“Native-like aggregation” of the acylphosphatase from Sulfolobus solfataricus and its biological implications

1. Introduction 

More than 40 human diseases are associated with the deposition of normally soluble and functional proteins into insoluble aggregates having typical characteristics, such as a long and unbranched fibrillar morphology when analyzed with atomic force microscopy [1] or electron microscopy [2], the ability to bind specific dyes [3], [4] and the presence of extensive β-sheet structure [2]. These pathologies are usually referred to as protein deposition diseases and the fibrillar aggregates appearing at their onset are usually called amyloid fibrils, at least when they accumulate in the extracellular space [5].

One of the key issues to elucidate the pathogenesis of protein deposition diseases is the mechanism by which such proteins convert from their native state (N) into amyloid fibrils. At least two major pathways have been identified to date [5]. In some cases aggregation starts from fully or partially unfolded conformational states. These can be either native states of intrinsically disordered proteins or non-native states experienced by a folded protein through an unfolding event, either partial or total. This unfolding event leads to the exposure of aggregation prone peptide segments, which are able to form intermolecular interactions, thus triggering aggregation [5]. More recently, a different pathway to amyloid formation has been identified, attracting increasing attention. According to this recently described process, aggregation of normally globular proteins can occur starting from conformational states directly accessible from native states via thermal fluctuations, mutations, cis-trans proline isomerisation, or disruption of quaternary structure, but in all cases with no need of transitions across the major energy barrier for unfolding [6]. Such conformational states, which we will refer to as native-like states (N∗), are responsible for an increased aggregation propensity of the protein, ultimately leading to amyloid formation [6]. They can be either transiently formed conformations accessible through fluctuations of the native state [7] or conformational states permanently populated due to mutation or other events, i.e. states with a thermodynamic stability higher than that of the fully folded state [8].

2. Aggregation from native-like states is a concrete possibility in vivo 

That aggregation processes in vivo for normally globular proteins can involve native-like states of the precursor proteins, rather than requiring processes of unfolding as a first obligatory step, is suggested by a number of experimental observations collected on different grounds. The first evidence originates from the analysis of bacterial inclusion bodies forming after the over-expression of normally globular proteins. These are increasingly recognised to consist of protein aggregates with morphological, structural and tinctorial properties typical of amyloid structures, and thus useful model systems to study amyloid formation in vivo [9], [10], [11], [12]. The coexistence of amyloid-like aggregates and aggregates containing active enzymes or functional proteins within inclusion bodies [11], [13], [14], [15] suggests that in vivo proteins can effectively aggregate from native or native-like states and form, as a first aggregation step, assemblies in which the folded and functional structure is largely maintained. Second, molecular chaperones can target compact native states in addition to polypeptide chains with substantially unfolded regions [16], [17], suggesting that the cellular machinery dedicated to prevent aggregation in vivo recognises folded states as potentially dangerous conformational states to guard. Furthermore, it has been recently proposed that the protein quality control machinery of Escherichia coli induces protein solubility through a promoted degradation of aggregation-prone but functional protein species, thus achieving a minimization of aggregation independently of the conformational state and biological function of the targeted proteins [18]. Third, the structural inspection of all-β proteins has led to the identification of sequence and structural adaptations that protect the β-strands occupying peripheral positions in natural β-sheets (edge β-strands) [19]. Such protections have their rationale in the fact that edge β-strands could give rise to intermolecular interactions with edge β-strands of other protein molecules, thus triggering aggregation through a propagation of the initial β-sheets. The observation that such adaptations are effective once the edge β-strands occupy their peripheral position in a folded β-sheet, rather than in an unfolded conformation, is by itself a strong indication that aggregation from fully folded states is a constant challenge for natural proteins working in vivo and that a specific evolutionary pressure exists to keep this process under control [19].

This evidence remains circumstantial and does not prove that amyloid formation occurring in pathological conditions involves the assembly of native-like folds as a first step. However, in physiological crowded environments in which normally globular proteins spend most of their lifetime in a folded conformation, the possibility that native-like states can aggregate and initiate the process of amyloid deposition is worth our attention. Our ability to determine whether aggregation in vivo involves directly native-like states or requires unfolding is not a purely mechanistically detail. It is rather a fundamental issue and has enormous implications for identifying the molecular targets to design appropriate drugs for protein deposition diseases and for understanding, more generally, how proteins have evolved to be soluble and how the cellular machinery works to prevent effectively protein aggregation and maintain protein homeostasis in vivo.

