Aggregation Propensity Research Articles

Holistic in silico developability assessment of novel classes of small proteins using publicly available sequence-based predictors.

The development of novel therapeutic proteins is a lengthy and costly process, with an average attrition rate of 91% (Thomas et al. Clinical Development Success Rates and Contributing Factors 2011-2020, 2021). To increase the probability of success and ensure robust drug supply beyond approval, it is essential to assess the developability profile of new potential drug candidates as early and broadly as possible in development (Jain et al. MAbs, 2023. https://doi.org/10.1016/j.copbio.2011.06.002 ). Predicting these properties in silico is expected to be the next leap in innovation as it would enable significantly reduced development timelines combined with broader screens at lower costs. However, developing predictive algorithms typically requires substantial datasets generated under very defined conditions, a limiting factor especially for new classes of therapeutic proteins that hold immense clinical promise. Here we describe a strategy for assessing the developability of a novel class of small therapeutic Anticalin® proteins using machine learning in conjunction with a knowledge-driven approach. The knowledge-driven approach considers developability attributes such as aggregation propensity, charge variants, immunogenicity, specificity, thermal stability, hydrophobicity, and potential post-translational modifications, to calculate a holistic developability score. Based on sequence-derived descriptors as input parameters we established novel statistical models designed to predict the developability scores for Anticalin proteins. The best models yielded low root mean square errors across the entire dataset and were further validated by removing input data from individual screening campaigns and predicting developability scores for those drug candidates. The adoption of the described workflow will enable significantly streamlined preclinical development of Anticalin drug candidates and could potentially be applied to other therapeutic protein scaffolds.

Phosphoryl group wires stabilize pathological tau fibrils as revealed by multiple quantum spin counting NMR.

Hyperphosphorylation of the protein tau is one of the biomarkers of neurodegenerative diseases in the category of tauopathies. However, the molecular level, mechanistic, role of this common post-translational modification (PTM) in enhancing or reducing the aggregation propensity of tau is unclear, especially considering that combinatorial phosphorylation of multiple sites can have complex, non-additive, effects on tau protein aggregation. Since tau proteins stack in register and parallel to elongate into pathological fibrils, phosphoryl groups from adjacent tau strands with 4.8 Å separation must find an energetically favorable spatial arrangement. At first glance, this appears to be an unfavorable configuration due to the proximity of negative charges between phosphate groups from adjacent neighboring tau fibrils. However, this study tests a counterhypothesis that phosphoryl groups within the fibril core-forming segments favorably assemble into highly ordered, hydrogen-bonded, one-dimensionally extended wires under biologically relevant conditions. We selected two phosphorylation sites associated with neurodegeneration, serine 305 (S305p) and tyrosine 310 (Y310p), on a model tau peptide jR2R3-P301L (tau295-313) spanning the R2/R3 splice junction of tau, that readily aggregate into a fibril with characteristics of a seed-competent mini prion. Using multiple quantum spin counting (MQ-SC) by 31P solid-state NMR of phosphorylated jR2R3-P301L tau peptide fibrils, enhanced by dynamic nuclear polarization, we find that at least six phosphorous spins must neatly arrange in 1D within fibrils or in 2D within a protofibril to yield the experimentally observed MQ-coherence orders of four. We found that S305p stabilizes the tau fibrils and leads to more seeding-competent fibrils compared to jR2R3 P301L or Y310p. This study introduces a new concept that phosphorylation of residues within a core forming tau segment can mechanically facilitate fibril registry and stability due a hitherto unrecognized role of phosphoryl groups to form highly ordered, extended, 1D wires that stabilize pathological tau fibrils.

