Artificial intelligence in molecular de novo design: Integration with experiment

In this mini review, we capture the latest progress of applying artificial intelligence (AI) techniques based on deep learning architectures to molecular de novo design with a focus on integration with experimental validation. We will cover the progress and experimental validation of novel generative algorithms, the validation of QSAR models and how AI-based molecular de novo design is starting to become connected with chemistry automation. While progress has been made in the last few years, it is still early days. The experimental validations conducted thus far should be considered proof-of-principle, providing confidence that the field is moving in the right direction.     

Principles and Methods in Computational Membrane Protein Design

After decades of progress in computational protein design, the design of proteins folding and functioning in lipid membranes appears today as the next frontier. Some notable successes in the de novo design of simplified model membrane protein systems have helped articulate fundamental principles of protein folding, architecture and interaction in the hydrophobic lipid environment. These principles are reviewed here, together with the computational methods and approaches that were used to identify them. We provide an overview of the methodological innovations in the generation of new protein structures and functions and in the development of membrane-specific energy functions. We highlight the opportunities offered by new machine learning approaches applied to protein design, and by new experimental characterization techniques applied to membrane proteins. Although membrane protein design is in its infancy, it appears more reachable than previously thought.     

The Stability Landscape of de novo TIM Barrels Explored by a Modular Design Approach

The ability to design stable proteins with custom-made functions is a major goal in biochemistry with practical relevance for our environment and society. Understanding and manipulating protein stability provide crucial information on the molecular determinants that modulate structure and stability, and expand the applications of de novo proteins. Since the (β/⍺)₈-barrel or TIM-barrel fold is one of the most common functional scaffolds, in this work we designed a collection of stable de novo TIM barrels (DeNovoTIMs), using a computational fixed-backbone and modular approach based on improved hydrophobic packing of sTIM11, the first validated de novo TIM barrel, and subjected them to a thorough folding analysis. DeNovoTIMs navigate a region of the stability landscape previously uncharted by natural TIM barrels, with variations spanning 60 degrees in melting temperature and 22 kcal per mol in conformational stability throughout the designs. Significant non-additive or epistatic effects were observed when stabilizing mutations from different regions of the barrel were combined. The molecular basis of epistasis in DeNovoTIMs appears to be related to the extension of the hydrophobic cores. This study is an important step towards the fine-tuned modulation of protein stability by design.     

Smart de novo Macromolecular Structure Modeling from Cryo-EM Maps

The study of macromolecular structures has expanded our understanding of the amazing cell machinery and such knowledge has changed how the pharmaceutical industry develops new vaccines in recent years. Traditionally, X-ray crystallography has been the main method for structure determination, however, cryogenic electron microscopy (cryo-EM) has increasingly become more popular due to recent advancements in hardware and software. The number of cryo-EM maps deposited in the EMDataResource (formerly EMDatabase) since 2002 has been dramatically increasing and it continues to do so. De novo macromolecular complex modeling is a labor-intensive process, therefore, it is highly desirable to develop software that can automate this process. Here we discuss our automated, data-driven, and artificial intelligence approaches including map processing, feature extraction, modeling building, and target identification. Recently, we have enabled DNA/RNA modeling in our deep learning-based prediction tool, DeepTracer. We have also developed DeepTracer-ID, a tool that can identify proteins solely based on the cryo-EM map. In this paper, we will present our accumulated experiences in developing deep learning-based methods surrounding macromolecule modeling applications.     

Deep generative modeling for volume reconstruction in cryo-electron microscopy

Advances in cryo-electron microscopy (cryo-EM) for high-resolution imaging of biomolecules in solution have provided new challenges and opportunities for algorithm development for 3D reconstruction. Next-generation volume reconstruction algorithms that combine generative modelling with end-to-end unsupervised deep learning techniques have shown promise, but many technical and theoretical hurdles remain, especially when applied to experimental cryo-EM images. In light of the proliferation of such methods, we propose here a critical review of recent advances in the field of deep generative modelling for cryo-EM reconstruction. The present review aims to (i) provide a unified statistical framework using terminology familiar to machine learning researchers with no specific background in cryo-EM, (ii) review the current methods in this framework, and (iii) outline outstanding bottlenecks and avenues for improvements in the field.     

