TFIIS, an elongation factor encoded by DST1 in Saccharomyces cerevisiae, stimulates transcript cleavage in arrested RNA polymerase II. Two components of the RNA polymerase II machinery, Med13 (Srb9) and Spt8, were isolated as two‐hybrid partners of the conserved TFIIS N‐terminal domain. They belong to the Cdk8 module of the Mediator and to a subform of the SAGA co‐activator, respectively. Co‐immunoprecipitation experiments showed that TFIIS can bind the Cdk8 module and SAGA in cell‐free extracts. spt8Δ and dst1Δ mutants were sensitive to nucleotide‐depleting drugs and epistatic to null mutants of the RNA polymerase II subunit Rpb9, suggesting that their elongation defects are mediated by Rpb9. rpb9Δ, spt8Δ and dst1Δ were lethal in cells lacking the Rpb4 subunit. The TFIIS N‐terminal domain is also strictly required for viability in rpb4Δ, although it is not needed for binding to RNA polymerase II or for transcript cleavage. It is proposed that TFIIS and the Spt8‐containing form of SAGA co‐operate to rescue RNA polymerase II from unproductive elongation complexes, and that the Cdk8 module temporarily blocks transcription during transcript cleavage.
All DNA‐dependent RNA polymerases (Pol's) are endowed with an intrinsic ribonuclease activity that cleaves a few nucleotides from the 3′ end of the elongating transcript. This cleavage activity probably helps to backtrack elongation complexes that are out of register with the transcript 3′ end, allowing them to resume transcription (Fish and Kane (2002) and references therein). Transcript cleavage operates in bacterial and archaeal Pol's, and in the three eukaryotic enzymes (Pol I, II and III). In bacteria, it is activated by the GreA and GreB factors (Borukhov et al, 1993). In the eukaryotic Pol III enzyme, cleavage is catalysed by the enzyme alone and depends on its Rpc11 subunit (Chédin et al, 1998). In Pol II, it depends on the Rpb9 subunit (akin to Rpc11) and on a factor initially referred to as S‐II in human cells (Natori et al, 1973; Izban and Luse, 1992; Reines, 1992) or P37 in yeast (Sawadogo et al, 1980), but now generally called TFIIS (Fish and Kane, 2002). Pol I also has a cleavage activity that does not depend on TFIIS (Tschochner, 1996) but may require Rpa12, a subunit paralogous to Rpb9 and Rpc11 (Nogi et al, 1993; Van Mullem et al, 2002a).
Yeasts (Saccharomyces cerevisiae and Schizosaccharomyces pombe) have only one form of TFIIS, in contrast to the multiple subforms present in vertebrates (Labhart and Morgan, 1998). The corresponding null mutants (dst1Δ) have little or no growth defects (Archambault et al, 1992; Exinger and Lacroute, 1992; Nakanishi et al, 1992; Williams and Kane, 1996), suggesting that TFIIS is only required under specific conditions or that it is functionally redundant with other transcription factors (Davie and Kane, 2000; Lindstrom and Hartzog, 2001; Ubukata et al, 2003). The structure of the yeast TFIIS–Pol II complex was recently determined at a resolution of 3.8 Å and reveals its probable mode of action (Kettenberger et al, 2003). In short, TFIIS binds the Rpb1/Rpb9 ‘jaw’ of Pol II and inserts into the Pol II pore, contacting the catalytic site by its highly conserved C‐end bearing an invariant RSADE motif. The two carboxylic amino acids of that motif are thought to contribute to metal coordination at the level of the enzyme active site. Remarkably, the RSADE motif is present in Rpa12 and Rpc11, but not in Rpb9. This suggests that these two subunits may themselves directly contact the catalytic sites in Pol I and Pol III, as their Pol's do not require TFIIS for cleavage.
The N‐terminal domain of TFIIS (approximately corresponding to its first 132 amino acids) stays on the outer surface of Pol II, where it is available for interaction with other components of the transcription complex (Kettenberger et al, 2003). This N‐terminal part is conserved in all eukaryotes sequenced so far. It also has significant homology to the N‐end of MED26 (a component of the human Mediator formerly called CRSP70 or ARC70; Ryu et al, 1999; Bourbon et al, 2004) and of the three subforms of human Elongin A (Aso et al, 1995; Yamazaki et al, 2002). This domain was found here to engage in specific interactions with Spt8 and Srb9 (now called Med13; Bourbon et al, 2004). Spt8 defines a subform of the SAGA co‐activator (Belotserkovskaya et al, 2000; Pray‐Grant et al, 2002; Sterner et al, 2002; Wu and Winston, 2002). Med13 (Srb9) is associated with the evolutionarily conserved kinase Srb10 (Hengartner et al, 1998), now called Cdk8 (Bourbon et al, 2004). Spt8 and Med13 had not been connected so far to each other or to TFIIS, but we shall present evidence suggesting that both may be genuine partners of this factor.