Understanding alpha-synuclein aggregation propensity in animals and humans

Alpha-synuclein (α-syn) aggregation plays a critical role in the pathogenicity of Parkinson's Disease (PD). This study aims to evaluate the aggregation propensity of α-syn fragment peptides designed using the variability found in humans and animals. Thioflavin T (ThT) and transmission electron microscopy (TEM) were used to validate the formation of fibrils to identify important amino acid residues. Human α-syn fragments 51–75, 37–61, 62–86, 76–100, and 116–140 demonstrate a significantly higher tendency to aggregate compared to fragments 1–25, 26–50, and 91–115. All species analyzed of the α-syn 37–61 and 62–86 regions were shown to form fibrils on both ThT and TEM. The α-syn 37–61 and 62–86 fragment regions exhibited a high susceptibility to aggregation, with fibril formation observed in all species. The A53T mutation in several α-syn 37–61 fragments may enhance their propensity for aggregation, suggesting a correlation between this mutation and the capacity for fibril formation. Furthermore, the presence of the non-amyloid-β component (NAC) region, specifically in α-syn 62–86, was consistently observed in several fragments that displayed fibril formation, indicating a potential correlation between the NAC region and the process of fibril formation in α-syn. Finally, the combination of a high quantity of valine and a low quantity of acidic amino acids in these fragments may serve as indicators of α-syn fibril formation.

Preparation of T8 and double-decker silsesquioxane-based Janus-type molecules: molecular modeling and DFT insights

We present a methodology for the synthesis of inorganic-organic Janus-type molecules based on mono-T8 and difunctionalized double-decker silsesquioxanes (DDSQs) via hydrosilylation reactions, achieving exceptionally high yields and selectivities. The synthesized compounds were extensively characterized using various spectroscopic techniques, and their sizes and spatial arrangements were predicted through molecular modelling and density functional theory (DFT) calculations. Quantum chemical calculations were employed to examine the interactions among four molecules of the synthesized compounds. These computational results allowed us to determine the propensity for molecular aggregation, identify the functional groups involved in these interactions, and understand the changes in interatomic distances during aggregation. Understanding the aggregation behaviour of silsesquioxane molecules is crucial for tailoring their properties for specific applications, such as nanocomposites, surface coatings, drug delivery systems, and catalysts. Through a combination of experimental and computational approaches, this study provides valuable insights into the design and optimization of silsesquioxane-based Janus-type molecules for enhanced performance across various fields.

Wet Conformation of Prion-Like Domain and Intimate Correlation of Hydration and Conformational Fluctuations.

Proteins with prion-like domains (PLDs) are involved in neurodegeneration-associated aggregation and are prevalent in liquid-like membrane-less organelles. These PLDs contain amyloidogenic stretches but can maintain dynamic disordered conformations, even in the condensed phase. However, the molecular mechanism underlying such intricate conformational properties of PLDs remains elusive. Here we employed molecular dynamics simulations to investigate the conformational properties of a prototypical PLD system (i.e., FUS PLD). According to our simulation results, PLD adopts a wet collapsed conformation, wherein most residues maintain sufficient hydration with the abundance of internal water. These internal water molecules can rapidly exchange between the protein interior and the bulk, enabling intensive coupling of the entire protein with its hydration environment. The dynamic exchange of water molecules is intimately correlated to the overall conformational fluctuations of PLD. Furthermore, the abundance of dynamic internal water suppresses the formation of aggregation-prone ordered structures. These results collectively elucidate the crucial role of internal water in sustaining the dynamic disordered conformation of the PLD and inhibiting its aggregation propensity.

Key charged residues influence the amyloidogenic propensity of the helix-1 region of serum amyloid A

Increased plasma levels of serum amyloid A (SAA), an acute-phase protein that is secreted in response to inflammation, may lead to the accumulation of amyloid in various organs thereby obstructing their functions. Severe cases can lead to a systemic disorder called AA amyloidosis. Previous studies suggest that the N-terminal helix is the most amyloidogenic region of SAA. Moreover, computational studies implicated a significant role for Arg-1 and the residue-specific interactions formed during the fibrillization process. With a focus on the N-terminal region of helix-1, SAA1–13, mutational analysis was employed to interrogate the roles of the amino acid residues, Arg-1, Ser-5, Glu-9, and Asp-12. The truncated SAA1–13 fragment was systematically modified by substituting the key residues with alanine or uncharged but structurally similar amino acids. We monitored the changes in the amyloidogenic propensities, associated conformational markers, and morphology of the amyloids resulting from the mutation of SAA1–13. Mutating out Arg-1 resulted in much reduced aggregation propensity and a lack of detectable β-structures alluding to the importance of salt-bridge interactions involving Arg-1. Our data revealed that by systematically mutating the key amino acid residues, we can modulate the amyloidogenic propensity and alter the time-dependent conformational variation of the peptide. When the behaviors of each mutant peptide were analyzed, they provided evidence consistent with the aggregation pathway predicted by MD simulation studies. Here, we detail the important temporal molecular interactions formed by Arg-1 with Ser-5, Glu-9, and Asp-12 and discuss its mechanistic implications on the self-assembly of the helix-1 region of SAA.