In our lab we have expended considerable effort in elucidating in vitro the process of native-like aggregation of the acylphosphatase from Sulfolobus solfataricus (Sso AcP), a protein that is particularly suitable for investigating this particular process for reasons that will become apparent in the following sections. In this review we shall summarize the main aspects of this process believing that this protein paradigms many of the characteristics of native-like aggregation.

3. Sso AcP adopts a native-like state under conditions promoting aggregation 

Sso AcP is a small, α/β enzyme belonging to the acylphosphatase-like structural family (Fig. 1). The structure of the protein is characterized by a 5-stranded β-sheet facing against two α-helices [20]. The overall Sso AcP fold displays the same βαββαβ topology previously described for other members of the acylphosphatase-like family and typical of the ferrodoxin-like fold [21], [22], [23], [24], [25]. By contrast to related acylphosphatases, however, Sso AcP contains an unstructured, 11-residue N-terminal segment [20], which plays a major role in the aggregation of the protein (Fig. 1). The protein is an enzyme able to hydrolyze, similarly to all other acylphosphatases, the phosphoanhydride bond of acylphosphates [20]. Since it derives from a hyperthermophilic organism it also shows a high conformational stability, amounting to 49.7±2.4kJmol−1 at pH 5.5, 25°C [20].

View Large Image Download to PowerPoint

Fig. 1. Structure of Sso AcP [20]. Regions shown to play important roles in inducing aggregation of the protein are highlighted in dark grey [27]. These span the N-terminal unstructured segment (residues 1–12) and the fourth β-strand (residues 83–91).

Sso AcP was found to aggregate in 50mM acetate buffer, pH 5.5, 25°C, in the presence of 15–25% (v/v) 2,2,2-trifluoroethanol (TFE) [26]. Before aggregation starts, Sso AcP was found to adopt, under these conditions, a native-like state largely indistinguishable from the fully native state. This is clearly indicated by a number of observations. First, the analysis of the rates of folding and unfolding processes under conditions promoting aggregation shows that folding is manifold faster than unfolding (Fig. 2A) [26]. This causes the folded state to predominate in the equilibrium between the folded and any unfolded or partially unfolded state existing under such conditions. Second, the protein is enzymatically active under conditions promoting aggregation, before aggregation occurs (Fig. 2B) [26]. This confirms further that self-assembly of the protein starts from a conformational state bearing native-like features. Third, the far- and near-UV circular dichroism (CD) spectra of a mutant lacking the N-terminal segment and resisting aggregation are identical, when acquired under conditions promoting aggregation, to the corresponding spectra from wild-type Sso AcP in conditions in which aggregation does not occur [27]. It therefore appears that the secondary structure content and the packing around the aromatic residues of the Sso AcP globular unit do not undergo detectable changes upon transfer to aggregating conditions (Fig. 2C and D). Fourth, at the pH value of 7.5, at which wild-type Sso AcP aggregates slowly even in the presence of TFE, the pattern of sites cleaved by proteases under conditions of limited proteolysis is similar when the protein is analyzed in the absence or in the presence of 15% or 25% (v/v) TFE (Fig. 2E) [27]. This suggests again an unchanged three-dimensional structure upon TFE addition. Finally, the recent observation that the [1H, 15N] HSQC spectrum of Sso AcP does not change substantially upon TFE addition at pH 5.5, with only minor modifications for the chemical shifts of a few residues, is the latest indication that Sso AcP maintains a folded and largely native-like structure under conditions that promote aggregation (unpublished results).

