Search

In this paper, we propose to incorporate the local attention in WaveNet-CTC to improve the performance of Tibetan speech recognition in multitask learning. With an increase in task number, such as simultaneous Tibetan speech content recognition, dialect identification, and speaker recognition, the accuracy rate of a single WaveNet-CTC decreases on speech recognition. Inspired by the attention mechanism, we introduce the local attention to automatically tune the weights of feature frames in a window and pay different attention on context information for multitask learning. The experimental results show that our method improves the accuracies of speech recognition for all Tibetan dialects in three-task learning, compared with the baseline model. Furthermore, our method significantly improves the accuracy for low-resource dialect by 5.11% against the specific-dialect model.

Immunotherapy has shown great potential in the treatment of cancer; however, only a fraction of patients respond to treatment, and many experience autoimmune‐related side effects. The pharmaceutical industry has relied on mathematical models to study the behavior of candidate drugs and more recently, complex, whole‐body, quantitative systems pharmacology (QSP) models have become increasingly popular for discovery and development. QSP modeling has the potential to discover novel predictive biomarkers as well as test the efficacy of treatment plans and combination therapies through virtual clinical trials. In this work, we present a QSP modeling platform for immuno‐oncology (IO) that incorporates detailed mechanisms for important immune interactions. This modular platform allows for the construction of QSP models of IO with varying degrees of complexity based on the research questions. Finally, we demonstrate the use of the platform through two example applications of immune checkpoint therapy.

Bioaugmentation is an optimization method with great potential to improve the treatment effect by introducing specific strains into the biological treatment system. In this study, a comprehensive review of the mechanism of bioaugmentation from the aspect of microbial community structure, the optimization methods facilitating application as well as feasible approaches of scale-up application has been provided. The different contribution of indigenous and exogenous strains was critically analyzed, the relationship between microbial community variation and system performance was clarified. Operation regulation and immobilization technologies are effective methods to deal with the possible failure of bioaugmentation. The gradual expansion from lab-scale, pilot scale to full-scale, the transformation and upgrading of wastewater treatment plants through the combination of direct dosing and biofilm, and the application of side-stream reactors are feasible ways to realize the full-scale application. The future challenges and prospects in this field were also proposed.

Bacterial colonization of biotic and abiotic surfaces and antibiotic resistance are grand challenges with paramount societal impacts. However, in the face of increasing bacterial resistance to all known antibiotics, efforts to discover new classes of antibiotics have languished, creating an urgent need to accelerate the antibiotic discovery pipeline. A major deterrent in the discovering of new antibiotics is the limited permeability of molecules across the bacterial envelope. Notably, the Gram-negative bacteria have nutrient specific protein channels (or porins) that restrict the permeability of non-essential molecules, including antibiotics. Here, we have developed the Computational Antibiotic Screening Platform (CLASP) for screening of potential drug molecules through the porins. The CLASP takes advantage of coarse grain (CG) resolution, advanced sampling techniques, and a parallel computing environment to maximize its performance. The CLASP yields comprehensive thermodynamic and kinetic output data of a potential drug molecule within a few hours of wall-clock time. Its output includes the potential of mean force profile, energy barrier, the rate constant, and contact analysis of the molecule with the pore-lining residues, and the orientational analysis of the molecule in the porin channel. In our first CLASP application, we report the transport properties of six carbapenem antibiotics-biapenem, doripenem, ertapenem, imipenem, meropenem, and panipenem-through OccD3, a major channel for carbapenem uptake in Pseudomonas aeruginosa. The CLASP is designed to screen small molecule libraries with a fast turnaround time to yield structure-property relationships to discover antibiotics with high permeability. The CLASP will be freely distributed to enable accelerated antibiotic drug discovery.

Choline and geranic acid (CAGE)-based ionic liquids have been recently developed for applications in drug delivery. Understanding the microscopic structures of CAGE in the presence of water is critical for its continued use in biomedical applications as it will undoubtedly come into contact with water in physiological fluids. Water can drastically impact the physicochemical properties of the ionic liquids, including CAGE. Computational and experimental characterization, namely viscosity, conductivity, and self-diffusion coefficient, were employed here to understand the properties of equimolar CAGE (1:1 choline/geranic acid) in the presence of varying amounts of water. It was found that under stored conditions, 1:1 CAGE contained up to 0.20 mole fraction water. Experimental and computational studies indicate that microscopic intraionic interactions within CAGE are not substantially changed until the water content exceeds 0.65 mole fraction. At this point, we theorize that the geranate ions undergo reorganization to minimize contact between the hydrophobic tails and the water molecules. This is evidenced by the plateau in viscosity at this mole fraction, and the increased interactions between the tails of the anions. This suggests that CAGE could be used without predrying in most applications and can be diluted to induce the organization of the anions where desired.

