Double-mutant Cycle Research Articles

London Dispersion Favors Sterically Hindered Diarylthiourea Conformers in Solution.

We present an experimental and computational study on the conformers of N,N′‐diphenylthiourea substituted with different dispersion energy donor (DED) groups. While the unfolded anti–anti conformer is the most relevant for thiourea catalysis, intramolecular noncovalent interactions counterintuitively favor the folded syn–syn conformer, as evident from a combination of low‐temperature nuclear magnetic resonance measurements and computations. In order to quantify the noncovalent interactions, we utilized local energy decomposition analysis and symmetry‐adapted perturbation theory at the DLPNO‐CCSD(T)/def2‐TZVPP and sSAPT0/6‐311G(d,p) levels of theory. Additionally, we applied a double‐mutant cycle to experimentally study the effects of bulky substituents on the equilibria. We determined London dispersion as the key interaction that shifts the equilibria towards the syn–syn conformers. This preference is likely a factor why such thiourea derivatives can be poor catalysts.

Limits to detecting epistasis in the fitness landscape of HIV.

The rapid evolution of HIV is constrained by interactions between mutations which affect viral fitness. In this work, we explore the role of epistasis in determining the mutational fitness landscape of HIV for multiple drug target proteins, including Protease, Reverse Transcriptase, and Integrase. Epistatic interactions between residues modulate the mutation patterns involved in drug resistance, with unambiguous signatures of epistasis best seen in the comparison of the Potts model predicted and experimental HIV sequence "prevalences" expressed as higher-order marginals (beyond triplets) of the sequence probability distribution. In contrast, experimental measures of fitness such as viral replicative capacities generally probe fitness effects of point mutations in a single background, providing weak evidence for epistasis in viral systems. The detectable effects of epistasis are obscured by higher evolutionary conservation at sites. While double mutant cycles in principle, provide one of the best ways to probe epistatic interactions experimentally without reference to a particular background, we show that the analysis is complicated by the small dynamic range of measurements. Overall, we show that global pairwise interaction Potts models are necessary for predicting the mutational landscape of viral proteins.

Exosite binding modulates the specificity of the immunomodulatory enzyme ScpA, a C5a inactivating bacterial protease

The C5a peptidase from Streptococcus pyogenes (ScpA) is a highly specific enzyme with potential therapeutic value. ScpA is a good model for studying determinants of specificity in the multidomain immunomodulatory enzymes (IMEs), which comprise a large family of bacterial surface proteases. The surface exposed region of ScpA has 5 main domains which includes 3 C-terminal Fn3-like domains (Fn1, Fn2 and Fn3) (Kagawa et al. 2009). Progressive deletion of the Fn3-like domains from the C-ter resulted in loss of enzyme activity and showed an important role for the Fn2 domain in enzyme function. Functional investigation of specific acidic residues on the Fn2 domain identified 3 residues 30–50 Å from the catalytic site (D783, E864 and D889) which impacted to differing degrees on binding and on catalysis, supporting the presence of an exosite on the Fn2. In particular, residue D783 was observed to impact on both substrate binding affinity and the activity of ScpA. A double mutant cycle analysis showed energetic coupling between the targeted ScpA residues and residues in the core portion (residues 1–67) of the C5a substrate. The data supports the presence of a communication network between the active site and the exosite on Fn2. These findings provide a basis for rational engineering of this important enzyme family to enhance stability, activity and/or specificity.

Molecular pathology of the R117H cystic fibrosis mutation is explained by loss of a hydrogen bond.

The phosphorylation-activated anion channel cystic fibrosis transmembrane conductance regulator (CFTR) is gated by an ATP hydrolysis cycle at its two cytosolic nucleotide-binding domains, and is essential for epithelial salt-water transport. A large number of CFTR mutations cause cystic fibrosis. Since recent breakthrough in targeted pharmacotherapy, CFTR mutants with impaired gating are candidates for stimulation by potentiator drugs. Thus, understanding the molecular pathology of individual mutations has become important. The relatively common R117H mutation affects an extracellular loop, but nevertheless causes a strong gating defect. Here, we identify a hydrogen bond between the side chain of arginine 117 and the backbone carbonyl group of glutamate 1124 in the cryo-electronmicroscopic structure of phosphorylated, ATP-bound CFTR. We address the functional relevance of that interaction for CFTR gating using macroscopic and microscopic inside-out patch-clamp recordings. Employing thermodynamic double-mutant cycles, we systematically track gating-state-dependent changes in the strength of the R117-E1124 interaction. We find that the H-bond is formed only in the open state, but neither in the short-lived 'flickery' nor in the long-lived 'interburst' closed state. Loss of this H-bond explains the strong gating phenotype of the R117H mutant, including robustly shortened burst durations and strongly reduced intraburst open probability. The findings may help targeted potentiator design.

