Poly(A)-tail profiling reveals an embryonic switch in translational control

Ribosomal protein and other housekeeping genes are enriched among short-tailed genes in yeast.

“This enrichment was strong in yeast, despite previous reports that ribosomal-protein genes tend to have long tails10,11.”

• cites [10] Beilharz, T. H. & Preiss, T. Widespread use of poly(A) tail length control to accentuate expression of the yeast transcriptome. RNA 13, 982–997 (2007) sourceNot checked yet

  Not checked in the free run.

• cites [11] Lackner, D. H. et al. A network of multiple regulatory layers shapes gene expression in fission yeast. Mol. Cell 26, 145–155 (2007) sourceNot checked yet

  Not checked in the free run.

At 6 hpf in zebrafish, ribosomal protein mRNAs switched from being enriched in shorter-tail genes to being enriched in longer-tail genes.

“At 6 hpf in zebrafish, ribosomal protein mRNAs had switched from being enriched in shorter-tail genes to being enriched in longer-tail genes (Extended Data Table 2), perhaps because these were mostly newly synthesized transcripts, which tended to have longer tails at this stage (Extended Data Fig. 5).”

• cites Extended Data Table 2Not checked yet

  Not checked in the free run.

• cites Extended Data Fig. 5Not checked yet

  Not checked in the free run.

The types of genes with shorter or longer tails differ between embryonic samples and other samples.

“The types of genes with shorter or longer tails differed between the embryonic samples and the other samples (Extended Data Table 2).”

• cites Extended Data Table 2Not checked yet

  Not checked in the free run.

Different mRNA transcripts from the same gene can have different poly(A)-tail lengths.

“Variation was also observed for different mRNA transcripts from the same gene (Fig. 2c).”

mRNAs encoding ribosomal proteins and other housekeeping proteins tend to have shorter poly(A) tails.

“Poly(A)-tail lengths were conserved between orthologous mRNAs, with mRNAs encoding ribosomal proteins and other “housekeeping” proteins tending to have shorter tails.”

• Also stated in: § Tail-length diversity within each species · boosted

  “the most striking and pervasive enrichment was for ribosomal protein and other ‘housekeeping’ genes among the short-tailed genes (Extended Data Table 2).”

• cites Extended Data Table 2Not checked yet

  Not checked in the free run.

Poly(A)-tail lengths are conserved between orthologous mRNAs.

“Poly(A)-tail lengths were conserved between orthologous mRNAs, with mRNAs encoding ribosomal proteins and other “housekeeping” proteins tending to have shorter tails.”

• Also stated in: § Tail-length diversity within each species · hedged

  “Comparison of tail lengths for orthologous genes in human (HeLa and HEK293T) and mouse (3T3 and liver) cells revealed moderately strong correlations, indicating that tail lengths are conserved [Extended Data Table 1, Spearman R (Rs) as high as 0.46].”

• cites Extended Data Table 1Not checked yet

  Not checked in the free run.

Independent measurements of poly(A)-tail lengths for eight yeast genes, including four ribosomal protein genes, are more consistent with PAL-seq measurements than with previous PASTA measurements.

“we used an independent method to measure the poly(A)-tail lengths of eight yeast genes, including four ribosomal protein genes. The results were much more consistent with our measurements than with the previous measurements (Extended Data Fig. 3 and 4).”

• cites Extended Data Fig. 3 and 4Not checked yet

  Not checked in the free run.

In early zebrafish and frog embryos, poly(A)-tail lengths are coupled to translational efficiency.

“As expected, tail lengths were coupled to translational efficiency in early zebrafish and frog embryos.”

• Also stated in: § Transient coupling of tail length and translation in embryos · boosted

  “We found that in early embryos (cleavage and blastula stages) of both fish and frog, mean poly(A)-tail length correlated strongly with TE (Fig. 3a, Rs from 0.62 to 0.77).”

Poly(A)-tail lengths increase through cleavage, blastula, and gastrula stages of zebrafish and analogous stages of frog development.

