7 July 2004. Gathering in Colorado’s Rocky Mountains, scientists from around the world discussed their independent research developments, only to realize that they have more in common than previously imagined. One in a series of FASEB Summer Research Conferences, the meeting on “Protein Misfolding, Amyloid and Conformational Disease,” held June 12-17, was elegantly choreographed by Jonathan Weissman of UCSF’s Howard Hughes Medical Institute and University of Cambridge’s Robin Carell. The topics covered over five days in Snowmass included Alzheimer’s, Parkinson’s, and prion diseases, as well as protein structure and misfolding mechanisms. The talks, discussions, and posters shed light on striking similarities among amyloids in their formation and toxicities, especially the emergence of oligomers as a common primary toxic species of amyloids. The following summary represents the range of topics covered, from amyloid fibril structure to autophagy, something to whet the appetites of protein chemists and cell biologists alike (see ARF related conference news story).
Mad About Prions
John Collinge, University College, London, inaugurated the first session by reviewing his group’s animal model of prion propagation and neurotoxicity. These investigators generated a mouse model in which they could turn neuronal expression of prions on or off at will. Blocking prion gene expression in the adult blocked prion disease and reversed spongiosis, even after scrapie inoculation. At a stage when cell loss would typically occur, the expression of cellular prion protein (PrPc) ceases while extra-neuronal prion deposits persist. Interestingly, the deposition of extra-neuronal prions is not toxic, implying that prion aggregation per se is not required for disease pathogenesis. Intra-neuronal prion expression and replication, and prion oligomerization, may be key elements required for cytotoxicity, Collinge reported. Moreover, mice injected with human vCJD exhibited subclinical pathology and were resistant to infection, leading the researchers to conclude that host PrPc primary structure influences infectivity.
For those more inclined to study yeast, Sup35 was the prion of choice. Sup35 is a protein translation suppressor. Aggregation of Sup35 produces the [PSI+] phenotype, which is marked by failure to terminate translation. Jonathan Weissman showed that his lab can create different [PSI+] strains by varying the polymerization temperature, and these different strains have different infectivities. Specifically, fibers made at 4°C did not convert to [PSI+] as well as those prepared at 37°C.
So how do these fibrils grow, and are oligomers obligatory intermediates in fibril formation? To address these questions, Weissman and his group studied discrete steps in the assembly process. They showed first that the vast majority of Sup35 is monomeric prior to polymerization and oligomers are rare; this points to monomer addition as the mechanism of fibril elongation. To confirm this, the researchers used total internal reflection (TIR) microscopy coupled with single molecule fluorescence to study monomer addition. To do so, the researchers added fluorescently labeled Sup35 monomers and observed their addition to fibrils attached to microscope slides.
Physical disturbance modulated this growth rate. In fact, agitating the sample dramatically accelerated polymerization, although the elongation rate only increased at later stages in the process. From this, and from the sigmoidal shape of the fibril growth curve, Weissman concluded that agitation fragments fibers, freeing up ends for addition.
Claudio Soto, University of Texas Medical Branch, Galveston, updated the audience on new advances with automated protein misfolding cyclic amplification (PMCA), his technique for mimicking prion infectivity in vitro. Soto reported a significant improvement in amplification factor, from ~10-30x in the original method to ~6,500x with newer techniques. The technique used hamster brain homogenates as the source of PrPc rather than recombinant PrPc. Small oligomers of misfolded protein act as seed to form larger aggregates that are later sonicated to form small pieces, which grow, multiply, and are sonicated again, each cycle exponentially amplifying the protein in a manner analogous to a PCR reaction. In many ways, this PMCA-produced PrPSc behaved identically to brain PrPSc. Both had the same electrophoretic pattern on SDS gels, amino acid composition, proteinase K resistance, insolubility, ability to propagate, β -sheet structure content, and deglycosylation behavior. Despite these similarities, Soto stressed the importance of his ongoing studies comparing the PMCA-generated prion infectivity in animals with that of natural prions, and similarities in the subsequent pathology. This is a Holy Grail of sorts because, to date, no artificially generated PrPSc has proved to be as infectious as endogenous forms.