Novel insights into Parkin-mediated mitochondrial dysfunction and neuroinflammation in Parkinson's disease

Mutations in PRKN cause the second most common genetic form of Parkinson's disease (PD)—a debilitating movement disorder that is on the rise due to population aging in the industrial world. PRKN codes for an E3 ubiquitin ligase that has been well established as a key regulator of mitophagy. Together with PTEN-induced kinase 1 (PINK1), Parkin controls the lysosomal degradation of depolarized mitochondria. But Parkin's functions go well beyond mitochondrial clearance: the versatile protein is involved in mitochondria-derived vesicle formation, cellular metabolism, calcium homeostasis, mitochondrial DNA maintenance, mitochondrial biogenesis, and apoptosis induction. Moreover, Parkin can act as a modulator of different inflammatory pathways. In the current review, we summarize the latest literature concerning the diverse roles of Parkin in maintaining a healthy mitochondrial pool. Moreover, we discuss how these recent discoveries may translate into personalized therapeutic approaches not only for PRKN-PD patients but also for a subset of idiopathic cases.     

Heteroarylamide smoothened inhibitors: Discovery of N-[2,4-dimethyl-5-(1-methylimidazol-4-yl)phenyl]-4-(2-pyridylmethoxy)benzamide (AZD8542) and N-[5-(1H-imidazol-2-yl)-2,4-dimethyl-phenyl]-4-(2- pyridylmethoxy)benzamide (AZD7254)

Aberrant hedgehog (Hh) pathway signaling is implicated in multiple cancer types and targeting the Smoothened (SMO) receptor, a key protein of the Hh pathway, has proven effective in treating metastasized basal cell carcinoma. Our lead optimization effort focused on a series of heteroarylamides. We observed that a methyl substitution ortho to the heteroaryl groups on an aniline core significantly improved the potency of this series of compounds. These findings predated the availability of SMO crystal structure in 2013. Here we retrospectively applied quantum mechanics calculations to demonstrate the o-Me substitution favors the bioactive conformation by inducing a dihedral twist between the heteroaryl rings and the core aniline. The o-Me also makes favorable hydrophobic interactions with key residue side chains in the binding pocket. From this effort, two compounds (AZD8542 and AZD7254) showed excellent pharmacokinetics across multiple preclinical species and demonstrated in vivo activity in abrogating the Hh paracrine pathway as well as anti- tumor effects.     

Structural insights into UbiD reversible decarboxylation

The ubiquitous UbiX-UbiD system is associated with a wide range of microbial (de)carboxylation reactions. Recent X-ray crystallographic studies have contributed to elucidating the enigmatic mechanism underpinning the conversion of α,β-unsaturated acids by this system. The UbiD component utilises a unique cofactor, prenylated flavin (prFMN), generated by the bespoke action of the associated UbiX flavin prenyltransferase. Structure determination of a range of UbiX/UbiD representatives has revealed a generic mode of action for both the flavin-to-prFMN metamorphosis and the (de)carboxylation. In contrast to the conserved UbiX, the UbiD superfamily is associated with a versatile substrate range. The latter is reflected in the considerable variety of UbiD quaternary structure, dynamic behaviour and active site architecture. Directed evolution of UbiD enzymes has taken advantage of this apparent malleability to generate new variants supporting in vivo hydrocarbon production. Other applications include coupling UbiD to carboxylic acid reductase to convert alkenes into α,β-unsaturated aldehydes via enzymatic CO₂ fixation.     