“Poly(A)-tail lengths increased when progressing through cleavage, blastula, and gastrula stages of zebrafish embryonic development [2, 4 and 6 h post-fertilization (hpf), respectively] and analogous stages of frog development (Fig. 2a,b and d).”

Within each species, mean poly(A)-tail lengths vary among genes, with the 10th and 90th percentiles differing by 1.4- to 1.6-fold.

“For genes within each species, mean tail lengths varied, with the 10th and 90th percentiles differing by 1.4- to 1.6-fold.”

Similar differences between mammalian, fly, plant, and yeast cells are observed when comparing tail-length averages for individual genes.

“Similar differences between mammalian, fly, plant, and yeast cells were observed when comparing tail-length averages for individual genes (Fig. 2b).”

In early zebrafish embryos, the relationship between poly(A)-tail length and tandem 3′-UTR isoform length is more pronounced when a predicted cytoplasmic polyadenylation element (CPE) is present in the unique region of the longer isoform.

“In early zebrafish embryos this relationship between 3′-UTR isoforms was even more pronounced when a predicted cytoplasmic polyadenylation element (CPE)4,19 was present in the unique region of the longer isoform (Extended Data Fig. 6b).”

• cites [4] Richter, J. D. Cytoplasmic polyadenylation in development and beyond. Microbiol. Mol. Biol. Rev. 63, 446–456 (1999) sourceNot checked yet

  Not checked in the free run.

• cites [19] Aanes, H. et al. Zebrafish mRNA sequencing deciphers novelties in transcriptome dynamics during maternal to zygotic transition. Genome Res. 21, 1328–1338 (2011) sourceNot checked yet

  Not checked in the free run.

• cites Extended Data Fig. 6bNot checked yet

  Not checked in the free run.

Intragenic comparisons reveal a weak positive relationship between poly(A)-tail length and the length of tandem 3′-UTR isoforms.

“Support for the latter relationship was also observed in intragenic comparisons, which revealed a weak positive relationship between tail length and the length of tandem 3′-UTR isoforms (Extended Data Fig. 6a).”

• cites Extended Data Fig. 6aNot checked yet

  Not checked in the free run.

In 2 hpf zebrafish, a 2-fold increase in tail length corresponds to a greater than 6-fold increase in translational efficiency when doubling the tail from 20 to 40 nucleotides.

“In these early embryonic stages a 2-fold increase in tail length corresponded to a large increase in TE—greater than 6-fold when doubling the tail from 20 to 40 nucleotides in 2 hpf zebrafish (Fig. 3a).”

In early embryos (cleavage and blastula stages) of both fish and frog, mean poly(A)-tail length correlates strongly with translational efficiency (Rs from 0.62 to 0.77).

“We found that in early embryos (cleavage and blastula stages) of both fish and frog, mean poly(A)-tail length correlated strongly with TE (Fig. 3a, Rs from 0.62 to 0.77).”

Poly(A)-tail length and mRNA half-life are variably correlated in yeast depending on the source of half-life measurements (Rs from −0.44 to 0.23).

“Tail-length and half-life were slightly negatively correlated in HeLa and 3T3 cells (Rs = −0.048 and −0.16, respectively) and variably correlated in yeast, depending on the source of the half-life measurements (Rs = ranging from −0.44 to 0.23).”

• cites Extended Data Table 1Not checked yet

  Not checked in the free run.

Poly(A)-tail length and mRNA half-life are slightly negatively correlated in HeLa and 3T3 cells (Rs = −0.048 and −0.16, respectively).

“Tail-length and half-life were slightly negatively correlated in HeLa and 3T3 cells (Rs = −0.048 and −0.16, respectively) and variably correlated in yeast, depending on the source of the half-life measurements (Rs = ranging from −0.44 to 0.23).”

• cites Extended Data Table 1Not checked yet

  Not checked in the free run.

No strong, easily interpretable correlations were observed between poly(A)-tail length and mRNA features (3′ UTR length, ORF length, total length, splice-site number, splice-site density) or expression (steady-state accumulation and nuclear-to-cytoplasmic ratio).

