Variants in Apaf-1 segregating with major depression promote apoptosome function.

APAF1, encoding the protein apoptosis protease activating factor 1 (Apaf-1), has recently been established as a chromosome 12 gene conferring predisposition to major depression in humans. The molecular phenotypes of Apaf-1 variants were determined by in vitro reconstruction of the apoptosome complex in which Apaf-1 activates caspase 9 and thus initiates a cascade of proteolytic events leading to apoptotic destruction of the cell. Cellular phenotypes were measured using a yeast heterologous expression assay in which human Apaf-1 and other proteins necessary to constitute a functional apoptotic pathway were overexpressed. Apaf-1 variants encoded by APAF1 alleles that segregate with major depression in families linked to chromosome 12 shared a common gain-of-function phenotype in both assay systems. In contrast, other Apaf-1 variants showed neutral or loss-of-function phenotypes. The depression-associated alleles thus have a common phenotype that is distinct from that of non-associated variants. This result suggests an etiologic role for enhanced apoptosis in major depression.

Driving affinity selection by centrifugal force.

We describe a new approach to affinity selection based on the application of centrifugal force to macromolecules in solution. The method relies on the well known macromolecular hydrodynamic principles of centrifugation. It can be automated and operated in a centralized fashion, or it can be decentralized and used by single researchers or networks of researchers with a minimal additional capital investment. In this method, a centrifugal driving force is used to establish a differential and selective concentration gradient between a therapeutic target and potential ligands in compound libraries. This concentration gradient, in turn, drives the binding of ligands. Once formed, the differential concentration gradient of target macromolecules and ligands is fractionated to capture the self-sorting binding events. Ligand binding is defined by the individual ligand binding constants, so tight binding ligands will essentially distribute identically with the protein target, and weaker binding ligands will not. The level of affinity needed to operationally define tight binding can be adjusted by selecting the initial concentration conditions or centrifugal force. A variety of rapid, commonly available, detection methods can be used to assess binding in the fractionated samples. The method can be broadly applied in drug discovery efforts to examine most types of cell-cell, protein-protein, and protein-small molecule interactions. We describe the application of this method to systems of small molecule interactions with several macromolecules of therapeutic interest.

Abrogation of G2 checkpoint specifically sensitize p53 defective cells to cancer chemotherapeutic agents.

BACKGROUND: Chkl is a checkpoint gene that is activated after DNA damage. It phosphorylates and inactivates Cdc25C at the late G2 phase. The inactivation of Cdc25C and consequently, the inactivation of Cdc2, are required for the G2 arrest induced by DNA damage. METHODS: We treated 184B5 cell line and its E6 transformed cell lines with adriamycin in the presence of staurosporine or UCNO1 and examined G2 arrest and cell death. RESULTS: We found that adriamycin induced a p53 and p21 response as well as a G1 arrest in 184B5 cells, but not in its E6 transformed cells. Staurosporine or UCNO1 abrogated the G2 arrest induced by adriamycin in both cell lines. In addition, staurosporine or UCNO1 specifically sensitized p53 incompetent cells to adriamycin. CONCLUSION: G2/M checkpoint abrogators can potentially enhance the cytotoxic effect of conventional chemotherapeutic reagents specifically to tumor cells.

Design of adenosine kinase inhibitors from the NMR-based screening of fragments.

A strategy is described for designing high-affinity ligands using information derived from the NMR-based screening of fragments. The method involves the fragmentation of an existing lead molecule, identification of suitable replacements for the fragments, and incorporation of the newly identified fragments into the original scaffold. Using this technique, novel substituents were rapidly identified and incorporated into lead inhibitors of adenosine kinase that exhibited potent in vitro and in vivo activities. This approach is a valuable strategy for modifying existing leads to improve their potency, bioavailability, or toxicity profile and thus represents a useful technique for lead optimization.

Domain structure analysis of elongation factor-3 from Saccharomyces cerevisiae by limited proteolysis and differential scanning calorimetry.