Modulation of α-synuclein aggregation amid diverse environmental perturbation

Intrinsically disordered protein α-synuclein (αS) is implicated in Parkinson’s disease due to its aberrant aggregation propensity. In a bid to identify the traits of its aggregation, here we computationally simulate the multi-chain association process of αS in aqueous as well as under diverse environmental perturbations. In particular, the aggregation of αS in aqueous and varied environmental condition led to marked concentration differences within protein aggregates, resembling liquid-liquid phase separation (LLPS). Both saline and crowded settings enhanced the LLPS propensity. However, the surface tension of αS droplet responds differently to crowders (entropy-driven) and salt (enthalpy-driven). Conformational analysis reveals that the IDP chains would adopt extended conformations within aggregates and would maintain mutually perpendicular orientations to minimize inter-chain electrostatic repulsions. The droplet stability is found to stem from a diminished intra-chain interactions in the C-terminal regions of αS, fostering inter-chain residue-residue interactions. Intriguingly, a graph theory analysis identifies small-world-like networks within droplets across environmental conditions, suggesting the prevalence of a consensus interaction patterns among the chains. Together these findings suggest a delicate balance between molecular grammar and environment-dependent nuanced aggregation behavior of αS.

Modulation of α-synuclein aggregation amid diverse environmental perturbation.

Intrinsically disordered protein α-synuclein (αS) is implicated in Parkinson's disease due to its aberrant aggregation propensity. In a bid to identify the traits of its aggregation, here we computationally simulate the multi-chain association process of αS in aqueous as well as under diverse environmental perturbations. In particular, the aggregation of αS in aqueous and varied environmental condition led to marked concentration differences within protein aggregates, resembling liquid-liquid phase separation (LLPS). Both saline and crowded settings enhanced the LLPS propensity. However, the surface tension of αS droplet responds differently to crowders (entropy-driven) and salt (enthalpy-driven). Conformational analysis reveals that the IDP chains would adopt extended conformations within aggregates and would maintain mutually perpendicular orientations to minimize inter-chain electrostatic repulsions. The droplet stability is found to stem from a diminished intra-chain interactions in the C-terminal regions of αS, fostering inter-chain residue-residue interactions. Intriguingly, a graph theory analysis identifies small-world-like networks within droplets across environmental conditions, suggesting the prevalence of a consensus interaction patterns among the chains. Together these findings suggest a delicate balance between molecular grammar and environment-dependent nuanced aggregation behavior of αS.

O-GlcNAc Modification of α-Synuclein Can Alter Monomer Dynamics to Control Aggregation Kinetics.

The intrinsically disordered protein α-Synuclein is identified as a major toxic aggregate in Parkinson's as well as several other neurodegenerative diseases. Recent work on this protein has focused on the effects of posttranslational modifications on aggregation kinetics. Among them, O-GlcNAcylation of α-Synuclein has been observed to inhibit the aggregation propensity of the protein. Here, we investigate the monomer dynamics of two O-GlcNAcylated α-Synucleins, α-Syn(gT72), and α-Syn(gS87) and correlate them with the aggregation kinetics. We find that, compared to the unmodified protein, glycosylation at T72 makes the protein less compact and more diffusive, while glycosylation at S87 makes the protein more compact and less diffusive. Based on a model of the earliest steps in aggregation, we predict that T72 should aggregate slower than unmodified protein, which is confirmed by ThT fluorescence measurements. In contrast, S87 should aggregate faster, which is not mirrored in ThT kinetics of later fibril formation but does not rule out a higher rate of formation of small oligomers. Together, these results show that posttranslational modifications do not uniformly affect aggregation propensity.