Application of the Double-Mutant Cycle Strategy to Protein Aggregation Reveals Transient Interactions in Amyloid-β Oligomers.

Transient oligomeric intermediates in the peptide or protein aggregation pathway are suspected to be the key toxic species in many amyloid diseases, but deciphering their molecular nature has remained a challenge. Here we show that the strategy of "double-mutant cycles", used effectively in probing protein-folding intermediates, can reveal transient interactions during protein aggregation. It does so by comparing the changes in thermodynamic parameters between the wild type, and single and double mutants. We demonstrate the strategy by probing the possible transient salt bridge partner of lysine 28 (K28) in the oligomeric states of amyloid β-40 (Aβ40), the putative toxic species in Alzheimer's disease. In mature fibrils, the binding partner is aspartate 23. This interaction differentiates Aβ40 from the more toxic Aβ42, where K28's binding partner is the C-terminal carboxylate. We selectively acetylated K28 and amidated the C-terminus of Aβ40, creating four distinct variants. Spectroscopic measurements of the kinetics and thermodynamics of aggregation show that K28 and the C-terminus interact transiently in the early phases of the Aβ40 aggregation pathway. Hydrogen-deuterium exchange mass spectrometry (using a simple analysis method that we introduce here that takes into account the isotopic mass distribution) supports this interpretation. It is also supported by cellular toxicity measurements, suggesting possible similarities in the mechanisms of toxicity of Aβ40 oligomers (which are more toxic than Aβ40 fibrils) and Aβ42. Our results show that double-mutant cycles can be a powerful tool for probing transient interactions during protein aggregation.

Dissection of the Polar and Non‐Polar Contributions to Aromatic Stacking Interactions in Solution

AbstractAromatic stacking interactions have been a matter of study and debate due to their crucial role in chemical and biological systems. The strong dependence on orientation and solvent together with the relatively small interaction energies have made evaluation and rationalization a challenge for experimental and theoretical chemists. We have used a supramolecular cage formed by two tris(pyridylmethyl)amines units to build chemical Double Mutant Cycles (DMC) for the experimental measurement of the free energies of π‐stacking interactions. Extrapolating the substituent effects to remove the contribution due to electrostatic interactions reveals that there is a substantial contribution to the measured stacking interaction energies which is due to non‐polar interactions (−3 to −6 kJ mol−1). The perfectly flat nature of the surface of an aromatic ring gives π‐stacking an inherent advantage over non‐polar interactions with alkyl groups and accounts for the wide‐spread prevalence of stacking interactions in Nature.

Dissection of the Polar and Non-Polar Contributions to Aromatic Stacking Interactions in Solution.

Aromatic stacking interactions have been a matter of study and debate due to their crucial role in chemical and biological systems. The strong dependence on orientation and solvent together with the relatively small interaction energies have made evaluation and rationalization a challenge for experimental and theoretical chemists. We have used a supramolecular cage formed by two tris(pyridylmethyl)amines units to build chemical Double Mutant Cycles (DMC) for the experimental measurement of the free energies of π‐stacking interactions. Extrapolating the substituent effects to remove the contribution due to electrostatic interactions reveals that there is a substantial contribution to the measured stacking interaction energies which is due to non‐polar interactions (−3 to −6 kJ mol−1). The perfectly flat nature of the surface of an aromatic ring gives π‐stacking an inherent advantage over non‐polar interactions with alkyl groups and accounts for the wide‐spread prevalence of stacking interactions in Nature.

