1-induced DNA bending is required for transcriptional activation by th
http://www.100md.com
《核酸研究医学期刊》
Waksman Institute and Department of Molecular Biology and Biochemistry, Rutgers University, Piscataway, NJ 08854, USA
*To whom correspondence should be addressed. Tel: +1 732 445 2905; Fax: +1 732 445 5735; Email: vershon@waksman.rutgers.edu
ABSTRACT
The yeast Mcm1 protein is a founding member of the MADS-box family of transcription factors that is involved in the regulation of diverse sets of genes through interactions with distinct cofactor proteins. Mcm1 interacts with the Mat1 protein to activate the expression of the -cell type-specific genes. To understand the requirement of the cofactor 1 for Mcm1–1-dependent transcriptional activation we analyzed the recruitment of Mcm1 to the promoters of -specific genes in vivo and found that Mcm1 is able to bind to the promoters of -specific genes in the absence of 1. This suggests the function of 1 is more complex than simply recruiting Mcm1. Several MADS-box transcription factors, including Mcm1, induce DNA bending and there is evidence the proper bend may be required for transcriptional activation. We analyzed Mcm1-dependent bending of a Mcm1–1 binding site in the presence and absence of 1 and found that Mcm1 alone shows a reduced DNA-bend at this site compared with other Mcm1 binding sites. However, the addition of 1 markedly increases the DNA-bend and we present evidence this bend is required for full transcriptional activation. These results support a model in which proper DNA-bending by the Mcm1–1 complex is required for transcriptional activation.
INTRODUCTION
The MADS-box family of DNA-binding proteins are involved in transcriptional regulation in a wide array of eukaryotes, including yeast, plants and mammals (1). Mcm1, a MADS-box transcription factor in yeast, is an essential protein involved in regulating diverse sets of genes; including, cell mating-type, cell cycle, osmotic regulation and arginine metabolism (2–5). In order to carry out these regulatory roles Mcm1 interacts with several different cofactors that help determine its function. For example, in cell-type determination, Mcm1 binds P-elements in the promoters of a-specific genes to activate their transcription in the a-cell type (6). However, in the -cell type, Mcm1 interacts with the 2 protein at these promoter elements to repress this same set of genes (7,8). Mcm1 also interacts with the 1 protein to bind P’Q promoter elements to activate the expression of -specific genes (9,10). Therefore, cofactor interaction markedly alters the function and specificity of the Mcm1 MADS-box protein.
A comparison of P’Q elements found upstream of -specific genes to the consensus P-element derived from the promoters of a-specific genes shows they are similar (Fig. 1). However, the P’Q significantly differs from the consensus P-element on the side that is directly flanked by Q, the binding site for 1 (9). Since Mcm1 is unable to activate transcription on its own from a P’Q site, the 1 protein must help Mcm1 by either increasing its ability to bind to the site or to recruit the transcriptional machinery .
Figure 1. A comparison of the consensus P-element to STE3 P’Q. The STE3 P’Q shows strong conservation with the consensus P-element except for the degenerate side, bases 5–8 (bold), which is flanked by the Q site, the binding site for 1. The predicted DNA contacts by Mcm1 with the outer regions of the site based on the crystal structure of Mcm1 are indicated (13). Base-specific contacts are indicated with a solid line, phosphate contacts are indicated with dashed lines. The asterisk indicates the axis of symmetry in the Mcm1 binding site. Positions –7 and 7 are important for DNA bending of the consensus P-element (16).
The crystal structures of several MADS-box proteins show that many produce a significant bend in the DNA when they bind their sites (11–13). Biochemical and genetic analysis suggest that this bending can play an important role in transcriptional activation (14–18). For example, the mcm1-V34A mutant, which reduces the apparent bend angle of a P-element from 95° to 82°, causes a >10-fold decrease in transcriptional activation, yet has only a minor effect on DNA-binding affinity (17,19). Other mutants with larger defects in binding affinity have less of an effect on activation than the V34A mutant, suggesting that bending may play a role in activation. Interestingly, when the mcm1-V34A mutant was assayed for the ability to activate a P’Q element in complex with 1, it showed a less severe defect in transcriptional activation than many other mutants, demonstrating that either activation by Mcm1–1 does not require DNA bending or that 1 helps mediate the bending (19).