Bacteria can survive antibiotic treatment both by acquiring antibiotic resistance genes and through mechanisms of tolerance that are based on phenotypic changes and the formation of metabolically inactive cells. Here, we report an Enterococcus faecalis strain (E. faecalis UM001B) that was isolated from a cystic fibrosis patient and had no increase in resistance but extremely high-level tolerance to ampicillin, vancomycin, and tetracycline. Specifically, the percentages of cells that survived 3.5-h antibiotic treatment (at 100 μg · ml−1) were 25.4% ± 4.3% and 51.9% ± 4.0% for ampicillin and tetracycline, respectively; vancomycin did not exhibit any significant killing. Consistent with the changes in antibiotic susceptibility, UM001B was found to have reduced penetration of ampicillin and vancomycin and accumulation of tetracycline compared to the reference strain ATCC 29212. Based on whole-genome sequencing, four amino acid substitutions were identified in one of the tetracycline efflux pump repressors (TetRs), compared to ATCC 29212. Results of molecular simulations and experimental assays revealed that these mutations could lead to higher levels of tetracycline efflux activity. Consistently, replicating these mutations in Escherichia coli MG1655 increased its tolerance to tetracycline. Overall, these findings provide new insights into the development of multidrug tolerance in E. faecalis, which can facilitate future studies to better control enterococcal infections. IMPORTANCEEnterococcus faecalis represents a major group of pathogens causing nosocomial infections that are resistant to multiple classes of antibiotics. An important challenge associated with E. faecalis infection is the emergence of multidrug-tolerant strains, which have normal MICs but do not respond to antibiotic treatment. Here, we report a strain of E. faecalis that was isolated from a cystic fibrosis patient and demonstrated high-level tolerance to ampicillin, vancomycin, and tetracycline. Whole-genome sequencing revealed critical substitutions in one of the tetracycline efflux pump repressors that are consistent with the increased tolerance of E. faecalis UM001B to tetracycline. These findings provide new information about bacterial antibiotic tolerance and may help develop more effective therapeutics.

[Image: see text] Progress in immunotherapy has resulted in explosively increased new therapeutic interventions and they have shown promising results in the treatment of cancer. Animal testing is performed to provide preliminary efficacy and safety data for drugs under development prior to clinical trials. However, translational challenges remain for preclinical studies such as study design and the relevance of animal models to humans. Hence, only a small fraction of cancer patients showed response. The explosion of drug candidates and therapies makes preclinical assessment of every plausible option impossible, but it can be easily tested using Quantitative System Pharmacology (QSP) models. Here, we developed a QSP model for humanized mice. Tumor growth dynamics, T cell dynamics, cytokine release, immune checkpoint expression, and drug administration were modeled and calibrated using experimental data. Tumor growth inhibition data were used for model validation. Pharmacokinetics of T cell engager (TCE), tumor growth profile, T cell expansion in the blood and infiltration into tumor, T cell dissemination from primary tumor, cytokine release profile, and expression of additional PD-L1 induced by IFN-γ were modeled and calibrated using a variety of experimental data and showed good consistency. Mouse-specific response to T cell engager monotherapy also showed the key features of in vivo efficacy of TCE. This novel QSP model, designed for human peripheral blood mononuclear cells (PBMC) engrafted xenograft mice, incorporating the most critical components of the mouse model with key cancer and immune cells, can become an integral part of preclinical drug development.

BackgroundT cells have been recognized as core effectors for cancer immunotherapy. How to restore the anti-tumor ability of suppressed T cells or improve the lethality of cytotoxic T cells has become the main focus in immunotherapy. Bispecific antibodies, especially bispecific T cell engagers (TCEs), have shown their unique ability to enhance the patient’s immune response to tumors by stimulating T cell activation and cytokine production in an MHC-independent manner. Antibodies targeting the checkpoint inhibitory molecules such as programmed cell death protein 1 (PD-1), PD-ligand 1 (PD-L1) and cytotoxic lymphocyte activated antigen 4 are able to restore the cytotoxic effect of immune suppressed T cells and have also shown durable responses in patients with malignancies. However, both types have their own limitations in treating certain cancers. Preclinical and clinical results have emphasized the potential of combining these two antibodies to improve tumor response and patients’ survival. However, the selection and evaluation of combination partners clinically is a costly endeavor. In addition, despite advances made in immunotherapy, there are subsets of patients who are non-responders, and reliable biomarkers for different immunotherapies are urgently needed to improve the ability to prospectively predict patients’ response and improve clinical study design. Therefore, mathematical and computational models are essential to optimize patient benefit, and guide combination approaches with lower cost and in a faster manner.MethodIn this study, we continued to extend the quantitative systems pharmacology (QSP) model we developed for a bispecific TCE to explore efficacy of combination therapy with an anti-PD-L1 monoclonal antibody in patients with colorectal cancer.ResultsPatient-specific response to TCE monotherapy, anti-PD-L1 monotherapy and the combination therapy were predicted using this model according to each patient’s individual characteristics.ConclusionsIndividual biomarkers for TCE monotherapy, anti-PD-L1 monotherapy and their combination have been determined based on the QSP model. Best treatment options for specific patients could be suggested based on their own characteristics to improve clinical trial efficiency. The model can be further used to assess plausible combination strategies for different TCEs and immune checkpoint inhibitors in different types of cancer.