Elongation-factor-3 (EF-3) is an essential factor of the fungal protein synthesis machinery. In this communication the structure of EF-3 from Saccharomyces cerevisiae is characterized by differential scanning calorimetry (DSC), ultracentrifugation, and limited tryptic digestion. DSC shows a major transition at a relatively low temperature of 39 degrees C, and a minor transition at 58 degrees C. Ultracentrifugation shows that EF-3 is a monomer; thus, these transitions could not reflect the unfolding or dissociation of a multimeric structure. EF-3 forms small aggregates, however, when incubated at room temperature for an extended period of time. Limited proteolysis of EF-3 with trypsin produced the first cleavage at the N-side of Gln775, generating a 90-kDa N-terminal fragment and a 33-kDa C-terminal fragment. The N-terminal fragment slowly undergoes further digestion generating two major bands, one at approximately 75 kDa and the other at approximately 55 kDa. The latter was unusually resistant to further tryptic digestion. The 33-kDa C-terminal fragment was highly sensitive to tryptic digestion. A 30-min tryptic digest showed that the N-terminal 60% of EF-3 was relatively inaccessible to trypsin, whereas the C-terminal 40% was readily digested. These results suggest a tight structure of the N-terminus, which may give rise to the 58 degrees C transition, and a loose structure of the C-terminus, giving rise to the 39 degrees C transition. Three potentially functional domains of the protein were relatively resistant to proteolysis: the supposed S5-homologous domain (Lys102-Ile368), the N-terminal ATP-binding cassette (Gly463-Lys622), and the aminoacyl-tRNA-synthase homologous domain (Glu820-Gly865). Both the basal and ribosome-stimulated ATPase activities were inactivated by trypsin, but the ribosome-stimulated activity was inactivated faster.

Crystal structure of ErmC', an rRNA methyltransferase which mediates antibiotic resistance in bacteria.

The prevalent mechanism of bacterial resistance to erythromycin and other antibiotics of the macrolide-lincosamide-streptogramin B group (MLS) is methylation of the 23S rRNA component of the 50S subunit in bacterial ribosomes. This sequence-specific methylation is catalyzed by the Erm group of methyltransferases (MTases). They are found in several strains of pathogenic bacteria, and ErmC is the most studied member of this class. The crystal structure of ErmC' (a naturally occurring variant of ErmC) from Bacillus subtilis has been determined at 3.0 A resolution by multiple anomalous diffraction phasing methods. The structure consists of a conserved alpha/beta amino-terminal domain which binds the cofactor S-adenosyl-l-methionine (SAM), followed by a smaller, alpha-helical RNA-recognition domain. The beta-sheet structure of the SAM-binding domain is well-conserved between the DNA, RNA, and small-molecule MTases. However, the C-terminal nucleic acid binding domain differs from the DNA-binding domains of other MTases and is unlike any previously reported RNA-recognition fold. A large, positively charged, concave surface is found at the interface of the N- and C-terminal domains and is proposed to form part of the protein-RNA interaction surface. ErmC' exhibits the conserved structural motifs previously found in the SAM-binding domain of other methyltransferases. A model of SAM bound to ErmC' is presented which is consistent with the motif conservation among MTases.

Structure of the E2 DNA-binding domain from human papillomavirus serotype 31 at 2.4 A.

The papillomaviruses are a family of small double-stranded DNA viruses which exclusively infect epithelial cells and stimulate the proliferation of those cells. A key protein within the papillomavirus life-cycle is known as the E2 (Early 2) protein and is responsible for regulating viral transcription from all viral promoters as well as for replication of the papillomavirus genome in tandem with another protein known as E1. The E2 protein itself consists of three functional domains: an N-terminal trans-activation domain, a proline-rich linker, and a C-terminal DNA-binding domain. The first crystal structure of the human papillomavirus, serotype 31 (HPV-31), E2 DNA-binding domain has been determined at 2.4 A resolution. The HPV DNA-binding domain monomer consists of two beta-alpha-beta repeats of approximately equal length and is arranged as to have an anti-parallel beta-sheet flanked by the two alpha-helices. The monomers form the functional in vivo dimer by association of the beta-sheets of each monomer so as to form an eight-stranded anti-parallel beta-barrel at the center of the dimer, with the alpha-helices lining the outside of the barrel. The overall structure of HVP-31 E2 DNA-binding domain is similar to both the bovine papillomavirus E2-binding domain and the Epstein-Barr nuclear antigen-1 DNA-binding domain.

NMR-based discovery of lead inhibitors that block DNA binding of the human papillomavirus E2 protein.

