Faculty Bibliography

MOTIVATION/BACKGROUND:Digital pathology supports analysis of histopathological images using deep learning methods at a large-scale. However, applications of deep learning in this area have been limited by the complexities of configuration of the computational environment and of hyperparameter optimization, which hinder deployment and reduce reproducibility. RESULTS:Here, we propose HEAL, a deep learning-based automated framework for easy, flexible, and multi-faceted histopathological image analysis. We demonstrate its utility and functionality by performing two case studies on lung cancer and one on colon cancer. Leveraging the capability of Docker, HEAL represents an ideal end-to-end tool to conduct complex histopathological analysis and enables deep learning in a broad range of applications for cancer image analysis. SUPPLEMENTARY INFORMATION/BACKGROUND:Supplementary data are available at Bioinformatics online.

Aims/UNASSIGNED:Histology, the microscopic study of normal tissues, is a crucial element of most medical curricula. Learning tools focused on histology are very important to learners who seek diagnostic competency within this important diagnostic arena. Recent developments in machine learning (ML) suggest that certain ML tools may be able to benefit this histology learning platform. Here, we aim to explore how one such tool based on a convolutional neural network, can be used to build a generalizable multi-classification model capable of classifying microscopic images of human tissue samples with the ultimate goal of providing a differential diagnosis (a list of look-alikes) for each entity. Methods/UNASSIGNED:We obtained three institutional training datasets and one generalizability test dataset, each containing images of histologic tissues in 38 categories. Models were trained on data from single institutions, low quantity combinations of multiple institutions, and high quantity combinations of multiple institutions. Models were tested against withheld validation data, external institutional data, and generalizability test images obtained from Google image search. Performance was measured with macro and micro accuracy, sensitivity, specificity, and f1-score. Results/UNASSIGNED:In this study, we were able to show that such a model's generalizability is dependent on both the training data source variety and the total number of training images used. Models which were trained on 760 images from only a single institution performed well on withheld internal data but poorly on external data (lower generalizability). Increasing data source diversity improved generalizability, even when decreasing data quantity: models trained on 684 images, but from three sources improved generalization accuracy between 4.05% and 18.59%. Maintaining this diversity and increasing the quantity of training images to 2280 further improved generalization accuracy between 16.51% and 32.79%. Conclusions/UNASSIGNED:This pilot study highlights the significance of data diversity within such studies. As expected, optimal models are those that incorporate both diversity and quantity into their platforms.s.

Immune checkpoint inhibitors (ICI), which target immune regulatory pathways to unleash antitumor responses, have revolutionized cancer immunotherapy. Despite the remarkable success of ICI immunotherapy, a significant proportion of patients whose tumors respond to these treatments develop immune-related adverse events (irAEs) resembling autoimmune diseases. Although the clinical spectrum of irAEs is well characterized, their successful management remains empiric. This is in part because the pathogenic mechanisms involved in the break-down of peripheral tolerance and induction of irAEs remain elusive. Herein, we focused on regulatory T cells (Tregs) in individuals with irAEs because these cells are vital for maintenance of peripheral tolerance, appear expanded in the peripheral blood of individuals with cancer and abundantly express checkpoint molecules, hence representing direct targets of ICI immunotherapy. Our data demonstrate an intense transcriptomic reprogramming of CD4+CD25+CD127- Tregs in the blood of individuals with advanced metastatic melanoma who develop irAEs following ICI immunotherapy, with a characteristic inflammatory, apoptotic and metabolic signature. This inflammatory signature was shared by Tregs from individuals with different types of cancer developing irAEs and individuals with autoimmune diseases. Our findings suggest inflammatory Treg reprogramming is a feature of immunotherapy-induced irAEs, and this may facilitate translational approaches aiming to induce robust antitumor immunity without disturbing peripheral tolerance.