In this work, interactions between amphiphilic amino methyl coumarin and dipalmitoyl-sn-glycero-3-phosphocholine/dipalmitoyl-sn-glycero-3-phosphoserine (DPPC/DPPS) lipid bilayer were investigated. A combination of experimental techniques (zeta potential, fluorescence spectroscopy, and differential scanning calorimetry) along with molecular dynamics simulations was employed to examine the influence of alkyl tail length and concentration of the amphiphilic coumarin on the lipid bilayer. Alkyl tails comprising 5(C5), 9(C9), and 12(C12) carbon atoms were conjugated to amino methyl coumarin via a single-step process. The binding and insertion mechanisms of the amphiphilic coumarins were studied in increasing concentrations for short-tailed (C5) and long-tailed (C12) coumarins. The simulation results show that C5 coumarin molecules penetrate the lipid bilayer, but owing to the short alkyl tail, they interact primarily with the lipid head groups resulting in lipid bilayer thinning; however, at high concentrations, the C5 coumarins undergo continuous insertion–ejection from the outer leaflet of the lipid bilayer. In contrast, C12 coumarins interact favorably with the hydrophobic lipid tails and lack the ejection–reinsertion behavior. Instead, the C12 coumarin molecules undergo flip-flops between the outer and inner leaflets of the lipid bilayer. At high concentrations, the high-frequency flip-flops lead to lipid destabilization, causing the lipid bilayer to rupture. The simulation results are in excellent agreement with the toxicity of amphiphilic coumarin activity in cancer cells. The efficacy of amphiphilic coumarins in liposomal lipid bilayers demonstrates the promise of these molecules as a tool in the treatment of cancer.

Cancer immunotherapy has recently drawn remarkable attention as promising results in the clinic have shown its ability to improve the overall survival, and T cells are considered to be one of the primary effectors for cancer immunotherapy. Enhanced and restored T cell tumoricidal activity has shown great potential for killing cancer cells. Bispecific T cell engagers (TCEs) are a growing class of molecules that are designed to bind two different antigens on the surface of T cells and cancer cells to bring them in close proximity and selectively activate effector T cells to kill target cancer cells. New T cell engagers are being investigated for the treatment of solid tumors. The activity of newly developed T cell engagers showed a strong correlation with tumor target antigen expression. However, the correlation between tumor-associated antigen expression and overall response of cancer patients is poorly understood. In this study, we used a well-calibrated quantitative systems pharmacology (QSP) model extended to bispecific T cell engagers to explore their efficacy and identify potential biomarkers. In principle, patient-specific response can be predicted through this model according to each patient’s individual characteristics. This extended QSP model has been calibrated with available experimental data and provides predictions of patients’ response to TCE treatment. Electronic supplementary material The online version of this article (10.1208/s12248-020-00450-3) contains supplementary material, which is available to authorized users.

The continued emergence of antibiotic-resistant organisms has severely depleted our arsenal of effective antimicrobials. Ionic liquids (ILs) show great promise as antibacterial agents but understanding the mechanism of attack on bacterial cells is key to ensuring that design of IL-based biocides impart maximum efficacy with minimal toxicity, while also avoiding the potential for the target organisms to become resistant. Here we report the antibacterial attributes of a set of choline and geranate (CAGE)-based ILs and identify the mechanism by which they interact with the Gram-negative cell wall of Escherichia coli. CAGE is envisaged as an antimicrobial agent to treat topical infections in skin. Our earlier work has shown that CAGE is highly effective across a breadth of bacterial, fungal, and viral species and is benign to human cells. This combination makes CAGE an ideal antimicrobial for human use. Four CAGE variants with varying ratios of choline and geranic acid were synthesized and tested for their an...

