Genome organization around nuclear speckles drives mRNA splicing efficiency.

The nucleus is highly organized, such that factors involved in the transcription and processing of distinct classes of RNA are confined within specific nuclear bodies1,2. One example is the nuclear speckle, which is defined by high concentrations of protein and noncoding RNA regulators of pre-mRNA splicing3. What functional role, if any, speckles might play in the process of mRNA splicing is unclear4,5. Here we show that genes localized near nuclear speckles display higher spliceosome concentrations, increased spliceosome binding to their pre-mRNAs and higher co-transcriptional splicing levels than genes that are located farther from nuclear speckles. Gene organization around nuclear speckles is dynamic between cell types, and changes in speckle proximity lead to differences in splicing efficiency. Finally, directed recruitment of a pre-mRNA to nuclear speckles is sufficient to increase mRNA splicing levels. Together, our results integrate the long-standing observations of nuclear speckles with the biochemistry of mRNA splicing and demonstrate a crucial role for dynamic three-dimensional spatial organization of genomic DNA in driving spliceosome concentrations and controlling the efficiency of mRNA splicing.

The impact of package selection and versioning on single-cell RNA-seq analysis.

Standard single-cell RNA-sequencing analysis (scRNA-seq) workflows consist of converting raw read data into cell-gene count matrices through sequence alignment, followed by analyses including filtering, highly variable gene selection, dimensionality reduction, clustering, and differential expression analysis. Seurat and Scanpy are the most widely-used packages implementing such workflows, and are generally thought to implement individual steps similarly. We investigate in detail the algorithms and methods underlying Seurat and Scanpy and find that there are, in fact, considerable differences in the outputs of Seurat and Scanpy. The extent of differences between the programs is approximately equivalent to the variability that would be introduced in benchmarking scRNA-seq datasets by sequencing less than 5% of the reads or analyzing less than 20% of the cell population. Additionally, distinct versions of Seurat and Scanpy can produce very different results, especially during parts of differential expression analysis. Our analysis highlights the need for users of scRNA-seq to carefully assess the tools on which they rely, and the importance of developers of scientific software to prioritize transparency, consistency, and reproducibility for their tools.

Nuclear to cytoplasmic transport is a druggable dependency in MYC-driven hepatocellular carcinoma.

The MYC oncogene is often dysregulated in human cancer, including hepatocellular carcinoma (HCC). MYC is considered undruggable to date. Here, we comprehensively identify genes essential for survival of MYChigh but not MYClow cells by a CRISPR/Cas9 genome-wide screen in a MYC-conditional HCC model. Our screen uncovers novel MYC synthetic lethal (MYC-SL) interactions and identifies most MYC-SL genes described previously. In particular, the screen reveals nucleocytoplasmic transport to be a MYC-SL interaction. We show that the majority of MYC-SL nucleocytoplasmic transport genes are upregulated in MYChigh murine HCC and are associated with poor survival in HCC patients. Inhibiting Exportin-1 (XPO1) in vivo induces marked tumor regression in an autochthonous MYC-transgenic HCC model and inhibits tumor growth in HCC patient-derived xenografts. XPO1 expression is associated with poor prognosis only in HCC patients with high MYC activity. We infer that MYC may generally regulate and require altered expression of nucleocytoplasmic transport genes for tumorigenesis.

Efficient and accurate detection of viral sequences at single-cell resolution reveals putative novel viruses perturbing host gene expression.

There are an estimated 300,000 mammalian viruses from which infectious diseases in humans may arise. They inhabit human tissues such as the lungs, blood, and brain and often remain undetected. Efficient and accurate detection of viral infection is vital to understanding its impact on human health and to make accurate predictions to limit adverse effects, such as future epidemics. The increasing use of high-throughput sequencing methods in research, agriculture, and healthcare provides an opportunity for the cost-effective surveillance of viral diversity and investigation of virus-disease correlation. However, existing methods for identifying viruses in sequencing data rely on and are limited to reference genomes or cannot retain single-cell resolution through cell barcode tracking. We introduce a method that accurately and rapidly detects viral sequences in bulk and single-cell transcriptomics data based on highly conserved amino acid domains, which enables the detection of RNA viruses covering up to 1012 virus species. The analysis of viral presence and host gene expression in parallel at single-cell resolution allows for the characterization of host viromes and the identification of viral tropism and host responses. We applied our method to identify putative novel viruses in rhesus macaque PBMC data that display cell type specificity and whose presence correlates with altered host gene expression.

kallisto, bustools, and kb-python for quantifying bulk, single-cell, and single-nucleus RNA-seq.