Antibodies to double-stranded DNA (dsDNA) are prevalent in systemic lupus erythematosus (SLE), particularly in patients with lupus nephritis, yet the nature and regulation of antigenic cell-free DNA (cfDNA) are poorly understood. Null mutations in the secreted DNase DNASE1L3 cause human monogenic SLE with anti-dsDNA autoreactivity. We report that >50% of sporadic SLE patients with nephritis manifested reduced DNASE1L3 activity in circulation, which was associated with neutralizing autoantibodies to DNASE1L3. These patients had normal total plasma cfDNA levels but showed accumulation of cfDNA in circulating microparticles. Microparticle-associated cfDNA contained a higher fraction of longer polynucleosomal cfDNA fragments, which bound autoantibodies with higher affinity than mononucleosomal fragments. Autoantibodies to DNASE1L3-sensitive antigens on microparticles were prevalent in SLE nephritis patients and correlated with the accumulation of cfDNA in microparticles and with disease severity. DNASE1L3-sensitive antigens included DNA-associated proteins such as HMGB1. Our results reveal autoantibody-mediated impairment of DNASE1L3 activity as a common nongenetic mechanism facilitating anti-dsDNA autoreactivity in patients with severe sporadic SLE.

Over-accumulation of oxalate in humans may lead to nephrolithiasis and nephrocalcinosis. Humans lack endogenous oxalate degradation pathways (ODP), but intestinal microbes can degrade oxalate using multiple ODPs and protect against its absorption. The exact oxalate-degrading taxa in the human microbiota and their ODP have not been described. We leverage multi-omics data (>3000 samples from >1000 subjects) to show that the human microbiota primarily uses the type II ODP, rather than type I. Further, among the diverse ODP-encoding microbes, an oxalate autotroph, Oxalobacter formigenes, dominates this function transcriptionally. Patients with Inflammatory Bowel Disease (IBD) frequently suffer from disrupted oxalate homeostasis and calcium oxalate nephrolithiasis. We show that the enteric oxalate level is elevated in IBD patients, with highest levels in Crohn's disease patients with both ileal and colonic involvement consistent with known nephrolithiasis risk. We show that the microbiota ODP expression is reduced in IBD patients, which may contribute to the disrupted oxalate homeostasis. The specific changes in ODP expression by several important taxa suggest that they play distinct roles in IBD-induced nephrolithiasis risk. Lastly, we colonize mice that are maintained in the gnotobiotic facility with O. formigenes, using either a laboratory isolate or an isolate we cultured from human stools, and observed a significant reduction in host fecal and urine oxalate levels, supporting our in silico prediction of the importance of the microbiome, particularly O. formigenes in host oxalate homeostasis.

During self-renewal, cell-type-defining features are drastically perturbed in mitosis and must be faithfully reestablished upon G1 entry, a process that remains largely elusive. Here, we characterized at a genome-wide scale the dynamic transcriptional and architectural resetting of mouse pluripotent stem cells (PSCs) upon mitotic exit. We captured distinct waves of transcriptional reactivation with rapid induction of stem cell genes and transient activation of lineage-specific genes. Topological reorganization at different hierarchical levels also occurred in an asynchronous manner and showed partial coordination with transcriptional resetting. Globally, rapid transcriptional and architectural resetting associated with mitotic retention of H3K27 acetylation, supporting a bookmarking function. Indeed, mitotic depletion of H3K27ac impaired the early reactivation of bookmarked, stem-cell-associated genes. However, 3D chromatin reorganization remained largely unaffected, suggesting that these processes are driven by distinct forces upon mitotic exit. This study uncovers principles and mediators of PSC molecular resetting during self-renewal.

