Generating functional protein variants with variational autoencoders

The vast expansion of protein sequence databases provides an opportunity for new protein design approaches which seek to learn the sequence-function relationship directly from natural sequence variation. Deep generative models trained on protein sequence data have been shown to learn biologically meaningful representations helpful for a variety of downstream tasks, but their potential for direct use in the design of novel proteins remains largely unexplored. Here we show that variational autoencoders trained on a dataset of almost 70000 luciferase-like oxidoreductases can be used to generate novel, functional variants of the luxA bacterial luciferase. We propose separate VAE models to work with aligned sequence input (MSA VAE) and raw sequence input (AR-VAE), and offer evidence that while both are able to reproduce patterns of amino acid usage characteristic of the family, the MSA VAE is better able to capture long-distance dependencies reflecting the influence of 3D structure. To confirm the practical utility of the models, we used them to generate variants of luxA whose luminescence activity was validated experimentally. We further showed that conditional variants of both models could be used to increase the solubility of luxA without disrupting function. Altogether 6/12 of the variants generated using the unconditional AR-VAE and 9/11 generated using the unconditional MSA VAE retained measurable luminescence, together with all 23 of the less distant variants generated by conditional versions of the models; the most distant functional variant contained 35 differences relative to the nearest training set sequence. These results demonstrate the feasibility of using deep generative models to explore the space of possible protein sequences and generate useful variants, providing a method complementary to rational design and directed evolution approaches.     

Monte Carlo simulation for Neptun 10 PC medical linear accelerator and calculations of output factor for electron beam

AimExact knowledge of dosimetric parameters is an essential pre-requisite of an effective treatment in radiotherapy. In order to fulfill this consideration, different techniques have been used, one of which is Monte Carlo simulation.Materials and methodsThis study used the MCNP-4C^b to simulate electron beams from Neptun 10 PC medical linear accelerator. Output factors for 6, 8 and 10 MeV electrons applied to eleven different conventional fields were both measured and calculated.ResultsThe measurements were carried out by a Wellhofler-Scanditronix dose scanning system. Our findings revealed that output factors acquired by MCNP-4C simulation and the corresponding values obtained by direct measurements are in a very good agreement.ConclusionIn general, very good consistency of simulated and measured results is a good proof that the goal of this work has been accomplished.     

Synthesis and Antimicrobial Activity of Some 2-Pyridone Nucleosides Containing a Sulfonamide Moiety

Glycosylation of 2-pyridonesulfonamide 1a,b with glycosyl/galactosyl bromide gave the corresponding glycosides 2a,b, 3a,b, 6a,b, and 7a,b, respectively. Deacetylation of the resulting glycosides gave the corresponding glycosides 4a,b, 5a,b, 8a,b, and 9a,b, respectively, in good yields. Furthermore, reaction of 2-pyridonesulfonamide 1b with lactosyl bromide gave a mixture the corresponding N, O-lactosides 10 and 11, which were deacetylated to give the corresponding glycosides 12 and 13, respectively. The structures of the new synthesized compounds were characterized by using IR, ¹H, ¹³C NMR spectra, and microanalysis. Selected members of these compounds were screened for antimicrobial activity.     

Binding thermodynamics of phosphorylated inhibitors to triosephosphate isomerase and the contribution of electrostatic interactions

Electrostatic interactions have a central role in some biological processes, such as recognition of charged ligands by proteins. We characterized the binding energetics of yeast triosephosphate isomerase (TIM) with phosphorylated inhibitors 2-phosphoglycollate (2PG) and phosphoglycolohydroxamate (PGH). We determined the thermodynamic parameters of the binding process (K_b, ΔG_b, ΔH_b, ΔS_b and ΔC_p) with different concentrations of NaCl, using fluorimetric and calorimetric titrations in the conventional mode of ITC and a novel method, multithermal titration calorimetry (MTC), which enabled us to measure ΔC_p in a single experiment. We ruled out specific interactions of Na⁺ and Cl^- with the native enzyme and did not detect significant linked protonation effects upon the binding of inhibitors. Increasing ionic strength (I) caused K_b, ΔG_b and ΔH_b to become less favorable, while ΔS_b became less unfavorable. From the variation of K_b with I, we determined the electrostatic contribution of TIM−2PG and TIM−PGH to ΔG_b at I = 0.06 M and 25 °C to be 36% and 26%, respectively. The greater affinity of PGH for TIM is due to a more favorable ΔH_b compared to 2PG (by 19-24 kJ mol^-1 at 25 °C). This difference is compatible with PGH establishing up to five more hydrogen bonds with TIM. Both binding ΔC_ps were negative, and less negative with increasing ionic strength. ΔC_ps at I = 0.06 M were much more negative than predicted by surface area models. Water molecules trapped in the interface when ligands bind to protein could explain the highly negative ΔCps. Thermodynamic binding functions for TIM−2PG changed more with ionic strength than those for TIM−PGH. This greater dependence is consistent with linked, but compensated, protonation equilibriums yielding the dianionic species of 2PG that binds to TIM, process that is not required for PGH.     