The E2 protein is required for the replication of human papillomaviruses (HPVs), which are responsible for anogenital warts and cervical carcinomas. Using an NMR-based screen, we tested compounds for binding to the DNA-binding domain of the HPV-E2 protein. Three classes of compounds were identified which bound to two distinct sites on the protein. Biphenyl and biphenyl ether compounds containing a carboxylic acid bind to a site near the DNA recognition helix and inhibit the binding of E2 to DNA. Benzophenone-containing compounds which lack a carboxylic acid group bind to the beta-barrel formed by the dimer interface and exhibit negligible effects on the binding of E2 to DNA. Structure-activity relationships from the biphenyl and biphenyl ether compounds were combined to produce a compound [5-(3'-(3",5"-dichlorophenoxy)-phenyl)-2,4-pentadienoic acid] with an IC50 value of approximately 10 microM. This compound represents a useful lead for the development of antiviral agents that interfere with HPV replication and further illustrates the usefulness of the SAR by NMR method in the drug discovery process.

Solution structure of an rRNA methyltransferase (ErmAM) that confers macrolide-lincosamide-streptogramin antibiotic resistance.

The Erm family of methyltransferases is responsible for the development of resistance to the macrolide-lincosamide-streptogramin type B (MLS) antibiotics. These enzymes methylate an adenine of 23S ribosomal RNA that prevents the MLS antibiotics from binding to the ribosome and exhibiting their antibacterial activity. Here we describe the three-dimensional structure of an Erm family member, ErmAM, as determined by NMR spectroscopy. The catalytic domain of ErmAM is structurally similar to that found in other methyltransferases and consists of a seven-stranded beta-sheet flanked by alpha-helices and a small two-stranded beta-sheet. In contrast to the catalytic domain, the substrate binding domain is different from other methyltransferases and adopts a novel fold that consists of four alpha-helices.

Initial characterization of autoprocessing and active-center mutants of CMV proteinase.

Human cytomegalovirus (CMV) encodes a unique serine proteinase that is required in the maturation of the viral capsid. The CMV proteinase can undergo autocatalytic activation and is subject to proteolytic self-inactivation. Mutant enzyme forms were prepared to eliminate the initial autoprocessing site and thus form an active single-chain protein for structure-function studies. Two mutants of CMV proteinase were cloned and expressed in Escherichia coli. The A143V mutant was a conservative substitution at the first internal cleavage site. The S132A mutant modified one of the triad of residues responsible for catalytic activity. Through the use of computer-controlled high-cell-density fermentations the mutant proteins were expressed in E. coli at approximately 170 mg/L as both soluble (approximately 40% of total) and inclusion-body forms (approximately 60% of total). The soluble enzyme was purified by standard methods; inclusion-body protein was isolated by standard methods after refolding and solubilization in guanidine or urea. Sedimentation equilibrium and sedimentation velocity analyses reveal that the enzyme undergoes concentration-dependent aggregation. It exhibits a monomer <==> dimer equilibrium (Kd = 1 microM) at low concentrations and remains dimeric at high concentrations (28 mg/ml). Differential scanning calorimetry data for protein thermal unfolding fit best to a non-two-state model with two components (Tm = 52.3 and 55.3 degrees C) which subsequently aggregate upon unfolding. Analysis of the short-UV circular dichroism spectra of protein forms resulting from expression as soluble molecules (not refolded) reveals that the two mutants have very similar secondary structures which comprise a mixed structural motif of 20% alpha-helix, 26% beta-sheet, and 53% random coil. Though soluble and active (A143V mutant only), CD analysis revealed that protein refolded from inclusion bodies did not exhibit spectra identical to that of protein expressed only in soluble form.

Autocatalyzed protein folding.

Proline isomerization, an intrinsically slow process, kinetically traps intermediates in slow protein folding reactions. Thus, enzymes that catalyze proline isomerization (prolyl isomerases) often catalyze protein folding. We have investigated the folding kinetics of FKBP, a prolyl isomerase. The main conclusion is that FKBP catalyzes its own folding. Altogether, the FKBP refolding kinetics are resolved into three exponential phases: a fast phase, tau 3; an intermediate phase, tau 2; and a slow phase, tau 1. Unfolding occurs in a single phase, the unfolding branch of phase tau 2. In the presence of native FKBP, both the intermediate (tau 2) and slow (tau 1) phases are faster, suggesting that folding phases tau 1 and tau 2 involve proline cis-trans isomerization. In the absence of added native FKBP, autocatalytic folding of FKBP is detected. For refolding starting with all the FKBP unfolded initially, the slowest folding phase (tau 1) is almost 2-fold faster at a final concentration of 14 microM FKBP than at 2 microM FKBP, suggesting that catalytically active FKBP formed in the fast (tau 3) or intermediate (tau 2) folding phases catalyzes the slow folding phase (tau 1). Moreover, autocatalysis of folding is inhibited by FK506, an inhibitor of the FKBP prolyl isomerase activity. The results show that the slow phase in FKBP folding is an autocatalyzed formation of native FKBP from kinetically trapped species with non-native proline isomers. While the magnitude of the catalytic effects reported here are modest, FKBP folding may provide a prototype for autocatalysis of kinetically trapped macromolecular conformational changes in other systems.

