Portal de Periódicos da CAPES

For Protein Misassembly, It's the “I” Decade

1996; Cell Press; Volume: 86; Issue: 5 Linguagem: Inglês

10.1016/s0092-8674(00)80143-9

1097-4172

Ronald Wetzel,

Protein Structure and Dynamics

Insoluble aggregates of normally well-behaved proteins are featured in a variety of human disease states, including various forms of amyloidosis (15Sipe J.D Annu. Rev. Biochem. 1992; 61: 947-975Crossref PubMed Scopus (398) Google Scholar) and the prion diseases (12Prusiner S.B DeArmond S.J Amyloid. 1995; 2: 39-65Crossref Scopus (25) Google Scholar). Protein aggregation and precipitation is also a commonplace observation in biotechnology, in both the inclusion body formation that can occur in the cell during heterologous expression of cDNAs and in attempts to refold these proteins in vitro (11Mitraki A King J Biotechnology. 1989; 7: 690-697Crossref Scopus (424) Google Scholar). In addition, it is now well-accepted that a primary function of many molecular chaperones is to thwart misassembly and aggregation during the protein folding process (4Hartl F.U Science. 1996; 381: 571-580Google Scholar). There is a voluminous literature on the interaction of chaperones with proteins in various states of folding. There has been generally less interest in the characterization of aggregation processes or products in non-chaperoned folding, perhaps in part because of historical assumptions about the nature of the phenomenon. The conventional wisdom regarding protein aggregation and precipitation has occupied two extreme positions. Some processes, like aggregation during folding of a protein in vitro, or inclusion body formation in bacteria, are usually regarded as being driven by nonspecific, hydrophobic interactions operating on random coil states or on collapsed, molten globule states. Other processes, such as the extracellular aggregation of some proteins into amyloid fibrils, have been visualized according to the sickle hemoglobin model, in which mutations alter the local surface properties of the native, folded state to introduce new packing interactions for noncovalent polymerization. The first view tends to discourage attempts at mechanistic understanding and therapeutic intervention; the second suggests that the straightforward route to both lies in the determination of the high resolution structures of native states. Recent results suggest, however, that many examples of protein aggregation occur by mechanisms involving structured folding intermediates. There are several important implications of such mechanisms. One is that there exists a class of diseases involving aberrant protein folding. Another is that such processes can involve interactions that exhibit significant structural specificity which cannot be deduced by the examination of native states. This is not a new idea. In 1974, Michel Goldberg and colleagues put forward a model for the unfolding-induced aggregation of multidomain proteins in vitro (11Mitraki A King J Biotechnology. 1989; 7: 690-697Crossref Scopus (424) Google Scholar). This model, shown in Figure 1, invokes structured folding intermediates with domains or subdomains folded as they are in the native state, but which undergo intermolecular, rather than intramolecular, interactions with each other during folding or unfolding—leading to the formation of polymers of partially folded states held together by noncovalent, native-like interactions. In the 1980s a few studies on protein aggregation in vitro and in bacteria during folding and unfolding suggested that these processes, in fact, could involve specific interactions of folding intermediates (11Mitraki A King J Biotechnology. 1989; 7: 690-697Crossref Scopus (424) Google Scholar). For example, David Brems and colleagues at Upjohn showed that a folding intermediate is involved in the aggregation of growth hormone in vitro and that the hydrophobic face of a particular helix is important to this interaction. The Jonathan King group at MIT identified a series of temperature sensitive folding mutations that function not by destabilizing the protein's native state but by placing a particular folding intermediate in jeopardy, leading to the irreversible formation of inclusion bodies. In this article I review the highlights of work over the past few years that provides further support of misassembly mechanisms involving structured