Spt8 and Med13 are two‐hybrid partners of the N‐terminal domain of TFIIS
Figure 1 presents the outcome of a two‐hybrid screening using the entire TFIIS sequence fused to the Gal4BD(1–147) domain. The corresponding pVV70 plasmid was used as a bait against a random library of yeast genomic fragments fused to the Gal4AD(768–881) domain (Fromont‐Racine et al, 1997; Flores et al, 1999). From a total of about 107 transformants obtained in strain Y190, 118 clones were selected by their ability to grow in the presence of 100 mM 3‐amino‐triazole, and were then shown to activate the lacZ reporter gene, as detected in a β‐galactosidase assay (Figure 1C). Based on the DNA sequence of their inserts, they were allocated to 36 independent clones, defined, respectively, by 25 and 11 distinct but overlapping fragments of Spt8 and Med13. Thus, we consistently identified the same two partners and the same domains on these partners, which strongly argues for a saturating and specific two‐hybrid screening.
Spt8 belongs to a subform of the yeast SAGA Pol II co‐activator (Grant et al, 1997). This protein is essentially formed of WD40‐like domains, with two acidic patches (Figure 1B). The 25 Spt8 clones isolated in the two‐hybrid screening shared a 102 amino‐acid segment between positions 364 and 465. This minimal TFIIS interacting domain includes one acidic patch and one of the WD40‐like motifs. Med13 is a moderately conserved component of the Mediator (Borggrefe et al, 2002; Boube et al, 2002). It belongs to the Cdk8 module, where Cdk8 is a conserved kinase initially identified by its ability to phosphorylate the C‐terminal domain (CTD) of the largest Pol II subunit (Hengartner et al, 1998). The 11 Med13 clones defined a 195 amino‐acid segment (320–514) as the minimal TFIIS interacting domain. This domain is included in a region conserved in all fungal Med13 (Figure 1B). It bears no detectable similarity to the Spt8 target domain defined above.
Four domains can be recognised on the TFIIS structure and are collinear to its amino‐acid sequence (Kettenberger et al, 2003; see also Figure 1A). The corresponding fragments were fused to the Gal4BD(1–147) domain and tested separately against Spt8 and Med13 in a two‐hybrid assay. As shown in Figure 1C, the N‐terminal part of TFIIS (amino acids 1–132) is necessary and sufficient to interact with both partners. NMR data have shown that this domain is made of four closely packed α helices (Booth et al, 2000). They are not included in the crystal structure (Kettenberger et al, 2003) but are evidently exposed on the outer surface of Pol II (Figure 1A), and are thus available for interactions with other components of the transcription machinery.
The N‐half of this domain (amino acids 1–74) is important for the two‐hybrid interaction since its deletion abolishes the two‐hybrid response (Figure 1C). However, a fragment bearing amino acids 1–74 alone failed to interact with Spt8 or Med13 (data not shown). Interestingly, this region harbours a highly conserved motif corresponding to the α2, α3 and α4 helices (Figure 1D). A Psi‐Blast survey of the human genome showed that this motif is present in nine distinct gene products, including three TFIIS isoforms (Labhart and Morgan, 1998), three forms of Elongin A (Aso et al, 1995; Booth et al, 2000; Yamazaki et al, 2002) and the MED26 component of the Mediator (Ryu et al, 1999; Bourbon et al, 2004).
TFIIS co‐purifies with Med13 and its associated Cdk8 kinase in cell‐free extracts
The two‐hybrid data above suggested that Med13 may be a functional partner of TFIIS. We therefore examined if the immunopurification of Med13 from cell‐free extracts leads to a co‐purification of TFIIS and vice versa. Med13∷13Myc was barely detectable when introduced as a chromosomal allele, but could be readily detected when expressed from a replicative plasmid and then pulled down a significant amount of TFIIS∷3HA (Figure 2A). This occurred no matter whether cells expressed Spt8 (SPT8+) or not (spt8Δ). In a reciprocal experiment, TFIIS∷13Myc pulled down Med13∷3HA (Figure 2B).
