Parallel reverse genetic screening in mutant human cells using transcriptomics

Detecting subtle transcriptomic perturbations induced by lncRNAs knock-down in single-cell CRISPRi screening using a new sparse supervised autoencoder neural network

Single-cell CRISPR-based transcriptome screens are potent genetic tools for concomitantly assessing the expression profiles of cells targeted by a set of guides RNA (gRNA), and inferring target gene functions from the observed perturbations. However, due to various limitations, this approach lacks sensitivity in detecting weak perturbations and is essentially reliable when studying master regulators such as transcription factors. To overcome the challenge of detecting subtle gRNA induced transcriptomic perturbations and classifying the most responsive cells, we developed a new supervised autoencoder neural network method. Our Sparse supervised autoencoder (SSAE) neural network provides selection of both relevant features (genes) and actual perturbed cells. We applied this method on an in-house single-cell CRISPR-interference-based (CRISPRi) transcriptome screening (CROP-Seq) focusing on a subset of long non-coding RNAs (lncRNAs) regulated by hypoxia, a condition that promote tumor aggressiveness and drug resistance, in the context of lung adenocarcinoma (LUAD). The CROP-seq library of validated gRNA against a subset of lncRNAs and, as positive controls, HIF1A and HIF2A, the 2 main transcription factors of the hypoxic response, was transduced in A549 LUAD cells cultured in normoxia or exposed to hypoxic conditions during 3, 6 or 24 h. We first validated the SSAE approach on HIF1A and HIF2 by confirming the specific effect of their knock-down during the temporal switch of the hypoxic response. Next, the SSAE method was able to detect stable short hypoxia-dependent transcriptomic signatures induced by the knock-down of some lncRNAs candidates, outperforming previously published machine learning approaches. This proof of concept demonstrates the relevance of the SSAE approach for deciphering weak perturbations in single-cell transcriptomic data readout as part of CRISPR-based screening.

Rabies virus P protein binds to TBK1 and interferes with the formation of innate immunity-related liquid condensates

Viruses must overcome the interferon-mediated antiviral response to replicate and propagate into their host. Rabies virus (RABV) phosphoprotein P is known to inhibit interferon induction. Here, using a global mass spectrometry approach, we show that RABV P binds to TBK1, a kinase located at the crossroads of many interferon induction pathways, resulting in innate immunity inhibition. Mutations of TBK1 phosphorylation sites abolish P binding. Importantly, we demonstrate that upon RABV infection or detection of dsRNA by innate immunity sensors, TBK1 and its adaptor proteins NAP1 and SINTBAD form dynamic cytoplasmic condensates that have liquid properties. These condensates can form larger aggregates having ring-like structures in which NAP1 and TBK1 exhibit locally restricted movement. P binding to TBK1 interferes with the formation of these structures. This work demonstrates that proteins of the signaling pathway leading to interferon induction transiently form liquid organelles that can be targeted by viruses.

Application of modular response analysis to medium- to large-size biological systems

The development of high-throughput genomic technologies associated with recent genetic perturbation techniques such as short hairpin RNA (shRNA), gene trapping, or gene editing (CRISPR/Cas9) has made it possible to obtain large perturbation data sets. These data sets are invaluable sources of information regarding the function of genes, and they offer unique opportunities to reverse engineer gene regulatory networks in specific cell types. Modular response analysis (MRA) is a well-accepted mathematical modeling method that is precisely aimed at such network inference tasks, but its use has been limited to rather small biological systems so far. In this study, we show that MRA can be employed on large systems with almost 1,000 network components. In particular, we show that MRA performance surpasses general-purpose mutual information-based algorithms. Part of these competitive results was obtained by the application of a novel heuristic that pruned MRA-inferred interactions a posteriori. We also exploited a block structure in MRA linear algebra to parallelize large system resolutions.

