Using Deep Learning to Automate Eosinophil Counting in Pediatric Ulcerative Colitis Histopathological Images.

Background: Accurate identification of inflammatory cells from mucosal histopathology images is important in diagnosing ulcerative colitis. The identification of eosinophils in the colonic mucosa has been associated with disease course. Cell counting is not only time-consuming but can also be subjective to human biases. In this study we developed an automatic eosinophilic cell counting tool from mucosal histopathology images, using deep learning. Method: Four pediatric IBD pathologists from two North American pediatric hospitals annotated 530 crops from 143 standard-of-care hematoxylin and eosin (H & E) rectal mucosal biopsies. A 305/75 split was used for training/validation to develop and optimize a U-Net based deep learning model, and 150 crops were used as a test set. The U-Net model was then compared to SAU-Net, a state-of-the-art U-Net variant. We undertook post-processing steps, namely, (1) the pixel-level probability threshold, (2) the minimum number of clustered pixels to designate a cell, and (3) the connectivity. Experiments were run to optimize model parameters using AUROC and cross-entropy loss as the performance metrics. Results: The F1-score was 0.86 (95%CI:0.79-0.91) (Precision: 0.77 (95%CI:0.70-0.83), Recall: 0.96 (95%CI:0.93-0.99)) to identify eosinophils as compared to an F1-score of 0.2 (95%CI:0.13-0.26) for SAU-Net (Precision: 0.38 (95%CI:0.31-0.46), Recall: 0.13 (95%CI:0.08-0.19)). The inter-rater reliability was 0.96 (95%CI:0.93-0.97). The correlation between two pathologists and the algorithm was 0.89 (95%CI:0.82-0.94) and 0.88 (95%CI:0.80-0.94) respectively. Conclusion: Our results indicate that deep learning-based automated eosinophilic cell counting can obtain a robust level of accuracy with a high degree of concordance with manual expert annotations.

Artificial intelligence image-based prediction models in IBD exhibit high risk of bias: A systematic review.

BACKGROUND: There has been an increase in the development of both machine learning (ML) and deep learning (DL) prediction models in Inflammatory Bowel Disease. We aim in this systematic review to assess the methodological quality and risk of bias of ML and DL IBD image-based prediction studies. METHODS: We searched three databases, PubMed, Scopus and Embase, to identify ML and DL diagnostic or prognostic predictive models using imaging data in IBD, to Dec 31, 2022. We restricted our search to include studies that primarily used conventional imaging data, were undertaken in human participants, and published in English. Two reviewers independently reviewed the abstracts. The methodological quality of the studies was determined, and risk of bias evaluated using the prediction risk of bias assessment tool (PROBAST). RESULTS: Forty studies were included, thirty-nine developed diagnostic models. Seven studies utilized ML approaches, six were retrospective and none used multicenter data for model development. Thirty-three studies utilized DL approaches, ten were prospective, and twelve multicenter studies. Overall, all studies demonstrated high risk of bias. ML studies were evaluated in 4 domains all rated as high risk of bias: participants (6/7), predictors (1/7), outcome (3/7), and analysis (7/7), and DL studies evaluated in 3 domains: participants (24/33), outcome (10/33), and analysis (18/33). The majority of image-based studies used colonoscopy images. CONCLUSION: The risk of bias was high in AI IBD image-based prediction models, owing to insufficient sample size, unreported missingness and lack of an external validation cohort. Models with a high risk of bias are unlikely to be generalizable and suitable for clinical implementation.

Molecular neurobiology of loss: a role for basolateral amygdala extracellular matrix.

Psychological loss is a common experience that erodes well-being and negatively impacts quality of life. The molecular underpinnings of loss are poorly understood. Here, we investigate the mechanisms of loss using an environmental enrichment removal (ER) paradigm in male rats. The basolateral amygdala (BLA) was identified as a region of interest, demonstrating differential Fos responsivity to ER and having an established role in stress processing and adaptation. A comprehensive multi-omics investigation of the BLA, spanning multiple cohorts, platforms, and analyses, revealed alterations in microglia and the extracellular matrix (ECM). Follow-up studies indicated that ER decreased microglia size, complexity, and phagocytosis, suggesting reduced immune surveillance. Loss also substantially increased ECM coverage, specifically targeting perineuronal nets surrounding parvalbumin interneurons, suggesting decreased plasticity and increased inhibition within the BLA following loss. Behavioral analyses suggest that these molecular effects are linked to impaired BLA salience evaluation, leading to a mismatch between stimulus and reaction intensity. These loss-like behaviors could be rescued by depleting BLA ECM during the removal period, helping us understand the mechanisms underlying loss and revealing novel molecular targets to ameliorate its impact.