“No strong, easily interpretable correlations between tail length and mRNA features (3′ UTR length, ORF length, total length, splice-site number, splice-site density) or expression (steady-state accumulation and nuclear-to-cytoplasmic ratio) were observed (Extended Data Table 1).”

• cites Extended Data Table 1Not checked yet

  Not checked in the free run.

The strong coupling between tail length and translational efficiency observed in blastula largely disappears in gastrulating embryos (Rs = 0.13 for both fish and frog).

“The strong coupling observed in the blastula largely disappeared in gastrulating embryos (Fig. 3a, Rs = 0.13 for both fish and frog).”

In 3T3 cells, the slopes of mean tail-length versus number of bound ribosomes are generally small and mostly slightly negative.

“The slopes of these lines were generally small, and most were slightly negative (Fig. 4b); positive slopes would have been expected if longer tails enhanced translation.”

The disappearance of coupling at gastrulation is not because of the more restricted tail-length range observed at gastrulation.

“This disappearance was not because of the more restricted tail-length range observed at gastrulation (Extended Data Fig. 7b).”

• cites Extended Data Fig. 7bNot checked yet

  Not checked in the free run.

There is no positive correlation of meaningful magnitude between mean poly(A)-tail length and translational efficiency in HeLa cells, HEK293T cells, 3T3 cells, mouse liver, budding yeast, or fission yeast (Rs = −0.10, 0.07, −0.04, 0.00, −0.12, and −0.15, respectively).

“we observed no positive correlation of a meaningful magnitude between mean poly(A)-tail length and TE in HeLa cells, HEK293T cells, 3T3 cells, mouse liver, budding yeast, or fission yeast (Rs = −0.10, 0.07, −0.04, 0.00, −0.12, and −0.15, respectively) (Fig. 3b).”

Injecting miR-155 into one-cell zebrafish embryos causes RPFs from many predicted targets to decrease relative to RPFs from no-site control mRNAs.

“Injecting miR-155 caused RPFs from many of its predicted targets to decrease relative to RPFs from no-site control mRNAs (Fig. 5a).”

At 6 hpf, decreases in RPFs from miR-155 targets were accompanied by nearly commensurate mRNA reductions.

“In contrast, decreases in RPFs were accompanied by nearly commensurate mRNA reductions at 6 hpf, indicating that by this stage the outcome of repression had shifted to mostly mRNA destabilization (Fig. 5a).”

At 2 and 4 hpf, miR-155 target mRNA levels did not change relative to controls despite decreased RPFs.

“Despite the decrease in RPFs, target mRNA levels did not change relative to the controls at 2 and 4 hpf, indicating that at these stages miR-155 targeting caused mostly translational repression.”

Analogous results to miR-155 were obtained after injecting miR-132.

“Analogous results were obtained after injecting a different miRNA, miR-132 (Fig. 5c, Extended Data Fig. 8).”

• cites Extended Data Fig. 8Not checked yet

  Not checked in the free run.

Predicted miR-155 targets had shortened tails at 2 and 4 hpf.

“Predicted miR-155 targets had shortened tails at 2 and 4 hpf (Fig. 5b), which explained most of the miRNA-induced translational repression observed at these stages (Fig. 5c).”

• AssumptionRests on an assumption: The predicted miR-155 targets are bona fide targets of miR-155.

Despite their zygotic transcription, late-blastula embryos still coupled tail length with translation.

“Despite their zygotic transcription, late-blastula embryos still coupled tail length with translation.”

The measured poly(A)-tail lengths were similar between the two RNA preparation types.

“The measured lengths in both types of RNA preparation were similar, despite the possibility that total RNA might have included more long-tailed mRNAs due to a population of nascent mRNAs that had full-length tails and were awaiting export.”

• AssumptionThe authors state this as an assumption: Not all nuclear mRNAs have full-length tails, and the nuclear population of mRNAs awaiting export is a small fraction of cellular mRNAs.

