β conformation in fibrils,23,24 turn formation is a key step.25–33 The precise organization of apoA-I fibrils is unknown. Nevertheless, it is intriguing that both the amyloidogenic G26R and W50R mutations are located near one of the β-strand minima that we observe. Having seen this, we decided to analyze how these two and two somewhat similar amino acid substitutions affect the β-strand probability profile. For this, S36A and K40L were selected, based on their locations (i) in the vicinity of β-strand minima and (ii) between the G26 and W50 residues in the primary structure of apoA-I. The S36A mutation of apoA-I has been linked to severe hypoalphalipoproteinemia,17 but seems not to be amyloidogenic.18 In contrast, there has, to our knowledge, not been any previous study of substitutions at position 40, which thus constitutes a potentially novel site for amyloidogenic apoA-I mutations. The selection of the K40L substitution was to increase the hydrophobicity of an already hydrophobic part of the molecule, potentially poised to drive aggregation.

Monomer simulations of the mutants

We simulate the mutants G26R, S36A, K40L, and W50R using the same setup as for WT. We find that all four have an overall secondary-structure composition similar to that of WT (Table 1), as one might expect from single-point mutations. The only significant exception is a reduced α-helix content for W50R. Upon helix formation, W50 becomes part of a hydrophobic cluster, which may explain why the substitution of this residue by a charged arginine leads to a loss in helix stability. The decrease in helix content for W50R can be traced around the 45–62 region (Fig. 2). Outside this region, the α-helix propensities are very similar for all five variants. Incidentally, since the simulations of the different variants are completely independent, this agreement provides a confirmation that the disordered ensembles indeed are well sampled.

Table 1.

Simulated Total α-helix and β-strand Contents (in %) for apoA-I(1–93) Variants in Disordered Form, Based on DSSP Secondary-Structure Assignments.19

Variant α-helix content β-strand content
WT 23.7 ± 0.2 4.5 ± 0.1
G26R 23.6 ± 0.2 4.6 ± 0.4
S36A 23.3 ± 0.1 4.8 ± 0.2
K40L 23.8 ± 0.3 4.8 ± 0.1
W50R 21.3 ± 0.2 4.5 ± 0.3
Open in a new tab

Figure 2.

Figure 2

Open in a new tab

Simulated residue-specific α- and β-structure probabilities for apoA-I(1–93) variants in disordered form, based on DSSP secondary-structure assignments.19 The top panel shows α-helix profiles for all five variants studied. Each of the other four panels compares the β-strand profile of one of the mutants to that of WT. Vertical lines indicate the locations of the mutation sites. The amino acid sequence is shown below the figure.

The β-strand probability profile is, in relative numbers, more sensitive to the mutations (Fig. 2). The G26R and W50R mutations both alter the β-strand probability over regions of >20 residues. Effects of the G26R mutation can be seen primarily in the 23–47 segment, but also downstream of this region. This observation matches well with data on full-length apoA-I obtained by electron paramagnetic resonance (EPR) spectroscopy, which showed that this mutation alters the β-strand propensity in the 27–56 region.34

A closer inspection reveals that the four mutations studied, all located near a β-strand minimum, have quite different effects on the β-strand probability profile (Fig. 2). The G26R and W50R mutations have the property of weakening or eliminating the corresponding β-strand minima (25–26 and 53–54, respectively). This effect is not observed for S36A, and only to a smaller extent for K40L, although the latter leads to some changes near the mutation site (residues 36–40 and 44–48).

To resolve the three-dimensional organization of the disordered monomers, we construct probability maps of residue-pair contact formation (Fig. 3). Two residues are defined in contact if their Cα atoms are within 10 Å. Almost all residue pairs with a contact probability >0.3 are found in a narrow band along the diagonal. Because of the dynamic and weakly helical character of the molecule, long-range contacts are rare. The scarcity of long-range contacts is particularly pronounced for the negatively charged C-terminal region. The contact maps also show that, while large hairpin-like structures are missing in the disordered ensembles, small turns can form at certain positions along the chain. For WT, turns occur that are centered at the β-strand minima 25–26 and 53–54. The G26R and W50R mutations remove the 25–26 and 53–54 turns, respectively, which is perfectly consistent with the observed effects of these mutations on the β-strand profile (Fig. 2).

Figure 3.

Figure 3

Open in a new tab

Simulated probability maps of residue-pair contact formation for apoA-I(1–93) variants in disordered form. Two residues are in contact if their Cα atoms are within 10 Å. Each of the four figures shows the results for one mutant (above the diagonal) and WT (below the diagonal). White lines indicate the positions of the respective mutations. In the text, turns centered at residues 25–26 and 53–54 are discussed. Such turns give rise to contacts that fall in the two transparent light bands perpendicular to the diagonal.

The main mutational effects suggested by our simulations of the five apoA-I(1–93) variants in their disordered form can be summarized as follows. The W50R mutation decreases the α-helix propensity in the 45–62 region, and thereby the overall α-helix content. The G26R and W50R mutations affect the β-strand probability over relatively large regions, and remove turns centered at 25–26 and 53–54, respectively. The effects of the S36A and K40L mutations are weaker, although K40L causes some changes in the β-strand probability in the vicinity of the mutation site.