Effects of molecule hydrophobicity and structural flexibility of appended bispecific antibody on Protein A chromatography

Appended bispecific antibody (aBsAb) with two single chain variable fragments (scFv) linked at the c-terminus of its heavy chains is one of the promising formats in bispecific therapeutics. The presence of hydrophobic and flexible scFv fragments render aBsAb molecules higher molecule hydrophobicity and structural flexibility compared to monoclonal antibody (mAb), thus making its purification more challenging. We set out to investigate how the unique molecular properties of aBsAb affect its performance on Protein A chromatography. We showed that aBsAb has a high propensity for chromatography-induced aggregation due to its high molecule hydrophobicity, and this couldn't be improved by the addition of common chaotropic salts. Moreover, the presence of chaotropic salts, such as arginine hydrochloride (Arg-HCl), retarded aBsAb elution during Protein A chromatography rather than facilitating which was widely observed in mAb Protein A elution. Nevertheless, we were able to overcome the aggregation issue by optimizing elution condition and improved aBsAb purity from 29 % to 93 % in Protein A eluate with a high molecular weight (HMW) species of less than 5 %. We also showed that the high molecular flexibility of aBsAb leads to different hydrodynamic sizes of the aBsAb molecule post Protein A elution, neutralization, and re-acidification, which are pH dependent. This is different from mAbs where their sizes do not change post neutralization even with re-exposure to acid. The above unique observations of aBsAb in Protein A chromatography were clearly explained from the perspectives of its high molecular hydrophobicity and structural flexibility.

Red-Light-Photosensitized Tyrosine 10 Nitration of β-Amyloid1-42 Diverts the Protein from Forming Toxic Aggregates.

Several studies have highlighted the presence of nitration damage following neuroinflammation in Alzheimer's disease (AD). Accordingly, post-transcriptional modifications of β-amyloid (Aβ), including peptide nitration, have been explored as a marker of the disease. However, the implications of Aβ nitration in terms of aggregation propensity and neurotoxicity are still debated. Here, we show new data obtained using a photoactivatable peroxynitrite generator (BPT-NO) to overcome the limitations associated with chemical nitration methods. We found that the photoactivation of BPT-NO with the highly biocompatible red light selectively induces the nitration of tyrosine 10 of freshly solubilized full-length Aβ1-42. Photonitrated Aβ1-42 was, therefore, investigated for aggregation states and functions. It resulted that photonitrated Aβ1-42 did not aggregate into small oligomers but rather self-assembled into large amorphous aggregates. When tested on neuronal-like SH-SY5Y cells and microglial C57BL/6 BV2 cells, photonitrated Aβ1-42 showed to be free of neurotoxicity and able to induce phagocytic microglia cells. We propose that light-controlled nitration of the multiple forms in which Aβ occurs (i.e., monomers, oligomers, fibrils) could be a tool to assess in real-time the impact of tyrosine nitration on the amyloidogenic and toxic properties of Aβ1-42.

Clickable tryptophan modification for late-stage diversification of native peptides.

As the least abundant residue in proteins, tryptophan widely exists in peptide drugs and bioactive natural products and contributes to drug-target interactions in multiple ways. We report here a clickable tryptophan modification for late-stage diversification of native peptides, via catalyst-free C2-sulfenylation with 8-quinoline thiosulfonate reagents in trifluoroacetic acid (TFA). A wide range of groups including trifluoromethylthio (SCF3), difluoromethylthio (SCF2H), (ethoxycarbonyl)difluoromethylthio (SCF2CO2Et), alkylthio, and arylthio were readily incorporated. The rapid reaction kinetics of Trp modification and full tolerance with other 19 proteinogenic amino acids, as well as the super dissolving capability of TFA, render this method suitable for all kinds of Trp-containing peptides without limitations from sequences, hydrophobicity, and aggregation propensity. The late-stage modification of 15 therapeutic peptides (1.0 to 7.6 kilodaltons) and the improved bioactivity and serum stability of SCF3- and SCF2H-modified melittin analogs illustrated the effectiveness of this method and its potential in pharmacokinetic property improvement.