Since Med13 belongs to the Cdk8 module of the Mediator, we also examined whether Cdk8 itself might co‐purify with TFIIS. Indeed, Cdk8∷13Myc co‐purified with the immunoprecipitated TFIIS∷3HA and TFIIS∷3HA co‐purified with Cdk8∷13Myc (Figure 2C and D). Moreover, this co‐purification was substantially reduced in a med13Δ context (Figure 2D). Along with our two‐hybrid data, this indicates that, in vivo, TFIIS may associate with the Cdk8 module and that this association is, to a large extent, dependent on Med13. We note that med13Δ also had a minor effect on the electrophoretic mobility of TFIIS, suggesting a change in its phosphorylation pattern.
dst1Δ and med13Δ mutants have different phenotypes
TFIIS is encoded by the DST1 gene. dst1Δ null mutants have no detectable defect except their sensitivity to 6‐azauracil and mycophenolate, two nucleotide‐depleting drugs that are thought to impair elongation (Exinger and Lacroute, 1992). med13Δ and cdk8Δ are not sensitive to mycophenolate and do not aggravate the sensitivity of dst1Δ (data not shown). Moreover, the med13Δ dst1Δ spt8Δ triple mutant grew like its dst1Δ spt8Δ parent, and with the same sensitivity to mycophenolate.
The only known phenotype that relates Med13 to TFIIS is their opposite effect on rpb1 mutants with partial deletion of the Pol II CTD, such as the rpb1Δ104 mutant used in this study (Allison and Ingles, 1989). Indeed, med13Δ and cdk8Δ were isolated as suppressors of such rpb1 mutants (Liao et al, 1995), while dst1Δ is lethal in this context (Lindstrom and Hartzog, 2001). The triple mutant dst1Δ med13Δ rpb1Δ104 is also lethal (data not shown), suggesting that the integrity of TFIIS may be required for the suppressor effect of med13Δ on Pol II CTD mutants.
TFIIS co‐purifies with Spt8 and other SAGA subunits in cell‐free extracts
SAGA is a Pol II co‐activator bearing the Gcn5 histone acetyl‐transferase (Grant et al, 1997). It was recently shown to exist in two subforms that essentially differ by the presence or absence of the Spt8 subunit (Belotserkovskaya et al, 2000; Pray‐Grant et al, 2002; Sterner et al, 2002; Wu and Winston, 2002). Since our two‐hybrid data suggested that TFIIS may directly interact with Spt8, we constructed a double‐mutant strain where Spt8∷13Myc and TFIIS∷3HA fusions were expressed from the chromosomal locus. Under these conditions, Spt8∷13Myc only pulled down a barely detectable amount of TFIIS∷3HA (data not shown). The expression of the two proteins from replicative vectors significantly enhanced the co‐immunopurification signal, and this occurred no matter whether cells expressed Med13 (MED13+) or not (med13Δ) (Figure 3A). In a reciprocal experiment, Spt8∷3HA co‐purified with the immunoprecipitated TFIIS∷13Myc (Figure 3B). Thus, Spt8 has affinity for TFIIS in terms of two‐hybrid interactions and co‐purification from a yeast cell‐free extract.
These data raise the question of whether Spt8 binds TFIIS in its own right or as part of the Spt8‐containing form of SAGA. Figure 3C and D tentatively suggest that the latter may be true since TFIIS∷3HA pulled down Gcn5∷13Myc and Spt7∷13Myc. Conversely, Gcn5∷13Myc and Spt7∷13Myc pulled down TFIIS∷3HA. However, there was no clear indication that this co‐purification depended on Spt8: the genetic inactivation of Spt8 reduced the co‐purification, but this might be an indirect effect as there was also less TFIIS∷3HA in the corresponding crude extracts (Figure 3D, compare lanes 7–8 and 9–10).