View Large Image Download to PowerPoint

Fig. 2. (A) Natural logarithm of folding and unfolding rate of Sso AcP as a function of TFE concentration. The plot shows folding rate (●), unfolding rate (○) and aggregation rate (□). The gray area represents the range of TFE concentrations in which ordered aggregates able to bind ThT, ANS, and CR are formed. Readapted from Ref. [26]. (B) Enzymatic activity of Sso AcP under different concentrations of TFE. The protein was pre-incubated for 30min in the absence of TFE (○) and in the presence of different concentrations of TFE (●). The gray area represents the range of TFE concentrations in which ordered aggregates able to bind ThT, ANS, and CR are formed. Readapted from Ref. [26]. (C) Far-UV CD spectra for native wild-type (continuous line), native ΔN11 Sso AcP (dashed line) and ΔN11 Sso AcP incubated under aggregating conditions (dotted line) for 4h. Readapted from Ref. [27] with permission from Elsevier. (D) Near-UV CD spectra of wild-type Sso AcP in the absence of TFE (continuous line) and of ΔN11 Sso AcP in 20% TFE (dashed line). Reprinted from Ref. [27] with permission from Elsevier. (E) Location of proteolytic sites obtained for Sso AcP. The three diagrams show the results obtained in 0%, 15%, and 25% TFE with trypsin (blue), chymotrypsin (green), subtilisin (purple), thermolysin (pink), elastase (yellow), and Glu-C (grey). The figure also shows the location of β-strands and α-helices along the sequence according to the X-ray structure. Reprinted from Ref. [27] with permission from Elsevier.

All these biophysical and biochemical analyses indicate that under conditions promoting aggregation (in the presence of TFE) Sso AcP populates a native-like state, with a secondary structure and a packing very similar to that of the native state in conditions in which aggregation does not occur (in the absence of TFE). However, important differences exist between the dynamics of native Sso AcP in the absence and presence of TFE. Under aggregating conditions the native-like state is characterized by a lower conformational stability than the fully native state. The values of free energy change following unfolding (ΔGU−F) are 22.6±7.3 and 48.0±1.2kJmol−1 under conditions that do and do not promote aggregation, respectively [27]. Hydrogen/deuterium exchange of the backbone amides, monitored with both NMR (unpublished results) and electro-spray ionisation mass spectrometry [27], indicate that most of the amides that are protected in native Sso AcP in the absence of TFE exchange faster in the presence of the cosolvent and that a larger number of amides appear to have exchanged after 1h.

This appears to be, to date, the most significant difference observed upon transfer of native Sso AcP from non-aggregating to aggregating conditions: increased dynamics and thermodynamic instability in the absence of large structural changes. In principle, this causes an increased probability to populate native-like states on the “native” side of the major energy barrier for unfolding, but also fully or partially unfolded states on the other side of the barrier. Thus, the observation of a native-like state with increased dynamics and instability does not per se demonstrate that aggregation proceeds through native-like states rather than requiring unfolding. In the next section we will see that evidence for a process of native-like aggregation came rather from a kinetic analysis of the aggregation process and from a structural investigation of the early aggregates.

4. Native-like Sso AcP aggregates into native-like aggregates 

The dynamic, native-like state populated by Sso AcP under aggregating conditions self-assembles through at least two sequential phases (Fig. 3A–C) [28]. The first phase is completed within one minute and can be followed as a small change of the far-UV CD spectrum, particularly as a decrease of the mean residue ellipticity at 208nm (Fig. 3E), and as an intense increase of light scattering (Fig. 3B and inset therein). These assemblies, which will be referred to as early aggregates in this review, show enzymatic activity (Fig. 3C). The enzymatic activity can be directly attributed to the early aggregates rather than to native-like monomers possibly present in solution, as it disappears after removal of aggregated material by filtration (Fig. 3C) [28].

View Large Image Download to PowerPoint

Fig. 3. (A–C) Change of (A) ThT fluorescence, (B) light scattering intensity and (C) enzymatic activity during aggregation of Sso AcP. The squares in (B) and (C) indicate the light scattering intensity and enzymatic activity of filtered solutions containing native Sso AcP in the absence of TFE. The increase in light scattering and the decrease in activity in the first 100s correspond to the formation of the early aggregates. The diamond in (C) indicates the residual enzymatic activity after filtration of the solution of the enzymatic activity assay containing both substrate and protein. The triangle in (C) indicates the enzymatic activity after centrifugation and filtration of the solution containing Sso AcP in the presence of TFE. Readapted from [28] with permission from Elsevier. The inset in panel (B) represents the time course of static light scattering within the first 200s. (D and E) FTIR (D) and far-UV CD (E) spectra of native Sso AcP (continuous line), Sso AcP early aggregates (dashed line) and protofibrils (dotted line). Reprinted from Ref. [28] with permission from Elsevier. (F) Increase of CR absorption, ThT fluorescence and ANS fluorescence for native Sso AcP (N), the early aggregates (E.A.) and the protofibrils (PF.). Data were normalized to attribute 100% to the maximum value for each assay. Reprinted from Ref. [28] with permission from Elsevier.