STAT1 gain-of-function heterozygous cell models reveal diverse interferon-signature gene transcriptional responses

Signal transducer and activator of transcription 1 (STAT1) gain-of-function (GOF) is an autosomal dominant immune disorder marked by wide infectious predisposition, autoimmunity, vascular disease, and malignancy. Its molecular hallmark, elevated phospho-STAT1 (pSTAT1) following interferon (IFN) stimulation, is seen consistently in all patients and may not fully account for the broad phenotypic spectrum associated with this disorder. While over 100 mutations have been implicated in STAT1 GOF, genotype-phenotype correlation remains limited, and current overexpression models may be of limited use in gene expression studies. We generated heterozygous mutants in diploid HAP1 cells using CRISPR/Cas9 base-editing, targeting the endogenous STAT1 gene. Our models recapitulated the molecular phenotype of elevated pSTAT1, and were used to characterize the expression of five IFN-stimulated genes under a number of conditions. At baseline, transcriptional polarization was evident among mutants compared with wild type, and this was maintained following prolonged serum starvation. This suggests a possible role for unphosphorylated STAT1 in the pathogenesis of STAT1 GOF. Following stimulation with IFNα or IFNγ, differential patterns of gene expression emerged among mutants, including both gain and loss of transcriptional function. This work highlights the importance of modeling heterozygous conditions, and in particular transcription factor-related disorders, in a manner which accurately reflects patient genotype and molecular signature. Furthermore, we propose a complex and multifactorial transcriptional profile associated with various STAT1 mutations, adding to global efforts in establishing STAT1 GOF genotype-phenotype correlation and enhancing our understanding of disease pathogenesis.

Integrative analysis of microRNA and mRNA expression profiles of monocyte-derived dendritic cells differentiation during experimental cerebral malaria

Heterogeneity and high plasticity are common features of cells from the mononuclear phagocyte system: monocytes (MOs), macrophages, and dendritic cells (DCs). Upon activation by microbial agents, MO can differentiate into MO-derived DCs (MODCs). In previous work, we have shown that during acute infection with Plasmodium berghei ANKA (PbA), MODCs become, transiently, the main CD11b+ myeloid population in the spleen (SP) and once recruited to the brain play an important role in the development of experimental cerebral malaria (ECM). Here, we isolated 4 cell populations: bone marrow (BM) MOs (BM-MOs) and SP-MOs from uninfected mice; BM inflammatory MOs (BM-iMOs) and SP-MODCs from PbA-infected mice and used a system biology approach to a holistic transcriptomic comparison and provide an interactome analysis by integrating differentially expressed miRNAs (DEMs) and their differentially expressed gene targets (DEGs) data. The Jaccard index (JI) was used for gauging the similarity and diversity among these cell populations. Whereas BM-MOs, BM-iMOs, and SP-MOs presented high similarity of DEGs, SP-MODCs distinguished by showing a greater number of DEGs. Moreover, functional analysis identified an enrichment in canonical pathways, such as DC maturation, neuroinflammation, and IFN signaling. Upstream regulator analysis identified IFNγ as the potential upstream molecule that can explain the observed DEMs-Target DEGs intersections in SP-MODCs. Finally, directed target analysis and in vivo/ex vivo assays indicate that SP-MODCs differentiate in the SP and IFNγ is a main driver of this process.

Incremental data integration for tracking genotype-disease associations

Functional annotation of genes remains a challenge in fundamental biology and is a limiting factor for translational medicine. Computational approaches have been developed to process heterogeneous data into meaningful metrics, but often do not address how findings might be updated when new evidence comes to light. To address this challenge, we describe requirements for a framework for incremental data integration and propose an implementation based on phenotype ontologies and Bayesian probability updates. We apply the framework to quantify similarities between gene annotations and disease profiles. Within this scope, we categorize human diseases according to how well they can be recapitulated by animal models and quantify similarities between human diseases and mouse models produced by the International Mouse Phenotyping Consortium. The flexibility of the approach allows us to incorporate negative phenotypic data to better prioritize candidate genes, and to stratify disease mapping using sex-dependent phenotypes. All our association scores can be updated and we exploit this feature to showcase integration with curated annotations from high-precision assays. Incremental integration is thus a suitable framework for tracking functional annotations and linking to complex human pathology.