The nanometer-scale structure of amyloid-beta visualized by atomic force microscopy.

Amyloid-beta (A beta) is the major protein component of neuritic plaques found in Alzheimer's disease. Evidence suggests that the physical aggregation state of A beta directly influences neurotoxicity and specific cellular biochemical events. Atomic force microscopy (AFM) is used to investigate the three-dimensional structure of aggregated A beta and characterize aggregate/fibril size, structure, and distribution. Aggregates are characterized by fibril length and packing densities. The packing densities correspond to the differential thickness of fiber aggregates along a zeta axis (fiber height above the x-y imaging surface). Densely packed aggregates ( > or = 100 nm thick) were observed. At the edges of these densely packed regions and in dispersed regions, three types of A beta fibrils were observed. These were classified by fibril thickness into three size ranges: 2-3 nm thick, 4-6 nm thick, and 8-12 nm thick. Some of the two thicker classes of fibrils exhibited pronounced axial periodicity. Substructural features observed included fibril branching or annealing and a height periodicity which varied with fibril thickness. When identical samples were visualized with AFM and electron microscopy (EM) the thicker fibrils (4-6 nm and 8-12 nm thick) had similar morphology. In comparison, the densely packed regions of approximately > or = 100 nm thickness observed by AFM were difficult to resolve by EM. The small, 2- to 3-nm-thick, fibrils were not observed by EM even though they were routinely imaged by AFM. These studies demonstrate that AFM imaging of A beta fibrils can, for the first time, resolve nanometer-scale, zeta-axis, surface-height (thickness) fibril features. Concurrent x-y surface scans of fibrils reveal the surface submicrometer structure and organization of aggregated A beta. Thus, when AFM imaging of A beta is combined with, and correlated to, careful studies of cellular A beta toxicity it may be possible to relate certain A beta structural features to cellular neurotoxicity.

Solution structure of the DNA-binding domain of a human papillomavirus E2 protein: evidence for flexible DNA-binding regions.

The three-dimensional structure of the DNA-binding domain of the E2 protein from human papillomavirus-31 was determined by using multidimensional heteronuclear nuclear magnetic resonance (NMR) spectroscopy. A total of 1429 NMR-derived distance and dihedral angle restraints were obtained for each of the 83-residue subunits of this symmetric dimer. The average root mean square deviations of 20 structures calculated using a distance geometry-simulated annealing protocol are 0.59 and 0.90 angstroms for the backbone and all heavy atoms, respectively, for residues 2-83. The structure of the human virus protein free in solution consists of an eight-stranded beta-barrel and two pairs of alpha-helices. Although the overall fold of the protein is similar to the crystal structure of the bovine papillomavirus-1 E2 protein when complexed to DNA, several small but interesting differences were observed between these two structures at the subunit interface. In addition, a beta-hairpin that contacts DNA in the crystal structure of the protein-DNA complex is disordered in the NMR structures, and steady-state 1H-15N heteronuclear NOE measurements indicate that this region is highly mobile in the absence of DNA. The recognition helix also appears to be flexible, as evidenced by fast amide exchange rates. This phenomenon has also been observed for a number of other DNA-binding proteins and may constitute a common theme in protein/DNA recognition.

Clusterin (apoJ) alters the aggregation of amyloid beta-peptide (A beta 1-42) and forms slowly sedimenting A beta complexes that cause oxidative stress.

Clusterin (apoJ), a multifunctional apolipoprotein made by cells in the brain and many other locations, is associated with aggregated amyloid beta-peptide (A beta) in senile and diffuse plaques of Alzheimer's disease (AD). We observed that purified human serum clusterin partially blocked the aggregation of synthetic A beta 1-42, as shown by centrifugal assays (14,000g x 10 min) and by atomic force (scanning probe) microscopy. Slowly sedimenting A beta complexes were formed in the presence of clusterin, which included aggregates > 200 kDa that resist dissociation by low concentrations of SDS. Clusterin enhanced the oxidative stress caused by A beta, as assayed by oxidative stress in PC12 cells with MTT, which is widely used to estimate neurotoxicity. These indications of enhanced neurotoxicity by the MTT assay were observed in the highly aggregated rapidly sedimenting fraction, but also in more slowly sedimenting "soluble" forms. This novel activity of slowly sedimenting A beta may enhance the neurotoxicity of A beta deposits in AD brains, because soluble complexes have a potential for diffusing to damage distal neurons.