Circular dichroism spectroscopy

We use CD spectroscopy to experimentally analyze the secondary structure of the apoA-I(1–93) variants and estimate their α-helix content at neutral pH (pH 7.4). CD spectroscopy is a frequently used method to analyze the secondary-structure composition of proteins at physiological-resembling conditions. The shape of the obtained spectra depends on the relative proportion of β-strand, α-helix and random coil,35 and the molar ellipticity values at 222 nm can be used to estimate the α-helix content of the analyzed protein.36

In Figure 4, we compare the obtained CD spectra for the five apoA-I(1–93) fragments (WT, G26R, S36A, K40L, and W50R) with that for the full-length apoA-I(1–243) WT protein. For comparison, we also show CD spectra taken after the 19-day-incubation period studied in the ThT experiments (see below). The minima at 208 and 222 nm that are typical for proteins with relatively high α-helix content can be observed for the apoA-I(1–243) WT protein, whereas the spectra of the apoA-I(1–93) fragments display a clearly different pattern with less pronounced amplitudes at 222 nm. The apoA-I(1–93) fragments also exhibit some inter-variability (see individual comparisons of peptide variants with apoA-I(1–93) WT in Fig. 4). In particular, the spectra for W50R and K40L are, unlike those of G26R and S36A, significantly different from that of WT. These observations suggest that G26R and S36A have similar structural compositions as WT before incubation, whereas both K40L and W50R are structurally different to apoA-I(1–93) WT already at this point.

Figure 4.

Figure 4

Open in a new tab

CD spectroscopy analyses of the apoA-I(1–93) variants WT, G26R, S36A, K40L, and W50R and the full-length WT apoA-I(1–243) protein (at protein concentrations of 0.14 mg mL−1, pH 7.4) before (solid lines) and after (dotted lines) incubation at 37°C or 19 days. Spectra were obtained in the region 200 to 260 nm. CD spectrum of full-length apoA-I WT is included for comparison.

We next determine the α-helix content based on the 222 nm values (Table 2). The calculated α-helix content of full-length apoA-I(1–243) WT protein (52.8%) is in agreement with earlier observations of lipid-free apoA-I in solution based on CD37,38 and EPR spectroscopy.2 The apoA-I(1–93) fragments are all significantly less helical. The α-helix contents are in the range 31–35%, except for W50R, which has only about 27% α-helical structure. Our simulation results agree well with the CD measurements, in particular with respect to the reduced α-helix content of W50R. Clear changes in the CD spectrum can also be seen for K40L. Hence, the K40L mutation may significantly alter the secondary-structure composition, although its effect on the estimated α-helix content is relatively small.

Table 2.

α-helix Content (in %) of apoA-I(1–93) Variants and Full-length apoA-I WT (1–243 WT), as Measured by CD

Variant G26R S36A K40L W50R WT 1–243 WT
α-helix content 35 32 31 27 34 53
Open in a new tab

Thioflavin T binding assay

We monitor the formation of amyloid fibrils over time using a ThT binding assay. ThT is a fluorescent compound that upon binding to amyloids changes it spectral characteristics, which can be detected by a spectrofluorometer (see, e.g., Ref.39). To ensure that the apoA-I(1–93) fragment is not amyloidogenic per se, we first compare apoA-I(1–93) WT to apoA-I(1–243) WT. As can be seen in Figure 5(A), the full-length control protein apoA-I(1–243) WT does not bind ThT during the 19-day-incubation period, which is in agreement with earlier findings.34,39 Similarly, no change in fluorescence is observed for the apoA-I(1–93) WT fragment during incubation. Of the mutated apoA-I(1–93) fragments, all but the K40L variant have low ThT binding at initiation of the experiment [Fig. 5(B,C)]. The high ThT fluorescence of the K40L (ca. twofold that of WT) is sustained, but not increased, throughout the incubation time, suggesting that this variant rapidly forms aggregates partly organized into amyloid-like substructures. This is supported by the findings from the CD analysis, where K40L is markedly different from WT before incubation and no further spectral change is observed during incubation (Fig. 4). Similarly, but at a later stage during incubation, both G26R and W50R adopt amyloid structure as shown by increases in ThT fluorescence. The transitions of the apoA-I(1–93) G26R and W50R peptides are initiated after about 8–10 and 12 days, respectively, and the binding is more pronounced for the G26R variant. The fourth apoA-I(1–93) mutant studied, S36A, behaves like the WT peptide; no change in fluorescence is observed over the incubation period. Finally, analyses of peptides by denaturing SDS-PAGE followed by Coomassie blue staining at start of incubation and after 19 days of incubation show that the peptides stay intact, without any significant undesired proteolytical cleavage [Fig. 5(D)].

Figure 5.

Figure 5

Open in a new tab

Fibril formation of apoA-I N-terminal fragments. A ThT binding assay was used to experimentally determine the propensity of full-length apoA-I (FL, 1–243 WT) and apoA-I(1–93) variants (WT, G26R, S36A, K40L, and W50R) to form amyloid during incubation (0–19 days at 37°C, pH 7.4). A: Comparison of WT apoA-I(1–243) and WT apoA-I(1–93). B: Comparison of apoA-I(1–93) variants (WT, G26R, S36A, K40L, and W50R). C: Comparison of apoA-I(1–93) variants (WT, G26R, S36A, K40L, and W50R) at 0 and 19 days. A significant change in fluorescence is observed for G26R and W50R (**P < 0.01, *P < 0.05, ns = not significant). D: Denaturing SDS-PAGE electrophoretic separation of proteins/fragments at start (0 days, top panel) and end (19 days, bottom panel) of incubation was used to analyze protein stability during the incubation. About 1.4 μg protein is loaded per lane.

Transmission electron microscopy analyses

To further analyze the formation of protein aggregates, we use negative-stain TEM. Morphologies observed after incubation at 37°C for 23 days are shown in Figure 6 (additional micrographs can be found in Supporting Information Fig. S1). We first look at apoA-I(1–243) WT and apoA-I(1–93) WT that both display only small amounts of aggregates. In contrast, large assemblies of apoA-I(1–93) peptides that carry the amyloidogenic G26R and W50R mutations can be observed. Similarly, the K40L mutation is found to lead to significant apoA-I(1–93) aggregation, thus supporting the findings from the ThT binding assay. In comparison to the G26R, W50R, and K40L variants, the S36A variant is far less prone to aggregate, although some aggregates can be observed. The different aggregation properties of the variants as observed by TEM thus corroborate the findings in the ThT binding analysis.