The early aggregates are not able to bind Thioflavin T (ThT), Congo red (CR) and 8-anilino-1-naphthalenesulfonate (ANS) (Fig. 2F). Furthermore, they do not show a remarkable increase in β-sheet structure, as inferred from Fourier transform infrared spectroscopy (FTIR) and far-UV CD (Fig. 3D and E) [28]. The far-UV CD and FTIR spectra of the early aggregates, albeit similar to those of native Sso AcP, highlight small changes reminiscent of the appearance of further β-turn or loop structure and of the presence of intermolecular β-sheet (Fig. 3D and E) [28]. The NMR analysis recently carried out on the Sso AcP aggregates has shown a native-like appearance of the [1H, 15N] HSQC spectrum and large D values in the DOSY spectra, indicating that in such assemblies the individual protein molecules adopt a native-like conformation and a large mobility, at least within the low molecular weight aggregates (unpublished results).

Taken together, these data show that the first aggregation phase consists in the direct self-assembly of native-like Sso AcP molecules to form early aggregates in which the individual molecules still retain a native-like secondary and tertiary structure, with a catalytic site that is not yet, albeit possibly distorted, unstructured or inactivated. The early aggregates are not yet characterized by structural and tinctorial characteristics typical of amyloid-like structure. The comparison between the folding, unfolding and aggregation rates at pH 5.5, 25°C and 20% (v/v) TFE indicates that aggregation is faster than unfolding, showing that formation of the early aggregates does not require a process of unfolding (Fig. 2A) [26].

5. Native-like aggregates convert into amyloid-like protofibrils 

The first rapid aggregation phase is followed, on the time scale, by a second slower phase, complete within 1h [28]. Such a second step is exponential and can be followed as a remarkable increase of ThT fluorescence (Fig. 3A), as a small decrease of light scattering (Fig. 3B) and as a disappearance of enzymatic activity (Fig. 3C). The aggregates formed after the second phase appear as spheres or short and thin fibrils, having a diameter of 3–5nm, as inferred from transmission electron microscopy [26], [28]. By contrast to early aggregates, these species are able to bind ThT, CR and ANS (Fig. 3F), are not enzymatically active (Fig. 3C) and possess extensive β-sheet structure as deduced from far-UV CD and FTIR spectroscopy (Fig. 3D and E) [26], [28]. These morphological, structural and tinctorial properties have been observed for other peptide and protein systems and define species often referred to as amyloid-like protofibrils [29], [30], [31], [32]. The binding to ANS suggests that hydrophobic clusters are exposed on the surface of these structures and may explain their tendency to assemble further and form larger clusters of aggregates visible in transmission electron microscopy images and detectable by DLS [26], [28].

The early aggregates diluted into conditions in which aggregation does not take place are inevitably committed to convert into amyloid-like protofibrils, whereas the native protein does not aggregate when incubated in the same conditions [28]. These kinetic tests have indicated that these protofibrillar species form directly from early aggregates with no need of their disassembly and that the latter species are thus on pathway to amyloid formation [28]. Furthermore, we showed that, while the rate constant of the first aggregation phase depends on protein concentration, the rate constant for the second aggregation phase is largely independent of the amount of protein in solution [28], [33]. These observations suggest that, while formation of early aggregates is an intermolecular process which requires interaction of monomers, protofibril formation involves a structural reorganisation within preformed early oligomers, from a native-like fold into an amyloid-like one [33].