In 3T3 cells, 1.0 RPKM in the authors' RNA-seq experiment corresponded to about 0.2 mRNA molecules per cell, based on comparison to published mRNA abundances.

“This estimate was conservative, in that a comparison to published mRNA abundances in 3T3 cells31 indicated that 1.0 RPKM from our experiment corresponded to about 0.2 mRNA molecules per 3T3 cell.”

• cites [31] Schwanhäusser, B. et al. Global quantification of mammalian gene expression control. Nature 473, 337–342 (2011) sourceNot checked yet

  Not checked in the free run.

There is a rapid developmental switch in the nature of translational control.

“However, this strong coupling diminished at gastrulation and was absent in non-embryonic samples, indicating a rapid developmental switch in the nature of translational control.”

PAL-seq accurately measures individual poly(A) tails of any physiological length.

“We developed a high-throughput sequencing method that accurately measures individual poly(A) tails of any physiological length (Fig. 1a).”

PAL-seq was applied to measure poly(A)-tail lengths of millions of individual RNAs from yeasts, cell lines, Arabidopsis leaves, mouse liver, and zebrafish and frog embryos.

“Here, we describe poly(A)-tail length profiling by sequencing (PAL-seq) and apply it to measure tail lengths of millions of individual RNAs isolated from yeasts, cell lines, Arabidopsis leaves, mouse liver, and zebrafish and frog embryos.”

The developmental switch in translational control explains why the predominant effect of microRNA-mediated deadenylation shifts from translational repression to mRNA destabilization.

“This switch complements an earlier switch to zygotic transcriptional control and explains why the predominant effect of microRNA-mediated deadenylation concurrently shifts from translational repression to mRNA destabilization.”

The developmental switch in translational control complements an earlier switch to zygotic transcriptional control.

“This switch complements an earlier switch to zygotic transcriptional control and explains why the predominant effect of microRNA-mediated deadenylation concurrently shifts from translational repression to mRNA destabilization.”

Beginning at gastrulation, translational control undergoes a mechanistic change that uncouples translational efficiency from poly(A)-tail length.

“Taken together, our results suggest that beginning at gastrulation, translational control undergoes a mechanistic change that uncouples TE from poly(A)-tail length.”

In all yeast and mammalian contexts examined (and presumably most other cellular contexts), mRNAs with longer poly(A)-tails are not more efficiently translated.

“This global intragenic analysis (Fig. 4b) supports the conclusion drawn from intergenic analyses (Fig. 3), that in all yeast and mammalian contexts examined (and presumably in most other cellular contexts), mRNAs with longer poly(A)-tails are not more efficiently translated.”

The only known exception to the pattern of miRNA targeting causing mostly mRNA destabilization is the transient translational repression observed in early zebrafish embryos.

“The only known exception is the transient translational repression observed in early zebrafish embryos36.”

• cites [36] Bazzini, A. A., Lee, M. T. & Giraldez, A. J. Ribosome profiling shows that miR-430 reduces translation before causing mRNA decay in zebrafish. Science 336, 233–237 (2012) sourceNot checked yet

  Not checked in the free run.

Increasing average poly(A)-tail length over the physiological range does not enhance translation in yeast and mammalian cells.

“The simplest interpretation of the weak or negative correlations we observed between tail length and TE in yeast and mammalian cells is that increasing average tail length over the physiological range does not enhance translation in these contexts.”

The increase in median tail length between the lightest and heaviest ribosome fractions in 3T3 cells (66 and 82 nt) does not indicate that longer tails enhance translation.

“Thus, the increase in median length observed between the lightest and heaviest fractions when considering bulk tail lengths (Fig. 4a; 66 and 82 nt, respectively) did not indicate a relationship between longer tails and enhanced translation but instead might have reflected the positive correlation between ORF length and tail length observed in 3T3 cells (Extended Data Table 1; Rs = 0.36).”

• cites Extended Data Table 1Not checked yet

  Not checked in the free run.

By 6 hpf, the outcome of miR-155 repression had shifted to mostly mRNA destabilization.