To more fully understand the relationship of DNA-bending and cofactor interaction for transcriptional activation, we analyzed bending of a P’Q element with wild-type and mutant Mcm1 proteins alone and in complex with 1. Our results demonstrate the 1 cofactor contributes to the DNA bend mediated by the complex and that a change in the bend angle results in a significant decrease in transcriptional activation. These results support a model of the requirement for a proper DNA-bend by the 1–Mcm1 complex for full transcriptional activation.
MATERIALS AND METHODS
Yeast and bacterial strains and plasmids
Chromatin immunoprecipitations were performed in either the yeast strain JMY041 (MATa ade2-1 trp1-1 his3-11,15 can1-100 ura3-1 leu2-3 mcm1::kanr/pSL1574-MCM1 CEN URA3) or EC7 (MAT ade2-1 trp1-1 his3-11,15 can1-100 ura3-1 leu2-3,112 mat1::TRP1 mcm1::kanr/pSL1574-MCM1 CEN URA3) derived from W303-1A and W303-1B, respectively. The strains were transformed with plasmids that contained either wild-type or mutant derivatives of untagged Mcm1 (pJM231, MCM1 CEN HIS3) or a V5-tagged derivative (pJM421, MCM1-V5 CEN HIS3), both of which complement a mcm1 strain for viability (17). Cells that had lost the plasmid pSL1574 with a wild-type copy of MCM1 were selected by growth in the presence of 5-FOA. In addition, these strains were transformed with pRS415, as an empty vector control, or a derivative of pEC43 (MAT1-Myc CEN LEU2) that expresses MAT1 from its native promoter with a 13-myc tag at the C-terminus. The ORF of MAT1 in pEC43 was engineered with silent restriction sites at the following positions, PstI (bp 61), NheI (bp 95), StuI (bp111), EagI (bp134), XbaI (bp175), NruI (bp 210), SpeI (bp250), EcoR1 (bp 280), SphI (bp 369), XmaI (bp 435), BamHI (bp 530) by recursive PCR as described previously (20). The engineered MAT1 construct complements a mat1 strain for mating and activates transcription of a heterologous STE3-lacZ promoter at the same level as an unmodified gene.
Transcription assays
Mcm1–1-dependent transcriptional activation was monitored in the yeast strain EC7 by monitoring lacZ expression from reporter constructs derived from pTBA23 (2 μ URA3) containing a CYC1-lacZ reporter promoter with the CYC1 UAS deleted (16). The transcription reporter constructs contained either a single P’Q wild-type site from STE3 (pEC25) or a P’Q site with a T–7 to G mutation (P’Q-T–7G) (pEC111) (9). ?-Galactosidase assays were performed in triplicate with independent isolates and the standard deviations of the calculated ?-galactosidase values were <10% (17).
