To follow the in vivo kinetics of pre-mRNA splicing in Saccharomyces cerevisiae
, we integrated reporters based on hybrid ACT1-PGK1
sequences (Hilleren and Parker 2003; Alexander et al., 2010
) into the genome at the HIS3
locus under either tetracycline-inducible (tetON) or tetracycline-repressible (tetOFF) control (Bellí et al., 1998
). Both the tetON and tetOFF strains express tetracycline-responsive repressor and tetracycline-responsive transactivator proteins, which provides a good dynamic range of gene expression (Bellí et al. 
; in this work, the tetracycline analog doxycyclin was used). The 1.3 kb Ribo1 gene (A; see the Supplemental Experimental Procedures
available online for full details) contains the budding yeast ACT1
intron with a short insertion to allow the transcripts to be distinguished from endogenous ACT1
transcripts in reverse transcriptase quantitative real-time PCR (RT-qPCR) assays. Variants of this reporter contain a point mutation at the 5′ splice site (5′SSRibo1), 3′ splice site (3′SSRibo1), or branch site (BSRibo1) or lack an intron (ILRibo1) (Hilleren and Parker 2003; Alexander et al., 2010
). Addition of doxycyclin to the growth medium of a tetON Ribo1 strain resulted in the transient low level accumulation of Ribo1 pre-mRNA at about 3 min, followed by spliced Ribo1 mRNA from 4 min, indicating splicing activity (B, Ribo1). Similarly, ILRibo1 mRNA was detectable from 3 min (B, ILRibo1).
Analysis of Ribo Reporter Induction
ChIP assays, using antibodies against the Rpb3p subunit of RNAPII, detected RNAPII recruitment to the promoter region of the Ribo1 and ILRibo1 genes by 3 min after doxycyclin addition, and the level of RNAPII at the promoter then remained above the uninduced level (C; qPCR amplicon 1, indicated in A). At 4 min (when spliced mRNA was first detectable), a strong, transient accumulation of RNAPII was observed around the 3′SS and just downstream at the 5′ end of exon 2 of Ribo1 (amplicons 3 and 4; D and 1E; see Figure S1
for the full dataset) but not in the corresponding region of ILRibo1 (amplicon 4; D and 1E). Note that the Rpb3p ChIP signal is low over the exon1/5′SS region (amplicon 2), showing that RNAPII levels on either side, at the promoter and 3′SS, can be distinguished in this assay (Ribo1 in E and Figures S1
A–S1C). From 5.5 min, there was also a persistent RNAPII signal toward the 3′ end of Ribo1 that was consistently higher than the signal at the 3′ end of ILRibo1, although the latter appeared earlier, from 3 min (amplicon 5; E). These results were highly reproducible and were similar upon derepression of tetOFF strains (data not shown). Although the timing of the first detection of transcripts varied slightly between different cultures, the transient accumulation of RNAPII in the 3′SS region always coincided with the appearance of spliced Ribo1 mRNA. In several experiments, two peaks of RNAPII were detected at or near the 3′ splice site, a few minutes apart (e.g., Figure S1
We next tested the effect of point mutations at the 5′SS or 3′SS (A), which abolish the first or second step of splicing, respectively. Unspliced 5′SSRibo1 transcripts accumulated from about 3 min and, as expected, no spliced mRNA was detectable (B, 5′SSRibo1). 3′SSRibo1 RNA is a substrate for the first but not the second step of splicing, and the lariat intron-exon 2 product of the first step accumulated with a delay of about 30 s after the appearance of pre-mRNA (B, 3′SSRibo1). ChIP assays showed RNAPII accumulation at the promoter of each mutant reporter gene (C), but to a lower level than with the Ribo1 gene, suggesting reduced transcriptional activity, and there was no accumulation of RNAPII around the 3′SS (D and 1E; amplicons 3 and 4). Thus, the transient RNAPII accumulation around the 3′SS region of the Ribo1 gene depends on the presence of a fully functional intron and/or completion of the splicing reaction, and neither spliceosome assembly nor the first step of splicing is sufficient to cause this. However, with the 3′SSRibo1 reporter, we observed a persistent accumulation of RNAPII over the exon1/5′SS (amplicon 2), suggesting that in the absence of a functional 3′SS, there is a change in the dynamics of transcript elongation, with RNAPII slowing its elongation rate or pausing over exon1/5′SS.