Differential Proteome and Interactome Analysis Reveal the Basis of Pleiotropy Associated With the Histidine Methyltransferase Hpm1p

The methylation of histidine is a post-translational modification whose function is poorly understood. Methyltransferase histidine protein methyltransferase 1 (Hpm1p) monomethylates H243 in the ribosomal protein Rpl3p and represents the only known histidine methyltransferase in Saccharomyces cerevisiae. Interestingly, the hpm1 deletion strain is highly pleiotropic, with many extraribosomal phenotypes including improved growth rates in alternative carbon sources. Here, we investigate how the loss of histidine methyltransferase Hpm1p results in diverse phenotypes, through use of targeted mass spectrometry (MS), growth assays, quantitative proteomics, and differential crosslinking MS. We confirmed the localization and stoichiometry of the H243 methylation site, found unreported sensitivities of Δhpm1 yeast to nonribosomal stressors, and identified differentially abundant proteins upon hpm1 knockout with clear links to the coordination of sugar metabolism. We adapted the emerging technique of quantitative large-scale stable isotope labeling of amino acids in cell culture crosslinking MS for yeast, which resulted in the identification of 1267 unique in vivo lysine–lysine crosslinks. By reproducibly monitoring over 350 of these in WT and Δhpm1, we detected changes to protein structure or protein–protein interactions in the ribosome, membrane proteins, chromatin, and mitochondria. Importantly, these occurred independently of changes in protein abundance and could explain a number of phenotypes of Δhpm1, not addressed by expression analysis. Further to this, some phenotypes were predicted solely from changes in protein structure or interactions and could be validated by orthogonal techniques. Taken together, these studies reveal a broad role for Hpm1p in yeast and illustrate how crosslinking MS will be an essential tool for understanding complex phenotypes.     

Clinical whole exome sequencing revealed de novo heterozygous stop-gain and missense variants in the STXBP1 gene associated with epilepsy in Saudi families

Intellectual disability and developmental encephalopathies are mostly linked with infant epilepsy. Epileptic encephalopathy is a term that is used to define association between developmental delay and epilepsy. Mutations in the STXBP1 (Syntaxin-binding protein 1) gene have been previously reported in association with multiple severe early epileptic encephalopathies along with many neurodevelopmental disorders. Among the disorders produced due to any mutations in the STXBP1 gene is developmental and epileptic encephalopathy 4 (OMIM: 612164), is an autosomal dominant neurologic disorder categorized by the onset of tonic seizures in early infancy (usually in the first months of life). In this article, we report two Saudi families one with de novo heterozygous stop-gain mutation c.364C > T and a novel missense c. 305C > A p.Ala102Glu in exon 5 of the STXBP1 gene (OMIM: 602926) lead to development of epileptic encephalopathy 4. The variants identified in the current study broadened the genetic spectrum of STXBP1 gene related with diseases, which will help to add in the literature and benefit to the studies addressing this disease in the future.     

Quantitative Metaproteomics and Activity-based Protein Profiling of Patient Fecal Microbiome Identifies Host and Microbial Serine-type Endopeptidase Activity Associated With Ulcerative Colitis

The gut microbiota plays an important yet incompletely understood role in the induction and propagation of ulcerative colitis (UC). Organism-level efforts to identify UC-associated microbes have revealed the importance of community structure, but less is known about the molecular effectors of disease. We performed 16S rRNA gene sequencing in parallel with label-free data-dependent LC-MS/MS proteomics to characterize the stool microbiomes of healthy (n = 8) and UC (n = 10) patients. Comparisons of taxonomic composition between techniques revealed major differences in community structure partially attributable to the additional detection of host, fungal, viral, and food peptides by metaproteomics. Differential expression analysis of metaproteomic data identified 176 significantly enriched protein groups between healthy and UC patients. Gene ontology analysis revealed several enriched functions with serine-type endopeptidase activity overrepresented in UC patients. Using a biotinylated fluorophosphonate probe and streptavidin-based enrichment, we show that serine endopeptidases are active in patient fecal samples and that additional putative serine hydrolases are detectable by this approach compared with unenriched profiling. Finally, as metaproteomic databases expand, they are expected to asymptotically approach completeness. Using ComPIL and de novo peptide sequencing, we estimate the size of the probable peptide space unidentified (“dark peptidome”) by our large database approach to establish a rough benchmark for database sufficiency. Despite high variability inherent in patient samples, our analysis yielded a catalog of differentially enriched proteins between healthy and UC fecal proteomes. This catalog provides a clinically relevant jumping-off point for further molecular-level studies aimed at identifying the microbial underpinnings of UC.     