folding intermediates, especially with respect to human diseases involving protein deposition. Many of these studies also begin to address key questions suggested by the involvement of partially folded states: structural details on the aggregation intermediates and the aggregates themselves, specificity of aggregate formation, and strategies for inhibiting these processes. In principle, the formation of aggregate from a folding intermediate can take place either in the folding or unfolding direction (Figure 2, top). Interestingly, both pathways are probably represented in cases of protein deposition in vivo (17Wetzel, R. (1992). In Stability of Protein Pharmaceuticals: In Vivo Pathways of Degradation and Strategies for Protein Stabilization, T.J. Ahern and M.C. Manning, eds. (New York: Plenum Press), pp. 43–88.Google Scholar). In the case of cytoplasmic inclusion body formation in bacteria, off-pathway aggregation probably most often occurs in the folding direction, with the extent of deposition depending on a kinetic competition between productive folding and aggregation for a poorly soluble, transient intermediate. Amyloid formation, in which fibrils are deposited outside the cell, presumably occurs at some point after proteins have had an opportunity to complete most or all of the folding process. In this case, aggregate formation seems to involve unfolding intermediates that can be populated, in equilibrium with the native state, under physiological conditions. The in vitro correlate for the former process is the loss of molecules due to aggregation during refolding of denatured proteins; the correlate for the latter is the loss of molecules to aggregation during thermal unfolding of native proteins. Amyloid fibril formation associated with different human disease states occurs with a variety of different proteins and peptides that have no obvious common properties in amino acid sequence, three-dimensional structure, or function (15Sipe J.D Annu. Rev. Biochem. 1992; 61: 947-975Crossref PubMed Scopus (398) Google Scholar). Despite these differences, amyloid fibrils are remarkably similar in size and shape as viewed in the electron microscope, and also share dye-binding and optical properties that suggest significant structural similarity (15Sipe J.D Annu. Rev. Biochem. 1992; 61: 947-975Crossref PubMed Scopus (398) Google Scholar). There is now substantial evidence that unfolding intermediates are the building blocks for amyloid fibril formation from globular proteins. For example, the kinetics of amyloid fibril formation in vitro by the wild type version of the protein transthyretin are most favorable under conditions in which the protein is in a non-native state (7Kelly J.W Curr. Opin. Struct. Biol. 1996; 6: 11-17Crossref PubMed Scopus (556) Google Scholar). Furthermore, amyloid disease-associated mutations in transthyretin (7Kelly J.W Curr. Opin. Struct. Biol. 1996; 6: 11-17Crossref PubMed Scopus (556) Google Scholar) and immunoglobulin light chain VL domain (5Helms L.R Wetzel R J. Mol. Biol. 1996; 257: 77-86Crossref PubMed Scopus (96) Google Scholar) have been shown to destabilize the native states of these proteins, allowing them to more easily enter global, cooperative unfolding transitions that can generate partially folded states that are the presumptive amyloidogenic intermediates. The involvement of non-native states in amyloid formation is also supported by data suggesting that the secondary structural content of the protein in the aggregated state is different from the native state. For example, the fundamental structural unit of all amyloid fibrils is thought to be stacked, anti-parallel β-sheet (15Sipe J.D Annu. Rev. Biochem. 1992; 61: 947-975Crossref PubMed Scopus (398) Google Scholar), but some amyloid-forming proteins, such as apolipoprotein A-1, are rich in α-helix in the native state. Similarly, the cellular form of the prion protein is rich in α-helix when isolated, while the aggregated, infectious form contains more β-sheet (12Prusiner S.B DeArmond S.J Amyloid. 