dst1Δ and spt8Δ mutants have related phenotypes
As already mentioned, dst1Δ has little or no growth defect but is sensitive to nucleotide‐depleting drugs such as mycophenolate (Exinger and Lacroute, 1992; see also Figure 4A). Null mutants of several other nonessential components of the Pol II transcription machinery are also sensitive to these inhibitors (Desmoucelles et al, 2002). They include the spt3Δ, spt7Δ and spt8Δ mutants of the SAGA co‐activator and the rpb4Δ and rpb9Δ mutants lacking the nonessential Pol II subunits Rpb4 or Rpb9. Figure 4A illustrates the different levels of sensitivity to mycophenolate displayed by these mutants. This drug sensitivity is not a general property of SAGA, since null mutants of Gcn5 or of the Ada2 and Ada3 subunits of SAGA (ada2Δ, ada3Δ and gcn5Δ) are not sensitive.
dst1Δ and rpb9Δ are epistatic, that is, the double mutant is indistinguishable from rpb9Δ alone in terms of growth and drug sensitivity (Figure 4A; Van Mullem et al, 2002b). This also holds for the spt8Δ rpb9Δ double mutant (Figure 4A and B). Remarkably, null mutants of all the other nonessential subunits of SAGA (Ada2, Ada3, Gcn5, Spt3 and Spt7) are lethal in a rpb9Δ context, disregarding whether they are sensitive to mycophenolate (spt3Δ and spt7Δ) or not (ada2Δ, ada3Δ and gcn5Δ) (Figure 4B; Van Mullem et al, 2002b). Thus, epistasis with rpb9Δ is a unique property of spt8Δ among SAGA null mutants.
Crosses between rpb4Δ and rpb9Δ yield no double mutants, suggesting that these mutations are synthetic lethal (Li and Smerdon, 2002). We confirmed this interpretation by complementation tests showing that lethality is relieved in the presence of a plasmid bearing the wild‐type RPB4 or RPB9 genes. dst1Δ and spt8Δ were also synthetic lethal with rpb4Δ. In both cases, double mutants grew as microcolonies but did not form viable clones when further streaked on YPD. Moreover, the double mutants were rescued by complementation with the RPB4 and DST1/SPT8 wild‐type genes (Figure 5). Again, this was not a general property of SAGA since spt3Δ and gcn5Δ are epistatic with rpb4Δ (Figure 4B). This is consistent with the fact that SAGA complexes purified from spt3Δ and gcn5Δ contain Spt8 (Sterner et al, 1999). In contrast, a spt7Δ deletion that leads to the disruption of SAGA (Sterner et al, 2002) is lethal in both rpb9Δ and rpb4Δ contexts.
From the data above, dst1Δ and spt8Δ behave very similarly in terms of lethality with rpb4Δ, epistasis to rpb9Δ and mycophenolate sensitivity. Yet, their physiological effects do not fully overlap. spt8 mutants suppress his4917δ and are protrophic for lysine in a LYS2–173R2 context, which reflects the less efficient transcription of solo δ and Ty1 elements in the his4‐917δ and LYS2‐173R2 alleles (Eisenmann et al, 1994; Wu and Winston, 2002). These spt phenotypes (suppression of Ty1) are not found in dst1Δ and are not aggravated in a dst1Δ spt8Δ double mutant (data not shown). As shown in Figure 4C, dst1Δ is sensitive to caffeine (trimethylxanthine). This phenotype is shared by rpb4Δ and rpb9Δ but not by spt8Δ (data not shown; Sterner et al, 1999). Finally, dst1Δ aggravates mycophenolate sensitivity in spt8Δ (Figure 4A), indicating that the drug sensitivity of Spt8 cannot be directly mediated by its binding to TFIIS.
The N‐terminal domain of TFIIS is critical in an rpb4Δ context
The N‐terminal domain of TFIIS is one of the most conserved components of that polypeptide (see Figure 1D) but does not participate in Pol II binding and is not required for transcript cleavage (Agarwal et al, 1991; Awrey et al, 1998). This is consistent with the observation that a dst1‐133,309 mutant lacking this domain is not sensitive to 6‐azauracil (Nakanishi et al, 1995; Ubukata et al, 2003) to mycophenolate (Figure 4D) or to caffeine (Figure 4C), and is in fact indistinguishable from wild type by all criteria examined so far. On the other hand, our two‐hybrid data suggest that this N‐terminal domain specifically interacts with Spt8. A deletion of that domain is therefore expected to disrupt the Spt8–TFIIS interaction. As shown in Figure 5B, a dst1‐133,309 mutant lacking this N‐terminal domain is lethal in an rpb4Δ context and behaves like dst1Δ and spt8Δ themselves. The same colethality was observed in rpb4Δ dst1‐R287Q,E291N double mutants affecting the invariant RSADE domain of TFIIS (Figure 5C). dst1‐R287Q,E291N is specifically defective in the transcript cleavage activity, but is not affected for Pol II binding (Ubukata et al, 2003). Taken together, these data show that, in the absence of Rpb4, transcription is strictly dependent on transcript cleavage and also requires a functional interaction between Spt8 and TFIIS.