The term "RNA-seq" refers to a collection of assays based on sequencing experiments that involve quantifying RNA species from bulk tissue, from single cells, or from single nuclei. The kallisto, bustools, and kb-python programs are free, open-source software tools for performing this analysis that together can produce gene expression quantification from raw sequencing reads. The quantifications can be individualized for multiple cells, multiple samples, or both. Additionally, these tools allow gene expression values to be classified as originating from nascent RNA species or mature RNA species, making this workflow amenable to both cell-based and nucleus-based assays. This protocol describes in detail how to use kallisto and bustools in conjunction with a wrapper, kb-python, to preprocess RNA-seq data.

Universal preprocessing of single-cell genomics data.

We describe a workflow for preprocessing a wide variety of single-cell genomics data types. The approach is based on parsing of machine-readable seqspec assay specifications to customize inputs for kb-python, which uses kallisto and bustools to catalog reads, error correct barcodes, and count reads. The universal preprocessing method is implemented in the Python package cellatlas that is available for download at: https://github.com/cellatlas/cellatlas/.

Flexible parsing, interpretation, and editing of technical sequences with splitcode.

Next-generation sequencing libraries are constructed with numerous synthetic constructs such as sequencing adapters, barcodes, and unique molecular identifiers. Such sequences can be essential for interpreting results of sequencing assays, and when they contain information pertinent to an experiment, they must be processed and analyzed. We present a tool called splitcode, that enables flexible and efficient parsing, interpreting, and editing of sequencing reads. This versatile tool facilitates simple, reproducible preprocessing of reads from libraries constructed for a large array of single-cell and bulk sequencing assays.

MYC-driven synthesis of Siglec ligands is a glycoimmune checkpoint.

The Siglecs (sialic acid-binding immunoglobulin-like lectins) are glycoimmune checkpoint receptors that suppress immune cell activation upon engagement of cognate sialoglycan ligands. The cellular drivers underlying Siglec ligand production on cancer cells are poorly understood. We find the MYC oncogene causally regulates Siglec ligand production to enable tumor immune evasion. A combination of glycomics and RNA-sequencing of mouse tumors revealed the MYC oncogene controls expression of the sialyltransferase St6galnac4 and induces a glycan known as disialyl-T. Using in vivo models and primary human leukemias, we find that disialyl-T functions as a "don't eat me" signal by engaging macrophage Siglec-E in mice or the human ortholog Siglec-7, thereby preventing cancer cell clearance. Combined high expression of MYC and ST6GALNAC4 identifies patients with high-risk cancers and reduced tumor myeloid infiltration. MYC therefore regulates glycosylation to enable tumor immune evasion. We conclude that disialyl-T is a glycoimmune checkpoint ligand. Thus, disialyl-T is a candidate for antibody-based checkpoint blockade, and the disialyl-T synthase ST6GALNAC4 is a potential enzyme target for small molecule-mediated immune therapy.

mTOR inhibition attenuates chemosensitivity through the induction of chemotherapy resistant persisters.

Chemotherapy can eradicate a majority of cancer cells. However, a small population of tumor cells often survives drug treatments through genetic and/or non-genetic mechanisms, leading to tumor recurrence. Here we report a reversible chemoresistance phenotype regulated by the mTOR pathway. Through a genome-wide CRISPR knockout library screen in pancreatic cancer cells treated with chemotherapeutic agents, we have identified the mTOR pathway as a prominent determinant of chemosensitivity. Pharmacological suppression of mTOR activity in cancer cells from diverse tissue origins leads to the persistence of a reversibly resistant population, which is otherwise eliminated by chemotherapeutic agents. Conversely, activation of the mTOR pathway increases chemosensitivity in vitro and in vivo and predicts better survival among various human cancers. Persister cells display a senescence phenotype. Inhibition of mTOR does not induce cellular senescence per se, but rather promotes the survival of senescent cells through regulation of autophagy and G2/M cell cycle arrest, as revealed by a small-molecule chemical library screen. Thus, mTOR plays a causal yet paradoxical role in regulating chemotherapeutic response; inhibition of the mTOR pathway, while suppressing tumor expansion, facilitates the development of a reversible drug-tolerant senescence state.

MYC oncogene elicits tumorigenesis associated with embryonic, ribosomal biogenesis, and tissue-lineage dedifferentiation gene expression changes.