BackgroundImmune checkpoint blockade therapy has clearly shown clinical activity in patients with triple-negative breast cancer, but less than half of the patients benefit from the treatments. While a number of ongoing clinical trials are investigating different combinations of checkpoint inhibitors and chemotherapeutic agents, predictive biomarkers that identify patients most likely to benefit remains one of the major challenges. Here we present a modular quantitative systems pharmacology (QSP) platform for immuno-oncology that incorporates detailed mechanisms of immune–cancer cell interactions to make efficacy predictions and identify predictive biomarkers for treatments using atezolizumab and nab-paclitaxel.MethodsA QSP model was developed based on published data of triple-negative breast cancer. With the model, we generated a virtual patient cohort to conduct in silico virtual clinical trials and make retrospective analyses of the pivotal IMpassion130 trial that led to the accelerated approval of atezolizumab and nab-paclitaxel for patients with programmed death-ligand 1 (PD-L1) positive triple-negative breast cancer. Available data from clinical trials were used for model calibration and validation.ResultsWith the calibrated virtual patient cohort based on clinical data from the placebo comparator arm of the IMpassion130 trial, we made efficacy predictions and identified potential predictive biomarkers for the experimental arm of the trial using the proposed QSP model. The model predictions are consistent with clinically reported efficacy endpoints and correlated immune biomarkers. We further performed a series of virtual clinical trials to compare different doses and schedules of the two drugs for simulated therapeutic optimization.ConclusionsThis study provides a QSP platform, which can be used to generate virtual patient cohorts and conduct virtual clinical trials. Our findings demonstrate its potential for making efficacy predictions for immunotherapies and chemotherapies, identifying predictive biomarkers, and guiding future clinical trial designs.

The self-assembly of outer membrane protein F (OmpF) in the outer membrane of Escherichia coli Gram-negative bacteria was studied using multiscale molecular dynamics simulations. To accommodate the long time scale required for protein assembly, coarse-grained parametrization of E. coli outer membrane lipids was first developed. The OmpF monomers formed stable dimers at specific protein–protein interactions sites irrespective of the lipid membrane environment. The dimer intermediate was asymmetric but provided a template to form a symmetric trimer. Superposition analysis of the self-assembled trimer with the X-ray crystal structure of the trimer available in the protein data bank showed excellent agreement with global root-mean-square deviation of less than 2.2 A. The free energy change associated with dimer formation was −26 ± 1 kcal mol–1, and for a dimer to bind to a monomer and to form a trimer yielded −56 ± 4 kcal mol–1. Based on thermodynamic data, an alternate path to trimer formation via interactio...

Lipopolysaccharides (LPSs) are vital components of the outer membrane of Gram-negative bacteria, and they act as extremely strong stimulators of innate immunity in diverse eukaryotic species. The primary immunostimulatory center of the LPS molecule is lipid A, a disaccharide-bound lipophilic domain. Considering the broad diversity in bacterial species, there are variations in the lipid A structures and their immunogenic potencies. In this work, we model the lipid A structures of eight commensal or human pathogenic bacterial species: Helicobacter pylori, Porphyromonas gingivalis, Bacteroides fragilis, Bordetella pertussis, Chlamydia trachomatis, Campylobacter jejuni, Neisseria meningitidis, and Salmonella minnesota. The membrane properties of these bacterial species were characterized and compared using molecular simulations. The structure-property relationships that emerge from this lipid A molecular library highlight the roles of acyl chain lengths, number of chains, phosphorylation state, membrane composition, and counterion charge in regulating the phase transition temperature of the membrane, diffusion coefficient of the lipids, and membrane thickness. The molecular and structural insights provided reveal the diversity in bacterial outer membrane lipids and their contribution to human disease and immunity.

The cell envelope of Gram-negative bacteria contains a lipopolysaccharide (LPS) rich outer membrane that acts as the first line of defense for bacterial cells in adverse physical and chemical environments. The LPS macromolecule has a negatively charged oligosaccharide domain that acts as an ionic brush, limiting the permeability of charged chemical agents through the membrane. Besides the LPS, the outer membrane has radially extending O-antigen polysaccharide chains and β-barrel membrane proteins that make the bacterial membrane physiologically unique compared to phospholipid cell membranes. Elucidating the interplay of these contributing macromolecular components and their role in the integrity of the bacterial outer membrane remains a challenge. To bridge the gap in our current understanding of the Gram-negative bacterial membrane, we have developed a coarse grained force field for outer membrane that is computationally affordable for simulating dynamical process over physiologically relevant time scales. The force field was benchmarked against available experimental and atomistic simulations data for properties such as membrane thickness, density profiles of the residues, area per lipid, gel to liquid-crystalline phase transition temperatures, and order parameters. More than 17 membrane compositions were studied with a combined simulation time of over 100 μs. A comparison of simulated structural and dynamical properties with corresponding experimental data shows that the developed force field reproduces the overall physiology of LPS rich membranes. The affordability of the developed model for long time scale simulations can be instrumental in determining the mechanistic aspects of the antimicrobial action of chemical agents as well as assist in designing antimicrobial peptides with enhanced outer membrane permeation properties.