Figure 6.

Figure 6

Open in a new tab

TEM analysis of apoA-I aggregate formation. Protein samples (full-length apoA-I and 1–93 fragments) incubated at 37°C, pH 7.4 for 23 days were analyzed by negative-stain TEM. Black bar is 2 μm.

Simulated β-sheet structures

In our simulations of the monomers, in addition to the disordered form discussed above, we observe a distinct second form (WT 3%, G26R 13%, S36A 5%, K40L 17%, W50R 37%), characterized by a significant β-strand content (>22%). The formation of large β-sheets occurs only rarely, and there is no direct experimental support for these states. Nevertheless, the fact that these states form spontaneously in simulations started from random initial conditions indicates that they might be readily accessible to the monomers. Therefore, we include a brief discussion of these conformations.

Inspection shows that the β-sheets typically have four or more strands, organized in a simple meander pattern. To characterize these states, we compute a residue-specific indicator of turn probability, Inline graphic, based on the contact probabilities in Figure 3. Specifically, for a given residue i, Inline graphic is calculated as the sum of all non-local contact probabilities pjk, Inline graphic, in a strip perpendicular to the main diagonal, Inline graphic.40 Because a β-hairpin corresponds to a band extending perpendicularly from the main diagonal in the contact map, peaks in Inline graphic indicate statistically preferred turn locations. Figure 7 shows two Inline graphic profiles for each apoA-I(1–93) variant. One is for the disordered form, as defined in Materials and Methods. The other profile represents an unfiltered average over the full simulation trajectories. In the disordered case, Inline graphic varies smoothly with i, with maxima occurring at positions 25–26, 35–41, 53–54, and 61–62. The exact location of the maximum in the interval 35–41 depends on variant. As expected (Figs. 2 and 3), the 25–26 and 53–54 maxima are missing for G26R and W50R, respectively.

Figure 7.

Figure 7

Open in a new tab

Position-dependence of the turn probability, Inline graphic (see text), in our apoA-I(1–93) simulations. Two averages are shown for each variant. One is over the disordered ensemble (“filtered”), as defined in Materials and Methods. The other is over the full simulation trajectories (“unfiltered”). Vertical lines indicate the three highest peaks of the curve based on all data.

Averaging instead over all data leads to markedly different Inline graphic profiles for G26R, K40L and W50R, whereas the changes are marginal for WT and S36A. For G26R, K40L and W50R, three or four clear, approximately equally spaced peaks appear, reflecting the occurrence of β-sheets with their turns at these locations. Interestingly, the turn locations differ among the three variants, which, however, could be due to insufficient sampling. It is therefore worth noting that the peaks tend to occur near local maxima of the disordered Inline graphic profile. This hints that the different β-sheet structures, rather than being accidental, reflect different local secondary-structure propensities of the three variants.

It is tempting to speculate that the β-sheet-containing conformations could play a key role in aggregation. In previous simulations of Aβ and α-synuclein, we observed β-sheet-containing conformations with similarities to the respective fibril folds.29,40–42 Unfortunately, at present, there are no detailed models available for apoA-I(1–93) fibrils.

Discussion

When diagnosed, the accumulation of apoA-I in organs such as heart, liver, kidney, larynx and skin in familial apoA-I amyloidosis has often already led to severe organ dysfunction, or complete failure, in the patient. Identifying novel amyloidogenic variants of apoA-I and understanding the molecular mechanisms of aggregation may allow for earlier diagnosis and facilitate therapeutic invention.

The fibril-associated apoA-I(1–93) peptide is known from previous CD measurements to be dynamic and weakly helical in monomeric form.15 Here, we have used atomic-level simulations to gain further insight into the conformational ensembles sampled by five apoA-I(1–93) variants. The simulations reproduce the main mutational effect seen in our CD experiments, namely a reduced α-helix content for W50R.

Having seen this agreement with CD data, we examined the simulated ensembles in more detail. We found that the position-dependent β-strand probability for WT displayed minima near the sites of the amyloidogenic G26R and W50R mutations, and selected the S36A and K40L mutations due to their proximity to similar β-strand minima. In our subsequent simulations of the mutants, we found that G26R, W50R and, to some extent, K40L significantly altered the β-strand profile, whereas S36A led to only minor changes, compared to WT. The G26R and W50R mutations weakened or eliminated the corresponding β-strand minima, and affected the β-strand probability far beyond their immediate locations, as has also been seen in EPR experiments on full-length G26R apoA-I.34

We studied the aggregation properties of these apoA-I(1–93) variants by ThT fluorescence and TEM experiments. Increased aggregation was observed for G26R, K40L, and W50R, whereas S36A, again, behaved more like WT. The enhanced aggregation of G26R and W50R may in part be due to their reduced net charge, which is known to be a major factor in amyloid formation.43,44 However, the relative aggregation propensities that we observe cannot be explained in terms of net charge alone, which in fact is increased by the K40L mutation. Another factor in amyloid formation is the ability of the monomer to adopt aggregation-competent conformations.45 The similar mutational effects seen in our monomer simulations and aggregation experiments (significant effects of the G26R, K40L, and W50R mutations, but weak effects for S36A) indicate that this factor indeed may play an important role in apoA-I(1–93) aggregation. Therefore, the simulations provide hints about potentially aggregation-enhancing conformational changes. Of particular interest with respect to amyloid formation are the mutational changes observed in the β-strand probability profile. To further elucidate the role of these conformational changes in aggregation, it would be of great interest to extend the simulations to the dimer and higher levels. This is, however, beyond the scope of the present study.

