Transcriptome-wide effects of inverted SINEs on gene expression and their impact on RNA polymerase II activity

Abstract

Background: Short interspersed elements (SINEs) represent the most abundant group of non-long-terminal repeat transposable elements in mammalian genomes. In primates, Alu elements are the most prominent and homogenous representatives of SINEs. Due to their frequent insertion within or close to coding regions, SINEs have been suggested to play a crucial role during genome evolution. Moreover, Alu elements within mRNAs have also been reported to control gene expression at different levels.

Results: Here, we undertake a genome-wide analysis of insertion patterns of human Alus within transcribed portions of the genome. Multiple, nearby insertions of SINEs within one transcript are more abundant in tandem orientation than in inverted orientation. Indeed, analysis of transcriptome-wide expression levels of 15 ENCODE cell lines suggests a cis-repressive effect of inverted Alu elements on gene expression. Using reporter assays, we show that the negative effect of inverted SINEs on gene expression is independent of known sensors of double-stranded RNAs. Instead, transcriptional elongation seems impaired, leading to reduced mRNA levels.

Conclusions: Our study suggests that there is a bias against multiple SINE insertions that can promote intramolecular base pairing within a transcript. Moreover, at a genome-wide level, mRNAs harboring inverted SINEs are less expressed than mRNAs harboring single or tandemly arranged SINEs. Finally, we demonstrate a novel mechanism by which inverted SINEs can impact on gene expression by interfering with RNA polymerase II.

Keywords: ADAR; Alu elements; Double-stranded RNA; Gene regulation; RNA Pol II; RNA editing; SINE; Transcription.

Fig. 1

iSINEs are less abundant than dSINEs in the human genome. a The proportional distribution of genome and “Aluome” to genic, intergenic, exonic and intronic partitions and different types of exons are shown. Alu elements are almost equally distributed to genic and intergenic regions. Within genic regions, however, a strong accumulation in non-coding regions, such as introns, 3′ UTRs, or non-coding RNAs, can be observed. b Comparison of the number of Alus in the genome in genic and intergenic regions. A large fraction of Alus accumulates in clusters, which are defined as a set of neighboring Alus that are at most 300 nucleotides apart from one another. Single SINEs (sSINE) thus have no other Alu in their vicinity. Pairs are clusters of size 2 and are further divided into dSINEs (direct tandem SINEs) and iSINEs (inverted SINEs). Based on their relative orientation to one another, iSINEs are grouped into tiSINEs (tail-to-tail) and hiSINEs (head-to-head). Clearly, iSINEs are less abundant than dSINEs

Fig. 2

iSINE-containing transcripts are less expressed in ENCODE datasets. a, b RNA-seq data for 15 different human cell lines available from the ENCODE project were analyzed and the expression level (FPKM fragments per kilobase of transcript per million mapped reads) was determined. Subsequently, the transcripts were sorted according to the presence or absence of Alu elements in exonic regions or the orientation of Alu elements. Shown is a pool of transcripts from all 15 cell lines grouped into seven categories: all = all transcripts, single = exactly one Alu element per transcript, iAlu = Alu elements in inverted orientation, head-head/tail-tail = Alu elements in sense/antisense or antisense/sense orientation, tandem = Alu elements in tandem orientation, no Alu = transcripts not containing any Alu. Horizontal numbers in the individual bars indicate the average FPKM for each class of transcript whereas the vertical numbers indicate the total number of transcripts found for each category in all 15 cell lines. b The statistical significance of differences in expression between gene sets was tested using the nonparametric Wilcoxon rank-sum test (two-sided and one-sided)

Fig. 3

iSINEs can repress reporter gene expression and RNA levels experimentally. The iSINE-containing 3′ UTRs of a the Nicn1 and b Inadl genes were inserted downstream of the firefly luciferase ORF. As controls, one of the Alu elements was flipped to make a duplicated SINE (dSINE) and as a second control one of the ALUs was removed (1SINE). Arrows show SINE orientation and absolute distances are indicated. Reporter genes harboring different SINEs derived from c the Nicn1 or d the Inadl 3′ UTRs were transfected into U2OS cells and gene expression was determined using a dual luciferase assay (DLA). Total RNA was extracted from transfected cells with e the Nicn1 or f the Inadl constructs and mRNA levels were measured using reverse transcription followed by quantitative PCR of total cDNA. Standard deviation is indicated by error bars. Asterisks indicate p values calculated with Student’s t-test: *p < 0.05, **p < 0.005, ***p < 0.0005