Human diseases are often caused or influenced by genetic factors. The link between a particular gene and a specific disease is well-established in some cases. However, the roles of many genes are still unclear and many diseases do not have an understood genetic mechanism. Dissecting such interactions requires using a range of experimental approaches and assessing the results in a holistic manner. Computational methods already exist for comparing phenotypes observed in models and patients, and they work well when the phenotypes are detailed. In this work we argue that algorithms should also be able to report meaningful assessments based on preliminary data, and to update reports in a coherent manner when new information comes to light. These requirements lead to specific mathematical properties, which define incremental integration. We implement these requirements in a computational framework. We study the extent individual rare human diseases might be recapitulated by animal models. We compute gene-disease associations using data from public resources, including previously unused negative data. Altogether, these examples illustrate the framework can use observations in model systems to track gene-disease associations in the human context.

Discovery of a Highly Selective Cell-Active Inhibitor of the Histone Lysine Demethylases KDM2/7

Histone lysine demethylases (KDMs) are of critical importance in the epigenetic regulation of gene expression, yet there are few selective, cell-permeable inhibitors or suitable tool compounds for these enzymes. We describe the discovery of a new class of inhibitor that is highly potent towards the histone lysine demethylases KDM2A/7A. A modular synthetic approach was used to explore the chemical space and accelerate the investigation of key structure-activity relationships, leading to the development of a small molecule with around 75-fold selectivity towards KDM2A/7A versus other KDMs, as well as cellular activity at low micromolar concentrations.

A small cassette enables conditional gene inactivation by CRISPR/Cas9

The availability of CRISPR/Cas9 technology has enabled the rapid establishment of gene knockouts in many cell types and even whole organisms. However, conditional inactivation of essential genes remains a challenge. We devised an approach named DECAI (DEgradation based on Cre-regulated- Artificial Intron). It utilizes a small cassette of just 201 nucleotides that is inserted into the coding exon of a target gene using CRISPR/Cas9 technology and homology-directed repair. As its sequence is derived from an artificial intron, the cassette is removed by the splicing machinery and thus leaves no trace in the "off-state". Upon activation with Cre recombinase ("on-state"), the intron is crippled and the target gene is disrupted by a series of stop codons. We exemplify the utility of this approach on several non-essential and essential human genes. Clones bearing the conditional knockout cassette are recovered at frequencies above 5% and cassette function can be traced at the genomic DNA and the mRNA level. Importantly, cassette activation leads to loss of gene expression as judged by flow cytometry, Western blot or immunofluorescence. Altogether, this highlights the broad utility of the approach for conditional gene inactivation and suggests that this tool could be used to study the loss-of-function phenotypes of essential genes.

Pooled CRISPR screening with single-cell transcriptome readout

CRISPR-based genetic screens are accelerating biological discovery, but current methods have inherent limitations. Widely used pooled screens are restricted to simple readouts including cell proliferation and sortable marker proteins. Arrayed screens allow for comprehensive molecular readouts such as transcriptome profiling, but at much lower throughput. Here we combine pooled CRISPR screening with single-cell RNA sequencing into a broadly applicable workflow, directly linking guide RNA expression to transcriptome responses in thousands of individual cells. Our method for CRISPR droplet sequencing (CROP-seq) enables pooled CRISPR screens with single-cell transcriptome resolution, which will facilitate high-throughput functional dissection of complex regulatory mechanisms and heterogeneous cell populations.

Parallel reverse genetic screening in mutant human cells using transcriptomics

Reverse genetic screens have driven gene annotation and target discovery in model organisms. However, many disease‐relevant genotypes and phenotypes cannot be studied in lower organisms. It is therefore essential to overcome technical hurdles associated with large‐scale reverse genetics in human cells. Here, we establish a reverse genetic approach based on highly robust and sensitive multiplexed RNA sequencing of mutant human cells. We conduct 10 parallel screens using a collection of engineered haploid isogenic cell lines with knockouts covering tyrosine kinases and identify known and unexpected effects on signaling pathways. Our study provides proof of concept for a scalable approach to link genotype to phenotype in human cells, which has broad applications. In particular, it clears the way for systematic phenotyping of still poorly characterized human genes and for systematic study of uncharacterized genomic features associated with human disease.