Interestingly, changes in APA did not correlate with defective in vitro cleavage activity associated with mutant alleles Figure 2B.
A likely explanation for this is that reduced cleavage activity impacted the usage of alternative sites equally and thus there was no apparent shift in the cleavage site. In contrast, in vitro polyadenylation activity did correlate with defective APA in all analysed cases. The reasons for this remain unclear but the mutants analysed with defective in vitro polyadenylation also display RNA-binding activity, which may be compromised giving rise to the defect in APA.
We observed a unique requirement for the Sen1 helicase for poly A site-choice on a subset of genes Figure 3A. Sen1 function has been well characterised in association with the NNS complex where it acts to promote transcription termination of non-coding RNA and of a small number of protein coding genes Steinmetz et al. The role played by Sen1 in alternative polyadenylation described here is distinct from the previously described roles in RNA processing because the observed sen mutant phenotype was not duplicated by nrd and nab mutations, and thus, Sen1 appeared to act independently of the RNA-binding proteins Nrd1 and Nab3.
Here, the sen allele caused shifting toward sites proximal to those utilised in the wild-type. Therefore, functional Sen1 appeared to promote the usage of downstream poly A sites and thus suppressed usage of proximal poly A sites rather than promoting termination as was previously seen with non-coding RNA transcription.
Interestingly, this activity was associated with genes that carried an above average length of the open reading frame suggesting that Sen1 may facilitate full-length expression of long genes. The mechanisms underlying Sen1 function in alternative polyadenylation remain unclear and need to be addressed in future studies.
We speculated that defective termination of antisense non-coding RNA transcription in sen mutants facilitated transcriptional interference that in turn caused an apparent suppression of the use of distal sites on the sense mRNA transcript. Such a role would, however, be expected to involve Nrd1 and Nab3 and therefore may not apply here. Another possibility may be that Sen1 helicase activity was required to resolve RNA structures that form in proximity to the RNAP II exit channel, which may provoke transcriptional pausing and promote preferential usage of proximal sites if unresolved.
Interestingly, an NNS-independent requirement for Sen1 has been reported previously in the protection of replication fork integrity that may be associated with pathology of Senataxin deficiencies Alzu et al. In the mammalian system cleavage factor II subunits have been associated with the selection of proximal sites Kamieniarz-Gdula and Proudfoot, ; Ogorodnikov et al. As no CFIm homologues are known in yeast the repressive role of the core cleavage and polyadenylation machinery may not be conserved.
Instead, the compact nature of the yeast genome with short intergenic regions may have favoured mechanisms that promote the preferential usage of proximal poly A site in order to avoid transcriptional interference between neighbouring genes Gullerova and Proudfoot, The genes that require Sen1 for selection at distal sites may form a distinct group that was characterised by long open reading frames and it will be interesting to see whether the Sen1 homologue Senataxin plays a similar role in humans.
Cordycepin effects were most comparable to shifts in pcf and ysh yeast strains. This indicated that changes in protein levels were unlikely to cause the cordycepin induced alternative polyadenylation, which was detected within 10 min of drug treatment. Our metabolomics analysis revealed accumulation of nucleotides and their biosynthetic intermediates in wild-type cells exposed to cordycepin.
An increase in the pool of cellular nucleotides may reduce cordycepin toxicity by dilution. This idea is in line with our previous observation that poly A polymerase mutants displayed cordycepin resistance due to increased levels of free ATP Holbein et al. The link between nucleotide levels and alternative polyadenylation is underscored by our observation that mycophenolic acid, which slows RNAP II transcription through reduced GTP levels, promoted usage of proximal sites and suppressed the effect of cordycepin.
Similarly, a slow rpb1 allele mirrored the effects of mycophenolic acid to suppress the effects of cordycepin on alternative polyadenylation. In addition to nucleotide levels and trans -acting factors, the structure of the chromatin template is an important parameter for RNAP II elongation. A stable nucleosome between potential cleavage positions is therefore predicted to provide a more proximal site with a kinetic advantage. Likewise, when the nucleosome depleted region is wider, the proximal site is expected to have less temporal advantage.
Consistent with this idea we showed that genes that switch their poly A site in response to cordycepin treatment differ in their chromatin landscape compared to those that retain their original poly A sites. Convergent genes that were subject to alternative polyadenylation following cordycepin treatment tended to have wider intergenic regions downstream of the coding sequence, which likely allowed for transcriptional extension without collision of the transcriptional machinery between neighbouring genes.