AggreProt: a web server for predicting and engineering aggregation prone regions in proteins.

Recombinant proteins play pivotal roles in numerous applications including industrial biocatalysts or therapeutics. Despite the recent progress in computational protein structure prediction, protein solubility and reduced aggregation propensity remain challenging attributes to design. Identification of aggregation-prone regions is essential for understanding misfolding diseases or designing efficient protein-based technologies, and as such has a great socio-economic impact. Here, we introduce AggreProt, a user-friendly webserver that automatically exploits an ensemble of deep neural networks to predict aggregation-prone regions (APRs) in protein sequences. Trained on experimentally evaluated hexapeptides, AggreProt compares to or outperforms state-of-the-art algorithms on two independent benchmark datasets. The server provides per-residue aggregation profiles along with information on solvent accessibility and transmembrane propensity within an intuitive interface with interactive sequence and structure viewers for comprehensive analysis. We demonstrate AggreProt efficacy in predicting differential aggregation behaviours in proteins on several use cases, which emphasize its potential for guiding protein engineering strategies towards decreased aggregation propensity and improved solubility. The webserver is freely available and accessible at https://loschmidt.chemi.muni.cz/aggreprot/.

Development of a Humanized Antibody Targeting Extracellular HSP90α to Suppress Endothelial-Mesenchymal Transition-Enhanced Tumor Growth of Pancreatic Adenocarcinoma Cells.

Extracellular HSP90α (eHSP90α) is a promoter of tumor development and malignant progression. Patients with malignancies, including pancreatic ductal adenocarcinoma (PDAC), have generally shown 5~10-fold increases in serum/plasma eHSP90α levels. In this study, we developed a humanized antibody HH01 to target eHSP90α and evaluated its anticancer efficacy. HH01, with novel complementarity-determining regions, exhibits high binding affinity toward HSP90α. It recognizes HSP90α epitope sites 235AEEKEDKEEE244 and 251ESEDKPEIED260, with critical amino acid residues E237, E239, D240, K241, E253, and K255. HH01 effectively suppressed eHSP90α-induced invasive and spheroid-forming activities of colorectal cancer and PDAC cell lines by blocking eHSP90α's ligation with the cell-surface receptor CD91. In mouse models, HH01 potently inhibited the tumor growth of PDAC cell grafts/xenografts promoted by endothelial-mesenchymal transition-derived cancer-associated fibroblasts while also reducing serum eHSP90α levels, reflecting its anticancer efficacy. HH01 also modulated tumor immunity by reducing M2 macrophages and reinvigorating immune T-cells. Additionally, HH01 showed low aggregation propensity, high water solubility, and a half-life time of >18 days in mouse blood. It was not cytotoxic to retinal pigmented epithelial cells and showed no obvious toxicity in mouse organs. Our data suggest that targeting eHSP90α with HH01 antibody can be a promising novel strategy for PDAC therapy.

Dimerization of Full‐length Aβ ‐42 Peptide: A Comparison of Different Force Fields and WaterModels

Among the two isoforms of amyloid‐ i.e., Aβ‐40 and Aβ‐42, Aβ‐42 is more toxic due to its increased aggregation propensity. The oligomerization pathways of amyloid‐β may be investigated by studying its dimerization process at an atomic level. Intrinsically disordered proteins (IDPs) lack well‐defined structures and are associated with numerous neurodegenerative disorders. Molecular dynamics simulations of these proteins are often limited by the choice of parameters due to inconsistencies in the empirically developed protein force fields and water models. To evaluate the accuracy of recently developed force fields for IDPs, we study the dimerization of full‐length Aβ‐42 in aqueous solution with three different combinations of AMBER force field parameters and water models such as ff14SB/TIP3P, ff19SB/OPC, and ff19SB/TIP3P using classical MD and Umbrella Sampling method. This work may be used as a benchmark to compare the performance of different force fields for the simulations of IDPs.