“In contrast, decreases in RPFs were accompanied by nearly commensurate mRNA reductions at 6 hpf, indicating that by this stage the outcome of repression had shifted to mostly mRNA destabilization (Fig. 5a).”

The shift in miRNA regulatory outcome between 4 and 6 hpf is not specific to miR-430 or its targets.

“Thus, the shift in miRNA regulatory outcome that occurs between 4 and 6 hpf is not specific to miR-430 or its targets.”

The tail shortening of predicted miR-155 targets at 2 and 4 hpf explains most of the miRNA-induced translational repression observed at these stages.

“Predicted miR-155 targets had shortened tails at 2 and 4 hpf (Fig. 5b), which explained most of the miRNA-induced translational repression observed at these stages (Fig. 5c).”

• AssumptionRests on an assumption: The predicted miR-155 targets are bona fide targets of miR-155.

Because tail lengths were no longer strongly coupled to translational efficiency, tail-length changes did not explain the decrease in mean translational efficiency observed at 6 hpf for miR-132 predicted targets.

“Because tail lengths were no longer strongly coupled to TE (Fig. 3), tail-length changes did not explain the decrease in mean TE observed at 6 hpf for miR-132 predicted targets (Fig. 5c).”

The shift between 4 and 6 hpf reflects a transition from the unusual regulatory regime operating in pre-gastrulation embryos more than the dynamics of miRNA action.

“With respect to mechanism, the observation of this shift between 4 and 6 hpf, even though the injected miR-155 was present and active much earlier than was miR-430, indicated that the shift reflected a transition from the unusual regulatory regime operating in pre-gastrulation embryos (in which TE is sensitive to tail length) more than it reflected the dynamics of miRNA action.”

At 2 and 4 hpf, miR-155 targeting caused mostly translational repression.

“Despite the decrease in RPFs, target mRNA levels did not change relative to the controls at 2 and 4 hpf, indicating that at these stages miR-155 targeting caused mostly translational repression.”

When poly(A)-tail length is uncoupled from translational efficiency, the translational repression often detected as a minor component of overall repression arises from a mechanism different from the one that dominates pre-gastrulation.

“We conclude that when poly(A)-tail length is uncoupled from TE, the translational repression often detected as a minor component of the overall repression33–35 arises from a mechanism different from the one that dominates pre-gastrulation.”

• cites [33] Baek, D. et al. The impact of microRNAs on protein output. Nature 455, 64–71 (2008) sourceNot checked yet

  Not checked in the free run.

• cites [34] Hendrickson, D. G. et al. Concordant regulation of translation and mRNA abundance for hundreds of targets of a human microRNA. PLoS Biol. 7, e1000238 (2009) sourceNot checked yet

  Not checked in the free run.

• cites [35] Guo, H., Ingolia, N. T., Weissman, J. S. & Bartel, D. P. Mammalian microRNAs predominantly act to decrease target mRNA levels. Nature 466, 835–840 (2010) sourceNot checked yet

  Not checked in the free run.

With their tail-length distribution shifted towards shorter tails, cleavage-stage embryos can most efficiently exploit the tail-length differences with the greatest impact.

“With their tail-length distribution also shifted towards shorter tails (Fig. 2b), cleavage-stage embryos can most efficiently exploit the tail-length differences with the greatest impact (Fig. 3a).”

To the extent that newly transcribed zygotic mRNA tended to have longer tails than maternally inherited mRNAs, the continued coupling in the hybrid state would increase relative output from newly minted mRNAs, sharpening the MZT.

“Indeed, to the extent that newly transcribed zygotic mRNA tended to have longer tails than did the maternally inherited mRNAs (Extended Data Fig. 5), the continued coupling observed in this hybrid state would act to increase the relative output from these newly minted mRNAs, thereby sharpening the MZT.”

• cites Extended Data Fig. 5Not checked yet

  Not checked in the free run.

The absence or presence of coupling between poly(A)-tail length and translational efficiency can be rationalized by the potential interplay among regulatory options available in the two regulatory regimes.

“This absence or presence of coupling between poly(A)-tail length and TE can be rationalized in light of the potential interplay among regulatory options available in the two regulatory regimes.”

