Leveraging metabolic modeling to identify functional metabolic alterations associated with COVID-19 disease severity.

OBJECTIVE: Since the COVID-19 pandemic began in early 2020, SARS-CoV2 has claimed more than six million lives world-wide, with over 510 million cases to date. To reduce healthcare burden, we must investigate how to prevent non-acute disease from progressing to severe infection requiring hospitalization. METHODS: To achieve this goal, we investigated metabolic signatures of both non-acute (out-patient) and severe (requiring hospitalization) COVID-19 samples by profiling the associated plasma metabolomes of 84 COVID-19 positive University of Virginia hospital patients. We utilized supervised and unsupervised machine learning and metabolic modeling approaches to identify key metabolic drivers that are predictive of COVID-19 disease severity. Using metabolic pathway enrichment analysis, we explored potential metabolic mechanisms that link these markers to disease progression. RESULTS: Enriched metabolites associated with tryptophan in non-acute COVID-19 samples suggest mitigated innate immune system inflammatory response and immunopathology related lung damage prevention. Increased prevalence of histidine- and ketone-related metabolism in severe COVID-19 samples offers potential mechanistic insight to musculoskeletal degeneration-induced muscular weakness and host metabolism that has been hijacked by SARS-CoV2 infection to increase viral replication and invasion. CONCLUSIONS: Our findings highlight the metabolic transition from an innate immune response coupled with inflammatory pathway inhibition in non-acute infection to rampant inflammation and associated metabolic systemic dysfunction in severe COVID-19.

Diagnosis of normal chest radiographs using an autonomous deep-learning algorithm.

AIM: To evaluate the suitability of a deep-learning (DL) algorithm for identifying normality as a rule-out test for fully automated diagnosis in frontal adult chest radiographs (CXR) in an active clinical pathway. MATERIALS AND METHODS: This multicentre study included 3,887 CXRs from four distinct NHS institutions. A convolutional neural network (CNN) was developed and trained prior to this study and was used to classify a subset of examinations with the lowest abnormality scores as high confidence normal (HCN). For each radiograph, the ground truth (GT) was established using two independent reviewers and an arbitrator in case of discrepancy. RESULTS: The DL algorithm was able to classify 15% of all examinations as HCN, with a corresponding precision of 97.7%. There were 0.33% of examinations classified incorrectly as HCN, with 84.6% of these examinations identified as borderline cases by the radiologist GT process. CONCLUSION: A DL algorithm can achieve a high level of precision as a fully automated diagnostic tool for reporting a subset of CXRs as normal. The removal of 15% of all CXRs has the potential to significantly reduce workload and focus radiology resources on more complex examinations. To optimise performance, site-specific deployment of algorithms should occur with robust feedback mechanisms for incorrect classifications.

A functional relation for field-scale nonaqueous phase liquid dissolution developed using a pore network model.

A pore network model with cubic chambers and rectangular tubes was used to estimate the nonaqueous phase liquid (NAPL) dissolution rate coefficient, Kdissai, and NAPL/water total specific interfacial area, ai. Kdissai was computed as a function of modified Peclet number (Pe') for various NAPL saturations (SN) and ai during drainage and imbibition and during dissolution without displacement. The largest contributor to ai was the interfacial area in the water-filled corners of chambers and tubes containing NAPL. When Kdissai was divided by ai, the resulting curves of dissolution coefficient, Kdiss versus Pe' suggested that an approximate value of Kdiss could be obtained as a weak function of hysteresis or SN. Spatially and temporally variable maps of Kdissai calculated using the network model were used in field-scale simulations of NAPL dissolution. These simulations were compared to simulations using a constant value of Kdissai and the empirical correlation of Powers et al. [Water Resour. Res. 30(2) (1994b) 321]. Overall, a methodology was developed for incorporating pore-scale processes into field-scale prediction of NAPL dissolution.

Surfactant treatment effects on lung structure and type II cells of preterm ventilated lambs.

We evaluated surfactant treatment effects on lung morphology and alveolar type II cells of preterm ventilated lambs. Lambs were ventilated for 10 h following treatment of the right lung with natural surfactant. Lung parenchyma from the surfactant-treated right and the untreated left lung was compared morphometrically. Mechanical ventilation without surfactant resulted in distention of alveolar ducts accompanied by shallowing and loss of well-defined alveoli without disruption of collagen or elastin fibers. Surfactant treatment almost completely prevented these changes. The percent of normal parenchyma was 82 +/- 7% in surfactant-treated lobes and 26 +/- 5% in the nontreated lobes (p < 0.05). Type II cells became flatter in lungs ventilated without surfactant, and cell shape was preserved by surfactant treatment. The volume densities of lamellar bodies and multivesicular bodies in alveolar type II cells were not changed by surfactant treatment. With or without surfactant treatment, mechanical ventilation was associated with a shift in lamellar body distribution to a smaller size and a decrease in glycogen content of type II cells. Surfactant treatment of the preterm lung prevents alveolar distortion and atelectasis, but does not result in changes in subcellular organelles in immature type II cells.