Microtubule-associated protein MAP1LC3C regulates lysosomal exocytosis and induces zinc reprogramming in renal cancer cells.

Microtubule-associated protein 1 light chain 3 gamma (MAP1LC3C or LC3C) is a member of the microtubule-associated family of proteins that are essential in the formation of autophagosomes and lysosomal degradation of cargo. LC3C has tumor-suppressing activity, and its expression is dependent on kidney cancer tumor suppressors, such as von Hippel-Lindau protein and folliculin. Recently, we demonstrated that LC3C autophagy is regulated by noncanonical upstream regulatory complexes and targets for degradation postdivision midbody rings associated with cancer cell stemness. Here, we show that loss of LC3C leads to peripheral positioning of the lysosomes and lysosomal exocytosis (LE). This process is independent of the autophagic activity of LC3C. Analysis of isogenic cells with low and high LE shows substantial transcriptomic reprogramming with altered expression of zinc (Zn)-related genes and activity of polycomb repressor complex 2, accompanied by a robust decrease in intracellular Zn. In addition, metabolomic analysis revealed alterations in amino acid steady-state levels. Cells with augmented LE show increased tumor initiation properties and form aggressive tumors in xenograft models. Immunocytochemistry identified high levels of lysosomal-associated membrane protein 1 on the plasma membrane of cancer cells in human clear cell renal cell carcinoma and reduced levels of Zn, suggesting that LE occurs in clear cell renal cell carcinoma, potentially contributing to the loss of Zn. These data indicate that the reprogramming of lysosomal localization and Zn metabolism with implication for epigenetic remodeling in a subpopulation of tumor-propagating cancer cells is an important aspect of tumor-suppressing activity of LC3C.

Assessing the effects of antipsychotic medications on schizophrenia functional analysis: a postmortem proteome study.

Antipsychotic drugs (APDs) are effective in treating positive symptoms of schizophrenia (SCZ). However, they have a substantial impact on postmortem studies. As most cohorts lack samples from drug-naive patients, many studies, rather than understanding SCZ pathophysiology, are analyzing the drug effects. We hypothesized that comparing SCZ-altered and APD-influenced signatures derived from the same cohort can provide better insight into SCZ pathophysiology. For this, we performed LCMS-based proteomics on dorsolateral prefrontal cortex (DLPFC) samples from control and SCZ subjects and used statistical approaches to identify SCZ-altered and APD-influenced proteomes, validated experimentally using independent cohorts and published datasets. Functional analysis of both proteomes was contrasted at the biological-pathway, cell-type, subcellular-synaptic, and drug-target levels. In silico validation revealed that the SCZ-altered proteome was conserved across several studies from the DLPFC and other brain areas. At the pathway level, SCZ influenced changes in homeostasis, signal-transduction, cytoskeleton, and dendrites, whereas APD influenced changes in synaptic-signaling, neurotransmitter-regulation, and immune-system processes. At the cell-type level, the SCZ-altered and APD-influenced proteomes were associated with two distinct striatum-projecting layer-5 pyramidal neurons regulating dopaminergic-secretion. At the subcellular synaptic level, compensatory pre- and postsynaptic events were observed. At the drug-target level, dopaminergic processes influenced the SCZ-altered upregulated-proteome, whereas nondopaminergic and a diverse array of non-neuromodulatory mechanisms influenced the downregulated-proteome. Previous findings were not independent of the APD effect and thus require re-evaluation. We identified a hyperdopaminergic cortex and drugs targeting the cognitive SCZ-symptoms and discussed their influence on SCZ pathology in the context of the cortico-striatal pathway.

Transcriptional profile of pyramidal neurons in chronic schizophrenia reveals lamina-specific dysfunction of neuronal immunity.