Chromatin immunoprecipitation assays
Chromatin immunoprecipitation assays of Mcm1 binding were performed based on the procedures described previously (21). Cultures (50 ml) were grown to an OD600 of 0.8 and were cross-linked with 1% formaldehyde for 30 min. The cells were washed twice in ice-cold TBS and frozen at –80°C. The cells were then suspended in 500 μl FA lysis buffer and 0.5 g acid washed glass beads and were vortexed 40 min at 4°C. The lysates were cleared by centrifugation for 1 min at 14 000 r.p.m. at 4°C. The supernatants were sonicated 6x for 5 s to shear the chromatin to 500 bp fragments and then centrifuged for 5 min at 14 000 r.p.m. at 4°C. A total of 50 μl of the supernatant was removed for total chromatin (TC) control. The remaining supernatant was pre-cleared with protein G–agarose beads and then incubated overnight with 1 μl anti-V5 antibody (Invitrogen). Protein G–agarose beads (50 μl) were incubated for 1 h and the beads were washed 1x in low salt wash (0.1% SDS, 1% Triton X-100, 150 mM NaCl 2x TE); 1x in high salt wash (0.1% SDS, 1% Triton X-100, 500 mM NaCl 2x TE); 1x in LiCl wash (0.25 M LiCl, 1% IGEPAL, 1x TE, 1% sodium deoxycholate); and 2x in 1x TE. Protein–DNA complexes were eluted from the beads by washing 2x with 250 μl elution buffer (1% SDS, 0.1 M NaHCO3). Cross-links were reversed by the addition of 20 μl 5 M NaCl and incubating at 65°C for 4 h. Protease digestion was performed by adding 10 μl 0.5 M EDTA, 20 μl Tris pH 7.4, 2 μl proteinase K (10 μg/μl) and incubating at 42°C for 45 min. The samples were then phenol–chloroform extracted and the DNA precipitated with ethanol.
Quantitative PCR was performed as described previously (22) using 1/50 of immunoprecipitated DNA fragments and 1/1000 of the TC material in a 50 μl total reaction volume. PCR amplification of STE6 or STE2 promoter region served as a positive control. Primers to amplify the promoters of ACT1 or YDL223C served as negative controls.
Protein purification
The wild-type and mutant Mcm1 proteins used in the in vitro assays were purified as described previously except that the proteins were expressed in Escherichia coli BL21 (codon +) and the maltose-binding protein (MBP) was not removed after thrombin cleavage (17). The Mcm1 proteins isolated were >95% homogeneous. The 1 protein was purified as an MBP–1 fusion from BL21 (codon +) bacteria cells expressing pSL2187 (23) grown in LB media containing ampicillin and chloramphenicol to an OD600 of 0.6 followed by induction with 0.3 mM isopropyl-?-D-thiogalactopyranoside for 20 h at 14°C. Cells were lysed and protein was purified to >95% homogeneity as described previously exclusive of cleavage of the MBP (23). Protein concentrations were normalized by Bradford assays and verified on Coomassie stained SDS–PAGE gels.
Electrophoretic mobility shift assays
Oligonucleotides containing STE3 P’Q wild type or P’Q-T–7G were end-labeled with ATP using polynucleotide kinase and purified by Qiagen nucleotide removal columns (Qiagen) according to the manufacturer. The oligonucleotides were made double-stranded by mixing with a 3-fold excess of the matching strand, incubating at 90°C for 20 min and slowly cooling to 25°C overnight in a water bath. The circular permutation assays were performed with BamHI, NheI, HindIII or EcoRI-generated end-labeled fragments as described previously (16) containing either a P-PAL site, STE3-P’Q or P’Q-T–7G site. All binding reactions with purified Mcm11–97 or MBP–1 were carried out in 10 mM Tris–HCl (pH 7.5), 40 mM NaCl, 4 mM MgCl2, 6% (w/v) glycerol, 10 mg/ml BSA, 10 μg/ml of sonicated salmon sperm DNA and 32P-labeled DNA probes (4500 c.p.m.) in a total volume of 30 μl at room temperature for 60 min. All protein dilutions were made in 20 mM Tris–HCl (pH 8), 50 mM NaCl, 1 mM EDTA, 1 mg/ml BSA, 5 mM ?-mercaptoethanol and 1 mM PMSF. Samples were analyzed on a 6% polyacrylamide gel . Gels were dried after electrophoresis, exposed to a phosphor screen and scanned on a Model Storm 840 Molecular Dynamics phosphorimager. Gels were quantified using IPlab Gel imaging software.