The existence of such phase is not surprising. It was found that when the native state is in a prevalently β-sheet structured conformation the resulting aggregates present cross-β structure perpendicular to the fibril axis and do not require further rearrangement [34], [35], [36]. However, when the native state consists of a conformation containing not only β-sheet structure, a reorganization of the structure from the initial native-like aggregates is required to reach the amyloid-like cross-β structure, which is at a minimum of the free energy profile [6].

6. The intermolecular interaction between the N-terminal segment and the globular unit promotes native-like aggregation 

In order to get insight into the process of Sso AcP self-assembly at a molecular level, the regions of the sequence that promote the formation of the early aggregates were investigated. It was first found that a mutant of Sso AcP lacking the first 11 unstructured residues and thus consisting only of the globular unit of the molecule (ΔN11 Sso AcP) is not able to aggregate into either the early or the later aggregates under conditions promoting aggregation of the wild-type protein, even though it maintains the same native-like fold [27]. However, peptides corresponding to the initial 11 unstructured residues (N11) or 14 residues (N14) of Sso AcP also do not aggregate in the same conditions. Solubility assays, far-UV CD and DLS show that the peptides are soluble, monomeric and unstructured [27]. Furthermore, ΔN11 Sso AcP does not aggregate even when it is incubated in the presence of 25 molar excesses of the N11 peptide (Fig. 4A) or in the presence of 4 molar excesses of the wild-type protein [33]. Taken together, these experiments show that neither the unstructured N-terminal segment nor the globular unit of Sso AcP can aggregate when separated from each other and that the two parts of the protein need to be covalently bound to trigger the process [33].

View Large Image Download to PowerPoint

Fig. 4. (A) Time course of ThT fluorescence during incubation of ΔN11 Sso AcP in aggregation promoting conditions in the presence of different molar excesses of N11 (red symbols). Data are shown for 2-fold (■), 5-fold (●), 10-fold (▴) and 25-fold (▾) molar excesses. The signal is shown as a ratio between ThT fluorescence in the presence (F) and in the absence (F0) of protein. The kinetic trace for wild-type Sso AcP in the same conditions with no peptides (black ●) is shown for comparison. The data show that N11 is not able to restore aggregation of ΔN11 Sso AcP. Readapted from Ref. [33] with permission from Elsevier. (B) Dependence of the rate of formation of early aggregates (k) on the change of conformational stability of Sso AcP upon mutation, (ΔΔGU−F). Mutations of residues belonging to the fourth β-strand (residues 83–91) significantly deviate from the correlation. Reprinted from Ref. [27] with permission from Elsevier. (C) Formation of early aggregates in a set of Sso AcP variants. Traces are shown for wild-type (black), V84P (blue), V84D (green) and Y86E (red). The plot shows that the introduction of protective mutations on the fourth β-strands abolishes native-like aggregation. Reprinted from Ref. [40] with permission of ACS Publications. (D) First phase of Sso AcP aggregation, monitored using static light scattering, in the presence of a 4-fold molar excess of N11 peptide (red ●) and of a control peptide having the same amino-acid content as N11 but presenting different primary sequence (blue ●). The trace shows that N11 peptide is able to slow down aggregation of Sso AcP while the control peptide is not. Reprinted from Ref. [33] with permission from Elsevier. (E) Proposed mechanism for the aggregation of Sso AcP (readapted from [33]). Three phases are depicted: (i) conversion of the native state (N) into native-like state (N∗); (ii) conversion of N∗ into early aggregates; (iii) conversion of early aggregates into amyloid-like protofibrils.

Limited proteolysis experiments performed with 6 different proteases showed that, in both the absence and presence of TFE, the regions of monomeric Sso AcP exposed to the solvent and/or flexible span the N-terminal segment, the β-hairpin between β-strand 2 and 3 and the β-strand 4 (Fig. 2E) [27]. It was hypothesised that either the second or the third region could have a degree of solvent-exposure and flexibility sufficient to participate to the formation of intermolecular interactions with the unstructured N-terminus of another Sso AcP molecule. Based on these results a number of mutants were produced with amino acid substitutions contained in the two suspected regions and others. A negative correlation was then found between the decrease in conformational stability of the native state following mutation and the rate constant for the formation of early aggregates of the corresponding protein variant, confirming the importance of flexibility within the native-like state of Sso AcP in promoting aggregation (Fig. 4B) [27]. However, only mutants of residues in the β-strand 4 and in the N-terminal segment significantly deviate from this correlation, suggesting that these regions play an additional, different role (Fig. 4B) [27].