While the pathophysiology of schizophrenia has been extensively investigated using homogenized postmortem brain samples, few studies have examined changes in brain samples with techniques that may attribute perturbations to specific cell types. To fill this gap, we performed microarray assays on mRNA isolated from anterior cingulate cortex (ACC) superficial and deep pyramidal neurons from 12 schizophrenia and 12 control subjects using laser-capture microdissection. Among all the annotated genes, we identified 134 significantly increased and 130 decreased genes in superficial pyramidal neurons, while 93 significantly increased and 101 decreased genes were found in deep pyramidal neurons, in schizophrenia compared to control subjects. In these differentially expressed genes, we detected lamina-specific changes of 55 and 31 genes in superficial and deep neurons in schizophrenia, respectively. Gene set enrichment analysis (GSEA) was applied to the entire pre-ranked differential expression gene lists to gain a complete pathway analysis throughout all annotated genes. Our analysis revealed overrepresented groups of gene sets in schizophrenia, particularly in immunity and synapse-related pathways, suggesting the disruption of these pathways plays an important role in schizophrenia. We also detected other pathways previously demonstrated in schizophrenia pathophysiology, including cytokine and chemotaxis, postsynaptic signaling, and glutamatergic synapses. In addition, we observed several novel pathways, including ubiquitin-independent protein catabolic process. Considering the effects of antipsychotic treatment on gene expression, we applied a novel bioinformatics approach to compare our differential expression gene profiles with 51 antipsychotic treatment datasets, demonstrating that our results were not influenced by antipsychotic treatment. Taken together, we found pyramidal neuron-specific changes in neuronal immunity, synaptic dysfunction, and olfactory dysregulation in schizophrenia, providing new insights for the cell-subtype specific pathophysiology of chronic schizophrenia.

Using protein turnover to expand the applications of transcriptomics.

RNA expression and protein abundance are often at odds when measured in parallel, raising questions about the functional implications of transcriptomics data. Here, we present the concept of persistence, which attempts to address this challenge by combining protein half-life data with RNA expression into a single metric that approximates protein abundance. The longer a protein's half-life, the more influence it can have on its surroundings. This data offers a valuable opportunity to gain deeper insight into the functional meaning of transcriptome changes. We demonstrate the application of persistence using schizophrenia (SCZ) datasets, where it greatly improved our ability to predict protein abundance from RNA expression. Furthermore, this approach successfully identified persistent genes and pathways known to have impactful changes in SCZ. These results suggest that persistence is a valuable metric for improving the functional insight offered by transcriptomics data, and extended application of this concept could advance numerous research fields.

Identification of candidate repurposable drugs to combat COVID-19 using a signature-based approach.

The COVID-19 pandemic caused by the novel SARS-CoV-2 is more contagious than other coronaviruses and has higher rates of mortality than influenza. Identification of effective therapeutics is a crucial tool to treat those infected with SARS-CoV-2 and limit the spread of this novel disease globally. We deployed a bioinformatics workflow to identify candidate drugs for the treatment of COVID-19. Using an "omics" repository, the Library of Integrated Network-Based Cellular Signatures (LINCS), we simultaneously probed transcriptomic signatures of putative COVID-19 drugs and publicly available SARS-CoV-2 infected cell lines to identify novel therapeutics. We identified a shortlist of 20 candidate drugs: 8 are already under trial for the treatment of COVID-19, the remaining 12 have antiviral properties and 6 have antiviral efficacy against coronaviruses specifically, in vitro. All candidate drugs are either FDA approved or are under investigation. Our candidate drug findings are discordant with (i.e., reverse) SARS-CoV-2 transcriptome signatures generated in vitro, and a subset are also identified in transcriptome signatures generated from COVID-19 patient samples, like the MEK inhibitor selumetinib. Overall, our findings provide additional support for drugs that are already being explored as therapeutic agents for the treatment of COVID-19 and identify promising novel targets that are worthy of further investigation.

Tobacco smoking induces metabolic reprogramming of renal cell carcinoma.