P300 Interacted With N-Myc and Regulated Its Protein Stability via Altering Its Post-Translational Modifications in Neuroblastoma

MYCN amplification is an independent risk factor for poor prognosis in neuroblastoma (NB), but its protein product cannot be directly targeted because of protein structure. Thus, this study aimed to explore novel ways to indirectly target N-Myc by regulating its post-translational modifications (PTMs) and therefore protein stability. N-Myc coimmunoprecipitation combined with HPLC–MS/MS identified 16 PTM residues and 114 potential N-Myc-interacting proteins. Notably, both acetylation and ubiquitination were identified on lysine 199 of N-Myc. We then discovered that p300, which can interact with N-Myc, modulated the protein stability of N-Myc in MYCN-amplified NB cell lines and simultaneously regulated the acetylation level and ubiquitination level on lysine-199 of N-Myc protein in vitro. Furthermore, p300 correlated with poor prognosis in NB patients. Taken together, p300 can be considered as a potential therapeutic target to treat MYCN-amplified NB patients, and other identified PTMs and interacting proteins also provide potential targets for further study.     

The functions of azurin of Pseudomonas aeruginosa and human mammaglobin-A on proapoptotic and cell cycle regulatory genes expression in the MCF-7 breast cancer cell line

Azurin protein of Pseudomonas aeruginosa is an anti-tumor agent against breast cancer and mammaglobin-A (MAM-A) protein is a specific antigen on the surface of MCF-7 for induction of cellular immune. The purpose of the present study was to investigate the effects of simultaneous expression of azurin and human MAM-A genes on the mRNA expression level of apoptosis-related and cell cycle genes in MCF-7 breast cancer cell line. The recombinant or empty plasmids were separately transferred into MCF-7 cells using Lipofectamine reagent. Flow cytometry was done to detect cell death and apoptosis. The expression of azurin and MAM-A genes were evaluated by IF assay, RT-PCR and western blot methods. Finally, apoptosis-related and cell cycle genes expression was examined in transformed and non-transformed MCF-7 cells by qPCR method. The successful expression of azurin and MAM-A genes in the MCF-7 cell were confirmed by RT-PCR, IF and western blotting. The apoptosis assay was showed a statistically significant (p < 0.05) difference after transfection. The expression of BAK, FAS, and BAX genes in transformed cells compare with non-transformed and transformed MCF-7 by pBudCE4.1 were increased statistically significant (p < 0.05) increases. Although, the increase of SURVIVIN and P53 expressions in transformed cells were not statistically significant (p > 0.05). Co-expression of azurin and MAM-A genes could induce apoptosis and necrosis in human MCF-7 breast cancer cells by up-regulation of BAK, FAS, and BAX genes. In future researches, it must be better the immune stimulation of pBudCE4.1-azurin-MAM-A recombinant vector in animal models and therapeutic approaches will be evaluated.     

In-planta production of the biodegradable polyester precursor 2-pyrone-4,6-dicarboxylic acid (PDC): Stacking reduced biomass recalcitrance with value-added co-product

2-Pyrone-4,6-dicarboxylic acid (PDC), a chemically stable intermediate that naturally occurs during microbial degradation of lignin by bacteria, represents a promising building block for diverse biomaterials and polyesters such as biodegradable plastics. The lack of a chemical synthesis method has hindered large-scale utilization of PDC and metabolic engineering approaches for its biosynthesis have recently emerged. In this study, we demonstrate a strategy for the production of PDC via manipulation of the shikimate pathway using plants as green factories. In tobacco leaves, we first showed that transient expression of bacterial feedback-resistant 3-deoxy-D-arabinoheptulosonate 7-phosphate synthase (AroG) and 3-dehydroshikimate dehydratase (QsuB) produced high titers of protocatechuate (PCA), which was in turn efficiently converted into PDC upon co-expression of PCA 4,5-dioxygenase (PmdAB) and 4-carboxy-2-hydroxymuconate-6-semialdehyde dehydrogenase (PmdC) derived from Comamonas testosteroni. We validated that stable expression of AroG in Arabidopsis in a genetic background containing the QsuB gene enhanced PCA content in plant biomass, presumably via an increase of the carbon flux through the shikimate pathway. Further, introducing AroG and the PDC biosynthetic genes (PmdA, PmdB, and PmdC) into the Arabidopsis QsuB background, or introducing the five genes (AroG, QsuB, PmdA, PmdB, and PmdC) stacked on a single construct into wild-type plants, resulted in PDC titers of ~1% and ~3% dry weight in plant biomass, respectively. Consistent with previous studies of plants expressing QsuB, all PDC producing lines showed strong reduction in lignin content in stems. This low lignin trait was accompanied with improvements of biomass saccharification efficiency due to reduced cell wall recalcitrance to enzymatic degradation. Importantly, most transgenic lines showed no reduction in biomass yields. Therefore, we conclude that engineering plants with the proposed de-novo PDC pathway provides an avenue to enrich biomass with a value-added co-product while simultaneously improving biomass quality for the supply of fermentable sugars. Implementing this strategy into bioenergy crops has the potential to support existing microbial fermentation approaches that exploit lignocellulosic biomass feedstocks for PDC production.     