Zinc is the metal cofactor of Borrelia burgdorferi peptide deformylase

Peptide deformylase (PDF, E.C. 3.5.1.88) catalyzes the removal of N-terminal formyl groups from nascent ribosome-synthesized polypeptides. PDF contains a catalytically essential divalent metal ion, which is tetrahedrally coordinated by three protein ligands (His, His, and Cys) and a water molecule. Previous studies revealed that the metal cofactor is a Fe²⁺ ion in Escherichia coli and many other bacterial PDFs. In this work, we found that PDFs from two iron-deficient bacteria, Borrelia burgdorferi and Lactobacillus plantarum, are stable and highly active under aerobic conditions. The native B. burgdorferi PDF (BbPDF) was purified 1200-fold and metal analysis revealed that it contains ∼1.1 Zn²⁺ ion/polypeptide but no iron. Our studies suggest that PDF utilizes different metal ions in different organisms. These data have important implications in designing PDF inhibitors and should help address some of the unresolved issues regarding PDF structure and catalytic function.     

The proportion of mutants in bacterial cultures

The proportion of mutants in a growing culture of organisms will depend upon (a) the rate at which the wild cells produce them (with or without growth), (b) the back mutation rate, and (c) the growth rates of the wild and mutant cells. If the mutation rate without growth and the back mutation rate are neglected, the growth of a mutant is expressed by See PDF for Equation and the ratio of the mutant to wild by See PDF for Equation in which λ = mutation frequency rate constant, "mutation rate," A = growth rate constant of wild cells W, B = growth rate constant of mutant cells M. If the term [B – (1 – 2λ)A] is positive, the proportion of mutants increases continuously. If it is negative, the proportion of mutants reaches a constant value See PDF for Equation If mutation is assumed to occur without growth at the rate C, then the corresponding equations are (11), (12), and (14). See PDF for Equation If (B + C – A) is negative and t = ∞, See PDF for Equation If C << A, See PDF for Equation     

Expression of functional mammal flavocytochrome b558 in yeast: Comparison with improved insect cell system

Activity of phagocyte NADPH-oxidase relies on the assembly of five proteins, among them the transmembrane flavocytochrome b₅₅₈ (Cytb₅₅₈) which consists of a heterodimer of the gp91^phox and p22^phox subunits. The Cytb₅₅₈ is the catalytic core of the NADPH-oxidase that generates a superoxide anion from oxygen by using a reducing equivalent provided by NADPH via FAD and two hemes. We report a novel strategy to engineer and produce the stable and functional recombinant Cytb₅₅₈ (rCytb₅₅₈). We expressed the gp91^phox and p22^phox subunits using the baculovirus insect cell and, for the first time, the highly inducible Pichia pastoris system. In both hosts, the expression of the full-length proteins reproduced native electrophoretic patterns demonstrating that the two polypeptides are present and, that gp91^phox undergoes co-translational glycosylation. Spectroscopic analyses showed that the rCytb₅₅₈ displayed comparable spectral properties to neutrophil Cytb₅₅₈. In contrast to rCytb₅₅₈ produced in the insect cells with higher yield, the enzyme expressed in yeast displayed a superoxide dismutase-sensitive NADPH-oxidase activity, indicating a superoxide generation activity. It was also blocked by an inhibitor of the respiratory burst oxidase, diphenylene iodonium (DPI). As in neutrophil NADPH-oxidase, activation occurred by the interactions with the soluble regulatory subunits suggesting comparable protein-protein contact patterns. We focus on the stability and function of the protein during solubilisation and reconstitution into liposomes. By comparing oxidase activities in different membrane types, we confirm that the lipid-protein environment plays a key role in the protein function.     