RESULTS
Mcm1 binds in vivo in the absence of 1
In vivo DNA footprint analysis led to the conclusion that Mcm1 is unable to bind P’Q elements in the absence of 1, suggesting, 1 may function primarily to recruit Mcm1 to P’Q elements in vivo (24). However, Mcm1 showed only a slightly reduced binding affinity (3-fold) in vitro for a P’Q site compared with a P-element, suggesting Mcm1 may be bound on its own in vivo (25). To characterize the in vivo binding of Mcm1 to P’Q elements alone and in complex with 1 chromatin immunoprecipitations were performed (Fig. 2). As expected, in the presence of 1, Mcm1 strongly bound to the promoters of -specific genes (Fig. 2, lane 7). Interestingly, there were also significant levels of Mcm1 binding to these sites even in the absence of 1 (Fig. 2, lane 3). Since Mcm1 bound to the promoters of -specific genes in vivo but failed to activate transcription, it suggests 1 has a greater role in transcriptional activation than simply recruiting Mcm1 to these promoters.
Figure 2. Mcm1 binds P’Q elements in the absence of 1 in vivo. Chromatin immunoprecipitations were performed with lysates containing untagged (lanes 1, 2, 5 and 6) or V5-epitope tagged Mcm1 (lanes 3, 4, 7 and 8) using anti-V5 antibody. A MATa strain was transformed with either an empty vector (pRS415, lanes 1–4) or wild-type 1 (pEC43, lanes 5–8). The panel shows an ethidium bromide stained gel of PCR amplifications of the indicated promoter region. PCR product using primers to the -specific genes STE3, SAG1 and MF1 are indicated. A promoter fragment of the a-specific STE6 gene bound by Mcm1 served as a positive control. YDL223C, a gene not regulated by Mcm1, was used as a negative control. Lanes 1, 4, 5 and 8 show PCR products using TC as a template, whereas lanes 2, 3, 6 and 7 used immunoprecipitated product (IP).
The 1 protein increases the bend of Mcm1 bound to a P’Q element
There is evidence that DNA bending may play a role in transcriptional regulation by MADS-box proteins (16–18). To determine if 1 affects the degree of DNA bending at a P’Q element with Mcm1 we performed EMSAs with circularly permuted fragments that contain P’Q sites. As demonstrated previously, Mcm1 induced an apparent bend angle of roughly 99° at a P-PAL element (Fig. 3, lanes 1–4) (16). However, the STE3 P’Q site was only bent with an apparent angle of 76° (Fig. 3, lanes 9–12). The decreased bend angle is presumably due to sub-optimal contacts at bases +5 to +8 in the degenerate side of the P’Q site. DNA-bending of the P’Q element was further reduced by mcm1-V34A, which contains a mutation at a residue required for DNA-bending . In the presence of both 1 and Mcm1 there was a shifted band with significantly slower mobility (Fig. 3, lanes 17–20). The presence of 1 significantly increased the bend of the P’Q site to an apparent angle of 94°. Interestingly, in the presence of 1 the mcm1-V34A mutant had a similar level of DNA bending as the wild-type protein (Fig. 3, compare lanes 17–20 with 21–24). These results indicate that the presence of 1 increases the DNA bend at a P’Q site and is able to suppress the bending defect caused by the V34A mutation.
Figure 3. The 1–Mcm1 complex induces a greater bend in the DNA than Mcm1 alone. An EMSA utilizing position permutation of either a P-PAL (lanes 1–8) or STE3-P’Q site (lanes 9–24) is shown. Lanes 1–4, 9–12 and 17–20 show binding by the wild-type Mcm1 protein. Lanes 5–8, 13–16 and 21–24 show binding by the mcm1-V34A mutant. Lanes 1–16 indicate binding by Mcm1 alone, whereas lanes 17–24 show binding in the presence of 1. The calculated bend angle for each binding complex is shown.
In addition to V34A, several other mutations in Mcm1 cause bending defects at a P-element (17,18). On their own, each of the mcm1 mutants showed a decrease in bending of the P’Q site (Fig. 4B–F, lanes 1–4). The relative decreases in bending were comparable with that seen for these mutants binding to a P-element (17,18). For example, the T66A mutation only caused a slight decrease in the apparent bend angle (Fig. 4E, lanes 1–4), whereas the V34A, S37A double mutant caused a substantial decrease (Fig. 4F, lanes 1–4). Although these mutants showed significant decreases in bending on their own, in complex with 1 none of the mutants significantly altered the apparent bend angle in comparison with the wild-type protein (Fig. 4A–F, lanes 9–12).