Role of Charge Heterogeneity on Physical Stability of Monoclonal Antibody Biotherapeutic Products.

Chemical modifications in monoclonal antibodies can change hydrophobicity, charge heterogeneity as well as conformation, which eventually can impact their physical stability. In this study, the effect of the individual charge variants on physical stability and aggregation propensity in two different buffer conditions used during downstream purification was investigated. The charge variants were separated using semi-preparative cation exchange chromatography and buffer exchanged in the two buffers with pH 6.0 and 3.8. Subsequently each variant was analysed for size heterogeneity using size exclusion chromatography and dynamic light scattering, conformational stability, colloidal stability, and aggregation behaviour under accelerated stability conditions. Size variants in each charge variant were similar in both pH conditions when analyzed without extended storage. However, conformational stability was lower at pH 3.8 than pH 6.0. All charge variants showed similar apparent melting temperature at pH 6.0. In contrast, at pH 3.8 variants A3, A5, B2, B3 and B4 display lower Tm, suggesting reduced conformational stability. Further, A2, A3 and A5 exhibit reduced colloidal stability at pH 3.8. In general, acidic variants are more prone to aggregation than basic variants. Typical industry practice today is to examine in-process intermediate stability with acidic species and basic species taken as a single category each. We suggest that perhaps stability evaluation needs to be performed at specie level as different acidic or basic species have different stability and this knowledge can be used for clever designing of the downstream process to achieve a stable product.

The Target Therapy Hyperbole: "KRAS (p.G12C)"-The Simplification of a Complex Biological Problem.

Kirsten Rat Sarcoma Viral Oncogene Homolog (KRAS) gene variations are linked to the development of numerous cancers, including non-small cell lung cancer (NSCLC), colorectal cancer (CRC), and pancreatic ductal adenocarcinoma (PDAC). The lack of typical drug-binding sites has long hampered the discovery of therapeutic drugs targeting KRAS. Since "CodeBreaK 100" demonstrated Sotorasib's early safety and efficacy and led to its approval, especially in the treatment of non-small cell lung cancer (NSCLC), the subsequent identification of specific inhibitors for the p.G12C mutation has offered hope. However, the CodeBreaK 200 study found no significant difference in overall survival (OS) between patients treated with Docetaxel and Sotorasib (AMG 510), adding another degree of complexity to this ongoing challenge. The current study compares the three-dimensional structures of the two major KRAS isoforms, KRAS4A and KRAS4B. It also investigates the probable structural changes caused by the three major mutations (p.G12C, p.G12D, and p.G12V) within Sotorasib's pocket domain. The computational analysis demonstrates that the wild-type and mutant isoforms have distinct aggregation propensities, resulting in the creation of alternate oligomeric configurations. This study highlights the increased complexity of the biological issue of using KRAS as a therapeutic target. The present study stresses the need for a better understanding of the structural dynamics of KRAS and its mutations to design more effective therapeutic approaches. It also emphasizes the potential of computational approaches to shed light on the complicated molecular pathways that drive KRAS-mediated oncogenesis. This study adds to the ongoing efforts to address the therapeutic hurdles presented by KRAS in cancer treatment.

Boosting the Accuracy and Chemical Space Coverage of the Detection of Small Colloidal Aggregating Molecules Using the BAD Molecule Filter.