Twists and Turns of Fibril Formation
How important is primary structure in determining fibril formation? Louise Serpell of the University of Sussex addressed this question using x-ray and electron diffraction to study amyloid fibrils. Understanding the molecular architecture of fibril formation is a key to unlocking the protein-misfolding problem in amyloid diseases. Serpell pointed out that amyloid proteins do not share sequences or native structure and yet all can form canonical fibrils. Moreover, Serpell proposed that fibril formation and stability are affected by primary structure, as revealed by the study of α-synuclein, Sup35, IAPP, Aβ11-25, and short designed peptides. All showed cross-β diffraction, but also variations in fiber morphology, perhaps due to differences in the amount of order within the fibrils. β-sheet structure was parallel or anti-parallel, depending on protein sequence and amphilicity; furthermore, spacing between β-sheets was dependent on the size of amino acid side chains. This led Serpell to conclude that side-chain interactions may determine which specific fibril morphologies form and how stable the fibrils are.
Yale University’s Andrew Miranker focused on IAPP fibril formation, showing that the lag phase for fibril formation is independent of concentration. IAPP is found in the halo region between insulin and the lipid bilayer in secretory β cells. Made up of liposomes composed of phospholipids, the lipid bilayer is thought to catalyze IAPP fibril formation at the N-terminus. Miranker showed that liposomes composed of anionic lipids were better at inducing fibril formation than those composed of cationic lipids. Interestingly, anionic lipids are upregulated in type II diabetes, and Miranker argued that charge is important in fibril formation and protein aggregation.
Sheena Radford of the University of Leeds, UK, presented data on the molecular mechanism of β 2-microglobulin (β 2M) amyloid formation. Amyloidosis caused by β 2M assembly develops as a complication of long-term kidney dialysis. So how does the monomeric protein fold, and is there more than one assembly pathway? Radford reported that low pH rapidly generated straight fibrils, while the presence of salt yields curvy fibrils. Could one be the progenitor of the other? Radford elucidated a bifurcated assembly pathway: monomers give rise to short rods and finally curvy fibrils, whereas a classical nucleation and growth process leads to long straight fibrils.
To examine the conformational properties that drive fibril formation, Radford analyzed the E strand of β -2M, mutating it such that it behaved more like random coil and perturbed fibril formation. From these data, Radford concluded that the ability to form amyloid depends exclusively on sequence, and that aromatic groups that stabilize the unfolded state may be critical in mediating fibril formation.
Before They Were Fibrils: Shift Toward Prefibrillar Assemblies
Chris Dobson of the University of Cambridge, UK, opened the session titled “Are oligomers the toxic species?” by presenting an overview of the generic nature of protein misfolding. He emphasized that the core structure of amyloid is determined primarily by a common polypeptide main chain, not by the variable side chains, as indicated by fiber x-ray and electron diffraction studies and protein modeling. Every polypeptide can access a variety of conformational states on a folding surface, and access to specific states is controlled in part by specific side-chain interactions. When cellular housekeeping mechanisms such as chaperones, quality control, and degradation mechanisms fail, proteins are free to interconvert among conformational states, including those that lead to oligomeric and fibrillar assemblies. The generic aggregation properties shared by amyloidogenic proteins are tied to a generic cytotoxicity of prefibrillar aggregates. What’s more, Dobson cited recent data showing that non-disease-related prefibrillar aggregates such as SH3 and HypF are just as toxic to cell cultures as proteins like Aβ.
This shift in focus to oligomers as the major pathogenic species in amyloidoses necessitates what David Teplow of Harvard University referred to as the “revised amyloid cascade hypothesis,” in which oligomerization is the seminal event in the pathogenesis of Alzheimer’s disease (AD). In this model, paranuclei (Aβ pentamer/hexamer units) form protofibrils and finally fibrils. Teplow’s studies of early Aβ oligomerization using photo-induced cross-linking of unmodified proteins (PICUP) implicate isoleucine41 and alanine42 as critical amino acids in paranuclei formation. Furthermore, Teplow showed that the oxidation state of methionine35 affects oligomerization and toxicity. In new studies, Teplow reported the use of limited proteolysis, mass spectrometry, and solution-state NMR to identify what may be the nucleation site for initial folding of the Aβ monomer. (see also ARF Live Discussion.)