1995; 2: 39-65Crossref Scopus (25) Google Scholar). These results suggest that some protein misassembly processes may require significant secondary structural rearrangements for oligomerization to proceed (Figure 1). One of the major challenges in this field is to move beyond these initial observations by determining at high resolution the structures of both bona fide aggregation intermediates and of final aggregated states. There are significant technical barriers to achieving this, however. For example, the metastable nature and inherently poor solubilities of aggregation-prone folding intermediates make solution-phase NMR analysis difficult. For the aggregates themselves, new solid state NMR methods are being developed which may ultimately provide important distance constraints to support construction of models of amyloid structure (9Lansbury P.J Costa P.R Griffiths J.M Simon E.J Auger M Halverson K.J Kocisko D.A Hendsch Z.S Ashburn T.T Spencer R.G et al.Nat. Struct. Biol. 1995; 2: 990-998Crossref PubMed Scopus (428) Google Scholar). In the solid state, aggregates like amyloid fibrils are, at best, paracrystalline, and thus cannot be solved at high resolution by conventional X-ray crystallographic methods. However, recent synchrotron X-ray studies of aligned fibrils have provided sufficient data to support construction of a model for an amyloid fibril. This important model features a continuous β-sheet helix, and strongly suggests the involvement of a non-native subunit in fibril formation (2Blake C Serpell L Structure, in press. 1996; Google Scholar). Despite their limitations in direct analysis of aggregation intermediates and the aggregated state, conventional X-ray crystallography and solution NMR analysis do have important roles to play in this field. For example, the high-resolution structures of a number of unusual "domain-swapped" dimeric proteins have been described (1Bennett M.J Schlunegger M.P Eisenberg D Prot. Sci. 1995; 4: 2455-2468Crossref PubMed Scopus (663) Google Scholar) which clearly support the Goldberg model of oligomerization by native-like interactions of folding intermediates (Figure 1). High resolution structures of an aggregating protein's native state also can provide valuable clues to the aggregation process. The recently described structure of a major fragment of the prion protein (13Riek R Hornemann S Wider G Billeter M Glockshuber R Wuthrich K Nature. 1996; 382: 180-182Crossref PubMed Scopus (1084) Google Scholar) identified the locations of the expected α-helices and also unveiled a small element of β-sheet structure which may turn out to be involved in the aggregation process. Refined X-ray crystal structures and NMR solution structures can also give valuable information on the most mobile—and thus perhaps most labile—elements of the folded structure, in turn suggesting where unfolding might start and what an early unfolding intermediate might look like. There are now several examples in which similar or identical proteins are capable of forming different aggregate types, as assessed by morphological or functional criteria. This has several important implications. In particular, if the biological effects of protein deposition in some human diseases are attributable to the specific interactions of the aggregate with particular molecules or cells, then different aggregated forms of the same protein may have different pathological consequences. Thus, the ability to form aggregates of functionally different morphologies may help explain how the misfolding or misassembly of a single prion protein sequence can give rise to various scrapie "strains" (8Kocisko D.A Priola S.A Raymond G.J Chesebro B Lansbury P.J Caughey B Proc. Natl. Acad. Sci. USA. 1995; 92: 3923-3927Crossref PubMed Scopus (316) Google Scholar). In fact, infectious prions are known to exhibit morphologically distinct aggregated states, although the exact relationship between any of these states and the neurotoxic and infectious properties of prions is not clear (12Prusiner S.B DeArmond S.J Amyloid. 