Neuroimmune dysfunction in frontotemporal dementia: Insights from progranulin and C9orf72 deficiency

Neuroimmune dysfunction is a cardinal feature of neurodegenerative diseases. But how immune dysregulation in the brain and peripheral organs contribute to neurodegeneration remains unclear. Here, we discuss the recent advances highlighting neuroimmune dysfunction as a key disease-driving factor in frontotemporal dementia (FTD). We provide an overview of the clinical observations supporting a high prevalence of autoimmune diseases in FTD patients with mutations in GRN or C9orf72. We then focus on a myriad of evidence from human genetic studies, mouse models, in vitro assays, and multi-omics platform, which indicate that haploinsufficiency in GRN and C9orf72 promotes neuroimmune dysfunction and contributes to neurodegeneration and premature death. These compelling data provide key insights to disease mechanisms, biomarker discovery, and therapeutic interventions for FTD (120 words).     

Deep generative models for 3D molecular structure

Deep generative models have gained recent popularity for chemical design. Many of these models have historically operated in 2D space; however, more recently explicit 3D molecular generative models have become of interest, which are the topic of this article. Dozens of published models have been developed in the last few years to generate molecules directly in 3D, outputting both the atom types and coordinates, either in one-shot or adding atoms or fragments step-by-step. These 3D generative models can also be guided by structural information such as a binding pocket representation to successfully generate molecules with docking score ranges similar to known actives, but still showing lower computational efficiency and generation throughput than 1D/2D generative models and sometimes producing unrealistic conformations. We advocate for a unified benchmark of metrics to evaluate generation and propose perspectives to be addressed in next implementations.     

Endothelial Acid Sphingomyelinase Promotes NLRP3 Inflammasome and Neointima Formation During Hypercholesterolemia

The NOD-like receptor pyrin domain 3 (NLRP3) inflammasome is activated during atherogenesis, but how this occurs is unclear. Here, we explored the mechanisms activating and regulating NLRP3 inflammasomes via the acid sphingomyelinase (ASM)-ceramide signaling pathway. As a neointima formation model, partial left carotid ligations were performed on endothelial cell (EC)-specific ASM transgene mice (Smpd1^trg/EC^cre) and their control littermates (Smpd1^trg/WT and WT/WT) fed on the Western diet (WD). We found neointima formation remarkably increased in Smpd1^trg/EC^cre mice over their control littermates. Next, we observed enhanced colocalization of NLRP3 versus adaptor protein ASC (the adaptor molecule apoptosis-associated speck-like protein containing a CARD) or caspase-1 in the carotid ECs of WD-treated Smpd1^trg/EC^cre mice but not in their control littermates. In addition, we used membrane raft (MR) marker flotillin-1 and found more aggregation of ASM and ceramide in the intima of Smpd1^trg/EC^cre mice than their control littermates. Moreover, we demonstrated by in situ dihydroethidium staining, carotid intimal superoxide levels were much higher in WD-treated Smpd1^trg/EC^cre mice than in their control littermates. Using ECs from Smpd1^trg/EC^cre and WT/WT mice, we showed ASM overexpression markedly enhanced 7-ketocholesterol (7-Ket)-induced increases in NLRP3 inflammasome formation, accompanied by enhanced caspase-1 activity and elevated interleukin-1β levels. These 7-Ket-induced increases were significantly attenuated by ASM inhibitor amitriptyline. Furthermore, we determined that increased MR clustering with NADPH oxidase subunits to produce superoxide contributes to 7-Ket-induced NLRP3 inflammasome activation via a thioredoxin-interacting protein-mediated controlling mechanism. We conclude that ceramide from ASM plays a critical role in NLRP3 inflammasome activation during hypercholesterolemia via MR redox signaling platforms to produce superoxide, which leads to TXNIP dissociation.     