MYC is a transcription factor frequently overexpressed in cancer. To determine how MYC drives the neoplastic phenotype, we performed transcriptomic analysis using a panel of MYC-driven autochthonous transgenic mouse models. We found that MYC elicited gene expression changes mostly in a tissue- and lineage-specific manner across B-cell lymphoma, T-cell acute lymphoblastic lymphoma, hepatocellular carcinoma, renal cell carcinoma, and lung adenocarcinoma. However, despite these gene expression changes being mostly tissue-specific, we uncovered a convergence on a common pattern of upregulation of embryonic stem cell gene programs and downregulation of tissue-of-origin gene programs across MYC-driven cancers. These changes are representative of lineage dedifferentiation, that may be facilitated by epigenetic alterations that occur during tumorigenesis. Moreover, while several cellular processes are represented among embryonic stem cell genes, ribosome biogenesis is most specifically associated with MYC expression in human primary cancers. Altogether, MYC's capability to drive tumorigenesis in diverse tissue types appears to be related to its ability to both drive a core signature of embryonic genes that includes ribosomal biogenesis genes as well as promote tissue and lineage specific dedifferentiation.

Inhibition of ERAD synergizes with FTS to eradicate pancreatic cancer cells.

BACKGROUND: Pancreatic ductal adenocarcinoma (PDAC), one of the most lethal cancers, is driven by oncogenic KRAS mutations. Farnesyl thiosalicylic acid (FTS), also known as salirasib, is a RAS inhibitor that selectively dislodges active RAS proteins from cell membrane, inhibiting downstream signaling. FTS has demonstrated limited therapeutic efficacy in PDAC patients despite being well tolerated. METHODS: To improve the efficacy of FTS in PDAC, we performed a genome-wide CRISPR synthetic lethality screen to identify genetic targets that synergize with FTS treatment. Among the top candidates, multiple genes in the endoplasmic reticulum-associated protein degradation (ERAD) pathway were identified. The role of ERAD inhibition in enhancing the therapeutic efficacy of FTS was further investigated in pancreatic cancer cells using pharmaceutical and genetic approaches. RESULTS: In murine and human PDAC cells, FTS induced unfolded protein response (UPR), which was further augmented upon treatment with a chemical inhibitor of ERAD, Eeyarestatin I (EerI). Combined treatment with FTS and EerI significantly upregulated the expression of UPR marker genes and induced apoptosis in pancreatic cancer cells. Furthermore, CRISPR-based genetic ablation of the key ERAD components, HRD1 and SEL1L, sensitized PDAC cells to FTS treatment. CONCLUSION: Our study reveals a critical role for ERAD in therapeutic response of FTS and points to the modulation of UPR as a novel approach to improve the efficacy of FTS in PDAC treatment.

Selective Labeling and Identification of the Tumor Cell Proteome of Pancreatic Cancer In Vivo.

Pancreatic ductal adenocarcinoma (PDAC) is among the deadliest cancers. Dissecting the tumor cell proteome from that of the non-tumor cells in the PDAC tumor bulk is critical for tumorigenesis studies, biomarker discovery, and development of therapeutics. However, investigating the tumor cell proteome has proven evasive due to the tumor's extremely complex cellular composition. To circumvent this technical barrier, we have combined bioorthogonal noncanonical amino acid tagging (BONCAT) and data-independent acquisition mass spectrometry (DIA-MS) in an orthotopic PDAC model to specifically identify the tumor cell proteome in vivo. Utilizing the tumor cell-specific expression of a mutant tRNA synthetase transgene, this approach provides tumor cells with the exclusive ability to incorporate an azide-bearing methionine analogue into newly synthesized proteins. The azide-tagged tumor cell proteome is subsequently enriched and purified via a bioorthogonal reaction and then identified and quantified using DIA-MS. Applying this workflow to the orthotopic PDAC model, we have identified thousands of proteins expressed by the tumor cells. Furthermore, by comparing the tumor cell and tumor bulk proteomes, we showed that the approach can distinctly differentiate proteins produced by tumor cells from those of non-tumor cells within the tumor microenvironment. Our study, for the first time, reveals the tumor cell proteome of PDAC under physiological conditions, providing broad applications for tumorigenesis, therapeutics, and biomarker studies in various human cancers.

ClinicNet: machine learning for personalized clinical order set recommendations.