A fundamental, evolutionarily conserved biological mechanism required for long-term memory formation is rapid induction of gene transcription upon learning in relevant brain areas. For episodic types of memories, two regions undergoing this transcription are the dorsal hippocampus (dHC) and prelimbic (PL) cortex. Whether and to what extent these regions regulate similar or distinct transcriptomic profiles upon learning remains to be understood. Here, we used RNA sequencing in the dHC and PL cortex of male rats to profile their transcriptomes in untrained conditions (baseline) and at 1 hour and 6 days after inhibitory avoidance learning. We found that, out of 33,713 transcripts, over 14,000 were significantly expressed at baseline in both regions and approximately 3,000 were selectively enriched in each region. Gene Ontology biological pathway analyses indicated that commonly expressed pathways included synapse organization, regulation of membrane potential, and vesicle localization. The enriched pathways in the dHC were gliogenesis, axon development, and lipid modification, while in the PL cortex included vesicle localization and synaptic vesicle cycle. At 1 hour after learning, 135 transcripts changed significantly in the dHC and 478 in the PL cortex; of these, only 34 were shared. Biological pathways most significantly regulated by learning in the dHC were protein dephosphorylation, glycogen and glucan metabolism, while in the PL cortex were axon development and axonogenesis. The transcriptome profiles returned to baseline by 6 days after training. Thus, a significant portion of dHC and PL cortex transcriptomic profiles is divergent and their regulation upon learning is largely distinct and transient.Significance StatementLong-term episodic memory formation requires gene transcription in several brain regions including the hippocampus and prefrontal cortex. The comprehensive profiles of the dynamic mRNA changes that occur in these regions following learning are not well understood. Here, we performed RNA sequencing in the dorsal hippocampus (dHC) and prelimbic (PL) cortex, a prefrontal cortex subregion, at baseline, 1 hour, and 6 days after episodic learning in rats. We found that at baseline, dHC and PL cortex differentially express a significant portion of mRNAs. Moreover, learning produces a transient regulation of region-specific profiles of mRNA, indicating that unique biological programs in different brain regions underlie memory formation.

During the germinal center (GC) reaction, B cells undergo extensive redistribution of cohesin complex and three-dimensional reorganization of their genomes. Yet, the significance of cohesin and architectural programming in the humoral immune response is unknown. Herein we report that homozygous deletion of Smc3, encoding the cohesin ATPase subunit, abrogated GC formation, while, in marked contrast, Smc3 haploinsufficiency resulted in GC hyperplasia, skewing of GC polarity and impaired plasma cell (PC) differentiation. Genome-wide chromosomal conformation and transcriptional profiling revealed defects in GC B cell terminal differentiation programs controlled by the lymphoma epigenetic tumor suppressors Tet2 and Kmt2d and failure of Smc3-haploinsufficient GC B cells to switch from B cell- to PC-defining transcription factors. Smc3 haploinsufficiency preferentially impaired the connectivity of enhancer elements controlling various lymphoma tumor suppressor genes, and, accordingly, Smc3 haploinsufficiency accelerated lymphomagenesis in mice with constitutive Bcl6 expression. Collectively, our data indicate a dose-dependent function for cohesin in humoral immunity to facilitate the B cell to PC phenotypic switch while restricting malignant transformation.

Lack of cellular differentiation is a hallmark of many human cancers, including acute myeloid leukemia (AML). Strategies to overcome such a differentiation blockade are an approach for treating AML. To identify targets for differentiation-based therapies, we applied an integrated cell surface-based CRISPR platform to assess genes involved in maintaining the undifferentiated state of leukemia cells. Here we identify the RNA-binding protein ZFP36L2 as a critical regulator of AML maintenance and differentiation. Mechanistically, ZFP36L2 interacts with the 3' untranslated region of key myeloid maturation genes, including the ZFP36 paralogs, to promote their mRNA degradation and suppress terminal myeloid cell differentiation. Genetic inhibition of ZFP36L2 restores the mRNA stability of these targeted transcripts and ultimately triggers myeloid differentiation in leukemia cells. Epigenome profiling of several individuals with primary AML revealed enhancer modules near ZFP36L2 that associated with distinct AML cell states, establishing a coordinated epigenetic and post-transcriptional mechanism that shapes leukemic differentiation.

Identifying new drug targets and developing safe and effective drugs is both challenging and risky. Furthermore, characterizing drug development risk, the probability that a drug will eventually receive regulatory approval, has been notoriously hard given the complexities of drug biology and clinical trials. This inherent risk is often misunderstood and mischaracterized, leading to inefficient allocation of resources and, as a result, an overall reduction in R&D productivity. Here we argue that the recent resurgence of Machine Learning (ML) in combination with the availability of data can provide a more accurate and unbiased estimate of drug development risk.