Fig. 4

Repression by iSINEs is dependent on secondary structures and sequence in a species-independent manner. a The iSINE-containing 3′ UTR of the Znf708 gene was inserted downstream of the firefly luciferase ORF. To generate perfect complementarity, one Alu Sg was replaced by a duplication of the Alu Sc, giving rise to a perfect inverted SINE (piSINE). A perfect tandem SINE (pdSINE) and 1SINE were used as controls and made by flipping or deleting the second SINE, respectively. b Dual luciferase assays of different SINE configurations derived from the Znf708 3′ UTR demonstrate that the reduction of gene expression correlates with the extent of double-strandedness. c–f To evaluate whether the observed reduction in gene expression is specific for iSINEs or dependent on RNA structure alone, UTRs that resemble the secondary structure of the Znf708 UTR but with different sequence context were designed. The Znf708 analogues and respective controls were transfected into U2OS cells and gene expression was quantified using a dual luciferase assay. See Additional file 1: Figure S2 for minimum free energy structures of the Znf708 UTR and the designed Znf708 analogues. g To generate reporter constructs harboring mouse SINEs, B1 elements of the mouse car5b gene were used to replace the Alu elements in Znf708. h The B1-harboring reporter genes were transfected into U2OS cells and a dual luciferase assay was performed after 24 h. Standard deviations are indicated. Asterisks indicate p values calculated with Student’s t-test: *p < 0.05, **p < 0.005, ***p < 0.0005

Fig. 5

iSINE-mediated gene repression is independent of ADARs and STAUFEN1. a To test whether human iSINEs lead to gene repression in mouse cells, Nicn1 constructs were transfected in wild-type MEFs and luciferase activity was detected after 24 h. b mRNA levels of SINE-containing reporter genes were detected using RT-qPCR. c, d Nicn1-containing constructs were transfected in Adar1 ^−/−/Adar2 ^−/− mouse cells and after 24 h c protein and d RNA levels were measured. e, f Stau1 ^−/− MEF cells were also transfected and the e protein expression and f RNA levels were measured. Standard deviation is indicated. Asterisks indicate p values calculated with Student’s t-test: *p < 0.05, **p < 0.005, ***p < 0.0005

Fig. 6

iSINEs do not destabilize mRNA but interfere with RNA polymerase II (polII). a mRNA transcription was blocked using Actinomycin D. Subsequently, mRNAs were collected 0, 1, and 2 h after transcriptional inhibition and mRNA levels were determined by RT-qPCR. b The Pol II-immunoprecipitated DNA was analyzed by real time PCR. Three different regions of the reporter gene—the 5′ coding region (A), stop codon (B), and near poly(A) signal (C)—were amplified. The amplicons are shown as grey boxes. Reporters harboring the c Nicn1 1SINE, iSINE, and dSINE or the d Znf708 1SINE, piSINE, and pdSINE were transfected in U2OS cells and the Pol II density along the genes was measured. Clearly, Pol II density decreases downstream of stable iSINEs. Error bars indicate standard deviation. Asterisks indicate p values calculated with Student’s t-test: *p < 0.05, **p < 0.005, ***p < 0.0005

Fig. 7

Pol II occupancy decreases downstream of endogenous iSINEs. a To test for Pol II occupancy we performed ChIP assays for three endogenous genes carrying a single 1SINE in their 3′ UTR and five endogenous genes carrying a paired iSINE in their last UTR exon. Pol II ChIP was performed and the co-precipitated DNA was quantified by qPCR. We amplified three different regions: region A, about 10 kb upstream of the single or inverted SINE; region B, upstream of the iSINE or 1SINE in close proximity; and region C, downstream of the iSINE or 1SINE. Whereas for the 1SINEs (b) we observed both a decrease and increase in Pol II occupancy, the Pol II occupancy clearly drops for four out of five genes carrying an iSINE (c). The average and the standard error of the mean (SEM) calculated from six biological replicates are plotted. d The average and SEM of the three 1SINE- and five iSINE-containing genes. Asterisks indicate p values (region A to region C) calculated with Student’s t-test: *p < 0.05, **p < 0.005, ***p < 0.0005; n.s. not significant