The structural inspection of Sso AcP and of the second acylphosphatase from Drosophila melanogaster (AcPDro2), which also aggregates directly from a native-like state [37], shows that such proteins differ from human muscle acylphosphatase (mAcP) and the N-terminal domain from E. coli HypF (HypF-N), which by contrast aggregate via unfolding [38], [39], for two structural features: (i) the higher number of unstructured regions and loops and (ii) the lower number of protections at the edge β-strands [40]. Accordingly, both deletion of the unstructured N-terminal segment and insertion of additional protections at the edge β-strand 4 of Sso AcP inhibit formation of the early aggregates (Fig. 4C), establishing a role of these two sequence regions in the process [27], [40]. Recent NMR experiments have shown that the regions of the sequence with the highest density of changes of 1H and 15N chemical shifts upon conversion of the native state into the early aggregates involve the N-terminus and the region encompassing β-strand-4, β-strand-5 and the loop between them. They also showed that other regions undergo other changes, although limited to a small number of residues within them. This confirms the involvement of the N-terminal segment and of the region comprising β-strand 4, and possibly the following segments, in the formation of the early aggregates.

If these experiments provided important information on the regions of the Sso AcP sequence promoting the formation of the early aggregates they did not, however, allow the formulation of a model able to account for the mechanism of the process at a molecular level. In follow-up studies it was found that a mutant in which the unfolded segment was moved from the N-terminus to the C-terminus exhibits the same aggregation properties as the wild-type protein [33]. The N- and C-terminus of Sso AcP are distant from each other, suggesting that specific intra-molecular interactions between the segment and the globular unit within the native-like monomeric state are unlikely to be necessary prerequisites for the process. In fact, such interactions would necessarily be affected by moving the unstructured segment to the distant C-terminus. Furthermore, aggregation of wild-type Sso AcP was found to be slower in the presence of the N11 peptide, whereas it was not affected by a peptide bearing the same amino-acid content as the N11 peptide but a scrambled, primary sequence (Fig. 4D) [33]. This suggests that the N11 peptide interacts with the wild-type protein inhibiting its aggregation. It also suggests that the interaction between the unstructured N-terminal segment and the globular unit of the molecule is intermolecular, as in the case of intra-molecular interactions the N11 peptide should speed up the process or produce no effect. The ability of the N-terminal segment to induce aggregation of the full length protein does not appear to be due to its overall physicochemical properties dictated by the amino acid composition, but rather to its primary sequence. Correspondingly, the interaction between this portion and the globular unit of the Sso AcP molecule is seemingly a specific one.

In contrast to the first phase of aggregation, the conversion of the early aggregates into the amyloid-like protofibrils seems to be a cooperative process involving the whole molecule. A significant correlation was found between the change in conformational stability of the native state following mutation and the rate constant for the conversion of the early aggregates into the later protofibrils, with no mutants deviating significantly from this trend [27]. This result indicated that flexibility within the early aggregates is an important factor promoting their conversion and that such conversion is a cooperative process involving the whole Sso AcP molecule [27].

These experiments led us to the formulation of a model for the aggregation of Sso AcP (Fig. 4E) [33]. In this model the major event in the self-assembly of native-like Sso AcP into native-like aggregates (first phase) is the establishment of a specific intermolecular interaction between the N-terminal segment of one Sso AcP molecule and the globular unit of another one, probably at the level of the β-strand 4. A structural flexibility involving the whole molecule is also a necessary prerequisite for the process to occur. This model is similar to a model proposed for the aggregation of the prion Ure2p, in which an unstructured segment interacts with the globular part of another molecule [41] and to the mechanism proposed for the assembly of serpins, in which a loop interacts with the β-sheet of another molecule, triggering self-assembly [42]. In the case of human transthyretin (TTR) it was shown that aggregation requires the dissociation of the native tetramer and the formation of a monomer in which two edge β-strands are unfolded. This leads to the exposure of other regions which trigger self-assembly [6], [43]. It is possible that also in the case of Sso AcP the edge β-strand 4 does not directly establishes interactions with the N-terminal segment but participates to the structural fluctuations from the native state to form native-like states. Further investigation will be required to shed light on this issue. The conversion of the early aggregates into the amyloid-like protofibrils involves the whole Sso AcP molecule and is a cooperative process.