Figure 4. The Mcm1–1 complex has increased bending at the P’Q-T–7G site. An EMSA utilizing position permutation of the binding sites, with each panel representing either wild-type Mcm1 (A) or mcm1-mutants V34A (B), S37A (C), K40A (D), T66A (E) and V34A/S37A (F). Lanes 1–4 and 9–12 of each panel show binding to a STE3-P’Q site. Lanes 5–8 and 13–16 in each panel show binding to a STE3-P’Q site with a T–7G substitution. Lanes 1–8 are in the absence of 1 and lanes 9–16 are in the presence of 1. The calculated bend angle for each complex is shown.
To further test if 1 affects the bending of the complex we examined binding to a mutant P’Q site that contains a base pair substitution, T–7G. In the context of a P-site this substitution significantly affects Mcm1-dependent DNA bending (17,18). The T–7G substitution also caused a decrease in bending by Mcm1 in the context of the P’Q site (Fig. 4A, lanes 5–8 versus 1–4). The difference in the apparent bend angle was comparable with the decrease observed for a P-element with a similar mutation (16). The S37A, K40A and T66A mutations in Mcm1 caused a further decrease in the apparent bend angle of the P’Q-T–7G site (Fig. 4C, D and F, lanes 1–4 versus 5–8). However, V34A and the V34A, S37A double mutant showed no further decrease in bending with the P’Q-T–7G site (Fig. 4B and F, lanes 1–8). This likely indicates a direct contact between V34 and the T–7 position, as seen previously for Mcm1 at a P-element (13,17). Interestingly, in complex with 1, the bend angle of the P’Q-T–7G site mediated by the WT, S37A, K40A or T66A mutants was not decreased, but was rather significantly increased above the levels of bending by the wild-type protein to the wild-type site (Fig. 4A, C, D and E, lanes 9–12 versus 13–16.) In contrast, V34A and the V34A, S37A double mutant did not show this further increase in bending but bent the DNA with an apparent bend angle that was similar to the one caused by the wild-type protein binding to the wild-type site (Fig. 4B and F, compare lanes 9–12 with 13–16). These data indicate that in complex with 1, the mcm1-V34A mutation suppresses the increased bending defect due to the P’Q-T–7G element.
1 interaction with the Mcm1–DNA complex is not altered by changes in DNA bending
We next compared the relative binding affinities of the Mcm1 mutants, alone and in complex with 1 at a wild-type P’Q and the P’Q-T–7G to determine if any of these amino acid substitutions affected the formation of the complex. The relative binding affinities of the mcm1 mutants for a P’Q site were all comparable with the affinities for a P-element (17). With the exception of K40A, the Mcm1 mutants bound to a P’Q with less than a 2.5-fold reduction compared with wild-type protein (data not shown). In addition, the P’Q-T–7G mutation did not greatly alter the binding affinity of wild-type or mutant Mcm1 proteins (data not shown).
To determine if 1 interaction with the Mcm1–DNA complex was affected by the degree of DNA bending we analyzed this interaction by EMSA. The formation of the binary Mcm1–DNA complex was normalized to account for differences in mcm1 DNA binding at P’Q and at P’Q-T–7G. The ability of 1 to interact with the Mcm1–DNA complex was unaffected, even by severe mcm1 DNA-bending mutants (Fig. 5). In addition, the P’Q-T–7G mutation did not cause a decrease in 1-binding affinity. The K40A mutant showed the opposite effect with 1 interaction, partially suppressing the binding defect. There was a decrease at a P’Q-T–7G with V34A and the V34A-S37A double mutant in 1 interaction showing an approximate 2-fold decrease. However, overall there was no correlation between the degree of bending of the binary Mcm1–DNA complex and the ability of 1 to interact with the complex.
