Spurious transcription and its impact on cell function

ABSTRACT

Most RNA polymerases can initiate transcription from diverse DNA template sequences with relatively few outright sequence restraints. Recent reports have demonstrated that failure to subdue the promiscuity of RNA polymerase in vivo can severely impede cell function. This phenomenon appears common to all cell types with undesirable effects ranging from growth inhibition in prokaryotes to cancer in higher organisms. Here we discuss similarities and differences in strategies employed by cells to minimise spurious transcription across life's domains.

Introduction

Promoters, the DNA sequences that allow RNA polymerases to initiate transcription, have few absolute DNA sequence constraints; many DNA sequences can serve as a promoter for any given RNA polymerase.^1–5 As a result, promoters arise in “unexpected” genomic contexts throughout life's different domains.^6–12 For example, promoters inside coding regions, or unsuitably orientated within non-coding DNA, are common.^6–12 In some instances, such promoters are properly regulated and generate functional transcripts.^13–19 In other cases, these promoters contribute to the phenomenon of pervasive transcription, a genome-wide background of low level RNA production, which could be beneficial in some situations.^6,7 For example, spurious promoters may act as an evolutionary source of new functional RNAs.⁷ However, some unexpected promoters appear to occur by happenstance, and are either transcriptionally silenced, or generate RNA species that are rapidly turned over.^20–22 If silencing systems fail, such transcripts can be generated at high levels.^{10–12,20–22} Since the synthesis of these RNAs is usually suppressed, and because the production of such transcripts can hinder correct cell function,²¹ we will refer to the RNAs as spurious. In this point-of-view we argue that spurious transcription is unavoidable in some circumstances given the promiscuous nature of RNA polymerases and the apparent inability of natural selection to remove all chance promoters. Consequently, all cell types have evolved mechanisms to suppress spurious transcription. We will also discuss the causes and consequences of unwanted transcription in bacteria, archaea, and eukaryotes.

Controlling spurious transcription at the level of initiation

The simplest way to prevent transcription in unwanted locations is to remove DNA sequences that can function as promoters. Natural selection appears to have been moderately successful in this regard; the occurrence of promoter-like sequences is indeed reduced within genes for many organisms.^8,9,23–25 However, because the absolute sequence requirements for transcription initiation are relatively few it may not be impossible to eradicate all such sequences. For example, the housekeeping RNA polymerase in bacteria requires only a partial match to the −10 hexamer consensus (5′-TATAAT-3′) and partial matches to one of several ancillary sequences, all with an elevated A/T-content, to initiate transcription.^1,3,23 Consequently, in Escherichia coli, genes with an A/T-content exceeding 60% contain many sequences capable of driving transcription both in vivo and in vitro.^20,21,26 Even if natural selection could eventually remove such promoters, rampant horizontal gene transfer ensures the task is never complete.²⁷ A similar situation may exist in archaea where there is also a close relationship between DNA A/T-content and transcription initiation.²⁴ Indeed, it is notable that A/T-rich regions of the archaeal Methanocaldococcus jannaschii genome lacking coding potential (e.g. DNA between convergent genes) are associated with transcription.²⁴ In eukaryotes, promoter sequences can be diverse, but A/T-rich DNA sequences disrupt nucleosome formation, and a common determinant for transcription initiation is the TATA box (consensus: 5′-TATA-3′).^5,28 Hence, a TATA box alone can stimulate transcription by human RNA polymerase II.²⁹ Consistent with this, spurious transcription initiation has been observed within many genes in Saccharomyces cerevisiae, and often coincides with the occurrence of a TATA box.¹¹ Furthermore, many promoters in eukaryotes are bidirectional, and generate antisense transcripts in addition to the expected sense RNA.²² In fact, bidirectional transcription is likely the ground state of a newly evolved promoter, and directionality evolves over time, likely due to acquisition of binding sites for asymmetric transcriptional regulators.³⁰

Inhibition of spurious transcription initiation

Natural selection has clearly produced organisms where promoter-like sequences within genes have been minimised. For example, in E. coli, Sulfolobus solfataricus and Schizosaccharomyces pombe, the average A/T-content of genes is often between 5% and 9% lower than that of intergenic DNA.^31–33 Even so, additional mechanisms are required to suppress transcription from promoters not removed by evolutionary pressure.^10,12,20,21 Prokaryotes and eukaryotes utilise analogous, but evolutionarily unrelated, repressive nucleoprotein structures to silence spurious transcription initiation (Fig. 1). In E. coli, the Histone-like nucleoid structuring (H-NS) protein specifically recognises A/T-rich DNA by virtue of an arginine side chain that interacts with the narrowed minor groove of A/T-rich DNA sequence^27,34 (Fig. 1A). Interactions between DNA-bound H-NS molecules drive polymerisation of the protein and create nucleoprotein complexes capable of repressing transcription.²⁷ Consequently, deletion of hns results in uncontrolled RNA synthesis inside genes that are A/T-rich.^20,21 Although hns is not widely conserved, other bacteria express functionally related proteins that preferentially bind A/T-rich DNA, e.g. Lsr2 in mycobacteria,³⁵ MvaT/U in pseudomonads,³⁶ and Rok in Bacillus subtilis.³⁷ It is likely that these H-NS analogues also prevent spurious transcription from intragenic promoters. Interestingly, the archaeal Cbp1 protein, a factor involved in chromosome packaging, is required to prevent transcription initiation within A/T-rich CRISPR loci,¹² suggesting it functions analogously to H-NS.