In contrast, no such widening of the intergenic region was observed for genes that are transcribed in tandem consistent with transcriptional interference being no issue at those genes. Taken together, our results support a model where kinetic competition governs the selection of alternative poly A sites. On the one hand, proximal sites are transcribed first and thus carry a temporal advantage. Transcription of distal sites, on the other hand, will be achieved only under conditions which delay rapid processing at proximal sites and which favour rapid elongation by RNAP II.
All samples were collected immediately by centrifugation, washed with 1 mL ice cold milliQ H 2 O containing sodium azide 0. Ten-fold serial dilutions were spotted on SC plates 0. An equivalent volume of DMSO was used as a control. Harvested cells were washed with ice cold water containing sodium azide 0. Two replicates were used for PAT-seq, six for metabolomics and five for proteomics. Cells were then resuspended in the same media with or without adenine hemisulphate and grown for 1 or 2 hr.
Total RNA extraction was carried out using the hot phenol method Schmitt et al. Samples were vortexed thoroughly and centrifuged at top speed for 5 min at room temperature. Pellets were air-dried at room temperature and resuspended in RNase-free dH 2 O. PAT-seq libraries were prepared as described previously Harrison et al. Samples were then transferred to ice for 5 min. Fragments selected were between bp and bp in length. The excised gel slices were placed inside 0.
Eluted material and gel slurry were transferred to a corning spin-x column and centrifuged at top speed for 3 min. Supernatant was then transferred to a 1.
A different TruSeq primer was used for each experimental condition. This is done to maintain the temperature during the initial stages of reverse transcription and prevent internal priming. A different TruSeq primer was used for each experimental condition and amplification as above was repeated for 10 cycles. Read counts were produced at the level of whole genes and at the level of peaks detected by the pipeline.
These counts were then used to test for differential gene expression and APA. Fitnoise is an implementation of Empirical Bayes moderated t-tests on weighted linear models as described in Smyth, Genes for which no relevant sample had at least 10 reads were removed before testing. Shifts in APA usage was analysed as previously described Turner et al.
Peaks were first assigned to genes with up to bp down-strand of the gene but not proceeding into another gene on the same strand were counted as belonging to that gene. Next, peak counts were log 2 transformed and linear models were fitted for each peak, using TMM normalisation, voom, and limma, producing as coefficients estimates of the log 2 abundance of reads at each peak in each of two experimental conditions.
In the simplest case of two peaks, s is equal to the proportion distal peak usage in B minus the proportion distal peak usage in A. Shift scores and associated standard errors are calculated for each gene with two or more peaks, then the topconfects R package Harrison et al. Confect values are only given for significantly non-zero shifts, with a False Discovery Rate of 0.
Ordering results by absolute confect values and reading down this list only as far as is desired, a False Coverage-statement Rate of 0. For mPAT experiments, we use the same s statistic as above, but with a simplified calculation method. Cell pellets obtained from 10 ml culture at OD of 0. Supernatant and cell debris were moved to a fresh tube and intermittent vortexing continued for 15 min.
The instrument was operated in both positive and negative ion mode. Parameters for the high-performance liquid chromatography and mass spectrometry analysis were as previously described Stoessel et al. Untargeted metabolomics data analysis was performed using the freely availably software packages mzMatch Scheltema et al.
A total of authentic metabolites were used for verification of retention times and to aid metabolite identification. Identification of metabolites with these standards have high confidence MSI level 1 and are highlighted yellow in Figure 5—source data 1. It is noted that one replicate from each time point was excluded during analysis due to the observation of unwanted phase separation in the samples likely due to excess residual water in the pellet.
Alternative polyadenylation was based on the cordycepin PAT-seq experiment. Where no primary peak was assigned, the gene was discarded. If the primary peak differed at t0 and t40, the gene was considered to undergo APA whereas if the primary peak remained the same after cordycepin treatment, no APA has occurred.
The reference annotation used was Ensembl Saccharomyces cerevisiae version Gene and intergenic region width were plotted as scaled density plots so that density peaks for APA and non-APA genes were comparable despite different gene numbers.