In the simulations, we also observed a distinct second form of the monomer, the β-sheet form. The properties of this form cannot be precisely determined from the present simulations. However, we note that the β-sheet form occurred more frequently for G26R, K40L, and W50R than it did for S36A and WT, indicating a possible correlation with aggregation propensity. Moreover, the β-sheets formed by G26R, K40L, and W50R seem to have different structures (Fig. 7). Such differences in the structural properties of the monomers could contribute to the diversity in aggregate morphology seen in our TEM images. The largely amorphous nature of the aggregates detected by TEM is in line with previously published work on apoA-I(1–93) at pH 4.15 Also, amorphous aggregates of apoA-I were observed in in vivo deposits by electron microscopy.46

The aggregation properties of intrinsically disordered proteins are a topic under intense study, with Aβ and α-synuclein as prominent examples. Structural insights into the aggregation mechanisms have been gained for many short peptides as well as for Aβ, with about 40 residues. In the case of Aβ, there is a range of evidence, from experimental and computational work, to suggest that the monomer transiently samples fibril-like conformations, with a potentially key role in aggregation.25–33 Although less is known about the 140-residue α-synuclein, there are indications that fibril-like conformations are readily accessible to this monomer as well.40,47–50 The present study is a step toward characterizing the apoA-I(1–93) peptide, whose structure in fibrils remains unknown. By comparing apoA-I(1–93) variants with different aggregation propensities, we identified structural properties that may have a key role in aggregation. The combined in silico and experimental approach taken here has the potential to facilitate the identification of additional, as of now not described, amyloidogenic variants of apoA-I. This knowledge could be useful in exploring therapeutic interventions at early disease stages, before the amyloid deposits lead to organ failure.

Materials and Methods

Simulations

We simulate the apoA-I(1–93) monomer using an all-atom protein model with implicit solvent and torsional degrees of freedom.51 The energy function consists of four main terms

graphic file with name pro0023-1559-m11.jpg (1)

The Inline graphic term represents excluded volume effects, whereas Inline graphic handles local interactions among neighboring atoms. Hydrogen bonds between backbone NH and CO groups, and between charged side-chain groups and the backbone are handled by Inline graphic. The last term, Inline graphic, represents interactions between side chains, both hydrophobic attraction and attraction/repulsion between charged side chains. The mathematical form and parameters of the model are given elsewhere.51 The same computational model has recently been applied to study other proteins similar in size to apoA-I(1–93), both natively folded52,53 and intrinsically disordered40,54 ones.

Our simulations are Monte Carlo-based and conducted using the software package PROFASI.55 The MC move set is as described previously.29,40 It includes a pivot-type update, which helps to speed up the simulation of flexible chains. For each of the five apoA-I(1–93) variants, we generate six independent replica-exchange runs56 with four replicas each, at the nominal temperatures 59°C, 73°C, 84°C, and 97°C. All simulations are started from random initial conditions.

Sampling the vast configuration space of a protein the size of apoA-I(1–93) is a challenge. In our simulations, we observe two distinct major forms of apoA-I(1–93). The molecule spends a large fraction of the simulation time in dynamic high-energy states, but in some runs one of the replicas gets trapped in low-energy states characterized by a significant β-strand content (e.g., see Supporting Information Fig. S2). The low-energy states mainly occupy the lowest of the four simulation temperatures. The fraction of time spent in low-energy states at this temperature (nominally 59°C) depends on apoA-I(1–93) variant (WT 3%, G26R 13%, S36A 5%, K40L 17%, W50R 37%). Unfortunately, this fraction and the properties of the low-energy states are statistically difficult to determine, because the formation of such states is a rare event. Because of this statistical limitation, we analyze the high- and low-energy states separately. We will refer to these two forms as the disordered and β-sheet forms, respectively. In the Results section, although we briefly discuss the shape of the observed β-sheets, our primary focus is on the disordered form, mainly because the observed properties of this form are in good overall agreement with experimental CD data. Also, we find that this dynamic form, unlike the β-sheet form, indeed can be analyzed in a statistically robust manner.

Our partitioning of the simulated ensembles into the disordered and β-sheet sub-ensembles is based on a simple cut in the β-strand content, b; a conformation is deemed disordered if b is below 22%. Histograms of b can be found in Supporting Information Figure S3. Values of b near the cut are suppressed for all our variants.

A well-known problem in protein simulations is to obtain a realistic temperature dependence. Throughout the analysis below, we restrict ourselves to data taken at the lowest simulation temperature, nominally 59°C, which is chosen for best agreement with experimental CD data measured at 25°C. The need to adjust the temperature scale is not surprising, as this was set by comparing simulated and experimental melting temperatures of a much smaller protein. Such shifts in the temperature scale are necessary with most current force fields,57 and similar corrections were also found necessary in our previous studies of both Aβ29 and α-synuclein.40

Production of recombinant protein

A bacterial expression system consisting of pEXP-5 plasmid in Escherichia coli strain BL21(DE3) pLysS cells (Invitrogen) was used to produce the full-length (243 amino acids) apoA-I WT protein, and the apoA-I(1–93) variants WT, G26R, S36A, K40L, and W50R, as previously described.39 Primer-directed PCR mutagenesis was used to create the stop codon at position 94 of full-length apoA-I WT, followed by introduction of the site-specific mutations using the same PCR-based method. The mutation was verified by dideoxy automated fluorescent sequencing (GATC Biotech). After purification of apoA-I proteins on Ni2+-chelated columns (GE Healthcare) and desalting to remove imidazole, treatment with Tobacco etch virus (TEV) protease (provided by the Lund University protein production platform LP3) was employed to cleave the His-tag. This was followed by a second Ni2+-column passage where TEV protease (containing a His-tag) and the cleaved His-tag were retained on the column. The flow-through containing cleaved apoA-I proteins was desalted into phosphate buffered saline, pH 7.4, 500 mM NaCl, concentrated with 3 kDa molecular weight cut-off Amicon Ultra centrifugal filter devices (Millipore) and stored at 4°C prior to use. Protein purity was confirmed by sodium dodecyl sulfate (SDS)-polyacrylamide gel electrophoresis with Coomassie blue staining and protein concentrations determined by Nanodrop (Thermo Scientific), using molecular weight and extinction coefficients of apoA-I proteins.