Electro-steric opening of the clc-2 chloride channel gate

The widely expressed two-pore homodimeric inward rectifier CLC-2 chloride channel regulates transepithelial chloride transport, extracellular chloride homeostasis, and neuronal excitability. Each pore is independently gated at hyperpolarized voltages by a conserved pore glutamate. Presumably, exiting chloride ions push glutamate outwardly while external protonation stabilizes it. To understand the mechanism of mouse CLC-2 opening we used homology modelling-guided structure–function analysis. Structural modelling suggests that glutamate E213 interacts with tyrosine Y561 to close a pore. Accordingly, Y561A and E213D mutants are activated at less hyperpolarized voltages, re-opened at depolarized voltages, and fast and common gating components are reduced. The double mutant cycle analysis showed that E213 and Y561 are energetically coupled to alter CLC-2 gating. In agreement, the anomalous mole fraction behaviour of the voltage dependence, measured by the voltage to induce half-open probability, was strongly altered in these mutants. Finally, cytosolic acidification or high extracellular chloride concentration, conditions that have little or no effect on WT CLC-2, induced reopening of Y561 mutants at positive voltages presumably by the inward opening of E213. We concluded that the CLC-2 gate is formed by Y561-E213 and that outward permeant anions open the gate by electrostatic and steric interactions.

Stereoelectronic effects in stabilizing protein-N-glycan interactions revealed by experiment and machine learning.

The energetics of protein-carbohydrate interactions, central to many life processes, cannot yet be predictably manipulated. This is mostly due to an incomplete quantitative understanding of the enthalpic and entropic basis of these interactions in aqueous solution. Here, we show that stereoelectronic effects contribute significantly to stabilizing protein–N-glycan interactions in the context of a cooperatively folding protein. Double-mutant cycle analyses of the folding data from 52 electronically-varied N-glycoproteins demonstrate an enthalpy-entropy compensation depending on the electronics of the interacting side-chains. Linear and non-linear models obtained using quantum mechanical calculations and machine learning explain up to 79 and 97 % of the experimental interaction energy variability as inferred from the R2 value of the respective models. Notably, protein-carbohydrate interaction energies strongly correlate with the molecular orbital energy gaps of the interacting substructures. This suggests that stereoelectronic effects must be given a greater weight than previously thought for accurately modelling the short-range dispersive van der Waals interactions between the N-glycan and the protein.

Swapping the Positions in a Cross-Strand Lateral Ion-Pairing Interaction between Ammonium- and Carboxylate-Containing Residues in a β-Hairpin.

Cross-strand lateral ion-pairing interactions are important for antiparallel β-sheet stability. Statistical studies suggested that swapping the position of cross-strand lateral residues should not significantly affect the interaction. Herein, we swapped the position of ammonium- and carboxylate-containing residues with different side-chain lengths in a cross-strand lateral ion-pairing interaction in a β-hairpin. The peptides were analyzed by 2D-NMR. The fraction folded population and folding free energy were derived from the chemical shift data. The ion-pairing interaction energy was derived using double mutant cycle analysis. The general trends for the fraction folded population and interaction energetics remained similar upon swapping the position of the interacting charged residues. The most stabilizing cross-strand interactions were between short residues, similar to the unswapped study. However, the fraction folded populations for most of the swapped peptides were higher compared to the corresponding unswapped peptides. Furthermore, subtle differences in the ion-pairing interaction energy upon swapping were observed, most likely due to the “unleveled” relative positioning of the interacting residues created by the inherent right-handed twist of the structure. These results should be useful for developing functional peptides that rely on lateral ion-pairing interactions across antiparallel β-strands.

Charge Interactions in a Highly Charge-Depleted Protein.

Electrostatic forces are important for protein folding and are favored targets of protein engineering. However, interactions between charged residues are difficult to study because of the complex network of interactions found in most proteins. We have designed a purposely simple system to investigate this problem by systematically introducing individual and pairs of charged and titratable residues in a protein otherwise free of such residues. We used constant pH molecular dynamics simulations, NMR spectroscopy, and thermodynamic double mutant cycles to probe the structure and energetics of the interaction between the charged residues. We found that the partial burial of surface charges contributes to a shift in pKa value, causing an aspartate to titrate in the neutral pH range. Additionally, the interaction between pairs of residues was found to be highly context dependent, with some pairs having no apparent preferential interaction, while other pairs would engage in coupled titration forming a highly stabilized salt bridge. We find good agreement between experiments and simulations and use the simulations to rationalize our observations and to provide a detailed mechanistic understanding of the electrostatic interactions.

