Sizing up the poly(A) tail: insights from deep sequencing

It is well established that the poly(A) tail, which decorates the 3’ end of most mature mRNAs in eukaryotes, plays important roles in mRNA stability and translation [1]. Regulation of poly(A) tail length impacts gene expression in processes such as early development, inflammation, learning and memory [2]. Although the poly(A) tail lengths of individual genes can be accurately measured using biochemical methods, genome-wide analyses have in the past yielded results with limited resolution [3, 4]. The advent of deep sequencing technologies brings a bonanza to RNA research. However, the difficulty in sequencing homopolymers (strings of identical nucleotides) [5] has hampered precise measurement of poly(A) tail length by deep sequencing. To work around this problem, Subtelny et al. and Chang et al. have recently developed two methods, named poly(A) tail length profiling by sequencing (PAL-seq) [6] and TAIL-seq [7], respectively. These two methods, both designed for the Illumina® platform, involve largely similar strategies to prepare cDNA libraries. The major difference lies at the sequencing step (summarized in Figure 1A).

Using these methods, Chang et al. examined the poly(A) tail lengths in HeLa and NIH3T3 cells, while Subtelny et al. analyzed cells or tissues from a variety of species, including yeast, plant, fly, zebrafish, xenopus, mouse and human. In line with previous reports, both studies found that different genes exhibit widely different poly(A) tail lengths, with genes in certain Gene Ontology (GO) groups tending to have shorter or longer poly(A) tails than genes in other groups. However, in contrast to prevailing perceptions [1], poly(A) tail lengths in the steady state appear to be quite short: ~30 nucleotides (nt) in yeast [6] and ~50 to 100 nt in most metazoan samples [6, 7]. Subtelny et al. additionally found a gradual lengthening of the poly(A) tail (from ~20 nt to ~50–60 nt) in early embryonic development of zebrafish and xenopus. It is notable, however, that the two studies have some discrepancies in reported poly(A) tail lengths for genes expressed in the same cell types, with the mean poly(A) lengths reported by Subtelny et al. being 20–40 nt longer than those by Chang et al. Some Gene Ontology (GO) analysis results also differ. For example, genes in the “ribosomal subunit” group were found to express transcripts with short poly(A) tails in NIH3T3 and HeLa cells by Subtelny et al., but not so by Chang et al. It is not clear whether cell conditions or technical reasons are behind these differences. Both studies found that alternative polyadenylation isoforms can have different poly(A) tail lengths; therefore, differences in defining gene 3’ ends could also cause some of the inconsistencies.

Although poly(A) tail shortening, also termed deadenylation, is the first step for the degradation of most mRNAs [8], neither study found connections between the poly(A) tail length and mRNA abundance. However, Chang et al. did observe a correlation between poly(A) tail length and mRNA half-life for relatively long mRNAs whose decay rates could be more reliably determined. Using TAIL-seq, the group also found widespread terminal uridylation (addition of uridine) and guanylation (addition of guanosine) of the poly(A) tail. Importantly, uridylation and guanylation frequencies have negative and positive correlations with mRNA half-lives, respectively, suggesting that these modifications are related to mRNA decay (Figure 1B). Although the fractions of mRNAs containing these modifications appear to be small (most genes have uridylation and guanylation in less than 10% and 5% of their transcripts, respectively), uridylation is more frequently associated with short poly(A) tails (< 25 nt) and guanylation is slightly biased to long poly(A) tails (>40 nt). Further studies are needed to determine whether it is the poly(A) tail length per seor terminal modification that determines or influences mRNA stability. Notably, poly(A) tail length was not found to correlate with mRNA half-life by Subtelny et al. Whether this discrepancy is related to uridylation or guanylation of transcripts remains to be examined.

The biggest surprise from these two studies perhaps is the finding that poly(A) tail length in most samples is not coupled with translational efficiency, as indicated by the lack of such correlation across different genes [6, 7], as well as across transcripts from the same gene [6]. Notable exceptions are samples from early developmental stages prior to or around the maternal-to-zygotic transition in zebrafish and xenopus, for which a clear correlation between poly(A) tail length and translational efficiency exists [6]. Therefore, Subtelny et al. proposed that there are two regulatory regimes concerning the impact of poly(A) tail length on translation (Figure 1B). Explaining previous observations regarding microRNA (miRNA) regulation in the zebrafish embryo [9], Subtelny et al. found that miRNAs elicit deadenylation and translational inhibition at the developmental stages when poly(A) tail length and translational efficiency are coupled (2 and 4 hours post fertilization, hpf), but induce mRNA degradation at a later stage (6 hpf). Together with previous studies showing that miRNAs in most cells function primarily by destabilizing their target mRNAs [10], this work indicates that the consequence of mRNA deadenylation elicited by miRNA depends on which regulatory regime is in play in a given cellular context (Figure 1B). Further studies are needed to elucidate the mechanistic details that differentiate the two regulatory regimes. How these mechanisms are exploited in other biological systems to mediate miRNA functions is also an open question.

In sum, despite some discrepancies that need to be reconciled by further experimentation, the studies by Subtelny et al. and Chang et al. present the first high resolution, global views of poly(A) tail length and its relevance to mRNA metabolism. Their methods provide powerful means to further unravel the functions of the poly(A) tail in gene regulation under physiological and pathological conditions.