Circular dichroism spectroscopy

CD measurements were performed on a Jasco J-810 spectropolarimeter equipped with a Jasco CDF-426S Peltier set to 25°C. ApoA-I was diluted to 0.14 mg mL−1 in PBS (final concentration was 25 mM phosphate, 500 mM NaCl, pH 7.4), placed in a 0.1 mm quartz cuvette and, after extensive purging with nitrogen, scanned in the region 200–260 nm (scan speed was 20 nm min−1). Averages of five scans were baseline-subtracted (PBS buffer; 25 mM phosphate, 500 mM NaCl) and the α-helix content was calculated from the molar ellipticity at 222 nm as previously described.39

Thioflavin T binding assay

WT and N-terminal fragments (0.14 mg mL−1) were incubated at 37°C and diluted with ThT stock at time of use. About 190 μL of protein was incubated for 15 min in the dark with 10 μL of a ThT (5 mM)/glycine (50 mM) solution (ThT stock: 1 mM stored in the dark at 4°C; glycine buffer stock: 0.1M at pH 8.5 stored at 4°C). ThT fluorescence was then measured using a VICTOR3 Multilabel Plate Counter (PerkinElmer, Waltham, MA) spectrofluorometer at an excitation of 450 nm and an emission wavelength of 545 nm, with excitation and emission slit widths of 10 nm.39

Transmission electron microscopy analyses

WT and N-terminal fragments (0.14 mg mL−1) were incubated at 37°C 23 days followed by negative stain electron microscopy. About 5 μL of apoA-I proteins were adsorbed onto carbon-coated grids for 60 s, and stained with 7 μL of 2% uranyl acetate for 20 s. The grids were rendered hydrophilic by glow discharge at low pressure in air. Specimens were observed in a FEI Tecnai Spirit BioTWIN electron microscope operated at 100 kV accelerating voltage, and images were recorded with a Veleta TEM CCD camera (Lund University Bioimaging Center).

Acknowledgments

The simulations were performed on resources provided by the Swedish National Infrastructure for Computing (SNIC) at LUNARC, Lund University. TEM analysis was performed at Lund University Bioimaging Center.

Glossary

apoA-I

apolipoprotein A-I

MC

Monte Carlo

TEM

transmission electron microscopy

ThT

thioflavin T.

Supporting Information

Additional Supporting Information may be found in the online version of this article.

pro0023-1559-sd1.pdf (15.3MB, pdf)

