Report

Regulation of Elongating RNA Polymerase II by Forkhead Transcription Factors in Yeast

See allHide authors and affiliations

Science  18 Apr 2003:
Vol. 300, Issue 5618, pp. 492-495
DOI: 10.1126/science.1081379

Abstract

The elongation phase of transcription by RNA polymerase II (RNAPII) is highly regulated and tightly linked to pre–messenger RNA (pre-mRNA) processing. Recent studies have implicated an early elongation checkpoint that facilitates the link to pre-mRNA processing. Here we show that the yeast forkhead transcription factors, Fkh1p and Fkh2p, associate with the coding regions of active genes and influence, in opposing ways, transcriptional elongation and termination. These events are coordinated with serine-5 and -2 phosphorylation of the heptad repeat of the carboxy-terminal domain (CTD) of RNAPII. Our results suggest that, in addition to their documented promoter function, Fkh1p and Fkh2p coordinate early transcription elongation and pre-mRNA processing. This may reflect a general feature of gene regulation in eukaryotes.

The winged-helix forkhead (Fkh) family of transcription factors is highly conserved in eukaryotes with roles in cell cycle control, cell death, proliferative responses, and differentiation (1). In yeast, FKH1 and FKH2 influence expression of a wide range of genes (2), including those that control the G2–M phase transition of the cell cycle (3, 4). Activation of genes of the CLB2 cluster, expressed at G2–M phase, requires co-operative binding of the Fkh2p and the Mcm1p transcription factors to the Swi Five Factor (SFF) site in the promoter. Fkh2p is the preferred component of the SFF factor, although Fkh1p substitutes in the absence of Fkh2p (5) and either factor is sufficient for activator recruitment (4). Previous studies have shown that the effects of ablation of each individual Fkh factor on the steady-state levels of CLB2 mRNA are different; Fkh1p is required to repress expression, whereas Fkh2p activates expression to normal levels (68). This suggests additional opposing functions for Fkh factors in regulating CLB2 expression. Structural and functional studies reveal a linker histone-like structure for the winged-helix Fkh domain and suggest interactions with chromatin other than sequence-specific DNA binding (9, 10). In agreement with a more general effect on gene expression, our results demonstrate that Fkh factors differentially regulate the elongation phase of transcription at CLB2 and at other loci.

We have analyzed the binding of Fkh factors to CLB2 with the use of chromatin immunoprecipitation (ChIP) in strains expressing epitopetagged Fkh factors (11). As expected, Fkh2p binds strongly and Fkh1p weakly to the upstream activating sequence (UAS) region on the promoter (5) (Fig. 1B). The majority of Fkh1p is associated with the first 600 base pairs (bp) of the transcribed region, whereas, in addition to UAS binding, a second peak of Fkh2p is present at the beginning of the coding region, extending toward the 3′ end of the gene. The unexpected presence of Fkh1p and Fkh2p in the CLB2 coding region suggests a role in transcriptional elongation, in addition to their well-known function in CLB2 activation involving UAS association.

Fig. 1.

Fkh factors associate with promoter and coding regions and influence RNAPII distribution at CLB2.(A) Schematic of CLB2 showing position (boxes) and size of the UAS and the PCR products (1 to 4), amplified with real-time PCR analysis of the distribution of (B) Fkh1-HA and Fkh2-myc and (C) Rbp3-HA (RNAPII) in W303-based strains indicated with ChIP.

We investigated the distribution of RNAPII over the CLB2 gene (Fig. 1C; fig. S1). In wild-type (WT), the distribution of RNAPII (Rbp3-HA) suggests that enzyme accumulates at the beginning of the coding region. However, RNAPII distribution was different in strains lacking Fkh1p (fkh1△) or Fkh2p (fkh2△). In fkh1△, RNAPII fails to accumulate at the beginning of the coding region, and instead the enzyme is concentrated toward the end of the gene. As expected, fkh2△ leads to a substantial reduction in the amount of RNAPII at CLB2, but the majority is found at the beginning of the coding region. These initial observations lead us to conclude, first, that Fkh transcription factors, believed to be localized exclusively to promoters, are also associated with transcribed regions, and second, that these factors influence in opposite ways the distribution of RNAPII over the gene. The observation that RNAPII accumulates at the beginning of the coding region of CLB2 indicates that a block to elongation occurs at this position and that this requires Fkh1p.