To test whether there was a significant difference between gene length in APA vs. Using the GenomicRanges package Lawrence et al.
This was not strand-specific so that intergenic regions are based on genes being on either strand. For tandem genes, genes undergo APA and genes do not. When limited to an intergenic region width of bp due to outliers of up to bp , genes undergo APA and do not.
To test whether there was a significant difference between tandem gene intergenic region length in APA vs. Pairs for which APA information was not available for both genes were discarded.
APA occurred in both genes for 15 gene pairs 30 genes , in one gene for gene pairs genes and in neither gene for gene pairs genes. Nucleosome positioning data was obtained from Kaplan et al. Log 2 normalised nucleosome occupancy per base pair values were used. Genes were characterised as above as tandem or convergent and then split based on APA. Nucleosome positioning data for bp either side of the translational end site for each gene was then averaged for each category.
A nucleosome occupancy value of 0 represents the genome-wide average. Values above zero indicate nucleosome enrichment relative to the genome-wide average and values below zero a relatively nucleosome depleted region. BY cells were grown in synthetic media 0. This involves the use of 4-thiouracil 4tU rather than 4-thiouradine 4sU , which, is readily taken up by yeast cells Munchel et al.
Once cell cultures reached an OD of 0. Samples were taken after 20 and 40 min. Due to the light-sensitive nature of 4tU, samples were treated and total RNA extracted in a light-proof setting. NaIO 4 was precipitated with mM sodium acetate pH 5.
The supernatant was transferred to a fresh tube and RNA was precipitated with 2. These variable bases lock the primer to the polyadenylation site for reverse transcription. PCR amplification, sequencing and data processing were carried out as for the mPAT with 21 and 12 gene-specific forward primers for the BY and rpb experiments respectively and primer sequences are indicated in Table 2.
For each read aligned to the target genes, every mismatch along this sequence compared to the reference was found and a list of SNPs created. From this, the total number of T to C nucleotide conversions, indicative of 4tU labelling, was calculated. The total reads in reads per million and those with T to C conversions for each read length was compared for each sample condition.
Reads aligned to the OM14 gene were plotted with values representative of two replicates. Cells were lysed by vortexing corex tubes 10 times for 1 min with 1 min on ice in between. From top to bottom, tubes contained 2. The protein content was estimated using the bicinchoninic acid method. The aqueous phase containing the peptides was collected, concentrated in a vacuum concentrator and reconstituted in buffer A 0.
The parameters for the mass spectrometric DIA acquisition were described previously Deo et al. The acquired DIA data was analysed in Spectronaut 8 Cassini Biognosys using an in-house generated spectral library established by acquiring the same samples with the same set-up in data-dependent acquisition mode.
To estimate PCR product sizes, band migration was determined relative to a bp ladder. Using cycling parameters 1 min mix, 1 min wait, 2 min measure, basal activity was measured for four cycles.
Data are representative of three independent experimental repeats, with 12 wells per condition in each. Graphs were prepared using Prism seven software. Our editorial process produces two outputs: i public reviews designed to be posted alongside the preprint for the benefit of readers; ii feedback on the manuscript for the authors, including requests for revisions, shown below. We also include an acceptance summary that explains what the editors found interesting or important about the work.
This study aims to provide a comprehensive analysis of factors governing polyadenylation site selection in yeast. Overall, the authors reveal that multiple but distinct inputs including polyadenylation machinery integrity, transcription elongation rate, nucleotide availability and chromatin landscape all contribute to controlling cleavage and polyadenylation.
The work provides the field with an understanding of how aspects of cellular state can control mRNA 3' end formation. Thank you for submitting your article "Genetic and pharmacological evidence for kinetic competition between alternative poly A sites in yeast" for consideration by eLife.
Your article has been reviewed by 3 peer reviewers, one of whom is a member of our Board of Reviewing Editors, and the evaluation has been overseen by James Manley as the Senior Editor.
The reviewers have opted to remain anonymous. The reviewers have discussed their reviews with one another, and the Reviewing Editor has drafted this to help you prepare a revised submission. Overall, the sentiments of the individual Reviewers were quite overlapping.
In a broad sense, the most important task to address through revisions is to improve the connections between the individual elements of the study. Reviewers judged there were some reasonable experiments that could be done to mitigate the disparateness this fact, in addition to adjustments to the text.