Figure 1.

Prokaryotes and eukaryotes use analogous mechanisms to prevent spurious transcription. The DNA is shown as an orange line with genes and promoters represented by block and bent arrows respectively. All other components are individually labelled. (A) In prokaryotic cells, the Histone-like nucleoid structuring protein (H-NS) can supress the activity of spurious promoters within genes, and can impede transcription elongation. If the elongating RNA polymerase complex includes a spurious non-coding transcript, transcription is often subject to premature termination by Rho. Resulting transcripts can be degraded by RNases. (B) In eukaryotic cells, nucleosomes can impede access to spurious promoters. The repressive properties of nucleosomes can be enhanced by methyltransferase proteins such as Set2 (in yeast) or SetD2 (in metazoa) act on histone H3 residue K36. The yeast Nrd1-Nab3-Sen1 (NNS) complex recognises spurious RNAs, by virtue of their different sequence properties, and can prematurely terminate transcription elongation. The cleavage and polyadenylation specificity factor (CPSF) is a multiprotein complex and recognises poly(A) sites. Ultimately, such transcripts can be degraded by the exosome.

In eukaryotes as diverse as humans and yeast, nucleosome occupancy inhibits transcription initiation, and canonical promoters are typically nucleosome-depleted^38–42 (Fig. 1B). Hence, antisense transcripts can arise near to canonical promoters or within the 3′ ends of genes.^38–42 Nucleosomes also play a key role in suppressing spurious transcription by virtue of their histone modifications. For example, deletion of the gene encoding yeast Set2, which catalyzes methylation of histone H3 residue K36, allows widespread spurious transcription initiation⁴³ (Fig. 1B). Similar effects are apparent in metazoan cells lacking the homologous SetD2 protein.⁴⁴ In yeast, these phenomena appear to be mediated via differential activation of Rpd3S, a histone deacetylase complex.^43,45–48 The methylation state of H3 K36 also controls recruitment of the Isw1b chromatin remodelling complex, which works with Chd1 to prevent histone exchange and maintain chromatin structure.⁴⁹ Hence, yeast strains lacking both isw1 and chd1 have a prominent spurious transcription phenotype.⁴⁹

Epigenetic DNA modifications occur in prokaryotic and eukaryotic cells.^50,51 However, little is known about the role of such nucleic acid changes in controlling spurious transcription initiation. To date, the best characterised consequences are those identified in mouse embryonic stem cells,⁵² where intragenic methylation of CpG dinucleotides within the body of genes is required to prevent intragenic transcription initiation by RNA polymerase II.⁵¹ Curiously, recruitment of Dnmt3B, the enzyme responsible for this DNA modification, is mediated by the methylation state of histone H3 K36.⁵² Consequently, mammalian SetD2 controls histone H3 K36 methylation and co-operates with Dmnt3B to prevent spurious transcription initiation.^44,52,53 Although DNA methylation is known to influence transcription initiation in bacteria,⁵⁴ there is no evidence this modification controls unwanted transcription.