New Global Insights on the Regulation of the Biphasic Life Cycle and Virulence Via ClpP-Dependent Proteolysis in Legionella pneumophila

Legionella pneumophila, an environmental bacterium that parasitizes protozoa, causes Legionnaires’ disease in humans that is characterized by severe pneumonia. This bacterium adopts a distinct biphasic life cycle consisting of a nonvirulent replicative phase and a virulent transmissive phase in response to different environmental conditions. Hence, the timely and fine-tuned expression of growth and virulence factors in a life cycle–dependent manner is crucial for survival and replication. Here, we report that the completion of the biphasic life cycle and bacterial pathogenesis is greatly dependent on the protein homeostasis regulated by caseinolytic protease P (ClpP)-dependent proteolysis. We characterized the ClpP-dependent dynamic profiles of the regulatory and substrate proteins during the biphasic life cycle of L. pneumophila using proteomic approaches and discovered that ClpP-dependent proteolysis specifically and conditionally degraded the substrate proteins, thereby directly playing a regulatory role or indirectly controlling cellular events via the regulatory proteins. We further observed that ClpP-dependent proteolysis is required to monitor the abundance of fatty acid biosynthesis–related protein Lpg0102/Lpg0361/Lpg0362 and SpoT for the normal regulation of L. pneumophila differentiation. We also found that the control of the biphasic life cycle and bacterial virulence is independent. Furthermore, the ClpP-dependent proteolysis of Dot/Icm (defect in organelle trafficking/intracellular multiplication) type IVB secretion system and effector proteins at a specific phase of the life cycle is essential for bacterial pathogenesis. Therefore, our findings provide novel insights on ClpP-dependent proteolysis, which spans a broad physiological spectrum involving key metabolic pathways that regulate the transition of the biphasic life cycle and bacterial virulence of L. pneumophila, facilitating adaptation to aquatic and intracellular niches.     

Terminase Subunits from the Pseudomonas-Phage E217

Pseudomonas phages are increasingly important biomedicines for phage therapy, but little is known about how these viruses package DNA. This paper explores the terminase subunits from the Myoviridae E217, a Pseudomonas-phage used in an experimental cocktail to eradicate P. aeruginosa in vitro and in animal models. We identified the large (TerL) and small (TerS) terminase subunits in two genes ～58 kbs away from each other in the E217 genome. TerL presents a classical two-domain architecture, consisting of an N-terminal ATPase and C-terminal nuclease domain arranged into a bean-shaped tertiary structure. A 2.05 Å crystal structure of the C-terminal domain revealed an RNase H-like fold with two magnesium ions in the nuclease active site. Mutations in TerL residues involved in magnesium coordination had a dominant-negative effect on phage growth. However, the two ions identified in the active site were too far from each other to promote two-metal-ion catalysis, suggesting a conformational change is required for nuclease activity. We also determined a 3.38 Å cryo-EM reconstruction of E217 TerS that revealed a ring-like decamer, departing from the most common nonameric quaternary structure observed thus far. E217 TerS contains both N-terminal helix-turn-helix motifs enriched in basic residues and a central channel lined with basic residues large enough to accommodate double-stranded DNA. Overexpression of TerS caused a more than a 4-fold reduction of E217 burst size, suggesting a catalytic amount of the protein is required for packaging. Together, these data expand the molecular repertoire of viral terminase subunits to Pseudomonas-phages used for phage therapy.