Substrate requirements for ErmC' methyltransferase activity.

ErmC' is a methyltransferase that confers resistance to the macrolide-lincosamide-streptogramin B group of antibiotics by catalyzing the methylation of 23S rRNA at a specific adenine residue (A-2085 in Bacillus subtilis; A-2058 in Escherichia coli). The gene for ErmC' was cloned and expressed to a high level in E. coli, and the protein was purified to virtual homogeneity. Studies of substrate requirements of ErmC' have shown that a 262-nucleotide RNA fragment within domain V of B. subtilis 23S rRNA can be utilized efficiently as a substrate for methylation at A-2085. Kinetic studies of the monomethylation reaction showed that the apparent Km of this 262-nucleotide RNA oligonucleotide was 26-fold greater than the value determined for full-size and domain V 23S rRNA. In addition, the Vmax for this fragment also rose sevenfold. A model of RNA-ErmC' interaction involving multiple binding sites is proposed from the kinetic data presented.

Cathepsin D from Alzheimer's-diseased and normal brains.

An acid protease activity from human brain was found to cleave a fluorogenic peptide substrate encompassing the amino terminus of Alzheimer's amyloid-beta peptide (A beta). The protease was isolated and determined to be cathepsin D based on chromatographic, immunological, and enzymatic data. Analysis of the cleavage sites indicated that cathepsin D hydrolyzed the methionine--aspartate bond generating the in vivo amino terminus of A beta. These data suggested that cathepsin D could be involved in amyloidogenic processing of the amyloid precursor protein. Consequently, cathepsin D from both Alzheimer's-diseased and control brains was compared to determine whether there were any differences which could account for an increase in A beta production in Alzheimer's disease. No differences were detected in isoform composition or tissue content of cathepsin D as measured by 2-D IEF-SDS-PAGE. Enzymological characterization of brain cathepsin D demonstrated that it could undergo a previously undescribed pH-dependent reversible activation. However, that activation appeared identical for both AD and normal brain enzymes. These data demonstrate that concentration, isoform distribution, and several enzymological characteristics of cathepsin D are not distinguishable between AD and normal brain. The pH dependence of cathepsin D activity suggests, however, that its intracellular localization may be important in considering the potential role of cathepsin D in Alzheimer's disease.

Evidence against a role for the Kunitz domain in amyloidogenic and secretory processing of the amyloid precursor protein.

The effect of the Kunitz proteinase inhibitor (KPI) on potential beta-amyloid precursor protein (beta PP)-processing activities from control and Alzheimer's disease (AD) brains was examined using fluorogenic substrates designed to mimic the secretory and amyloidogenic cleavages in beta PP. In addition, the level of secretion of KPI-containing beta PP751 and KPI-lacking beta PP695 from transfected cells was examined to assess the effect of the KPI on beta PP secretion. beta PP751 and beta PP695, obtained from conditioned media of transfected cells, had no effect on proteinase activities against the secretory and amyloidogenic substrates in extracts from control and AD brains. At similar concentrations beta PP751, but not beta PP695, completely inhibited the activity of trypsin against these substrates. Serine proteinase inhibitors had only modest effects on activities from brain, whereas cysteine modification completely inhibited them, indicating that these proteinase activities were not of the serine type. Thus, the results do not support a role for the KPI in the secretion of beta PP or in the amyloidogenic cleavage of beta PP. The amounts of beta PP695 and beta PP751 collected from the media of transfected cells after 48 h of growth were similar, indicating an equal rate of secretion. This result suggests that the KPI domain in beta PP751 did not inhibit the secretory cleavage in transfected cells.