Elongating RNA polymerases that meet obstacles on the DNA template become arrested and probably displace the 3′ end of the transcript relatively to the catalytic site of polymerisation. Escape from arrest requires an RNA cleavage process that, in the case of Pol II, is strongly stimulated by the elongation factor TFIIS (Fish and Kane, 2002). Elegant structural studies by Kettenberger et al (2003) have shown that TFIIS binds Pol II on its Rpb9/Rpb1 jaw and that its C‐terminal part reaches the internal active site of the enzyme, where the factor induces a conformational change switching Pol II to its RNA cleavage mode. However, they provide no function for the N‐terminal domain of TFIIS.
This N‐terminal domain forms a bulky four‐helix bundle (Booth et al, 2000) on the periphery of the Pol II structure, approximately facing the downstream DNA (Kettenberger et al, 2003), and is thus available for interaction with other Pol II factors. Its physiological role has remained a mystery, but studies on the human and yeast factor have shown that it is not needed for transcript cleavage, for the binding of TFIIS to Pol II or for the stimulation of elongation in vitro (Agarwal et al, 1991; Nakanishi et al, 1995; Awrey et al, 1998). Indeed, a deletion of this domain could not so far be associated to any growth phenotype in S. cerevisiae. In mammals, this domain has a clear homology to the N‐terminal region of the transcription factors MED26 and Elongin A (Aso et al, 1995; Ryu et al, 1999), and it is hard to believe that this evolutionary conservation is not associated with some specific function.
We report here that the N‐terminal domain of TFIIS interacts with Spt8 and Med13. Both are well‐defined components of the Pol II transcription machinery, but were so far not connected to each other or to TFIIS. Med13 belongs to the Cdk8 kinase module of the Mediator. The two‐hybrid interaction between Med13 and TFIIS is further supported by co‐immunoprecipitation data showing that TFIIS co‐purified with Med13 and its associated Cdk8 kinase. Moreover, co‐purification with Cdk8 was substantially reduced in a med13Δ mutant. In yeast and human cells, GST∷TFIIS fusions pull down a Pol II holoenzyme that contains Cdk8 (Pan et al, 1997; Hirst et al, 1999). Furthermore, this only requires the N‐terminal part of the human TFIIS (positions 1–103). Taken together, these data support the idea that TFIIS binds the Mediator at the level of its Cdk8 module, by a specific interaction between Med13 and the N‐terminal part of TFIIS. Since Cdk8 inhibits transcription via its CTD kinase activity (Hengartner et al, 1998), it may facilitate the TFIIS‐dependent cleaving process by temporarily holding back elongation (Figure 6A). Alternatively, the kinase could phosphorylate the N‐end of TFIIS, as the latter is phosphorylated in human cells (Horikoshi et al, 1985; Agarwal et al, 1991).
Spt8 defines a recently discovered subform of the SAGA Pol II co‐activator (Belotserkovskaya et al, 2000; Pray‐Grant et al, 2002; Sterner et al, 2002; Wu and Winston, 2002). Along with our two‐hybrid data, the co‐immunopurification of TFIIS with Spt8, Gcn5 and Spt7 argues for a direct interaction between TFIIS and the Spt8‐containing form of SAGA. However, the genetic inactivation of Spt8 did not impair the co‐purification of TFIIS with Gcn5 and Spt7. One possibility is that SAGA binds the Pol II/TFIIS transcription complex by additional contact points. In this context, we note that Spt7 is a two‐hybrid partner of the Rpb9 subunit of Pol II (M Werner and P Thuriaux, unpublished observation).