Early embryonic cells harness differential tail length for global gene control.

“In these circumstances, early embryonic cells apparently harness differential tail length for global gene control.”

In transcriptionally active yeast, mammalian, and mid-gastrulation cells, an additional layer of control in which translational efficiency depends on poly(A)-tail length is dispensable.

“Thus, an additional layer of control in which TE depends on poly(A)-tail length is dispensable.”

The pre-gastrulation zebrafish embryo is the only known context for which translational repression is the dominant outcome of miRNA-mediated regulation.

“They also explain why the pre-gastrulation zebrafish embryo is the only known context for which translational repression is the dominant outcome of miRNA-mediated regulation; in all other contexts examined, tail-length shortening causes mRNA destabilization with little or no effect on TE.”

In the conventional regulatory regime, coupling translational efficiency to poly(A)-tail length would compromise the utility of gene regulation through mRNA stability.

“More importantly, because this type of coupling would lower output from older mRNA molecules that, in the absence of cytoplasmic polyadenylation, would often have shorter poly(A) tails, the utility of gene regulation through mRNA stability would be compromised.”

The paper's results from yeast, cultured mammalian cells, and mouse liver refute the prevailing view that poly(A)-tail length broadly influences translational efficiency.

“Our results from yeast, cultured mammalian cells, and mouse liver refute the prevailing view that poly(A)-tail length broadly influences TE.”

Additional analyses indicated that mRNA degradation did not bias against longer poly(A) tails.

“Additional analyses indicated that mRNA degradation did not bias against longer poly(A) tails (Extended Data Fig. 2a).”

• cites Extended Data Fig. 2aNot checked yet

  Not checked in the free run.

Due to the start-codon ribosome accumulation bias, genes with shorter ORFs have artifactually higher translation efficiencies when all bound ribosomes are considered, as in conventional polysome gradient analysis.

“Because of this bias, genes with shorter ORFs have artifactually higher TEs if all the bound ribosomes are considered (as in conventional polysome gradient analysis).”

The weak relationships between tail length and half-life in HeLa and 3T3 would be expected if mRNAs with different half-lives have similar steady-state tail-length distributions with less stable mRNAs transiting more quickly.

“The weak relationships in HeLa and 3T3 cells would be expected if mRNAs with different half-lives have similar steady-state tail-length distributions with the less stable mRNAs transiting through the distributions more quickly.”

The paper's results add to the known differences between the regulatory regime operating in yeast, cultured mammalian cells, and mouse liver and that in early metazoan embryos.

“In doing so, they add to the known differences between the regulatory regime operating in these cells and that in early metazoan embryos.”

The primer-extension step quantitatively marks each cluster with biotin in proportion to the length of the poly(A) tail.

“This key step quantitatively marks each cluster with biotin in proportion to the length of the poly(A) tail (Fig. 1a, step 11).”

The cycloheximide effect may have distorted TE measurements in studies that used polysome gradient fractionation followed by microarray analysis, including those reporting a positive link between ribosome density and poly(A)-tail length.

“This cycloheximide effect might have distorted the TE measurements in studies that calculated ribosome densities using polysome gradient fractionation followed by microarray analysis, including those reporting a positive correlation between ribosome density and poly(A)-tail length10,11.”

• cites [10] Beilharz, T. H. & Preiss, T. Widespread use of poly(A) tail length control to accentuate expression of the yeast transcriptome. RNA 13, 982–997 (2007) sourceNot checked yet

  Not checked in the free run.

• cites [11] Lackner, D. H. et al. A network of multiple regulatory layers shapes gene expression in fission yeast. Mol. Cell 26, 145–155 (2007) sourceNot checked yet

  Not checked in the free run.

The cycloheximide effect could not have influenced the conclusions of the authors' polysome-gradient experiment because the analysis focused on intragenic comparisons.

“However, this effect could not have influenced the conclusions of our polysome-gradient experiment, because our analysis focused on intragenic comparisons (Fig. 4b).”