We set out to test the hypothesis that the Fkh factors differentially influence elongating RNAPII by investigating the effect of 6-azauracil (6AU) on the growth of fkh1△ and fkh2△ mutants. The phenotype of 6AU sensitivity is linked to transcription elongation (12) as treatment depletes pools of guanosine 5′-triphosphate (GTP). Consequently, RNAPII cannot elongate efficiently, and elongation factors such as TFIIS (Dst1p) and Spt4p become crucial for growth (13, 14). We show that fkh2△, like dst1△ and spt4△ strains, displays sensitivity to 6AU that is suppressed by guanine (Fig. 2A; fig. S2A). Furthermore, phenotypes in combined fkh2spt4△ or fkh2dst1△ mutants are more severe, consistent with a positive role for Fkh2p in transcription elongation. A full genetic analysis is shown in fig. S2. By contrast, fkh1△ is resistant to 6AU and suppresses the phenotypes of the fkh2△ and spt4△ strains, suggesting an opposing function.

Fig. 2.

Fkh factors influence transcription elongation. (A) 6AU phenotypes. Tenfold dilutions of strains indicated were spotted onto medium containing 75 (plate) or 100 μg/ml 6AU (table). Relative growth (+) is scored as 10-fold differences in number of cells surviving treatment (fig. S2). All strains scored 4+ on control plates containing guanine (G). (B) ChIP analysis of Fkh1-HA, Fkh2-myc, and RNAPII from strains cultured in the absence (–) or presence (+) of 100 μg/ml 6AU for 60 min, amplified with real-time PCR. Size and position of PCR products are indicated (1 to 3) on IMD2. No binding was observed without 6AU; only signals above background (0.1) are shown (fig. S1C).

WT strains survive 6AU treatment because the IMD2 gene is induced to replenish GTP pools (15). Northern blot analysis suggests that IMD2 is a target of Fkh regulation (fig. S2C). The IMD2 promoter lacks Fkh binding sites (16) and is not associated with Fkh factors. However, association of the Fkh factors with the coding region of IMD2 paralleled that observed at CLB2, but only when IMD2 transcription was induced with 6AU (Fig. 2B; fig. S1C). A similar pattern of Fkh factor association with the coding region (Fig. 3B; figs. S3 and S5C) and Fkh-dependent distribution of RNAPII (Fig. 3E; figs. S5B and S6B) was observed at three further loci, PMA1, MET16, and pTEF: KanR, not previously identified as targets of Fkh factors (2, 16), suggesting that many unrelated genes may be influenced by Fkh factors when transcribed.

Fig. 3.

Fkh factors regulate RNAPII at PMA1. (A) Schematic of PMA1 and distribution of (B) Fkh1-HA, Fkh2-myc, (C) Kin28p, (D) RNAPII phosphorylated at Ser2 or Ser5, and (E) RNAPII detected with real-time (B, C, and E) or conventional (D) PCR (fig. S4). The position and size of PCR products are indicated (1 to 4).

We sought to establish how Fkh1p and Fkh2p influence RNAPII. The heptad repeat (Tyr-Ser-Pro-Thr-Ser-Pro-Ser)27 of the C-terminal repeat domain (CTD) of the largest subunit of RNAPII becomes differentially phosphorylated at Ser5 and Ser2 during elongation and recruits activities for co-transcriptional processing of the transcript (17). We performed ChIP analysis with antibodies specific for CTD phosphorylated at either Ser5 or Ser2 (Fig. 3D; figs. S4 to S6). In WT, Ser5 phosphorylation decreased but Ser2 phosphorylation increased between the 5′ and 3′ end of the genes, as observed previously (18). In fkh2△, dst1△, and spt4△, the pattern of Ser5 and Ser2 phosphorylation is more constant across the gene. In these strains, the majority of RNAPII is located at the 5′ part of the coding region (Fig. 3E; figs. S5B and S6B), suggesting there is a defect in the release of RNAPII into the elongation phase of expression. This raises the possibility that failure to release RNAPII disturbs the temporal control of Ser5 and Ser2 phosphorylation, leading to a low level of hyperphosphorylated RNAPII at 3′ regions. By contrast, in the fkh1△ strain, phosphorylation of Ser5 at all three promoters is markedly reduced, and this modification remains low throughout the coding regions (Fig. 3D; figs. S4 to S6). Interestingly, Fkh1p is also required for normal association of Kin28p, the Ser5 kinase (19) at PMA1 (Fig. 3C) and IMD2 (fig. S6C). This supports a link between Fkh1p and early transcription events. Furthermore, Ser2 phosphorylation is barely detectable in fkh1△ and does not increase as RNAPII reaches the 3′ end of the genes (Fig. 3D; figs. S4 to S6). Ser2 phosphorylation is not dependent on Kin28p, suggesting that activity of the serine-2 kinase, or regulation of the CTD phosphatase, is also defective in fkh1△ (20).