The phosphorylation status of the CTD was also monitored by ChIP, using antibodies specific for phosphorylated serine 5 (pSer5) or phosphorylated serine 2 (pSer2) (Kim et al., 2009
). This showed that, as expected, RNAPII at the promoter of Ribo1 had mainly pSer5 (A). The RNAPII that accumulated transiently around the 3′SS at 4 min was also highly phosphorylated on Ser5, and pSer5 RNAPII accumulated transiently at the 3′SS again a few minutes later (Figure S2
C). Notably, there was little or no pSer5 detected between the promoter and the 3′SS (amplicon 2 in A, left panel). The paused RNAPII was also phosphorylated on Ser2 (A, right panel; Figure S2
C), with the 3′SS being the most 5′ position on the gene at which pSer2 was detected. This suggests that phosphorylation of Ser2 occurred on the paused RNAPII; however, the pSer2 accumulation may not display exactly the same timing as the pSer5 data. Toward the 3′ end of the gene, pSer5 declined, whereas pSer2 increased at later time points. Presenting the RNAPII phosphorylation signal as a proportion of the total RNAPII signal shows that RNAPII at the 3′SS was hyperphosphorylated compared to RNAPII at the promoter (Figure S2
, pSer5/RNAPII, compare A and C). This suggests new phosphorylation of Ser5 and Ser2 at the 3′SS.
Phosphorylation Status of RNAPII CTD during Induction
With the intronless ILRibo1 reporter, the pSer5 signal simply decreased from the promoter toward the 3′ end of the gene, as the pSer2 signal gradually increased (B). With 5′SSRibo1, RNAPII with pSer5 accumulated strongly at the promoter, despite the lower level of total RNAPII signal, and there was only a low level of pSer2 across the body of the gene, more like ILRibo1 than Ribo1 (). With 3′SSRibo1 (D), there was an accumulation of pSer5 at the promoter and also over the exon1/5′SS (amplicon 2), compatible with a slowing or pausing of RNAPII in this region. The level of pSer2 increased toward the 3′ end of the 3′SSRibo1 gene. Thus, the dynamics of RNAPII phosphorylation differ significantly with the splicing status of the gene.
If the RNAPII pause in the region of the 3′SS is determined by splicing, two predictions can be made: (1) splicing of Ribo1 transcripts should be cotranscriptional at this time and (2) suppression of the splicing defect of a mutant intron will lead to RNAPII pausing on the mutant gene.
To address the first point, we analyzed the cotranscriptional recruitment of U2 and U5 snRNPs by performing ChIP with antibodies to the snRNP components Prp11p and Prp8p, respectively. U2 snRNP was detectable at 3.5 min after doxycyclin addition, and the U5 snRNP was first detected at 4 min (), consistent with cotranscriptional spliceosome assembly at the time of the RNAPII pause (Görnemann et al., 2005
) and continuing thereafter. Furthermore, we have recently shown in a kinetic analysis of splicing and 3′ end formation that a significant amount of splicing of Ribo1 transcripts occurs prior to 3′ end cleavage and polyadenylation, indicating cotranscriptional splicing (Alexander et al., 2010
). It is therefore conceivable that cotranscriptionally recruited splicing factors might affect RNAPII and/or chromatin factors that are in close proximity.