If amyloid oligomers share a common structure then perhaps this implies a common mechanism of pathogenesis. Charles Glabe of the University of California, Irvine, proposed that oligomers are intrinsically toxic, since those from non-disease proteins are equally cytotoxic. An anti-oligomer antibody generated in Glabe’s lab recognizes amyloid oligomers regardless of sequence (Kayed et al., 2003), and neutralizes oligomer-induced toxicity in cells. In an attempt to elucidate a possible common mechanism of pathogenesis, Glabe focused on the plasma membrane as the target of both intracellular and extra-cellular amyloids. Data from his lab implicate oligomers as the primary toxic species, increasing membrane conductance and permeabilizing cell membranes. This loss of membrane integrity causes dysregulation of ion homeostasis, leading to cell death. However, Glabe hypothesizes that this increase in membrane conductance is not due to pore or channel formation, but rather an oligomer-induced alteration in lipid packing in the membrane.
Offering a different perspective, Martin Duennwald of the Whitehead Institute for Biomedical Research, Cambridge, Massachusetts, pointed to endoplasmic reticulum (ER) stress as a major cause for huntingtin (Htt) toxicity. Using a yeast model, Duennwald showed aggregation and toxicity that depended on polyQ length. In these studies, a repeat length of 38 was required for toxicity. When the cellular protein quality control systems become impaired, htt is more likely to accumulate. Specifically, Duennwald’s lab found the 26S proteasome and ubiquitin co-localized with misfolded htt inclusions. Duennwald also showed that misfolded htt hobbles the function of the ER-associated degradation (ERAD) system, which leads to cytosolic protein aggregates and impaired protein turnover. Thus, Duennwald proposed a mechanism by which Htt aggregates interfere with proteasome function, lead to accumulation of ERAD substrates in the ER, induction of the unfolded protein response (UPR), caspase-12 activation, and finally, apoptosis.
Cytosolic protein aggregates were also the topic of interest for Stanford University’s Ron Kopito, who introduced autophagy as a cellular defense against protein aggregates. Proteasomal degradation requires that the substrate unfold, and is thus not effective at degrading oligomerized proteins. When mechanisms to suppress aggregation are impaired, aggregates are swept along microtubules to a location near the nucleus, where they form an aggresome or inclusion body. Kopito’s data indicated that autophagins, or other autophagy-associated proteins, localize with htt inclusions, and that this recruitment of machinery is a general response to protein aggregation. The researchers turned on htt expression in transfected neuroblastoma cells, then turned it off after three days. They initially saw a rise in the number of cells with aggregates, followed by a 50 percent reduction in the number of aggregates five days later. Kopito attributed this clearance of aggregates to autophagy; when they knocked down autophagins, they eliminated autophagy and subsequent protein clearance (see ARF related news story). Kopito also presented an interesting, preliminary result suggesting that polyQ aggregates can be secreted into the extracellular space, where neighboring cells endocytose it. Upon internalization, the aggregates apparently nucleate aggregation of endogenous polyQ in a prion-like propagation.
Part 4: Going Against the Grain: Can Amyloids Be Good?
William Balch of the Scripps Research Institute in San Diego, California, put a positive spin on some amyloids by presenting them as a functional component of the secretory pathway. He studied melanosomes, cellular compartments that generate a natural fibril-like polymer used for deposition of the pigment melanin. Using differentiated retinal pigmented epithelial cells to study fibril formation, Balch inhibited Pmel17, the core component of melanosomes, and subsequently prevented the formation of thioflavin-S structures. In vitro, recombinant Pmel17 formed amyloid within seconds, the rapid polymerization consistent with a normal biologically optimized function. In vivo, melanosomes stained with amyloidophilic dyes. Balch proposed that Pmel17 forms a functional native amyloid structure that is optimized for melanosome biogenesis. Therefore, his alternative view of amyloids asserts that the amyloid fold may not be a remnant of the past, but rather is an evolutionarily conserved and optimized structural element of normal folding pathways.
This conference encouraged debate and discussion, which allowed participants to think beyond their immediate area of interest and search for a common threads among different amyloid diseases. This is not a comprehensive report on all of the talks and posters, rather, it is meant as an opportunity for other conference participants to fill in the gaps. All readers are invited to write in to discuss topics stimulated by this exciting meeting.—Erene Mina.
Erene Mina is a graduate student at the University of California, Irvine