Because the increase in Ser2 phosphorylation in coding regions is associated with pre-mRNA 3′ end processing (21), it seemed likely that fkh1△ is also defective in pre-mRNA 3′ end formation. We therefore looked for transcriptional interference between the tandem GAL10 and GAL7 genes that occurs when GAL10 termination is disrupted (22, 23) with the use of a tandem reporter construct containing the GAL10-7 intergenic region (Fig. 4A; fig. S7A). Here, transcription running through the GAL10 terminator reduces green fluorescent protein (GFP) produced by the downstream gene. WT expresses GFP from the GAL7 promoter regardless of whether the interfering upstream gene is active or not. This is also the case in fkh2△, dst1△, and spt4△, although the efficiency of expression may be reduced. In contrast, in fkh1△ expression from the downstream GAL7 promoter is reduced by 80% when the interfering upstream gene is expressed. To confirm a 3′ end formation defect for fkh1△, reverse transcription polymerase chain reaction (RT-PCR) was performed to detect steady-state RNA complementary to 3′ sequences of the IMD2 gene. This revealed extended transcripts in fkh1△ but not WT or fkh2△ (Fig. 4B). These data strongly suggest that Fkh1p plays a role in mRNA 3′ end formation.

Fig. 4.

Fkh1p influences transcription termination. (A) Schematic of the transcription interference assay constructs. Data show expression of pGAL7:gfp in strains indicated relative to WT in induced (with galactose) or repressed (with glucose) growth conditions (fig. S7A). (B) RT-PCR assay to detect steady-state RNA at the 3′ end of IMD2 with primers 1 to 4, at +100, +200, +300, and +400, relative to the translation termination codon, to prime cDNA synthesis on total RNA from strains indicated, followed by PCR amplification with primers indicated. Cells were induced with 100 μg/ml 6AU for 60 min before RNA extraction. (C) TRO analysis for IMD2. Pulse-labeled nascent RNA isolated from strains grown in 100 μg/ml 6AU was hybridized to 80 base (b) single-stranded DNAs (1 to 12), spanning the 5′ (1 to 4) and 3′ (5 to 12) regions of IMD2. A, actin (positive control); pG, pGEMplasmid (negative control).

To confirm that the hypophosphorylated RNAPII in fkh1△ is defective in pre-mRNA 3′ end formation, we used a transcription run-on (TRO) assay. Here, incorporation of a pulse of radiolabel into nascent transcripts indicates engaged RNAPII. We investigated the profile of incorporation of labeling at IMD2 (Fig. 4C) and pTEF::KanR (fig. S7B) on plasmids. As expected, in WT signals dropped to near background after the polyadenylation [poly(A)] site. By contrast, at the 3′ side of the poly(A) site, signals in fkh1△ are of higher intensity than in WT, consistent with active transcription past the poly(A) site, which confirms the 3′ end formation defect. Furthermore, at IMD2, the signal at probe 1 in fkh1△, upstream of the transcription start site, would be consistent with transcription running around the plasmid backbone, commonly observed in termination mutants (24). Given the relatively even TRO signals in fkh1△ compared with uneven distribution of RNAPII detected by ChIP, we predict that much of the RNAPII at the 3′ end is not actively engaged in transcription, a likely consequence of the lack of Ser5 and Ser2 phosphorylation.

We suggest that, at the beginning of genes, the opposing action of Fkh1p versus Fkh2p (together with Dst1p and Spt4p) may be part of the proposed checkpoint mechanism to coordinate transcription and pre-mRNA processing via CTD phosphorylation (17) (fig. S8). Furthermore, modulation of the levels (25, 26) or activity (6) of these factors might regulate the release of RNAPII into the elongation phase, allowing transcription in yeast to be matched to the environmental conditions and the availability of nutrients. An outstanding question is how, in the absence of high-affinity binding sites and a cooperative binding partner, the Fkh factors associate with chromatin in coding regions. Like Dst1p, the Fkh factors may directly associate with elongating RNAPII (27). An alternative possibility is that the Fkh factors are tethered to specific domains of chromatin in coding regions (28). It remains to be determined how the diverse activities of Fkh factors in metazoans will relate to this demonstrated role in transcriptional elongation.

Supporting Online Material

www.sciencemag.org/cgi/content/full/300/5618/492/DC1

Materials and Methods

SOMText

Figs. S1 to S8

References

References and Notes

View Abstract

Navigate This Article