7. Stabilisation of the native state is an effective therapeutic strategy to inhibit native-like aggregation 

The addition of ligands binding to the native state of Sso AcP, such as the phosphate ion, was found to slow down considerably both the conversion of the native-like state into the early aggregates as well as the subsequent conversion of these oligomers into the later protofibrils [44]. The rationale of this finding lies in the ability of this ligand to generate, through specific binding, a sort of crystallisation of the native state inhibiting both kinetically and thermodynamically not just the unfolding process, but also the structural fluctuations that are necessary to reach the native-like state and initiate aggregation. The conversion of the early native-like aggregates into the amyloid-like protofibrils is also inhibited by ligand binding for similar reasons. This result raises optimism in the utilisation of therapeutic strategies aimed at inhibiting pathways of aggregation such as those observed here for Sso AcP. Indeed, stabilisation of the native state through specific ligand binding, or via any other strategy, is anticipated to be an effective means to inhibit both native-like and unfolding-based aggregation processes and a promising avenue in the search of an effective cure of protein deposition diseases associated with initially globular proteins. Important results have been obtained with this strategy in the case of TTR aggregation [45]. However, the design of therapeutic strategies aimed at interfering directly with the formation of intermolecular interactions requires a knowledge of both the aggregation pathway followed by a protein and the molecular level at which this occurs.

8. Conclusions and future perspectives 

The aggregation process of Sso AcP recapitulates many of the fundamental features of the aggregation pathway starting from native-like states. The formation of the native-like state does not involve any global unfolding event but involves thermal fluctuations from the fully native state or other events that are however distinct from unfolding processes across the major energy barrier for unfolding. Aggregation of Sso AcP leads to the formation of assemblies in which the individual molecules maintain initially a structure similar to the initial aggregation prone state in its monomeric form, which in this case presents a native-like fold. Since the Sso AcP native fold does not contain only β-sheet structure, a second aggregation phase is necessary in which the amyloid-like fold is reached. Processes of native-like aggregation in which the proteins aggregate from native-like states and maintain either transiently or permanently a native-like structure in the aggregates are reported for other systems [6], [34], [35], [36], [41], [42], [43], [46], [47], [48].

Mechanisms of native-like aggregation such as those described here for Sso AcP and described elsewhere for other systems have been observed and studied in vitro, in an environment and context far from that existing in vivo. Nevertheless, the observation that native-like aggregates can also form in vivo, for example in bacterial inclusion bodies which are demonstrated to be good model systems to study the formation and structure of amyloid fibrils in living organisms [9], [10], [12], [13], [14], provides a strong indication that folded proteins have a significant propensity to aggregate. The finding that proteins have evolved structural and sequence adaptations to counteract aggregation of their folded states, such as that described to protect their edge β-strands from aggregation only when they are structured in native β-sheets [19], is perhaps one of the strongest indications that native-like aggregation is a challenge for any living system against which an evolutionary pressure has been exerted. Along the same lines, the increasing evidence that the cellular machinery dedicated to the maintenance of protein homeostasis is effective in inhibiting aggregation of folded states, in addition to counteract aggregation of proteins in their fully or partially unfolded states, suggests that native-like aggregation is per se a challenge in the cell [16], [17], [18].

Further investigation is needed to shed light on the structural and dynamical properties of the native-like amyloidogenic state and on the molecular mechanism by which this converts into native-like aggregates down to the amyloid-like protofibrils and fibrils. Novel experimental design and technology are also required to detect and characterise processes of this type in vivo, with an aim of understanding maintenance of protein homeostasis in biology and occurrence of protein aggregation in pathology. The elucidation of protein aggregation at the molecular level will have implications in the search for therapeutic strategies specific for the identified pathway. It will also have important implications in biotechnology as our anticipated ability to produce large amounts of functional proteins in the form of inclusion bodies can have important exploitable implications.