References

  1. Chetty PS, Mayne L, Lund-Katz S, Stranz D, Englander SW, Phillips MC. Helical structure and stability in human apolipoprotein A-I by hydrogen exchange and mass spectrometry. Proc Natl Acad Sci USA. 2009;106:19005–19010. doi: 10.1073/pnas.0909708106. [DOI] [PMC free article] [PubMed] [Google Scholar]
  2. Lagerstedt JO, Budamagunta MS, Liu GS, DeValle NC, Voss JC, Oda MN. The “beta-clasp” model of apolipoprotein A-I—a lipid-free solution structure determined by electron paramagnetic resonance spectroscopy. Biochim Biophys Acta. 2012;1821:448–455. doi: 10.1016/j.bbalip.2011.12.010. [DOI] [PMC free article] [PubMed] [Google Scholar]
  3. Chetty PS, Mayne L, Kan ZY, Lund-Katz S, Englander SW, Phillips MC. Apolipoprotein A-I helical structure and stability in discoidal high-density lipoprotein (HDL) particles by hydrogen exchange and mass spectrometry. Proc Natl Acad Sci USA. 2012;109:11687–11692. doi: 10.1073/pnas.1209305109. [DOI] [PMC free article] [PubMed] [Google Scholar]
  4. Oda MN, Budamagunta MS, Borja MS, Petrlova J, Voss JC, Lagerstedt JO. The secondary structure of apolipoprotein A-I on 9.6-nm reconstituted high-density lipoprotein determined by EPR spectroscopy. FEBS J. 2013;280:3416–3424. doi: 10.1111/febs.12334. [DOI] [PMC free article] [PubMed] [Google Scholar]
  5. Mei X, Atkinson D. Crystal structure of C-terminal truncated apolipoprotein A-I reveals the assembly of high density lipoprotein (HDL) by dimerization. J Biol Chem. 2011;286:38570–38582. doi: 10.1074/jbc.M111.260422. [DOI] [PMC free article] [PubMed] [Google Scholar]
  6. Cukalevski R, Lundqvist M, Oslakovic C, Dahlbäck B, Linse S, Cedervall T. Structural changes in apolipoproteins bound to nanoparticles. Langmuir. 2011;27:14360–14369. doi: 10.1021/la203290a. [DOI] [PubMed] [Google Scholar]
  7. Catte A, Patterson JC, Bashtovyy D, Jones MK, Gu F, Li L, Rampioni A, Sengupta D, Vuorela T, Niemelä P, Karttunen M, Marrink SJ, Vattulainen I, Segrest JP. Structure of spheroidal HDL particles revealed by combined atomistic and coarse-grained simulations. Biophys J. 2008;94:2306–2319. doi: 10.1529/biophysj.107.115857. [DOI] [PMC free article] [PubMed] [Google Scholar]
  8. Shih AY, Sligar SG, Schulten K. Maturation of high-density lipoproteins. J R Soc Interface. 2009;6:863–871. doi: 10.1098/rsif.2009.0173. [DOI] [PMC free article] [PubMed] [Google Scholar]
  9. Vuorela T, Catte A, Niemelä PS, Hall A, Hyvönen MT, Marrink SJ, Karttunen M, Vattulainen I. Role of lipids in spheroidal high density lipoproteins. PLOS Comput Biol. 2010;6:e1000964. doi: 10.1371/journal.pcbi.1000964. [DOI] [PMC free article] [PubMed] [Google Scholar]
  10. Teoh CL, Griffin MDW, Howlett GJ. Apolipoproteins and amyloid fibril formation in atherosclerosis. Protein Cell. 2011;2:116–127. doi: 10.1007/s13238-011-1013-6. [DOI] [PMC free article] [PubMed] [Google Scholar]
  11. Rowczenio D, Dogan A, Theis JD, Vrana JA, Lachmann HJ, Wechalekar AD, Gilbertson JA, Hunt T, Gibbs SDJ, Sattianayagam PT, Pinney JH, Hawkins PN, Gillmore JD. Amyloidogenicity and clinical phenotype associated with five novel mutations in apolipoprotein A-I. Am J Pathol. 2011;179:1978–1987. doi: 10.1016/j.ajpath.2011.06.024. [DOI] [PMC free article] [PubMed] [Google Scholar]
  12. Obici L, Franceschini G, Calabresi L, Giorgetti S, Stoppini M, Merlini G, Bellotti V. Structure, function and amyloidogenic propensity of apolipoprotein A-I. Amyloid. 2006;13:191–205. doi: 10.1080/13506120600960288. [DOI] [PubMed] [Google Scholar]
  13. Obici L, Bellotti V, Mangione P, Stoppini M, Arbustini E, Verga L, Zorzoli I, Anesi E, Zanotti G, Campana C, Viganò M, Merlini G. The new apolipoprotein A-I variant Leu174 → Ser causes hereditary cardiac amyloidosis, and the amyloid fibrils are constituted by the 93-residue N-terminal polypeptide. Am J Pathol. 1999;155:695–702. doi: 10.1016/S0002-9440(10)65167-X. [DOI] [PMC free article] [PubMed] [Google Scholar]
  14. Lagerstedt JO, Budamagunta MS, Oda MN, Voss JC. Electron paramagnetic resonance spectroscopy of site-directed spin labels reveals the structural heterogeneity in the N-terminal domain of ApoA-I in solution. J Biol Chem. 2007;282:9143–9149. doi: 10.1074/jbc.M608717200. [DOI] [PubMed] [Google Scholar]
  15. Raimondi S, Guglielmi F, Giorgetti S, Gaetano SD, Arciello A, Monti DM, Relini A, Nichino D, Doglia SM, Natalello A, Pucci P, Mangione P, Obici L, Merlini G, Stoppini M, Robustelli P, Tartaglia GG, Vendruscolo M, Dobson CM, Piccoli R, Bellotti V. Effects of the known pathogenic mutations on the aggregation pathway of the amyloidogenic peptide of apolipoprotein A-I. J Mol Biol. 2011;407:465–476. doi: 10.1016/j.jmb.2011.01.044. [DOI] [PubMed] [Google Scholar]
  16. Bashtovyy D, Jones MK, Anantharamaiah G, Segrest JP. Sequence conservation of apolipoprotein A-I affords novel insights into HDL structure-function. J Lipid Res. 2011;52:435–450. doi: 10.1194/jlr.R012658. [DOI] [PMC free article] [PubMed] [Google Scholar]
  17. Kiss RS, Kavaslar N, Okuhira Ki, Freeman MW, Walter S, Milne RW, McPherson R, Marcel YL. Genetic etiology of isolated low HDL syndrome: incidence and heterogeneity of efflux defects. Arterioscler Thromb Vasc Biol. 2007;27:1139–1145. doi: 10.1161/ATVBAHA.106.137646. [DOI] [PubMed] [Google Scholar]