Direct M3-M4 Interactions Control NMDA Receptor Gating and Desensitization

N-Methyl-D-aspartate receptors (NMDARs) are glutamate-gated channels that mediate excitatory neurotransmission in the central nervous system. They are heterotetrameric channels composed of two GluN1 and two GluN2 subunits, each containing four (M1-M4) membrane-embedded helices. Glutamate and glycine binding results in closure of the ligand binding domain and results in opening of the Na+/Ca2+-permeable pore by direct tethering to the gate at the M3 helix. The role of the peripheral M4 helix in modulating channel gating is unclear. To determine if the M4 helix controls gating by direct allosteric interactions with the gate, we performed a cysteine mutagenesis scan and western blot analyses to identify putative regions of direct contact between GluN1 M4 and GluN2A M3. We identified strong crosslinking at GluN2A-F636 with GluN1-G815 and GluN1-L808. We performed alanine mutagenesis on these sites and recorded single-molecule steady-state currents using cell-attached patch clamp. We performed kinetic modelling and used estimated rate constants to execute a double mutant cycle analysis on these recordings to determine the functional impact of these interactions. Each mutant significantly impacted gating open probability (GluN1/GluN2A 0.53 ± 0.10; GluN1G815A/GluN2A 0.32 ± 0.11; GluN1/GluN2AF636A 0.81 ± 0.04; GluN1G815A/GluN2AF636A 0.88 ± 0.04). We found GluN1-G815 made contact with GluN2A-F636 and contributed 1.9 kJ to desensitization. This indicates a functional interaction between the residues that coincides with the physical interaction observed between GluN1-G815 on the M4 helix with GluN2A-F636 on the M3 helix. Our results suggest a model whereby the GluN1 M4 helix makes direct, functionally important contacts with the GluN2A channel gate. This may serve as a mechanism by which intracellular changes in the c-terminal domain can communicate with the channel gate. The GluN1 M4 also houses several disease-associated sites. Mutations that disrupt M3-M4 coupling may be a novel mechanism of disease pathology.

A State-Specific Interaction Stabilizes NMDA Receptor Opening

NMDA receptors mediate glutamate-gated excitatory currents that set the decay time of the excitatory post synaptic currents in the central nervous system. The current time course reflects a multi-step reaction mechanism whose structural underpinnings remain insufficiently understood. Generally, glutamate binding to extracellular domains (LBDs) initiates LBD constriction. The intramolecular movements culminate with the opening of the channel gate at the apex of M3 transmembrane helices. Several 3D arrangements for functional NMDA receptor constructs have been delineated in high resolution, however, their corresponding functional states and the order of their functional interconversion remain poorly delineated. We used all-atom targeted molecular dynamics simulations and previously reported NMDA receptor structures to identify pairs of residues whose state-dependent interactions appear to stabilize the channel gate in an open position. Here we present evidence for a functional interaction between one such pair, GluN1-I642 (on M3 helix) and GluN2A-L550 (on S1-M1 linker). We used single-molecule currents to delineate kinetic mechanisms for wild-type receptors and full length receptors with a series of single residue substitutions at these positions. Mutations at these sites significantly reduced channel open duration (wild-type 4.5 ± 0.7 ms; GluN1I642L 1.3 ± 0.3 ms; GluN2AL550I 1.9 ± 0.4 ms; GluN1I642L/GluN2AL550I 0.7 ± 0.1 ms) and burst open probability (wild-type 0.73 ± 0.05; GluN1I642L 0.08 ± 0.03; GluN2AL550I 0.08 ± 0.03; GluN1I642L/GluN2AL550I 0.003 ± 0.001). Results from double-mutant cycle analyses indicate strong coupling energy between the side chains of these proximally positioned residues and this interaction energy contributes ∼2.4 kJ to the stability of the open state. Overall, these data provide functional evidence for a new state-dependent interaction in NMDA receptors and suggest that absent this interaction, NMDA receptors generate currents with accelerated decay as observed for the human pathogenic variant GluN1-I642L.

Double Mutant Cycles as a Tool to Address Folding, Binding, and Allostery.

