EINCR1 is an EGF inducible lincRNA overexpressed in lung adenocarcinomas

Abstract

Long non-coding RNAs are being increasingly recognised as important molecules involved in regulating a diverse array of biological functions. For example, many long non-coding RNAs have been associated with tumourigenesis and in this context their molecular functions often involves impacting on chromatin and transcriptional control processes. One important cellular control system that is often deregulated in cancer cells is the ERK MAP kinase pathway. Here we have investigated whether ERK pathway signaling in response to EGF stimulation, leads to changes in the production of long non-coding RNAs. We identify several different classes of EGF pathway-regulated lncRNAs. We focus on one of the inducible lincRNAs, EGF inducible long intergenic non-coding RNA 1 (EINCR1). EINCR1 is predominantly nuclear and shows delayed activation kinetics compared to other immediate-early EGF-inducible genes. In humans it is expressed in a tissue-specific manner and is mainly confined to the heart but it exhibits little evolutionary conservation. Importantly, in several cancers EINCR1 shows elevated expression levels which correlate with poor survival in lung adenocarcinoma patients. In the context of lung adenocarcinomas, EINCR1 expression is anti-correlated with the expression of several protein coding EGF-regulated genes. A potential functional connection is demonstrated as EINCR1 overexpression is shown to reduce the expression of EGF-regulated protein coding genes including FOS and FOSB.

Fig 1. EGF-stimulation leads to transcriptional induction of lncRNAs.

(A) Scatter plot of differential gene expression before (0 min) and after (30 min) EGF stimulation. The identities of several upregulated protein coding genes and the lincRNA EINCR1 are indicated. (B) Average profiles of sequencing read density in the 2000 bp window around the putative transcription start sites (TSS) of EGF-upregulated protein-coding genes and lincRNAs. (C) Heat map showing the sequencing read density in the 2000 bp window around the putative transcription start sites (indicated by arrows) of EGF-upregulated protein-coding genes and lincRNAs. Data are row Z-normalised. (D) Example genome browser views of RNAseq reads located around the DUSP5 (top) and EDN1 (bottom) loci. LncRNAs located immediately downstream from DUSP5 (“polymerase run on”) and on the anti-sense strand downstream from EDN1 (EINCR2) are shown. (E) Distribution of EGF-upregulated genes across gene classes defined by the ENCODE project. The percentage of transcripts falling into each class is indicated. (F) RT-qPCR analysis of the expression of the indicated up- and down-regulated lincRNAs following EGF induction for the indicated times. Data are shown according to the indicated colour coding (capped at log₂≤3.5) and are the average of three biological replicates (n = 3).

Fig 2. EINCR1 is an EGF-regulated, nuclear long intergenic non-coding RNA.

(A) Cumulative phyloCSF scores of EINCR1, GAPDH and XIST. (B) Top: Schematic of sucrose differential centrifugation of MCF7 cells in the presence of the protein synthesis inhibitor (cycloheximide; ‘freezes’ RNA on the ribosomes) or in the presence of 50 mM EDTA (dissociates ribosomal subunits). Ribosomal subunits are shown in red and RNA species in blue. Bottom: RT-qPCR analysis of relative RNA levels (after 60 min EGF stimulation) in the ribosomal (pellet; blue) and soluble (supernatant; red) fractions in the presence of cycloheximide (top) or EDTA (bottom). Data represent average ±SEM of the proportion in fractions from two biological repeats. (C) RT-qPCR of nuclear (N) and cytoplasmic (C) fractions of MCF10A cells. MCF10A cells were stimulated with EGF for 30 minutes before harvesting RNA. Individual data points from 3 independent repeats were normalized to the nuclear RNA level (taken as 1) and are shown on log₂ scale. Horizontal lines indicated mean value. * = P-value < 0.01 (t-test with multiple testing correction). (D) UCSC genome browser tracks of EINCR1 expression in MCF10A (top) in the presence and absence of EGF stimulation (nuclear RNA-seq, this study) and MCF7 (bottom) cells (cytoplasmic and nuclear fractions, ENCODE [1]).

Fig 3. EINCR1 transcription is regulated via the ERK MAPK pathway.