Social Environment Influences Performance in a Cognitive Task in Natural Variants of the Foraging Gene

In Drosophila melanogaster, natural genetic variation in the foraging gene affects the foraging behaviour of larval and adult flies, larval reward learning, adult visual learning, and adult aversive training tasks. Sitters (for ^s) are more sedentary and aggregate within food patches whereas rovers (for^R) have greater movement within and between food patches, suggesting that these natural variants are likely to experience different social environments. We hypothesized that social context would differentially influence rover and sitter behaviour in a cognitive task. We measured adult rover and sitter performance in a classical olfactory training test in groups and alone. All flies were reared in groups, but fly training and testing were done alone and in groups. Sitters trained and tested in a group had significantly higher learning performances compared to sitters trained and tested alone. Rovers performed similarly when trained and tested alone and in a group. In other words, rovers learning ability is independent of group training and testing. This suggests that sitters may be more sensitive to the social context than rovers. These differences in learning performance can be altered by pharmacological manipulations of PKG activity levels, the foraging (for) gene''s gene product. Learning and memory is also affected by the type of social interaction (being in a group of the same strain or in a group of a different strain) in rovers, but not in sitters. These results suggest that for mediates social learning and memory in D. melanogaster.     

Multi-timescale dynamics study of FKBP12 along the rapamycin-mTOR binding coordinate

Drugs can affect function in proteins by modulating their flexibility. Despite this possibility, there are very few studies on how drug binding affects the dynamics of target macromolecules. FKBP12 (FK506 binding protein 12) is a prolyl cis-trans isomerase and a drug target. The immunosuppressant drug rapamycin exerts its therapeutic effect by serving as an adaptor molecule between FKBP12 and the cell proliferation regulator mTOR (mammalian target of rapamycin). To understand the role of dynamics in rapamycin-based immunosuppression and to gain insight into the role of dynamics in the assembly of supramolecular complexes, we used ¹⁵N, ¹³C, and ²H NMR spin relaxation to characterize FKBP12 along the binding coordinate that leads to cell cycle arrest. We show that sequential addition of rapamycin and mTOR leads to incremental rigidification of the FKBP12 backbone on the picosecond-nanosecond timescale. Both binding events lead to perturbation of main-chain and side-chain dynamics at sites distal to the binding interfaces, suggesting tight coupling interactions dispersed throughout the FKBP12-rapamycin interface. Binding of the first molecule, rapamycin, quenches microsecond-millisecond motions of the FKBP12 80's loop. This loop provides much of the surface buried at the protein-protein interface of the ternary complex, leading us to assert that preorganization upon rapamycin binding facilitates binding of the second molecule, mTOR. Widespread microsecond-millisecond motions of the backbone persist in the drug-bound enzyme, and we provide evidence that these slow motions represent coupled dynamics of the enzyme and isomerization of the bound drug. Finally, the pattern of microsecond-millisecond dynamics reported here in the rapamycin complex is dramatically different from the pattern in the complex with the structurally related drug FK506. This raises the important question of how two complexes that are highly isomorphic based on high-resolution static models have such different flexibilities in solution.     

Automated Detection of B Cell and T Cell Acute Lymphoblastic Leukaemia Using Deep Learning

PurposeLeukaemia is diagnosed conventionally by observing the peripheral blood and bone marrow smear using a microscope and with the help of advanced laboratory tests. Image processing-based methods, which are simple, fast, and cheap, can be used to detect and classify leukemic cells by processing and analysing images of microscopic smear. The proposed study aims to classify Acute Lymphoblastic Leukaemia (ALL) by Deep Learning (DL) based techniques.ProceduresThe study used Deep Convolutional Neural Networks (DNNs) to classify ALL according to WHO classification scheme without using any image segmentation and feature extraction that involves intense computations. Images from an online image bank of American Society of Haematology (ASH) were used for the classification.FindingsA classification accuracy of 94.12% is achieved by the study in isolating the B-cell and T-cell ALL images using a pretrained CNN AlexNet as well as LeukNet, a custom-made deep learning network designed by the proposed work. The study also compared the classification performances using three different training algorithms.ConclusionsThe paper detailed the use of DNNs to classify ALL, without using any image segmentation and feature extraction techniques. Classification of ALL into subtypes according to the WHO classification scheme using image processing techniques is not available in literature to the best of the knowledge of the authors. The present study considered the classification of ALL only, and detection of other types of leukemic images can be attempted in future research.     