Quantitative measurement of intramolecular and intermolecular interactions in protein structure is an elusive task, not easy to address experimentally. The phenomenon denoted ‘energetic coupling’ describes short- and long-range interactions between two residues in a protein system. A powerful method to identify and quantitatively characterize long-range interactions and allosteric networks in proteins or protein–ligand complexes is called double-mutant cycles analysis. In this review we describe the thermodynamic principles and basic equations that underlie the double mutant cycle methodology, its fields of application and latest employments, and caveats and pitfalls that the experimentalists must consider. In particular, we show how double mutant cycles can be a powerful tool to investigate allosteric mechanisms in protein binding reactions as well as elusive states in protein folding pathways.

A conserved residue in the P2X4 receptor has a nonconserved function in ATP recognition

Highly conserved amino acids are generally anticipated to have similar functions across a protein superfamily, including that of the P2X ion channels, which are gated by extracellular ATP. However, whether and how these functions are conserved becomes less clear when neighboring amino acids are not conserved. Here, we investigate one such case, focused on the highly conserved residue from P2X4, E118 (rat P2X4 numbering, rP2X4), a P2X subtype associated with human neuropathic pain. When we compared the crystal structures of P2X4 with those of other P2X subtypes, including P2X3, P2X7, and AmP2X, we observed a slightly altered side-chain orientation of E118. We used protein chimeras, double-mutant cycle analysis, and molecular modeling to reveal that E118 forms specific contacts with amino acids in the “beak” region, which facilitates ATP binding to rP2X4. These contacts are not present in other subtypes because of sequence variance in the beak region, resulting in decoupling of this conserved residue from ATP recognition and/or channel gating of P2X receptors. Our study provides an example of a conserved residue with a specific role in functional proteins enabled by adjacent nonconserved residues. The unique role established by the E118-beak region contact provides a blueprint for the development of subtype-specific inhibitors of P2X4.

Cross-subunit interactions that stabilize open states mediate gating in NMDA receptors

NMDA receptors are excitatory channels with critical functions in the physiology of central synapses. Their activation reaction proceeds as a series of kinetically distinguishable, reversible steps, whose structural bases are currently under investigation. Very likely, the earliest steps include glutamate binding to glycine-bound receptors and subsequent constriction of the ligand-binding domain. Later, three short linkers transduce this movement to open the gate by mechanical pulling on transmembrane helices. Here, we used molecular and kinetic simulations and double-mutant cycle analyses to show that a direct chemical interaction between GluN1-I642 (on M3 helix) and GluN2A-L550 (on L1-M1 linker) stabilizes receptors after they have opened and thus represents one of the structural changes that occur late in the activation reaction. This native interaction extends the current decay, and its absence causes deficits in charge transfer by GluN1-I642L, a pathogenic human variant.

High-Throughput Affinity Measurements of Transcription Factor and DNA Mutations Reveal Affinity and Specificity Determinants.

Transcription factors (TFs) bind regulatory DNA to control gene expression, and mutations to either TFs or DNA can alter binding affinities to rewire regulatory networks and drive phenotypic variation. While studies have profiled energetic effects of DNA mutations extensively, we lack similar information for TF variants. Here, we present STAMMP (simultaneous transcription factor affinity measurements via microfluidic protein arrays), a high-throughput microfluidic platform enabling quantitative characterization of hundreds of TF variants simultaneously. Measured affinities for ∼210 mutants of a model yeast TF (Pho4) interacting with 9 oligonucleotides (>1,800 Kds) reveal that many combinations of mutations to poorly conserved TF residues and nucleotides flanking the core binding site alter but preserve physiological binding, providing a mechanism by which combinations of mutations in cis and trans could modulate TF binding to tune occupancies during evolution. Moreover, biochemical double-mutant cycles across the TF-DNA interface reveal molecular mechanisms driving recognition, linking sequence to function. A record of this paper's Transparent Peer Review process is included in the Supplemental Information.

Insight into the Autosomal-Dominant Inheritance Pattern of SOD1-Associated ALS from Native Mass Spectrometry

About 20% of all familial amyotrophic lateral sclerosis (ALS) cases are associated with mutations in superoxide dismutase (SOD1), a homodimeric protein. The disease has an autosomal-dominant inheritance pattern. It is, therefore, important to determine whether wild-type and mutant SOD1 subunits self-associate randomly or preferentially. A measure for the extent of bias in subunit association is the coupling constant determined in a double-mutant cycle type analysis. Here, cell lysates containing co-expressed wild-type and mutant SOD1 subunits were analyzed by native mass spectrometry to determine these coupling constants. Strikingly, we find a linear positive correlation between the coupling constant and the reported average duration of the disease. Our results indicate that inter-subunit communication and a preference for heterodimerization greatly increase the disease severity.