OBJECTIVE: This study assesses whether neural networks trained on electronic health record (EHR) data can anticipate what individual clinical orders and existing institutional order set templates clinicians will use more accurately than existing decision support tools. MATERIALS AND METHODS: We process 57 624 patients worth of clinical event EHR data from 2008 to 2014. We train a feed-forward neural network (ClinicNet) and logistic regression applied to the traditional problem structure of predicting individual clinical items as well as our proposed workflow of predicting existing institutional order set template usage. RESULTS: ClinicNet predicts individual clinical orders (precision = 0.32, recall = 0.47) better than existing institutional order sets (precision = 0.15, recall = 0.46). The ClinicNet model predicts clinician usage of existing institutional order sets (avg. precision = 0.31) with higher average precision than a baseline of order set usage frequencies (avg. precision = 0.20) or a logistic regression model (avg. precision = 0.12). DISCUSSION: Machine learning methods can predict clinical decision-making patterns with greater accuracy and less manual effort than existing static order set templates. This can streamline existing clinical workflows, but may not fit if historical clinical ordering practices are incorrect. For this reason, manually authored content such as order set templates remain valuable for the purposeful design of care pathways. ClinicNet's capability of predicting such personalized order set templates illustrates the potential of combining both top-down and bottom-up approaches to delivering clinical decision support content. CONCLUSION: ClinicNet illustrates the capability for machine learning methods applied to the EHR to anticipate both individual clinical orders and existing order set templates, which has the potential to improve upon current standards of practice in clinical order entry.

The MYC Oncogene Cooperates with Sterol-Regulated Element-Binding Protein to Regulate Lipogenesis Essential for Neoplastic Growth.

Lipid metabolism is frequently perturbed in cancers, but the underlying mechanism is unclear. We present comprehensive evidence that oncogene MYC, in collaboration with transcription factor sterol-regulated element-binding protein (SREBP1), regulates lipogenesis to promote tumorigenesis. We used human and mouse tumor-derived cell lines, tumor xenografts, and four conditional transgenic mouse models of MYC-induced tumors to show that MYC regulates lipogenesis genes, enzymes, and metabolites. We found that MYC induces SREBP1, and they collaborate to activate fatty acid (FA) synthesis and drive FA chain elongation from glucose and glutamine. Further, by employing desorption electrospray ionization mass spectrometry imaging (DESI-MSI), we observed in vivo lipidomic changes upon MYC induction across different cancers, for example, a global increase in glycerophosphoglycerols. After inhibition of FA synthesis, tumorigenesis was blocked, and tumors regressed in both xenograft and primary transgenic mouse models, revealing the vulnerability of MYC-induced tumors to the inhibition of lipogenesis.

Neural Networks for Clinical Order Decision Support.

Consistent and high quality medical decisions are difficult as the amount of literature, data, and treatment options grow. We developed a model to provide automated physician order decision support suggestions for inpatient care through a feed-forward neural network. Given a patient's current status based on information data-mined and extracted from the Electronic Health Record (EHR), our model predicts clinical orders a physician enters for a patient within 24 hours. As a reference benchmark of real-world standard-of-care clinical decision support, existing manually-curated order sets implemented in the hospital demonstrate precision: 0.21, recall: 0.48, AUROC: 0.75 relative to what clinicians actually order within 24 hours. Our feed-forward model provides an automated, scalable, and robust system that achieves precision: 0.41, recall: 0.61, AUROC: 0.80.

Oncogene KRAS activates fatty acid synthase, resulting in specific ERK and lipid signatures associated with lung adenocarcinoma.

KRAS gene mutation causes lung adenocarcinoma. KRAS activation has been associated with altered glucose and glutamine metabolism. Here, we show that KRAS activates lipogenesis, and this activation results in distinct proteomic and lipid signatures. By gene expression analysis, KRAS is shown to be associated with a lipogenesis gene signature and specific induction of fatty acid synthase (FASN). Through desorption electrospray ionization MS imaging (DESI-MSI), specific changes in lipogenesis and specific lipids are identified. By the nanoimmunoassay (NIA), KRAS is found to activate the protein ERK2, whereas ERK1 activation is found in non-KRAS-associated human lung tumors. The inhibition of FASN by cerulenin, a small molecule antibiotic, blocked cellular proliferation of KRAS-associated lung cancer cells. Hence, KRAS is associated with activation of ERK2, induction of FASN, and promotion of lipogenesis. FASN may be a unique target for KRAS-associated lung adenocarcinoma remediation.