Study of the Individual Cytochrome b5 and Cytochrome b5 Reductase Domains of Ncb5or Reveals a Unique Heme Pocket and a Possible Role of the CS Domain

NADH cytochrome b₅ oxidoreductase (Ncb5or) is found in animals and contains three domains similar to cytochrome b₅ (b₅), CHORD-SGT1 (CS), and cytochrome b₅ reductase (b₅R). Ncb5or has an important function, as suggested by the diabetes and lipoatrophy phenotypes in Ncb5or null mice. To elucidate the structural and functional properties of human Ncb5or, we generated its individual b₅ and b₅R domains (Ncb5or-b₅ and Ncb5or-b₅R, respectively) and compared them with human microsomal b₅ (Cyb5A) and b₅R (Cyb5R3). A 1.25 Å x-ray crystal structure of Ncb5or-b₅ reveals nearly orthogonal planes of the imidazolyl rings of heme-ligating residues His⁸⁹ and His¹¹², consistent with a highly anisotropic low spin EPR spectrum. Ncb5or is the first member of the cytochrome b₅ family shown to have such a heme environment. Like other b₅ family members, Ncb5or-b₅ has two helix-loop-helix motifs surrounding heme. However, Ncb5or-b₅ differs from Cyb5A with respect to location of the second heme ligand (His¹¹²) and of polypeptide conformation in its vicinity. Electron transfer from Ncb5or-b₅R to Ncb5or-b₅ is much less efficient than from Cyb5R3 to Cyb5A, possibly as a consequence of weaker electrostatic interactions. The CS linkage probably obviates the need for strong interactions between b₅ and b₅R domains in Ncb5or. Studies with a construct combining the Ncb5or CS and b₅R domains suggest that the CS domain facilitates docking of the b₅ and b₅R domains. Trp¹¹⁴ is an invariant surface residue in all known Ncb5or orthologs but appears not to contribute to electron transfer from the b₅R domain to the b₅ domain.     

Synthesis of N-Aryl Uracils and Hypoxanthines and Their Biological Properties

Abstract

1-Benzyluracils 2a,b were treated with iodobenzene in the presence of cuprous oxide in 2,4,6-trimethylpyridine at 180°C to give the N ¹-phenyl derivatives 3a and 3b in 47% and 55%, respectively. Similar reaction of 2a with 2-bromopyridine at 120°C gave the 3-(2-pyridinyl)uracil 4a in 42% yield. However, unusual product 5 as well as 3-(2-pyridinyl) derivative 4b were obtained in the case of 2b. The structure of 5 was identified as 1-(2,6-difluorobenzyl)-3-[(2,4-dimethyl-2-pyridinyl)methyl]uracil from spectroscopic data. Reaction of the hypoxanthines 7a,b with 2-bromopyridine gave the 1-(2-pyridinyl)hypoxanthines 8a,b in low yields. But N-phenylation of 7a,b were unsuccessful.     

LeafNet: A computer vision system for automatic plant species identification

AimsTaxon identification is an important step in many plant ecological studies. Its efficiency and reproducibility might greatly benefit from partly automating this task. Image-based identification systems exist, but mostly rely on hand-crafted algorithms to extract sets of features chosen a priori to identify species of selected taxa. In consequence, such systems are restricted to these taxa and additionally require involving experts that provide taxonomical knowledge for developing such customized systems. The aim of this study was to develop a deep learning system to learn discriminative features from leaf images along with a classifier for species identification of plants. By comparing our results with customized systems like LeafSnap we can show that learning the features by a convolutional neural network (CNN) can provide better feature representation for leaf images compared to hand-crafted features.MethodsWe developed LeafNet, a CNN-based plant identification system. For evaluation, we utilized the publicly available LeafSnap, Flavia and Foliage datasets.ResultsEvaluating the recognition accuracies of LeafNet on the LeafSnap, Flavia and Foliage datasets reveals a better performance of LeafNet compared to hand-crafted customized systems.ConclusionsGiven the overall species diversity of plants, the goal of a complete automatisation of visual plant species identification is unlikely to be met solely by continually gathering assemblies of customized, specialized and hand-crafted (and therefore expensive) identification systems. Deep Learning CNN approaches offer a self-learning state-of-the-art alternative that allows adaption to different taxa just by presenting new training data instead of developing new software systems.     

Truncation of Peptide Deformylase Reduces the Growth Rate and Stabilizes Solvent Production in Clostridium beijerinckii NCIMB 8052