Beta/A4-evoked degeneration of differentiated SH-SY5Y human neuroblastoma cells.

beta/A4 peptides are known to induce neurodegeneration in cultures of rat brain cells and rat neural cell lines (Yankner et al: Science 250:279-282, 1990; Behl et al: Biochem Biophys Res Commun 186:944-950, 1992). The current data show that these peptides induce similar neurodegeneration in SH-SY5Y neuroblastoma cells, extending characterization of beta/A4 toxicity to a human nerve cell line. Human SH-SY5Y cells respond to aggregated beta/A4 with changes in cell shape, membrane blebbing, antigenic modification, loss of attachment to the substrate, and cell death. beta/A4 peptides require aggregation for maximum toxic effects, as cellular degeneration is evoked by aggregated beta/A4 1-42 and 4-41 cysteine but not by monomeric beta/A4 1-40. Aged (pre-aggregated) beta/A4 1-40 also evoked neurodegeneration. Antigenic changes comprise upregulation of Alzheimer's-type tau epitopes, recognized by the PHF-1 and Alz-50 monoclonals. These particular changes in tau support the connectivity between this in vitro model and mechanisms leading to neurodegeneration in Alzheimer's disease. A significant feature of the SH-SY5Y response is that cells must be differentiated before they become sensitive to the degeneration evoked by beta/A4. Signaling pathways leading to beta/A4-evoked neurodegeneration thus are under experimental control, becoming complete only when proliferating cells withdraw from the cell cycle and develop a postmitotic phenotype.

Cleavage of fluorogenic substrates for APP-processing proteases by human brain extracts. Ca(2+)-substrate interaction is responsible for Ca2+ stimulation of the neural protease activity.

The proteases that cleave amyloid precursor protein (APP) leading to generation of amyloid A beta peptide are potential targets for therapeutical intervention of Alzheimer disease. We have been pursuing the identification and characterization of these proteases using as probes the fluorogenic substrates encompassing the cleavage sites of APP that we described recently (Wang, G. T., Krafft, G. A. [1992] Bioorg. Med. Chem. Lett. 2, 1665). This article describes results of experiments designed to examine the effect of Ca(2+) on the cleavage of these substrates by human brain extracts. Fluorogenic substrates encompassing either the N-terminal amyloidogenic cleavage site or the secretory cleavage site were synthesized in five formats with various peripheral residues. Incubation with extracts from normal brain tissue revealed that more negatively charged amyloidogenic substrates were less reactive and exhibited larger rate enhancement in the presence of Ca(2+). The results imply that Ca(2+) stimulation of substrate cleavage by brain proteases occurs primarily as a result of Ca(2+)-substrate interactions, and caution against interpretations that invoke the involvement of Ca(2+)-stimulated proteases in A beta formation.

Amyloid-beta aggregation: selective inhibition of aggregation in mixtures of amyloid with different chain lengths.

One of the clinical manifestations of Alzheimer's disease is the deposition of the 39-43 residue amyloid-beta (A beta) peptide in aggregated fibrils in senile plaques. Characterization of the aggregation behavior of A beta is one of the critical issues in understanding the role of A beta in the disease process. Using solution hydrodynamics, A beta was observed to form three types of species in phosphate-buffered saline: insoluble aggregates with sedimentation coefficients of approximately 50,000 S and molecular masses of approximately 10(9) Da, "soluble aggregates" with sedimentation coefficients of approximately 30 S and masses of approximately 10(6) Da, and monomer. When starting from monomer, the aggregation kinetics of A beta 1-40 (A beta 40) and A beta 1-42 (A beta 42), alone and in combination, reveal large differences in the tendency of these peptides to aggregate as a function of pH and other solution conditions. At pH 4.1 and 7.0-7.4, aggregation is significantly slower than at pH 5 and 6. Under all conditions, aggregation of the longer A beta 42 was more rapid than A beta 40. Oxidation of Met-35 to the sulfoxide in A beta 40 enhances the aggregation rate over that of the nonoxidized peptide. Aggregation was found to be dependent upon temperature and to be strongly dependent on peptide concentration and ionic strength, indicating that aggregation is driven by a hydrophobic effect. When A beta 40 and A beta 42 are mixed together, A beta 40 retards the aggregation of A beta 42 in a concentration-dependent manner. Shorter fragments have a decreasing ability to interfere with A beta 42 aggregation. Conversely, the rate of aggregation of A beta 40 can be significantly enhanced by seeding slow aggregating solutions with preformed aggregates of A beta 42. Taken together, the inhibition of A beta 42 aggregation by A beta 40, the seeding of A beta 40 aggregation by A beta 42 aggregates, and the chemical oxidation of A beta 40 suggest that the relative abundance and rates of production of different-length A beta and its exposure to radical damage may be factors in the accumulation of A beta in plaques in vivo.