Among the suggested revisions, the following stood out in discussion, but the authors should also consider the reviews in aggregate as there were a substantial number of concerns:. Can the authors utilize temperature sensitive mutants in some of the nucleotide biosynthesis enzymes to either mask the cordecypin effect or, if there are gain-in-function mutations can they show the PAS shift in the absence of cordecypin?
Can the authors test some of the defined RNAPII mutants that have been found to be increased in elongation efficiency? Do these cause a shift on their own? In figure one, the authors describe the PAT-seq data as representative of the mutant collection.
But from close inspection, it appears that the one non-temperature sensitive strain Clp1-pm is the only strain that didn't exhibit a significant PAS shift but all of the temp sensitive strains did. Further, the authors do not show what happened to the other two strains that did not exhibit temperature sensitivity yth1 and syc1. This leads to an important question: is the PAS shift simply due to yeast being shifted to a restrictive temperature that they cannot tolerate?
Can the authors address this? Conceptually, there is an abrupt pivot between figures 3 and 4. The authors uncover an interesting phenotype with Sen1 mutant but then explore the role of cordecypin in PAS choice.
I am having trouble understanding the connection. In my mind, the next figure should have investigated the mechanism of sen1 termination to understand the opposing phenotypes.
The authors arrive at a conclusion that cordecypin treatment triggers upregulation of nucleotide biosynthesis, which leads to enhances transcription elongation rate.
This seems to be an important conclusion as it opens up the final segment of the paper and is critical to their model. Validation of a few genes by conventional methods other than Seq could be done. Line — Sup Figure 1 in the text also refers to cordycepin, which is missing from text. Line related to Figure 1 C — rna14 and pcf11 mutants cause a complete switch in SUB2 APA, but there is no complete switch for the other mutants used and there is still growth for those mutants Figure 1B.
How do you reconcile these observations? Clp1-pm does not have an effect — this should be highlighted. What is its function in pA?
It is interesting that the mutant where the authors show a bigger effect in APA is for ysh1. The authors suggest that proximal PAS is skipped because the cleavage reaction is inefficient. Figure 2 B and line — this part is repetitive for yth1.
On the other hand, how do the authors reconcile the nab4 result with the lack of an effect shown in Figure 1B?
Figure 2 B is confusing with the red rectangle positioned at zero for yth1 graph only. Data need to be shown. The authors conclude that six of the 3'-end processing mutants rna, pcf, nab, ysh, fip, pap cause a switch to more distal cleavage site usage and longer 3'-UTRs, but little change is observed in clp1-pm, based on the PAT-seq data.
They suggest that the clp1-pm mutant, containing point mutations that affect ATP binding, does not impact poly A site selection. Given the similarity in the global 3'-end shift effects in Figure 1D, one could conclude that inactivation of the six 3'-end processing factors, regardless of subcomplex, impacts poly A site usage in a similar manner, predominantly increasing distal site selection.
Notably, mass spectrometry was performed on cells treated with cordycepin to address this possibility for the drug. In addition, the authors could test other alleles of 3'-end processing mutants that are documented not to significantly alter the levels of the core cleavage and polyadenylation complex.
These experiments may expose differences between the 3'-end processing factors in poly A site selection. The sen mutant is shown to cause significant switching to shorter 3'-UTRs for genes, suggesting Sen1 antagonizes cleavage site choice by the core 3'-end processing machinery at a subset of protein-coding genes. Given that genes were altered in cleavage site usage in both sen and pcf mutants in opposing directions, it would informative to assess whether a sen pcf double mutant exhibits a positive genetic interaction and restores growth of the cells and some of the altered poly A site usage back to wildtype.
The 3'-end processing factor mutants are all generated in the W strain background, whereas cells treated with cordycepin are all in the BY strain background. Publication types Research Support, N. Gov't, P. Substances Chromatin Deoxyadenine Nucleotides Poly A 3'-deoxyadenosine 5'-triphosphate Adenosine Triphosphate Polynucleotide Adenylyltransferase 2'-deoxyadenosine triphosphate.
Therefore, we isolated total RNA from wild-type and mutant strains and analyzed the length distribution of cellular poly A. With wild-type RNA, a uniformly distributed length of approximately 10—70 adenosines was observed Figure 2 A.