  18. Weers PMM, Patel AB, Wan LC, Guigard E, Kay CM, Hafiane A, McPherson R, Marcel YL, Kiss RS. Novel N-terminal mutation of human apolipoprotein AI reduces self-association and impairs LCAT activation. J Lipid Res. 2011;52:35–44. doi: 10.1194/jlr.M007500. [DOI] [PMC free article] [PubMed] [Google Scholar]
  19. Kabsch W, Sander C. Dictionary of protein secondary structure: pattern recognition of hydrogen-bonded and geometrical features. Biopolymers. 1983;22:2577–2637. doi: 10.1002/bip.360221211. [DOI] [PubMed] [Google Scholar]
  20. Buchan DWA, Minneci F, Nugent TCO, Bryson K, Jones DT. Scalable web services for the PSIPRED protein analysis workbench. Nucleic Acids Res. 2013;41:W349–W357. doi: 10.1093/nar/gkt381. [DOI] [PMC free article] [PubMed] [Google Scholar]
  21. Adachi E, Kosaka A, Tsuji K, Mizuguchi C, Kawashima H, Shigenaga A, Nagao K, Akaji K, Otaka A, Saito H. The extreme N-terminal region of human apolipoprotein A-I has a strong propensity to form amyloid fibrils. FEBS Lett. 2014;588:389–394. doi: 10.1016/j.febslet.2013.11.031. [DOI] [PubMed] [Google Scholar]
  22. Wong YQ, Binger KJ, Howlett GJ, Griffin MDW. Identification of an amyloid fibril forming peptide comprising residues 46–59 of apolipoprotein A-I. FEBS Lett. 2012;586:1754–1758. doi: 10.1016/j.febslet.2012.05.007. [DOI] [PubMed] [Google Scholar]
  23. Petkova AT, Ishii Y, Balbach JJ, Antzutkin ON, Leapman RD, Delaglio F, Tycko R. A structural model for Alzheimer’s β-amyloid fibrils based on experimental constraints from solid state NMR. Proc Natl Acad Sci USA. 2002;99:16742–16747. doi: 10.1073/pnas.262663499. [DOI] [PMC free article] [PubMed] [Google Scholar]
  24. Lührs T, Ritter C, Adrian M, Riek-Loher D, Bohrmann B, Döbeli H, Schubert D, Riek R. 3D structure of Alzheimer’s amyloid-β(1–42) fibrils. Proc Natl Acad Sci USA. 2005;102:17342–17347. doi: 10.1073/pnas.0506723102. [DOI] [PMC free article] [PubMed] [Google Scholar]
  25. Sciarretta KL, Gordon DJ, Petkova AT, Tycko R, Meredith SC. Aβ40-Lactam(D23/K28) models a conformation highly favorable for nucleation of amyloid. Biochemistry. 2005;44:6003–6014. doi: 10.1021/bi0474867. [DOI] [PubMed] [Google Scholar]
  26. Hoyer W, Grönwall C, Jonsson A, Ståhl S, Härd T. Stabilization of a β-hairpin in monomeric Alzheimer’s amyloid-β peptide inhibits amyloid formation. Proc Natl Acad Sci USA. 2008;105:5099–5104. doi: 10.1073/pnas.0711731105. [DOI] [PMC free article] [PubMed] [Google Scholar]
  27. Krone MG, Baumketner A, Bernstein SL, Wyttenbach T, Lazo ND, Teplow DB, Bowers MT, Shea JE. Effects of familial Alzheimer’s disease mutations on the folding nucleation of the amyloid β-protein. J Mol Biol. 2008;381:221–228. doi: 10.1016/j.jmb.2008.05.069. [DOI] [PMC free article] [PubMed] [Google Scholar]
  28. Reddy G, Straub JE, Thirumalai D. Influence of preformed Asp23-Lys28 salt bridge on the conformational fluctuations of monomers and dimers of Aβ peptides with implications for rates of fibril formation. J Phys Chem B. 2009;113:1162–1172. doi: 10.1021/jp808914c. [DOI] [PMC free article] [PubMed] [Google Scholar]
  29. Mitternacht S, Staneva I, Härd T, Irbäck A. Comparing the folding free-energy landscapes of Aβ42 variants with different aggregation properties. Proteins. 2010;78:2600–2608. doi: 10.1002/prot.22775. [DOI] [PubMed] [Google Scholar]
  30. Côté S, Derreumaux P, Mousseau N. Distinct morphologies for amyloid beta protein monomer: Aβ1−40, Aβ1−42, and Aβ1−40 (D23N) J Chem Theory Comput. 2011;7:2584–2592. doi: 10.1021/ct1006967. [DOI] [PubMed] [Google Scholar]
  31. Cruz L, Rao JS, Teplow DB, Urbanc B. Dynamics of metastable β-hairpin structures in the folding nucleus of amyloid β-protein. J Phys Chem B. 2012;116:6311–6325. doi: 10.1021/jp301619v. [DOI] [PMC free article] [PubMed] [Google Scholar]
  32. Ball KA, Phillips AH, Wemmer DE, Head-Gordon T. Differences in β-strand populations of monomeric Aβ40 and Aβ42. Biophys J. 2013;104:2714–2724. doi: 10.1016/j.bpj.2013.04.056. [DOI] [PMC free article] [PubMed] [Google Scholar]
  33. Rosenman DJ, Connors CR, Chen W, Wang C, García AE. Aβ monomers transiently sample oligomer and fibril-like configurations: ensemble characterization using a combined MD/NMR approach. J Mol Biol. 2013;425:3338–3359. doi: 10.1016/j.jmb.2013.06.021. [DOI] [PMC free article] [PubMed] [Google Scholar]
  34. Lagerstedt JO, Cavigiolio G, Roberts LM, Hong HS, Jin LW, Fitzgerald PG, Oda MN, Voss JC. Mapping the structural transition in an amyloidogenic apolipoprotein A-I. Biochemistry. 2007;46:9693–9699. doi: 10.1021/bi7005493. [DOI] [PMC free article] [PubMed] [Google Scholar]
  35. Greenfield N, Fasman GD. Computed circular dichroism spectra for the evaluation of protein conformation. Biochemistry. 1969;8:4108–4116. doi: 10.1021/bi00838a031. [DOI] [PubMed] [Google Scholar]
  36. Morrow JA, Segall ML, Lund-Katz S, Phillips MC, Knapp M, Rupp B, Weisgraber KH. Differences in stability among the human apolipoprotein E isoforms determined by the amino-terminal domain. Biochemistry. 2000;39:11657–11666. doi: 10.1021/bi000099m. [DOI] [PubMed] [Google Scholar]
  37. Brubaker G, Peng DQ, Somerlot B, Abdollahian DJ, Smith JD. Apolipoprotein A-I lysine modification: effects on helical content, lipid binding and cholesterol acceptor activity. Biochim Biophys Acta. 2006;1761:64–72. doi: 10.1016/j.bbalip.2006.01.007. [DOI] [PubMed] [Google Scholar]