1995; 2: 39-65Crossref Scopus (25) Google Scholar). In another example, aggregates of immunoglobulin variable domains are involved in both light chain amyloidosis (AL) and light chain deposition disease (LCDD), but the morphology of the deposited aggregates are clearly different. Not only are both AL and LCDD associated with destabilizing amino acid changes, but the nature of these sequence changes may also help dictate aggregate morphology and hence disease type (5Helms L.R Wetzel R J. Mol. Biol. 1996; 257: 77-86Crossref PubMed Scopus (96) Google Scholar). One way particular sequence changes might control aggregate morphology is by controlling the extent to which different aggregation-prone unfolding intermediates (such as Inat and Inon in Figure 1) are populated under physiological conditions. Although aggregate morphology informs us about structural differences at a macroscopic level, a more discerning test of relatedness—at the molecular level—may be the ability of an aggregate to provide a growth point, or seed, for further aggregation. A nucleated growth mechanism (Figure 2, bottom) has been demonstrated for several amyloid peptides in vitro (6Jarrett J.T Lansbury P.J Cell. 1993; 73: 1055-1058Abstract Full Text PDF PubMed Scopus (1829) Google Scholar). One peptide that exhibits nucleation and seeding in fibril formation is the Alzheimer's peptide Aβ. This peptide forms amyloid fibrils when incubated at pH 7.4, but more amorphous aggregates when incubated at pH 5.8 (20Wood S.J Maleeff B Hart T Wetzel R J. Mol. Biol. 1996; 256: 870-877Crossref PubMed Scopus (328) Google Scholar). Although these two aggregate types are composed of exactly the same peptide, the pH 5.8 aggregate—in contrast to the pH 7.4 aggregate—is incapable of seeding Aβ fibril formation at pH 7.4. If disease-related protein aggregation is mediated by precise packing interactions of structurally well-defined partially folded intermediates, it becomes more feasible to consider the possibility of interfering with these interactions by the traditional pharmaceutical approach of identifying small molecules which bind to and block interaction sites. Some possible points of attack are indicated in Figure 2. Stabilization of the native state of transthyretin by ligand-binding (1) has been reported to stabilize the protein against fibril formation in vitro (10Miroy G.J Lai Z Lashuel H Peterson S.A Strang C Kelly J.W Proc. Natl. Acad. Sci. USA, in press. 1996; Google Scholar). Small organic molecules can inhibit the off-pathway aggregation of proteins during folding in vitro (14Rudolph, R. (1996). In Protein Engineering: Principles and Practice, J.L Cleland and C.S. Craik, eds. (New York: Wiley-Liss), pp. 283–298.Google Scholar), probably by discouraging aggregate packing (2). In vitro fibril formation by Aβ can also be inhibited by small molecules (19Wood S.J MacKenzie L Maleeff B Hurle M.R Wetzel R J. Biol. Chem. 1996; 271: 4086-4092Abstract Full Text PDF PubMed Scopus (155) Google Scholar), which apparently bind to Aβ and in doing so either interfere with self-association (4) or shift a conformational equilibrium of the monomer away from the amyloidogenic form (3). Nucleation (3Evans K.C Berger E.P Cho C.-G Weisgraber K.H Lansbury Jr., P.T Proc. Natl. Acad. Sci. USA. 1995; 92: 763-767Crossref PubMed Scopus (336) Google Scholar) and seeding (18Wood S.J Chan W Wetzel R Biochemistry, in press. 1996; Google Scholar) of Aβ fibril formation (5) can be inhibited in vitro by apolipoprotein E, which may help to explain the genetic risk for developing Alzheimer's disease associated with certain apoE alleles. Noncovalent association of proteins into aggregates can set the stage for several modes of covalent crosslinking, which would be expected to make reversal in vivo much more difficult. Agents capable of selectively breaking these covalent crosslinks (16Vasan S Zhang X Zhang X Kapurniotu A Bernhagen J Teichberg S Basgen J Wagle D Shih D Terlecky I Bucala R Cerami A Egan J Ulrich P Nature. 1996; 382: 275-278Crossref PubMed Scopus (411) Google Scholar) might play an important role in the therapeutic disaggregation of such aggregates. It is well known that amino acid sequence controls the final form a protein takes when it folds, and the free energy by which that native state is favored. Only recently has it become apparent that amino acid sequence also controls the viability of the path(s) to the correct structure, by controlling and limiting the degree to which misassembly and aggregation irreversibly siphon protein molecules from the folding pathway. It is now clear that these side reactions of folding are not only of fundamental interest for the rules for correct translation of genetic information into functional proteins, they also control the features of a growing body of pathological protein deposition phenomena. At present we can only make out the rough features of this new landscape—the structural biology of non-native states. It may be necessary to devise new tools to bring it to sharper focus.