(A) A screenshot of the genomic features observed near the EINCR1 genomic locus compiled from UCSC genome browser data. Data for the histone modifications (H3K3me3, H3K9me3, H3K27ac, H3K27me3, H3K36me3 ChIP-seq are derived from ENCODE data from the MCF7 cell line [38]). RNA polII ChIP-seq data from HeLa cells (before and after EGF induction) are derived from [39]. The RNA-seq data are from this study. The promoter sequence of human RP11-7F17.7 (TSS±1000bp) is aligned with five primates, mouse and rat reference sequence using multiz aligner [40] (middle panel). The sequence immediately surrounding the TSS is shown at in the top panel and a putative TATA-like sequence unique to humans is indicated. The TSS is indicated by an arrow. Genomic locations are shown above each panel. (B) RT-qPCR analysis of EINCR1 expression following EGF treatment of MCF10A cells for 15 mins in the presence and absence of the MEK inhibitor U0126. n = 2, * = P-value < 0.05. (C) RT-qPCR analysis of FOS and EINCR1 after EGF stimulation of MCF10A cells for indicated times in the presence or absence of cycloheximide (CHX). Data are shown relative to the zero timepoint (taken as 1) and represent mean ± SEM from two independent repeats. (D) RT-qPCR analysis of EINCR1 after EGF stimulation of MCF10A cells for the indicated times in the presence or absence of dominant negative FOS (A-FOS). Data are shown relative to the zero timepoint (taken as 1) and represent mean ± SEM from two independent repeats.

Fig 4. EINCR1 is specifically expressed in normal human heart tissues and is up-regulated in cancer.

(A) Violin plots of the expression of RP11-7F17.7 in normal tissues [37]. (B) Heatmap showing the expression of EINCR1 and the indicated protein coding genes across different organs in human embryos [22]. Data are row-normalised to the maximum observed expression (blue scale). The maximum absolute expression of each gene is indicated by the green scale. (C) Boxplots of EINCR1 and MALAT1 expression in the indicated cancer categories. Expression in both normal (N; blue) and tumour (T; red) samples are shown. Numbers of each type of sample are provided below each cancer subtype. Median values are shown by horizontal lines. Statistically significant differences are indicated: ** = P-value <0.01; *** = P-value<0.001. (D) Kaplan-Meier plot of overall survival in patients with lung adenocarcinoma (n = 206, from the TCGA-LUAD dataset)), using sample groups with either the top or bottom 20% expression of EINCR1. Log rank probabilities between low and high expression are shown. (E) RT-qPCR analysis of EINCR1 and FOS expression after EGF stimulation of A549 cells for the indicated times. Data are shown relative to the zero timepoint (taken as 1) and represent mean ± SD (n = 3). * = P-value = 0.05.

Fig 5. High levels of EINCR1 transcription leads to changes in the expression profiles of EGF-regulated genes.

(A) Boxplots of EINCR1, FOSB and FOS expression in lung adenocarcinomas (n = 339 cancer samples from lung adenocarcinomas “not otherwise specified” [NOS], and 108 normal lung samples from TCGA data). Expression in both normal (N; blue) and tumour (T; red) samples are shown. Median values are shown by horizontal lines. Statistically significant differences are indicated: *** = P-value<0.001. (B and C) Scatterplots showing the expression of EINCR1 and either FOSB (B) or FOS (C) in lung adenocarcinomas NOS (n = 339). Samples containing either high EINCR1 but low protein coding gene expression (or vice versa) are circled. The units on each axis are the RPK10M (reads per 1 kb per 10 M reads) normalised by the maximum value of the corresponding gene and then multiplied by 10. (D) Model of the CRISPR-based upregulation system [26]. dCas9 –mutated Cas9, VP64—4x viral protein 16 transactivation domain, MBP—MS2 binding protein, p65 –transcriptional activation domain of p65. (E-I) RT-qPCR analysis of expression of EINCR1 (E) or the indicated protein coding genes (F-I) following EGF stimulation of MCF10A cells for the indicated times either in the presence of guide RNA targeting the EINCR1 promoter 218 bp upstream of the TSS (red line) or control non-targeting (NT) guides (black line). Data are shown relative to each sample in the absence of EGF (taken as 1) and are from three independent replicates (except IER3 where n = 2). * = P-value <0.05; ** = P-value <0.01; *** = P-value<0.001.