We considered the possibility that some inhibitory activity may be present in the RNA preparations of the mutant strains interfering with the activity of poly A polymerase and performed the labeling assay with decreasing amounts of total RNA.
Furthermore, reduced amounts of wild-type RNA gave increased labeling efficiency. Lane 2 shows the reaction of wild-type RNA without addition of inhibitors. RNAs were treated as indicated with yeast exopolyphosphatase scPpx1p , with heat-denatured exopolyphosphase denat.
Ppx1p or buffer mock before the labeling reaction was performed. It was suggested that the PHO pathway is involved in regulating the synthesis of poly P 6. These results support the idea that poly P metabolism is regulated by the PHO pathway and that constitutive activity of this pathway in the absence of Pho85p and Pho80p resulted in increased cellular poly P concentrations.
The inhibition was nearly complete with the highest concentration of poly P employed nM. However, in multiple repetitions of this experiment, we could not observe a clear correlation of the degree of inhibition and the chain length of poly P included in the assay. Ppx1p specifically degrades poly P and does not act on pyrophosphate or ATP No poly A was detected when reaction buffer replaced Ppx1p or when the Ppx1p was heat-denatured prior to use.
Interestingly, the labeling efficiency of wild-type RNA following Ppx1p pre-treatment was also enhanced, demonstrating the presence of poly P also in wild-type RNA, consistent with our poly P measurements in wild-type strains Figure 2 B.
Since the poly A -labeling assay uses CoTP, poly A polymerase activity is restricted to a single round of nucleoside addition. In the presence of 2 nM poly P , we observed a slight reduction of the length of the polyadenylation products during the time course. This effect was stronger with 20 nM poly P and nM resulted in almost complete inhibition. Similar effects were observed when poly P of varying length average of 28 and 61 phosphate residues was tested data not shown.
Poly P inhibition of poly A polymerase activity in vitro. A Reaction with S. C Effects of poly P binding on the solubility of Pap1p. D Polyadenylation assay as described in A. Next, we tested whether poly P also inhibited the activity of bovine poly A polymerase. Therefore, poly P was a less potent inhibitor of bovine poly A polymerase compared to the yeast enzyme.
Poly P is an efficient chelator of divalent metal ions 1. A possible explanation for the inhibitory effect on poly A polymerase activity could be the chelation of magnesium ions which are essential for Pap1p function.
However, the magnesium in our assays is present in large molar excess over inhibitory poly P concentrations calculated as the amount of free inorganic phosphate.
Therefore, we considered the possibility that poly P may inhibit the enzymes by direct binding. To test this, we performed filter-binding assays with radioactively labeled poly P and yeast or bovine poly A polymerase.
We determined an apparent dissociation constant of 0. It has to be pointed out, however, that one poly P molecule is likely to provide multiple binding sites for Pap1p. Therefore, it seems possible that binding was influenced by avidity effects. Irrespective of that, the tight binding of poly P to poly A polymerase was most likely responsible for the observed inhibitory effects of poly P in polyadenylation assays.
Binding of poly P to yeast and bovine poly A polymerase. Poly P binding by yeast A and bovine B poly A polymerase determined by filter-binding assay as a function of poly P concentration left panels. Scatchard plot representations of the obtained data right panels were used to derive the indicated apparent k D values. Cellular poly P contents correlates with cordycepin-hypersensitive growth.
B Poly P content of the indicated strains was determined following growth in YPD medium for 6 h as described. C Drop test of fold serial dilutions of the indicated strains on agar plates containing 10 mM high phosphate or 0.
Since 5-FOA selects against the URA3 marked plasmid, lack of growth in the presence of the drug indicated that cell viability was dependent on the wild-type copy of RNA14 present on the plasmid. To address the mechanisms of inhibition, we considered the possibility that poly P binding may result in unspecific aggregation or precipitation of the enzyme.
To test this idea we incubated Pap1p with increasing amounts of poly P, collected protein aggregates by centrifugation and resolved remaining soluble material by SDS—PAGE. As shown in Figure 3 C, poly P binding did not result in precipitation of the enzyme. Thus, unspecific aggregation can be excluded as a possible reason for the observed inhibition.