The ability to conduct effective high throughput screening (HTS) campaigns in drug discovery is often hampered by the detection of false positives in these assays due to small colloidally aggregating molecules (SCAMs). SCAMs can produce artifactual hits in HTS by nonspecific inhibition of the protein target. In this work, we present a new computational prediction tool for detecting SCAMs based on their 2D chemical structure. The tool, called the boosted aggregation detection (BAD) molecule filter, employs decision tree ensemble methods, namely, the CatBoost classifier and the light gradient-boosting machine, to significantly improve the detection of SCAMs. In developing the filter, we explore models trained on individual data sets, a consensus approach using these models, and, third, a merged data set approach, each tailored for specific drug discovery needs. The individual data set method emerged as most effective, achieving 93% sensitivity and 90% specificity, outperforming existing state-of-the-art models by 20 and 5%, respectively. The consensus models offer broader chemical space coverage, exceeding 90% for all testing sets. This feature is an important aspect particularly for early stage medicinal chemistry projects, and provides information on applicability domain. Meanwhile, the merged data set models demonstrated robust performance, with a notable sensitivity of 79% in the comprehensive 10-fold cross-validation test set. A SHAP analysis of model features indicates the importance of hydrophobicity and molecular complexity as primary factors influencing the aggregation propensity. The BAD molecule filter is readily accessible for the public usage on https://molmodlab-aau.com/Tools.html. This filter provides a new, more robust tool for aggregate prediction in the early stages of drug discovery to optimize hit rates and reduce associated testing and validation overheads.

A new lysine biosynthetic enzyme from a bacterial endosymbiont shaped by genetic drift and genome reduction.

The effect of population bottlenecks and genome reduction on enzyme function is poorly understood. Candidatus Liberibacter solanacearum is a bacterium with a reduced genome that is transmitted vertically to the egg of an infected psyllid-a population bottleneck that imposes genetic drift and is predicted to affect protein structure and function. Here, we define the function of Ca. L. solanacearum dihydrodipicolinate synthase (CLsoDHDPS), which catalyzes the committed branchpoint reaction in diaminopimelate and lysine biosynthesis. We demonstrate that CLsoDHDPS is expressed in Ca. L. solanacearum and expression is increased ~2-fold in the insect host compared to in planta. CLsoDHDPS has decreased thermal stability and increased aggregation propensity, implying mutations have destabilized the enzyme but are compensated for through elevated chaperone expression and a stabilized oligomeric state. CLsoDHDPS uses a ternary-complex kinetic mechanism, which is to date unique among DHDPS enzymes, has unusually low catalytic ability, but an unusually high substrate affinity. Structural studies demonstrate that the active site is more open, and the structure of CLsoDHDPS with both pyruvate and the substrate analogue succinic-semialdehyde reveals that the product is both structurally and energetically different and therefore evolution has in this case fashioned a new enzyme. Our study suggests the effects of genome reduction and genetic drift on the function of essential enzymes and provides insights on bacteria-host co-evolutionary associations. We propose that bacteria with endosymbiotic lifestyles present a rich vein of interesting enzymes useful for understanding enzyme function and/or informing protein engineering efforts.

Predictions of Colloidal Molecular Aggregation Using AI/ML Models.

To facilitate the triage of hits from small molecule screens, we have used various AI/ML techniques and experimentally observed data sets to build models aimed at predicting colloidal aggregation of small organic molecules in aqueous solution. We have found that Naïve Bayesian and deep neural networks outperform logistic regression, recursive partitioning tree, support vector machine, and random forest techniques by having the lowest balanced error rate (BER) for the test set. Derived predictive classification models consistently and successfully discriminated aggregator molecules from nonaggregator hits. An analysis of molecular descriptors in favor of colloidal aggregation confirms previous observations (hydrophobicity, molecular weight, and solubility) in addition to undescribed molecular descriptors such as the fraction of sp3 carbon atoms (Fsp3), and electrotopological state of hydroxyl groups (ES_Sum_sOH). Naïve Bayesian modeling and scaffold tree analysis have revealed chemical features/scaffolds contributing the most to colloidal aggregation and nonaggregation, respectively. These results highlight the importance of scaffolds with high Fsp3 values in promoting nonaggregation. Matched molecular pair analysis (MMPA) has also deciphered context-dependent substitutions, which can be used to design nonaggregator molecules. We found that most matched molecular pairs have a neutral effect on aggregation propensity. We have prospectively applied our predictive models to assist in chemical library triage for optimal plate selection diversity and purchase for high throughput screening (HTS) in drug discovery projects.