The wild-type strain of Clostridium beijerinckii NCIMB 8052 tends to degenerate (i.e., lose the ability to form solvents) after prolonged periods of laboratory culture. Several Tn1545 mutants of this organism showing enhanced long-term stability of solvent production were isolated. Four of them harbor identical insertions within the fms (def) gene, which encodes peptide deformylase (PDF). The C. beijerinckii fms gene product contains four diagnostic residues involved in the Zn²⁺ coordination and catalysis found in all PDFs, but it is unusually small, because it lacks the dispensable disordered C-terminal domain. Unlike previously characterized PDFs from Escherichia coli and Thermus thermophilus, the C. beijerinckii PDF can apparently tolerate N-terminal truncation. The Tn1545 insertion in the mutants is at a site corresponding to residue 15 of the predicted gene product. This probably removes 23 N-terminal residues from PDF, leaving a 116-residue protein. The mutant PDF retains at least partial function, and it complements an fms(Ts) strain of E. coli. Northern hybridizations indicate that the mutant gene is actively transcribed in C. beijerinckii. This can only occur from a previously unsuspected, outwardly directed promoter located close to the right end of Tn1545. The Tn1545 insertion in fms causes a reduction in the growth rate of C. beijerinckii, and, associated with this, the bacteria display an enhanced stability of solvent production. The latter phenotype can be mimicked in the wild type by reducing the growth rate. Therefore, the observed amelioration of degeneration in the mutants is probably due to their reduced growth rates.     

Potential antimicrobial agents from triazole-functionalized 2H-benzo[b][1,4]oxazin-3(4H)-ones

A series of substituted triazole functionalized 2H-benzo[b][1,4]oxazin-3(4H)-ones were synthesized by employing click chemistry and further characterized based on ¹H NMR, ¹³C NMR, IR and mass spectral studies. All the synthesized derivatives were screened for their in vitro antimicrobial activities. Further, molecular docking studies were accomplished to explore the binding interactions between 1,2,3-triazol-4-yl-2H-benzo[b][1,4]oxazin-3(4H)-one and the active site of Staphylococcus aureus (CrtM) dehydrosqualene synthase (PDB ID: 2ZCS). These docking studies revealed that the synthesized derivatives showed high binding energies and strong H-bond interactions with the dehydrosqualene synthase validating the observed antimicrobial activity data. Based on antimicrobial activity and docking studies, the compounds 9c, 9d and 9e were identified as promising antimicrobial leads.     

Evidence that Isolated Developing Chloroplasts Are Capable of Synthesizing Chlorophyll b from 5-Aminolevulinic Acid

Developing chloroplasts isolated from cucumber (Cucumis sativus L. var Beit Alpha) cotyledons are capable of incorporating [¹⁴C]5-aminolevulinic acid into chlorophyll (Chl) b and Chl a when incubated under photosynthetic illumination. Thin layer chromatography and high pressure liquid chromatography were employed to analyze the pigments. The specific radioactivity in Chl a was over three times higher than that found in Chl b. Both Chl a and b synthesizing activities in organello decayed rapidly at approximately the same rate. We conclude that concomitant synthesis of Chl a/b-binding apoprotein is not required for Chl b synthesis.     

SYNTHESIS AND BIOLOGICAL EVALUATION OF S-GLYCOSYLATED PYRIDINES

ABSTRACT

The formation of thioglycosides 7a-j and 10a-j via the reaction of sodium salts of thiopyridines 3a-e with glycosyl bromides 6a,b has been studied. Comparison with the products obtained from silylated thiopyridines 8a-e and peracetylated sugars 9a,b is made. ¹³C NMR was utilized to elucidate the proposed structures of the products.     

Interspecific variation in thermoregulation among three sympatric bats inhabiting a hot, semi-arid environment

Bats in hot roosts experience some of the most thermally challenging environments of any endotherms, but little is known about how heat tolerance and evaporative cooling capacity vary among species. We investigated thermoregulation in three sympatric species (Nycteris thebaica, Taphozous mauritianus and Sauromys petrophilus) in a hot, semi-arid environment by measuring body temperature (T _b), metabolic rate and evaporative water loss (EWL) at air temperatures (T _a) of 10?C42?°C. S. petrophilus was highly heterothermic with no clear thermoneutral zone, and exhibited rapid increases in EWL at high T _a to a maximum of 23.7?±?7.4?mg?g^?1?h^?1 at T _a????42?°C, with a concomitant maximum T _b of 43.7?±?1.0?°C. T. mauritianus remained largely normothermic at T _as below thermoneutrality and increased EWL to 14.7?±?1.3?mg?g^?1?h^?1 at T _a????42?°C, with a maximum T _b of 42.9?±?1.6?°C. In N. thebaica, EWL began increasing at lower T _a than in either of the other species and reached a maximum of 18.6?±?2.1?mg?g^?1?h^?1 at T _a?=?39.4?°C, with comparatively high maximum T _b values of 45.0?±?0.9?°C. Under the conditions of our study, N. thebaica was considerably less heat tolerant than the other two species. Among seven species of bats for which data on T _b as well as roost temperatures in comparison to outside T _a are available, we found limited evidence for a correlation between overall heat tolerance and the extent to which roosts are buffered from high T _a.