To test whether the inhibition of poly P on Pap1p activity was reversible, we pre-incubated Pap1p with nM poly P to allow binding of poly P and Pap1p. As expected, this resulted in strong inhibition of polyadenylation activity Figure 3 D, lanes 7— However, Pap1p activity could be almost completely restored by treatment of the Pap1p—poly P complexes with Ppx1p Figure 3 D, lanes 12— This demonstrated that the inhibition of Pap1p by poly P was reversible.
The Pap1p that we used in the poly A length analysis and in the polyadenylation assay was purified from E. Thus, purified Pap1p may already be partially bound to poly P causing a partial inhibition of the enzymatic activity.
To test this, we performed a polyadenylation assays with Pap1p that was pretreated with Ppx1p. Since the treatment did not noticeably increase the activity of Pap1p preparation data not shown , we conclude that most of Pap1p that we used in our experiments was not pre-bound by poly P.
This indicated that the conditions used for over-expression of Pap1p did not induce significant poly P accumulation in E. These observations suggested a correlation between cellular poly P accumulation and the drug-dependent growth defect. However, we could not observe cordycepin-hypersensitive growth associated with these mutant strains Figure 5 A. The differences between the measurements presented here and in Ogawa et al.
When low concentrations of phosphate 0. This observation suggested that abundant phosphate was required for the observed cordycepin hypersensitivity. Furthermore, the PHO pathway regulates poly P accumulation in yeast 6.
Increased cellular poly P levels may interfere with poly A polymerase activity, and this may cause hypersensitive growth in the presence of cordycepin, another inhibitor of RNA synthesis. It was previously shown that combination of mutations in rna14 and pap1 resulted in a synthetic growth defect 39 and rna14 mutant strains are cordycepin hypersensitive [ 46 ; our unpublished data].
We provide evidence for an unexpected link between poly P metabolism and polyadenylation in yeast. We propose that a synergistic effect of poly P-mediated inhibition of poly A polymerase together with inhibition of RNA synthesis by CoTP can cause lethality.
Poly P may therefore play a role in the regulation of poly A polymerase activity. The rational for this screen was that cordycepin will be converted to CoTP, which will then provoke RNA-chain termination in the nucleus.
Although CoTP will terminate RNA synthesis when incorporated during transcription, it seems possible that polyadenylation would be a more sensitive target since up to 70 consecutive adenosines are incorporated during this process.
A complete account of the results obtained from our cordycepin-dependent growth screening will be presented elsewhere S. Interestingly, poly P was proposed to act as a phosphate buffer that can be mobilized to maintain constant internal phosphate levels under conditions where external phosphate levels fluctuate 7 , 8 and the PHO pathway was linked to expression of genes that are involved in poly P homeostasis 6.
As a consequence, constitutive activation of the PHO pathway e. We attribute these conflicting poly P measurements to fluctuations of poly P levels in the yeast depending on the growth state and recently proposed a standardized procedure for poly P determination Consistent with our observations, previous biochemical work on the effects of poly P on yeast Pap1p revealed an inhibitory activity on tRNA primed poly A synthesis We show in this work by filter-binding assays that poly P directly binds to yeast Pap1p with a k D in the low nanomolar range 0.
This value is comparable to the k D of 0. Pap1p and Lon are thus the interactors with the highest reported affinities for poly P. Consistent with a physiological relevant role for the interaction of poly P and Pap1p, nuclear poly P levels were reported to reach micromolar concentrations 1 ; furthermore, nuclei predominantly contain poly P of an average chain length of approximately 45 phosphate residues 33 , and poly P with similar length poly P 28 and poly P 61 acted as efficient inhibitor of Pap1p in our in vitro experiments Figure 2 C.
We have not delineated the binding site of poly P on Pap1p, but potential interaction surfaces include regions of RNA primer binding or of ATP binding 57 , Notably, binding of poly P to bovine poly A polymerase and inhibition of its activity is approximately fold weaker compared to the yeast enzyme.
This may have important consequences on a potential in vivo interaction of poly P and poly A polymerases in higher eukaryotes. What is the biological role of poly P-mediated inhibition of poly A polymerase activity?
We predict that such a function must be connected to growth conditions that are accompanied by high poly P concentrations and that the control of poly P levels is at least partially under control of the PHO pathway. We recently showed that poly P content reaches a peak when cells experience the diauxic shift following depletion of glucose from the medium This growth phase is characterized by major changes in the gene expression program 59 ,
0コメント