Spt8 and TFIIS are not normally required for transcription in S. cerevisiae, since spt8Δ and dst1Δ null mutants have little or no effect on growth, and since a deletion of the Spt8‐interacting N‐terminal domain of TFIIS (dst1‐133,309) is indistinguishable from wild type. Remarkably, both factors are essential in rpb4Δcells lacking the Rpb4 subunit of Pol II. This is not a general property of SAGA mutants since gcn5Δ and spt3Δ mutants, for example, are viable in an rpb4Δ context. gcn5Δ lacks the histone acetylase of SAGA but, like spt3Δ, still produces an Spt8‐containing form of SAGA (Sterner et al, 2002; Wu and Winston, 2002). In the case of TFIIS, the lethality of rpb4Δ dst1‐R287Q,E291N (where dst1‐R287Q,E291N affects the invariant RSADE motif and specifically lacks the cleavage activity, see Ubukata et al, 2003) proves that transcript cleavage is critical in the absence of Rpb4. The fact that this lethality extends to the dst1‐133,309 mutant deleted for the N‐terminal domain of TFIIS strongly suggests that the interaction with Spt8 is also needed in an Rpb4‐less context.
spt8Δ and dst1Δ are epistatic with rpb9Δ mutants lacking Rpb9, the other nonessential Pol II subunit (Van Mullem et al (2002b) and this study). Again, this is not a general property of SAGA mutants since gcn5Δ rpb9Δ double mutants are lethal, suggesting that the histone acetylase activity of SAGA is critical in an Rpb9‐less context. Thus, Rpb4 and Rpb9 are not directly involved in RNA polymerisation but may stabilise two distinct conformations of Pol II. The absence of Rpb9 makes Pol II incompetent for transcript cleavage (as shown by Awrey et al, 1998), and we speculate that the absence of Rpb4 would instead make Pol II strictly dependent on transcript cleavage. Both conditions would lead to slow growth and to a strong sensitivity to nucleotide‐depleting drug, but for different reasons. For example, Pol II molecules lacking Rpb4 could be prone to transcription accidents, obliging them to resort to cleavage. This would account for the strict dependency of rpb4Δ on Rpb9 and TFIIS, since both are required for RNA cleavage (Awrey et al, 1997). As sketched out in Figure 6B, the ability to recruit the Spt8‐containing form of SAGA via the N‐terminal of TFIIS may also be essential under these conditions. This may not be restricted to the elongating Pol II, as TFIIS binds the promoter and coding part of GAL1 (Pokholok et al, 2002; D Prather, E Larschan and F Winston, personal communication). SAGA itself is recruited to the promoter (Cosma et al, 1999; Bhaumik and Green, 2001; Larschan and Winston, 2001). This, however, need not apply to the Spt8‐containing form, and an elongation role of SAGA is indeed suggested by the sensitivity of spt3Δ, spt7Δ and spt8Δ to nucleotide‐depleting drugs (Desmoucelles et al, 2002).
Materials and methods
Yeast strains were constructed by standard meiotic crosses and transformations (Table I). Most of the mutant strains were full deletions with a KanMX4 insertion (Euroscarf: http://www.uni-frankfurt.de/fb15/mikro/euroscarf/). In crosses involving the KanMX4 marker in the two parental strains, double mutant segregants were isolated from a nonparental ditype tetrad, and confirmed by PCR. Y14279 (med13Δ) and Y12666 (spt8Δ) derive from the BY4742 strain (MATα his3Δ1 leu2Δ0 lys2Δ0 ura3Δ0). Y04411 (dst1Δ), Y04228 (spt3Δ), Y03218 (spt7Δ) and Y02666 (spt8Δ) derive from BY4741 (MATa his3Δ1 leu2Δ0 met15Δ0 ura3Δ0). The other parental strains used were CMKy1 (Davie and Kane, 2000), FY1093 (spt7Δ; Wu and Winston, 2002) RPO21‐Δ104 (Allison and Ingles, 1989), MC11‐1 (Choder and Young, 1993), OG30‐4C (gcn5Δ∷HIS3) and YVV9 (rpb9Δ; Van Mullem et al, 2002b), SL21 (Shpakovski et al, 2000), YPH499 and YPH500 (Sikorski and Hieter, 1989). Mt8 is a dst1‐R287Q, E291N double‐mutant defective in the RNA cleavage activity (Ubukata et al, 2003). DY236‐6B is a MATa trp1Δ63 strain obtained by crossing FY67 (Winston et al, 1995) to YPH500. Strain Y190 (MATa gal4 gal80 his3 trp1‐901 ade2‐1 ura3‐52 leu2‐2,112 CYH1RURA3∷GAL1∷lacZ LYS2∷GAL4(UAS)∷HIS3) was used as host in two‐hybrid tests (Flores et al, 1999). In this strain, HIS3 and lacZ are used as reporter genes of the two‐hybrid interaction.