A paradigm for drug discovery employing encoded combinatorial libraries.

Very large combinatorial libraries of small molecules on solid supports can now be synthesized and each library element can be identified after synthesis by using chemical tags. These tag-encoded libraries are potentially useful in drug discovery, and, to test this utility directly, we have targeted carbonic anhydrase (carbonate dehydratase; carbonate hydro-lyase, EC 4.2.1.1) as a model. Two libraries consisting of a total of 7870 members were synthesized, and structure-activity relationships based on the structures predicted by the tags were derived. Subsequently, an active representative of each library was resynthesized (2-[N-(4-sulfamoylbenzoyl)-4'-aminocyclohexanespiro]-4-oxo-7 -hydroxy- 2,3-dihydrobenzopyran and [N-(4-sulfamoylbenzoyl)-L-leucyl]piperidine-3-carboxylic acid) and these compounds were shown to have nanomolar dissociation constants (15 and 4 nM, respectively). In addition, a focused sublibrary of 217 sulfamoylbenzamides was synthesized and revealed a clear, testable structure-activity relationship describing isozyme-selective carbonic anhydrase inhibitors.

Complex synthetic chemical libraries indexed with molecular tags.

Combinatorial methods of chemical synthesis allow the creation of molecular libraries having immense diversity. The utility of such libraries is dependent upon identifying the structures of the molecules so prepared. We describe the construction of a peptide combinatorial library, having 117,649 different members, synthesized on beads and indexed with inert chemical tags. These tags are used as a binary code to record the reaction history of each bead. The code can be read directly from a single bead by electron capture capillary gas chromatography. We demonstrate the correct selection of members of the library on the basis of binding to a monoclonal antibody.

A human-human hybridoma secreting anti-HLA class II antibody.

In this report, we describe the production and characterization of the first human-human hybridoma secreting antibody to HLA Class II determinants. The hybridoma (GMEC101), which has been stable in tissue culture for greater than 20 mo, secretes 10 to 50 micrograms/ml of IgM-kappa antibody. This antibody binds to a wide range of human cell lines, but not to the HLA-A,B,C, and DR-negative K562 cell line. Functionally, GMEC101 strongly inhibits a unidirectional mixed lymphocyte reaction (MLR) at the level of the stimulator cell. Neither the cellular ELISA binding nor the MLR inhibition is lost after a triple platelet absorption (which removes Class I but not Class II activity). Because the binding and MLR blocking show no correlation with the known DR or DQ specificities, we suggest that GMEC101 may be detecting a novel HLA Class II determinant.

Production of a human monoclonal antibody to HLA by human-human hybridoma technology. A preliminary report.

An Epstein-Barr-virus-transformed lymphoblastoid cell line (ECEBV) was derived from a multiply transfused renal dialysis patient. ECEBV was shown to secrete specific antibody in a cellular enzyme-linked immunosorbent assay (CELISA) and was hybridized with the mutagenized human fusion partner G M1500 resistant to 6-thioguanine and ouabain. Hybridomas surviving hypoxanthine-aminopterin-thymidine (HAT) and ouabain selection were cloned by limiting dilution. The hybridomas continue to secrete antibody which reacts with some human cells but not with others after 14 months in culture. None reacts with K562 (no HLA-A, -B, -C or -DR) or with Daudi (no HLA-A, -B, or -C). This is a preliminary report of the production of a human monoclonal antibody to HLA. Application of this technique could result in the large-scale production of human monoclonal antibodies for HLA typing, the production of anti-idiotype antibodies for use in transplant patients to prevent acute rejection, and for the study of the structure and function of HLA in man.

Detection of antibodies to cell surface antigens by a simplified cellular ELISA (CELISA).

An improved method for screening human hybridoma antibodies to cell surface antigens is described. The following modifications have been developed: rapid expansion of desired screening cell types by EBV transformation; use of only 5 X 10(4) cells/well; elimination of the need for glutaraldehyde fixation; elimination of the requirement for PLL to attach cells to microplates; preparation of a large number of plates which can be stored at 4 degrees for 3 months; Protein A-peroxidase ELISA assay yielding excellent replicates, low background "noise", and high OD readings for positive wells. The techniques we have developed should greatly simplify and shorten the assay procedures for detecting human antibodies to a variety of cell surface antigens.

Strong HLA-DR expression in T cell cultures after activation is necessary for IL-2-dependent proliferation.

We have examined the appearance of DR antigens on human T cells activated by PHA, using a monoclonal anti-DR framework antibody and a large panel of human HLA typing sera. Strong DR expression within the culture by day 8 was associated with the ability of cells to grow for long periods in IL-2 containing medium, whereas weak or absent DR expression was predictive of poor in vitro growth. All the cells responded equivalently to initial stimulation with PHA. These data support the hypothesis proposed by Moretta et al. to explain blocking of IL-2-dependent proliferation by an anti-DR antibody--that DR molecules may be involved in the transmission of signals by IL-2.