  38. Silva RAGD, Hilliard GM, Fang J, Macha S, Davidson WS. A three-dimensional molecular model of lipid-free apolipoprotein A-I determined by cross-linking/mass spectrometry and sequence threading. Biochemistry. 2005;44:2759–2769. doi: 10.1021/bi047717+. [DOI] [PubMed] [Google Scholar]
  39. Petrlova J, Duong T, Cochran MC, Axelsson A, Morgelin M, Roberts LM, Lagerstedt JO. The fibrillogenic L178H variant of apolipoprotein A-I forms helical fibrils. J Lipid Res. 2012;53:390–398. doi: 10.1194/jlr.M020883. [DOI] [PMC free article] [PubMed] [Google Scholar]
  40. Jónsson SÆ, Mohanty S, Irbäck A. Distinct phases of free α-synuclein—a Monte Carlo study. Proteins. 2012;80:2169–2177. doi: 10.1002/prot.24107. [DOI] [PubMed] [Google Scholar]
  41. Mitternacht S, Staneva I, Härd T, Irbäck A. Monte Carlo study of the formation and conformational properties of dimers of Aβ42 variants. J Mol Biol. 2011;410:357–367. doi: 10.1016/j.jmb.2011.05.014. [DOI] [PubMed] [Google Scholar]
  42. Jónsson SÆ, Mitternacht S, Irbäck A. Mechanical resistance in unstructured proteins. Biophys J. 2013;104:2725–2732. doi: 10.1016/j.bpj.2013.05.003. [DOI] [PMC free article] [PubMed] [Google Scholar]
  43. Chiti F, Stefani M, Taddei N, Ramponi G, Dobson CM. Rationalization of the effects of mutations on peptide and protein aggregation rates. Nature. 2003;424:805–808. doi: 10.1038/nature01891. [DOI] [PubMed] [Google Scholar]
  44. López de la Paz M, Serrano L. Sequence determinants of amyloid fibril formation. Proc Natl Acad Sci USA. 2004;101:87–92. doi: 10.1073/pnas.2634884100. [DOI] [PMC free article] [PubMed] [Google Scholar]
  45. Straub JE, Thirumalai D. Principles governing oligomer formation in amyloidogenic peptides. Curr Opin Struct Biol. 2010;20:187–195. doi: 10.1016/j.sbi.2009.12.017. [DOI] [PMC free article] [PubMed] [Google Scholar]
  46. Scalvini T, Martini PR, Obici L, Tardanico R, Biasi L, Gregorini G, Scolari F, Merlini G. Infertility and hypergonadotropic hypogonadism as first evidence of hereditary apolipoprotein A-I amyloidosis. J Urol. 2007;178:344–348. doi: 10.1016/j.juro.2007.03.005. [DOI] [PubMed] [Google Scholar]
  47. Kim HY, Heise H, Fernandez CO, Baldus M, Zweckstetter M. Correlation of amyloid fibril β-structure with the unfolded state of α-synuclein. ChemBioChem. 2007;8:1671–1674. doi: 10.1002/cbic.200700366. [DOI] [PubMed] [Google Scholar]
  48. Sandal M, Valle F, Tessari I, Mammi S, Bergantino E, Musiani F, Brucale M, Bubacco L, Samorì B. Conformational equilibria in monomeric α-synuclein at the single-molecule level. PLOS Biol. 2008;6:e6. doi: 10.1371/journal.pbio.0060006. [DOI] [PMC free article] [PubMed] [Google Scholar]
  49. Wu KP, Kim S, Fela DA, Baum J. Characterization of conformational and dynamic properties of natively unfolded human and mouse α-synuclein ensembles by NMR: implication for aggregation. J Mol Biol. 2008;378:1104–1115. doi: 10.1016/j.jmb.2008.03.017. [DOI] [PMC free article] [PubMed] [Google Scholar]
  50. Hervás R, Oroz J, Galera-Prat A, Goñi O, Valbuena A, Vera AM, Gómez-Sicilia À, Losada-Urzáiz F, Uversky VN, Menéndez M, Laurents DV, Bruix M, Carrión-Vázquez M. Common features at the start of the neurodegeneration cascade. PLOS Biol. 2012;10:e1001335. doi: 10.1371/journal.pbio.1001335. [DOI] [PMC free article] [PubMed] [Google Scholar]
  51. Irbäck A, Mitternacht S, Mohanty S. An effective all-atom potential for proteins. PMC Biophys. 2009;2:2. doi: 10.1186/1757-5036-2-2. [DOI] [PMC free article] [PubMed] [Google Scholar]
  52. Mohanty S, Meinke JH, Zimmermann O. Folding of Top7 in unbiased all-atom Monte Carlo simulations. Proteins. 2013;81:1446–1456. doi: 10.1002/prot.24295. [DOI] [PubMed] [Google Scholar]
  53. Bille A, Jónsson SÆ, Akke M, Irbäck A. Local unfolding and aggregation mechanisms of SOD1—a Monte Carlo exploration. J Phys Chem B. 2013;117:9194–9202. doi: 10.1021/jp404500b. [DOI] [PubMed] [Google Scholar]
  54. Cong X, Casiraghi N, Rossetti G, Mohanty S, Giachin G, Legname G, Carloni P. Role of prion disease-linked mutations in the intrinsically disordered N-terminal domain of the prion protein. J Chem Theory Comput. 2013;9:5158–5167. doi: 10.1021/ct400534k. [DOI] [PubMed] [Google Scholar]
  55. Irbäck A, Mohanty S. PROFASI: a Monte Carlo simulation package for protein folding and aggregation. J Comput Chem. 2006;27:1548–1555. doi: 10.1002/jcc.20452. [DOI] [PubMed] [Google Scholar]
  56. Swendsen RH, Wang JS. Replica Monte Carlo simulation of spin glasses. Phys Rev Lett. 1986;57:2607–2609. doi: 10.1103/PhysRevLett.57.2607. [DOI] [PubMed] [Google Scholar]
  57. Lindorff-Larsen K, Maragakis P, Piana S, Eastwood MP, Dror RO, Shaw DE. Systematic validation of protein force fields against experimental data. PLOS One. 2012;7:e32131. doi: 10.1371/journal.pone.0032131. [DOI] [PMC free article] [PubMed] [Google Scholar]

Associated Data

This section collects any data citations, data availability statements, or supplementary materials included in this article.

Supplementary Materials

pro0023-1559-sd1.pdf (15.3MB, pdf)

Articles from Protein Science : A Publication of the Protein Society are provided here courtesy of The Protein Society

RESOURCES