Termination of spurious transcription

Whilst all cell types take measures to block spurious transcription initiation, these inhibitory mechanisms are imperfect. Hence, bacteria and eukaryotes have each evolved mechanisms to rapidly terminate production of spurious transcripts. In both cases, the termination machinery recognises a property of spurious RNA production not associated with functional transcription. In bacteria, discrimination is based on the coupling of transcription and translation. Since appropriately positioned translation start codons and ribosome binding sites rarely occur by chance, most spurious transcripts are not translated. The Rho transcription termination factor, found in 90% of bacteria, recognises and terminates transcription of non-coding RNA⁵⁵ (Fig. 1A). Hence, chemical inhibition of Rho results in increased transcription beyond gene boundaries and within A/T-rich genes.^56–58 Interestingly, H-NS occupancy can enhance transcription termination by Rho⁵⁶ (Fig. 1A). Thus, H-NS serves a dual purpose in suppressing pervasive transcription: silencing spurious promoters, and enhancing termination of spurious transcripts. In eukaryotes, transcription and translation are not coupled. Hence, cryptic unstable transcripts arising between genes are identified by a different mechanism. For example, in mouse embryonic stem cells, spurious antisense transcripts can arise from bidirectional promoters, with the corresponding sense transcripts being functional RNAs. Poly(A) sites are enriched in the 5′ regions of the antisense RNAs, and stimulate premature termination of antisense transcripts by cleavage and polyadenylation specificity factor (CPSF) and associated proteins.⁵⁹ In contrast, binding sites for U1 snRNP are enriched in 5′ regions of the sense transcripts, and recruitment of U1 snRNP protects these RNAs from premature cleavage and polyadenylation.⁵⁹ In S. cerevisiae, the Nrd1-Nab3-Sen1 (NNS) complex has a key role⁶⁰ (Fig. 1B). To distinguish between spurious and functional transcripts, the yeast NNS complex also recognises specific nucleotide signatures in the RNA.⁶¹ Crucially, these sequences are depleted in mRNAs.⁶² In some instances, termination by the polyadenylation machinery may provide a back-up mechanism.⁶³ Sequences recognised by the polyadenylation machinery are enriched at the 3′ ends of genes in the antisense orientation, preventing read-through of spurious transcripts into genes.⁶³ In prokaryotes, intrinsic terminators downstream of genes can be bidirectional, but most are not.⁶⁴

Degradation of spurious transcripts

In both bacteria and eukaryotes, many spurious transcripts are rapidly degraded following transcription. This process is best understood in eukaryotes, where some spurious transcripts (as well as some functional transcripts) are degraded by the exosome complex. In S. cerevisiae, Nrd1 interacts with Trf4, a member of the TRAMP polyadenylation complex.⁶⁵ Thus, NNS-terminated transcripts are polyadenylated by TRAMP, which leads to degradation by the exosome⁶⁶ (Fig. 1B). There is also feedback from the exosome to the NNS complex, whereby the exosome component Rrp6 stimulates NNS-mediated transcription termination of a subset of RNAs.⁶⁷ The details of spurious transcript degradation are poorly understood in bacteria; the process has only been studied in the context of antisense RNAs. Thus, RNase III has been shown to degrade antisense RNAs in E. coli and Staphylococcus aureus,^66–68 and may target antisense RNAs paired with their cognate mRNA.^68–70 In Bacillus subtilis, RNase Y and RNase J1 play a larger role than RNase III in degradation of antisense RNAs⁷¹ (Fig. 1A).

The relationship between spurious transcription and impaired cell function

As described above, all cell types appear to permit low levels of pervasive transcription, but multiple systems exist to avoid high level production of spurious transcripts. When these control measures fail, cell function is impaired. For example, in many bacteria, deletion of hns results in a slow growth phenotype, and such strains rapidly acquire compensatory mutations to alleviate these effects.^21,72,73 The underlying mechanism involves titration of the limited RNA polymerase pool and a consequent down-regulation of housekeeping genes.²¹ Formation of R-loops following inhibition of Rho is also likely to be deleterious.⁷⁴ Adverse consequences of spurious transcription initiation or read-through in eukaryotes have also been reported.^75,76 Of particular note are observations identifying SetD2 as a tumour suppressor.^77–80 For example, loss of SetD2 activity in renal carcinoma cells causes inefficient transcription termination. As a result, transcription elongation complexes for spurious RNAs invade oncogenes and increase their expression.⁸¹ Similarly, in some melanomas, aberrant chromatin modifications are associated with intron derived RNAs and expression of a novel anaplastic lymphoma kinase isoform.⁸² Chromatin alterations, and the activation of otherwise cryptic promoters, are also common in gastric adenocarcinoma.⁸³ More anecdotally, there are many accounts of A/T-rich DNA sequences being associated with chromosome instability and the synthesis of poorly defined microRNAs.^84,85 This is significant, given the likelihood of such DNA sequences being enriched for spurious promoter elements.

Concluding remarks

The structure and function of housekeeping RNA polymerases is conserved throughout life.² In particular, RNA polymerase has a conserved propensity to initiate transcription with relatively low sequence specificity. Consequently, most organisms have evolved mechanisms to minimise the occurrence of spurious transcription (Fig. 1). In both bacteria and eukaryotes, derepression of spurious transcription leads to impaired cell function. In metazoans, this can manifest as disease. We argue that such spurious transcriptional events are an unavoidable consequence of DNA-based life where the flow of genetic information via an RNA intermediate requires a transcriptional apparatus that is unable to differentiate between promoters of functional RNAs and promoters that occur spuriously. Importantly, spurious transcription may also play a positive role, serving as a rich source for the evolution of functional transcripts.⁸⁶

Disclosure of potential conflicts of interest

No potential conflicts of interest were disclosed.

Acknowledgments

We thank Fred Winston for support and critical reading of the manuscript.