Forbidden Protein Angles: Triosephosphate Isomerase as a Model for Studying Peripheral Residue Contribution to Catalysis.

Over the past 3.5 billion years of evolution, enzymes have adopted a seemingly infinite number of conformations to suit life on earth. However, torsional angles of proteins have settled into limited zones of energetically favorable dihedrals observed in Ramachandran plots. Areas outside said zones are believed to be forbidden to all amino acids, save glycine, due to steric hindrance. Triosephosphate Isomerase (TIM), a homodimer with a catalytic rate approaching the diffusion limit, contains an active site Lysine 13 (K13) with a dihedral within the forbidden zone (phi=69/psi=−139). Both the amino acid and the dihedral angle are conserved across all species of TIM and known crystal structures regardless of ligand. Only crystal structures of the engineered monomeric version (1MSS) show accepted β‐sheet dihedral values of phi=−135/psi=170 but experiments show a 1,000‐fold loss in activity. Based on these results, we hypothesized that adopting the forbidden torsion angle for K13 contributes to catalysis. Using a combination of computational and experimental approaches, four residues that interact with K13 (N11, M14, E97, and Q64) were mutated to alanine from single to quadruple mutants. In Silico Molecular Dynamic (MD) simulations of 5×200ns per mutant species were made using 2JK2 unliganded Human TIM as a starting structure. Heat Maps, containing 400,000 K13 dihedral values per run reveal full or partial loss of forbidden zone angles depending on mutant combination. Initial experimental results with E97A single mutant showed a three‐fold decrease in kcat with little change in the KM value suggesting a modest catalytic role for this residue. For this mutant the heatmaps revealed the K13 angle spends less than 30% of its time in the forbidden zone compared to over 50% occupation in the wildtype suggesting the rate decrease may be due to the mutation affecting the positioning of K13. In addition to ongoing In Vivo characterization of the remaining mutants, double mutant cycles will be used to evaluate their energetic connections with K13. As modern protein design relies on naturally occurring protein folds, understanding how to recreate forbidden angles may enhance catalytic rates in unnatural enzymes and achieve chemical reactions not yet envisioned by nature nor organic chemistry.Support or Funding InformationWe would like to thank California State University Long Beach and the Carl E. Riley Endowed STEM scholarship for their generous funding.Heat map of where K13 dihedral occupies space on Ramachandran Plot over 200ns of MD simulation using Wild Type Human Triosephosphate Isomerase (RCSB:2JK2).Figure 1

A Double Mutant Cycle Involving the Charged Residues of Amyloid Beta

A critical difference between the fibril structures of Amyloid beta 40 (Aβ40) and its more toxic variant Abeta42 (Aβ42), associated with Alzheimer's disease, lies in the nature of a specific salt bridge involving Lysine-28 (K28). In Aβ42, it has a salt bridge with the C-terminal negative charge, while in Aβ40 it has the bridge with Aspartate-23 (D23). Interestingly, Aβ40 oligomers, which are aggregation intermediates known to be more toxic than the fibrils, do not have the D23-K28 salt bridge. So we ask whether transient interactions between K28 and the C-terminal of Aβ40, make the oligomers similar to the more toxic Aβ42. We use the powerful technique of ‘double mutant cycle’ to probe if K28 and the terminal COO- interact directly or indirectly during any stage of the aggregation process, and how that may influence the aggregation and toxicity of the peptide. We acetylate K28 and amidate the C-terminal, separately and also in the same mutant, and probe the aggregation kinetics, final solubility, secondary structure, membrane affinity and toxicity of the 4 variants. We find that the K28 acetylation has a very strong effect on the solubility, while amidation of the C-terminal has a mild but opposite effect. While K28 acetylation hardly affects the membrane affinity, amidation increases it sharply. However, the secondary structure of the amidated Abeta40 differs from the wild-type in the fibrillary state. Interestingly, both C-terminal amidation and K28-acetylation decrease the toxicity of the peptide considerably. Our results show that the K28 and the terminal COO- have independent but strong effects on the properties of the peptide. Moreover, the results prove the futility of focusing on reagents that just reduce aggregation propensity, since the toxicity of the peptide can vary independently.