Aliaksei Holik

Why weight? Modelling sample and observational level variability improves power in RNA-seq analyses

Variations in sample quality are frequently encountered in small RNA-sequencing experiments, and pose a major challenge in a differential expression analysis. Removal of high variation samples reduces noise, but at a cost of reducing power, thus limiting our ability to detect biologically meaningful changes. Similarly, retaining these samples in the analysis may not reveal any statistically significant changes due to the higher noise level. A compromise is to use all available data, but to down-weight the observations from more variable samples. We describe a statistical approach that facilitates this by modelling heterogeneity at both the sample and observational levels as part of the differential expression analysis. At the sample level this is achieved by fitting a log-linear variance model that includes common sample-specific or group-specific parameters that are shared between genes. The estimated sample variance factors are then converted to weights and combined with observational level weights obtained from the mean–variance relationship of the log-counts-per-million using ‘voom’. A comprehensive analysis involving both simulations and experimental RNA-sequencing data demonstrates that this strategy leads to a universally more powerful analysis and fewer false discoveries when compared to conventional approaches. This methodology has wide application and is implemented in the open-source ‘limma’ package.

Repression of Igf1 expression by Ezh2 prevents basal cell differentiation in the developing lung.

Epigenetic mechanisms involved in the establishment of lung epithelial cell lineage identities during development are largely unknown. Here, we explored the role of the histone methyltransferase Ezh2 during lung lineage determination. Loss of Ezh2 in the lung epithelium leads to defective lung formation and perinatal mortality. We show that Ezh2 is crucial for airway lineage specification and alveolarization. Using optical projection tomography imaging, we found that branching morphogenesis is affected in Ezh2 conditional knockout mice and the remaining bronchioles are abnormal, lacking terminally differentiated secretory club cells. Remarkably, RNA-seq analysis revealed the upregulation of basal genes in Ezh2-deficient epithelium. Three-dimensional imaging for keratin 5 further showed the unexpected presence of a layer of basal cells from the proximal airways to the distal bronchioles in E16.5 embryos. ChIP-seq analysis indicated the presence of Ezh2-mediated repressive marks on the genomic loci of some but not all basal genes, suggesting an indirect mechanism of action of Ezh2. We found that loss of Ezh2 de-represses insulin-like growth factor 1 (Igf1) expression and that modulation of IGF1 signaling ex vivo in wild-type lungs could induce basal cell differentiation. Altogether, our work reveals an unexpected role for Ezh2 in controlling basal cell fate determination in the embryonic lung endoderm, mediated in part by repression of Igf1 expression. SUMMARY: The histone methyltransferase Ezh2 inhibits basal cell differentiation in the mouse lung by depositing repressive marks on the promoter region of basal cell genes and by repressing Igf1 expression.

Cisplatin Increases Sensitivity to FGFR Inhibition in Patient-Derived Xenograft Models of Lung Squamous Cell Carcinoma

Lung squamous cell carcinoma (SqCC) is a molecularly complex and genomically unstable disease. No targeted therapy is currently approved for lung SqCC, although potential oncogenic drivers of SqCC have been identified, including amplification of the fibroblast growth factor receptor 1 (FGFR1). Reports from a recently completed clinical trial indicate low response rates in patients treated with FGFR tyrosine kinase inhibitors, suggesting inadequacy of FGFR1 amplification as a biomarker of response, or the need for combination treatment. We aimed to develop accurate models of lung SqCC and determine improved targeted therapies for these tumors. We show that detection of FGFR1 mRNA by RNA in situ hybridization is a better predictor of response to FGFR inhibition than FGFR1 gene amplification using clinically relevant patient-derived xenograft (PDX) models of lung SqCC. FGFR1-overexpressing tumors were observed in all histologic subtypes of non–small cell lung cancers (NSCLC) as assessed on a tissue microarray, indicating a broader range of tumors that may respond to FGFR inhibitors. In FGFR1-overexpressing PDX tumors, we observed increased differentiation and reduced proliferation following FGFR inhibition. Combination therapy with cisplatin was able to increase tumor cell death, and dramatically prolonged animal survival compared to single-agent treatment. Our data suggest that FGFR tyrosine kinase inhibitors can benefit NSCLC patients with FGFR1-overexpressing tumors and provides a rationale for clinical trials combining cisplatin with FGFR inhibitors. Mol Cancer Ther; 16(8); 1610–22. ©2017 AACR.

RNA-seq mixology: designing realistic control experiments to compare protocols and analysis methods

Abstract Carefully designed control experiments provide a gold standard for benchmarking different genomics research tools. A shortcoming of many gene expression control studies is that replication involves profiling the same reference RNA sample multiple times. This leads to low, pure technical noise that is atypical of regular studies. To achieve a more realistic noise structure, we generated a RNA-sequencing mixture experiment using two cell lines of the same cancer type. Variability was added by extracting RNA from independent cell cultures and degrading particular samples. The systematic gene expression changes induced by this design allowed benchmarking of different library preparation kits (standard poly-A versus total RNA with Ribozero depletion) and analysis pipelines. Data generated using the total RNA kit had more signal for introns and various RNA classes (ncRNA, snRNA, snoRNA) and less variability after degradation. For differential expression analysis, voom with quality weights marginally outperformed other popular methods, while for differential splicing, DEXSeq was simultaneously the most sensitive and the most inconsistent method. For sample deconvolution analysis, DeMix outperformed IsoPure convincingly. Our RNA-sequencing data set provides a valuable resource for benchmarking different protocols and data pre-processing workflows. The extra noise mimics routine lab experiments more closely, ensuring any conclusions are widely applicable.