Cotranscriptional Recruitment of Splicing Factors
To test the second prediction, we used the BSRibo1 reporter that has a point mutation at the branch site, which causes a first step splicing defect. This splicing defect can be largely suppressed by a mutant U2 snRNA that restores base pairing with the mutant branch site sequence (Parker et al., 1987
) (A). Plasmids encoding the wild-type or mutant U2 snRNA were introduced into a tetOFF BSRibo1 strain. After derepression in the wild-type U2 control strain, unspliced transcripts accumulated with no detectable splicing (B, left panel), and there was no RNAPII accumulation around the 3′SS (D and 4E; amplicons 3 and 4). This resembles the ILRibo1 and 5′SSRibo1 result (D). In a strain producing mutant U2 snRNA that complements the BSRibo1 mutation, splicing was substantially restored (B, right panel). Importantly, RNAPII accumulated transiently at the 3′SS at 6.5 min, the time when spliced mRNA was first detected, and a second peak of RNAPII appeared at the 3′SS at 8.5 min (D right panel; see Figure S3
for more detail and ChIP data for pSer5 and pSer2). Thus, it is clearly the actual process of splicing, rather than the intron sequence, that causes transient RNAPII pausing. Additionally, it may be significant that suppression of the BS splicing defect also resulted in higher levels of pSer2 RNAPII toward the 3′ end of exon 2 (amplicon 5; E and Figure S3
E) as was noted earlier for Ribo1, suggesting that this too may be splicing dependent.
It Is the Splicing Event Rather than the Intron Sequence that Causes RNAPII to Pause at the 3′SS
The observation of a second RNAPII pause in several experiments raised the possibility that pausing may be a recurring event. To test this, we performed a longer time course of Ribo1 induction, and, indeed, RNAPII was observed to accumulate strongly near the 3′SS three times, at approximately 3 min intervals ( and Figure S4
). Intriguingly, each RNAPII pause seems to occur at or shortly after a peak in pre-mRNA accumulation and increased mRNA production, suggesting bursts of splicing at these times.
RNAPII Pauses Repeatedly at the 3′ End of the Ribo1 Intron
The majority of intron-containing genes are constitutively expressed during normal growth in budding yeast. Therefore, in order to examine another intron-containing gene under similar induction conditions, the nonessential APE2
gene was deleted from its genomic locus in the tetON yeast strain, and the APE2
sequence, was integrated downstream of the doxycyclin-inducible promoter at the HIS3
locus (like Ribo1). An advantage of APE2
for this analysis is that both of its exons are longer than for Ribo1, allowing qPCR analysis of more, nonoverlapping, regions of the APE2
gene. After the addition of doxycyclin, RNAPII occupancy on the gene was monitored as for Ribo1. As shown in C, oscillations of RNAPII accumulation were observed in the region of the 3′SS, with a periodicity of 2.5 to 3 min. There was also increased RNAPII accumulation toward the 3′ end of the gene (C, and more detail in Figure S5
). Analysis of the phosphorylation status showed RNAPII with pSer5 in the promoter region and in the transient peaks over the 3′SS ( and Figure S5
). The pSer2 signal also increased from this point toward the 3′ end of the gene. Therefore, the APE2
and Ribo1 genes show similar patterns of RNAPII accumulation after induction.
RNAPII Pauses Repeatedly at the 3′ End of the APE2 Intron
Although RNAPII pauses only very briefly at the end of the Ribo1 and APE2
introns, the observation that pausing occurs repeatedly suggested that it may be possible to detect an elevated level of RNAPII over 3′ splice sites of constitutively expressed endogenous genes, without inducing synchronous transcription in the population of cells. However, with an asynchronous population of cells, the length, amplitude, and frequency of the pause or oscillation will determine how readily an elevated level of RNAPII will be detectable above background in the snapshot in time that is captured by the ChIP assay. Clearly, this may vary between genes. ChIP of RNAPII performed on four endogenous intron-containing genes, including APE2
, shows that the level of RNAPII is slightly elevated over the 3′SS of all four genes, although the resolution is poor (Figure S6
). As it had been shown that the level of phosphorylation of RNAPII is significantly elevated around the 3′SS of Ribo1 () and APE2
(), the amount of phosphorylated RNAPII was measured. Indeed, coincident peaks of enrichment of RNAPII with pSer5 and pSer2 are evident over the 3′ splice sites of all four intron-containing genes (A–7D). It is particularly clear for APE2
, which have longer first exons, that the peak of pSer5 at the 3′SS is distinct from the peak at the promoter. In contrast, different patterns were observed for two intronless genes, ADH1
, with pSer5 declining and pSer2 increasing from the 5′ ends to the 3′ ends of the genes (E and 7F).
Phosphorylated RNAPII Accumulates around the 3′ Splice Sites of Endogenous Yeast Intron-Containing Genes