BACKGROUNDClear cell renal cell carcinoma (ccRCC) is the most common histologically defined renal cancer. However, it is not a uniform disease and includes several genetic subtypes with different prognoses. ccRCC is also characterized by distinctive metabolic reprogramming. Tobacco smoking (TS) is an established risk factor for ccRCC, with unknown effects on tumor pathobiology.METHODSWe investigated the landscape of ccRCCs and paired normal kidney tissues using integrated transcriptomic, metabolomic, and metallomic approaches in a cohort of white males who were long-term current smokers (LTS) or were never smokers (NS).RESULTSAll 3 Omics domains consistently identified a distinct metabolic subtype of ccRCCs in LTS, characterized by activation of oxidative phosphorylation (OXPHOS) coupled with reprogramming of the malate-aspartate shuttle and metabolism of aspartate, glutamate, glutamine, and histidine. Cadmium, copper, and inorganic arsenic accumulated in LTS tumors, showing redistribution among intracellular pools, including relocation of copper into the cytochrome c oxidase complex. A gene expression signature based on the LTS metabolic subtype provided prognostic stratification of The Cancer Genome Atlas ccRCC tumors that was independent of genomic alterations.CONCLUSIONThe work identified the TS-related metabolic subtype of ccRCC with vulnerabilities that can be exploited for precision medicine approaches targeting metabolic pathways. The results provided rationale for the development of metabolic biomarkers with diagnostic and prognostic applications using evaluation of OXPHOS status. The metallomic analysis revealed the role of disrupted metal homeostasis in ccRCC, highlighting the importance of studying effects of metals from e-cigarettes and environmental exposures.FUNDINGDepartment of Defense, Veteran Administration, NIH, ACS, and University of Cincinnati Cancer Institute.

Signature-based approaches for informed drug repurposing: targeting CNS disorders.

CNS disorders, and in particular psychiatric illnesses, lack definitive disease-altering therapeutics. The limited understanding of the mechanisms driving these illnesses with the slow pace and high cost of drug development exacerbates this issue. For these reasons, drug repurposing - both a less expensive and time-efficient practice compared to de novo drug development - has been a promising strategy to overcome the paucity of treatments available for these debilitating disorders. While empirical drug-repurposing has been a routine practice in clinical psychiatry, innovative, informed, and cost-effective repurposing efforts using big data ("omics") have been designed to characterize drugs by structural and transcriptomic signatures. These strategies, in conjunction with ontological integration, provide an important opportunity to address knowledge-based challenges associated with drug development for CNS disorders. In this review, we discuss various signature-based in silico approaches to drug repurposing, its integration with multiple omics platforms, and how this data can be used for clinically relevant, evidence-based drug repurposing. These tools provide an exciting translational avenue to merge omics-based drug discovery platforms with patient-specific disease signatures, ultimately facilitating the identification of new therapies for numerous psychiatric disorders.

Combining Neurobehavioral Analysis and In Vivo Photoaffinity Labeling to Understand Protein Targets of Methamphetamine in Casper Zebrafish.

Photoaffinity labeling (PAL) remains one of the most widely utilized methods of determining protein targets of drugs. Although useful, the scope of this technique has been limited to in vitro applications because of the inability of UV light to penetrate whole organisms. Herein, pigment-free Casper zebrafish were employed to allow in vivo PAL. A methamphetamine-related phenethylamine PAL probe, designated here as 2, demonstrated dose-dependent effects on behavior similar to methamphetamine and permitted concentration-dependent labeling of protein binding partners. Click chemistry was used to analyze binding partners via fluoroimaging. Conjugation to a biotin permitted streptavidin pull-down and proteomic analysis to define direct binding partners of the methamphetamine probe. Bioinformatic analysis revealed the probe was chiefly bound to proteins involved in phagocytosis and mitochondrial function. Future applications of this experimental paradigm combining examination of drug-protein binding interactions alongside neurobehavioral readouts via in vivo PAL will significantly enhance our understanding of drug targets, mechanism(s) of action, and toxicity/lethality.

Identification of new drug treatments to combat COVID19: A signature-based approach using iLINCS.

The COVID-19 pandemic caused by the novel SARS-CoV-2 is more contagious than other coronaviruses and has higher rates of mortality than influenza. As no vaccine or drugs are currently approved to specifically treat COVID-19, identification of effective therapeutics is crucial to treat the afflicted and limit disease spread. We deployed a bioinformatics workflow to identify candidate drugs for the treatment of COVID-19. Using an "omics" repository, the Library of Integrated Network-Based Cellular Signatures (LINCS), we simultaneously probed transcriptomic signatures of putative COVID-19 drugs and signatures of coronavirus-infected cell lines to identify therapeutics with concordant signatures and discordant signatures, respectively. Our findings include three FDA approved drugs that have established antiviral activity, including protein kinase inhibitors, providing a promising new category of candidates for COVID-19 interventions.