The LIM-domain only protein 4 contributes to lung epithelial cell proliferation but is not essential for tumor progression

BackgroundThe lung is constantly exposed to environmental challenges and must rapidly respond to external insults. Mechanisms involved in the repair of the damaged lung involve expansion of different epithelial cells to repopulate the injured cellular compartment. However, factors regulating cell proliferation following lung injury remain poorly understood. Here we studied the role of the transcriptional regulator Lmo4 during lung development, in the regulation of adult lung epithelial cell proliferation following lung damage and in the context of oncogenic transformation.MethodsTo study the role of Lmo4 in embryonic lung development, lung repair and tumorigenesis, we used conditional knock-out mice to delete Lmo4 in lung epithelial cells from the first stages of lung development. The role of Lmo4 in lung repair was evaluated using two experimental models of lung damage involving chemical and viral injury. The role of Lmo4 in lung tumorigenesis was measured using a mouse model of lung adenocarcinoma in which the oncogenic K-Ras protein has been knocked into the K-Ras locus. Overall survival difference between genotypes was tested by log rank test. Difference between means was tested using one-way ANOVA after assuring that assumptions of normality and equality of variance were satisfied.ResultsWe found that Lmo4 was not required for normal embryonic lung morphogenesis. In the adult lung, loss of Lmo4 reduced epithelial cell proliferation and delayed repair of the lung following naphthalene or flu-mediated injury, suggesting that Lmo4 participates in the regulation of epithelial cell expansion in response to cellular damage. In the context of K-RasG12D-driven lung tumor formation, Lmo4 loss did not alter overall survival but delayed initiation of lung hyperplasia in K-RasG12D mice sensitized by naphthalene injury. Finally, we evaluated the expression of LMO4 in tissue microarrays of early stage non-small cell lung cancer and observed that LMO4 is more highly expressed in lung squamous cell carcinoma compared to adenocarcinoma.ConclusionsTogether these results show that the transcriptional regulator Lmo4 participates in the regulation of lung epithelial cell proliferation in the context of injury and oncogenic transformation but that Lmo4 depletion is not sufficient to prevent lung repair or tumour formation.

Transcriptome and H3K27 tri-methylation profiling of Ezh2-deficient lung epithelium.

The adaptation of the lungs to air breathing at birth requires the fine orchestration of different processes to control lung morphogenesis and progenitor cell differentiation. However, there is little understanding of the role that epigenetic modifiers play in the control of lung development. We found that the histone methyl transferase Ezh2 plays a critical role in lung lineage specification and survival at birth. We performed a genome-wide transcriptome study combined with a genome-wide analysis of the distribution of H3K27 tri-methylation marks to interrogate the role of Ezh2 in lung epithelial cells. Lung cells isolated from Ezh2-deficient and control mice at embryonic day E16.5 were sorted into epithelial and mesenchymal populations based on EpCAM expression. This enabled us to dissect the transcriptional and epigenetic changes induced by the loss of Ezh2 specifically in the lung epithelium. Here we provide a detailed description of the analysis of the RNA-seq and ChIP-seq data, including quality control, read mapping, differential expression and differential binding analyses, as well as visualisation methods used to present the data. These data can be accessed from the Gene Expression Omnibus database (super-series accession number GSE57393).

Dual inhibition of BCL-XL and MCL-1 is required to induce tumour regression in lung squamous cell carcinomas sensitive to FGFR inhibition

Genetic alterations in the fibroblast growth factor receptors (FGFRs) have been described in multiple solid tumours including bladder cancer, head and neck and lung squamous cell carcinoma (SqCC). However, recent clinical trials showed limited efficacy of FGFR-targeted therapy in lung SqCC, suggesting combination therapy may be necessary to improve patient outcomes. Here we demonstrate that FGFR therapy primes SqCC for cell death by increasing the expression of the pro-apoptotic protein BIM. We therefore hypothesised that combining BH3-mimetics, potent inhibitors of pro-survival proteins, with FGFR-targeted therapy may enhance the killing of SqCC cells. Using patient-derived xenografts and specific inhibitors of BCL-2, BCL-XL, and MCL-1, we identified a greater reliance of lung SqCC cells on BCL-XL and MCL-1 compared to BCL-2 for survival. However, neither BCL-XL nor MCL-1 inhibitors alone provided a survival benefit in combination FGFR therapy in vivo. Only triple BCL-XL, MCL-1, and FGFR inhibition resulted in tumour volume regression and prolonged survival in vivo, demonstrating the ability of BCL-XL and MCL-1 proteins to compensate for each other in lung SqCC. Our work therefore provides a rationale for the inhibition of MCL-1, BCL-XL, and FGFR1 to maximize therapeutic response in FGFR1-expressing lung SqCC.