YPD and SC are standard complete or synthetic growth media with 2% glucose. YPD‐Caffeine contains caffeine at 10 mM. SC‐W is tryptophan omission media. MPA contains mycophenolate (Sigma) freshly dissolved in methanol (10 mg/ml), added to SC or SC‐W. FOA (Boeke et al, 1984) is SC with 0.1% of 5‐fluoro orotic acid (Toronto Research Chemicals). Spt phenotypes were assessed by their ability to restore histidine prototrophy in his4‐917δ and to generate lysine auxotrophy in a LYS2‐173R2 context (Wu and Winston, 2002).
Plasmids (Table II) were prepared by standard subcloning or by the GATEWAY™ technique (Walhout et al, 2000), except for pSPT8‐84 and pMED13‐111, which were isolated from a random library of genomic fragments (about 0.7 kb) fused to the Gal4AD(768–881). Constructs based on polymerase chain reaction (PCR) were sequenced to avoid spurious mutations generated by the amplification process. pVV70–pVV75 are derivatives of pGBT9 with full‐length or partly deleted forms of DST1 fused to the Gal4BD(1–147) domain, constructed by directional cloning between SmaI and BamHI. Their expression was tested by Western blotting assay using anti‐Gal4BD(1–147) antibodies (Clontech). pVV80 (2 μ TRP1 DST1) was constructed by cloning the DST1 SmaI–SalI insert from pVV70 into the multicopy expression vector pGEN (Shpakovski et al, 1995). pVV81 (2 μ TRP1 dst1‐133,309) was obtained by directional cloning of a PCR‐amplified BamHI–ClaI dst1‐133,309 insert into the pGEN. pCM‐DST1 and pCM‐ΔN were obtained by subcloning the BamHI–MluI DST1 and dst1‐133,309 inserts of pVV80 and pVV81 into pCM185 (Gari et al, 1997). PYX212‐RPB4 was constructed by cloning a PCR‐amplified RPB4 coding sequence between the NcoI and SalI sites of the multicopy expression vector pYX212 (Yeast R&D Systems). pVV225–pVV234 were constructed using the GATEWAY™ technique (Walhout et al, 2000). Briefly, the SPT8, GCN5, SPT7, MED13, CDK8 and DST1 coding sequences were PCR‐amplified without their termination codon from oligonucleotides ending with attB1 and attB2 sites. Entry clones were generated by in vitro recombination with the attP1 and attP2 sites of the pDONR™201 vector (Invitrogen). Inserts were sequenced and then subcloned using LR reactions into appropriate pVV201, pVV203, pVV215 or pVV217 pGEN‐derived vectors (Van Mullem et al, 2003).
Immunopurification was carried out as previously described (Van Mullem et al, 2002b), starting from 1 mg of yeast crude extract and using about 0.8 μg of mouse monoclonal anti‐Myc (9E10 from Babco) or anti‐HA (12CA5) antibody. Beads were washed three times for 5 min in a modified IP buffer (20 mM HEPES, pH 7.5, 0.5 mM EDTA, 500 mM NaCl, 1 mM dithiothreitol, 20% glycerol, 0.1% Triton X‐100). Immunoprecipitated proteins were eluted, heated for 10 min at 95°C, separated by SDS–PAGE and revealed by monoclonal anti‐HA or anti‐Myc antibodies (Babco) using the ECL™ Western Blotting Detection kit (Amersham).
We thank André Sentenac for his kind support and Michel Werner for helpful comments during this work. Drs Fred Winston, Caroline Kane and Toshiyuki Nakanishi provided useful mutant strains, Audrey Suleau constructed plasmid pYX212‐RPB4 and Séverine Coddens helped in immunopurification experiments. We also thank an anonymous referee for illuminating critics of an earlier version of this paper. This work was partly supported by the Direction des Relations Internationales (CEA). MW was supported by short‐term fellowships of EMBO and FEBS. ES was supported by a Young Scientist Fellowship from INTAS, by the Russian Federation of Young Scientists and by research programmes from the Russian Academy of Sciences and the Russian Foundation of Basic Research. BVD and VVM were supported by the Belgian Fonds pour la Recherche dans l'Industrie et l'Agriculture and Fonds National de la Recherche Scientifique, respectively.
- Copyright © 2004 European Molecular Biology Organization