Figure 5. Mutations that affect Mcm1-dependent bending do not alter the interaction with 1. Wild-type and mutant Mcm1 proteins were normalized for DNA-binding affinity on their own. Serial dilutions (3-fold) of 1 were added to each of the Mcm1 mutants. The fold decrease for 1 interaction with the Mcm1–DNA complex is indicated relative to the Mcm1 wild-type protein.
Two mutants shown previously to be defective in interaction with 1, mcm1-T66E and mcm1-S73R, were analyzed for bending in the presence or absence of 1 to determine if defects in interaction would result in a bending deficiency (26). As reported previously, both mutants showed a significant defect in interaction with 1 (Fig. 6A) (26,27). In the absence of 1 the relative bending of the T66E mutant was significantly reduced compared with wild-type Mcm1, whereas the S73R mutant showed levels comparable with wild type (Fig. 6B). However, in complex with 1 both mutants showed apparent bend angles comparable with the wild-type protein. These data indicate that defects in the interaction between 1 and Mcm1 do not necessarily result in bending defects.
Figure 6. Defects in interaction with 1 do not result in bending defects in the Mcm1–1 complex. (A) An EMSA with WT-Mcm1, T66E or S73R bound to a P’Q site (lanes 2, 6 or 10, respectively) or in complex with 1 (lanes 3–5, 7–9 or 11–13, respectively). The Mcm1 protein concentration is constant in each lane with 3-fold serial dilutions 1. (B) A circular permutation assay is shown in the absence of 1 (lanes 1–12) or in the presence of 1 (lanes 13–24) with Mcm1-WT (lanes 1–4, 13–16), T66E (lanes 5–8, 17–20) and S73R (lanes 9–12, 21–24). The calculated bend angle for each complex is shown.
The 1–Mcm1 DNA bend must be at a proper level for full transcriptional activation
To determine if changes in the apparent bend angle caused by the T–7G mutation altered transcriptional activation we constructed heterologous reporter promoters driving lacZ transcription that contained either a single P’Q or P’Q-T–7G site. In a wild-type strain activation from the T–7G reporter was decreased greater than 4-fold compared with the wild-type site (Fig. 7). Since the DNA-binding affinity of the 1–Mcm1 complex for P’Q-T–7G was comparable with the wild-type site it suggests that the increase in the bend angle affects activation. We also analyzed the ability of mcm1-mutants to drive expression from a P’Q and P’Q-T–7G site. The S37A mutant showed levels comparable with wild-type protein at both the P’Q and P’Q-T–7G site (Fig. 7). Since both the wild-type and S37A proteins showed increased bending of the P’Q-T–7G, it is possible the change in the apparent bending angle causes a decrease in activation at the mutant site. Although the V34A mutant showed a small reduction in activation from a wild-type P’Q, this mutant was able to partially suppress the decrease in activation by the P’Q-T–7G site, which correlates with the ability to suppress the bending defect (Fig. 7). The T66E and S73R mutants both showed significant decreases in transcriptional activation at both sites, which is likely due to defects in interaction with 1 (Fig. 7) (18,26).
Figure 7. In vivo assays to determine the effect of DNA-bending on transcriptional activation. Reporter promoters containing either a P’Q (solid bars) or P’Q-T–7G (open bars) were used to drive lacZ expression in a WT or mutant Mcm1 background. Activation of the reporter in MAT cells was determined from the average of three independent assays that differed by <10%. Percentages of activation are shown relative to the WT Mcm1 protein at a WT-P’Q site.
DISCUSSION
The 1 and Mcm1 proteins combine together to form a complex that activates -specific genes by binding to promoter elements termed P’Q (2). To understand the requirements of both proteins for transcriptional activation we characterized the in vivo binding of Mcm1 to P’Q promoter elements found upstream of -specific genes with and without 1. Interestingly, 1 is not required for Mcm1 binding to the promoters of -specific genes in vivo. However, although Mcm1 binds to P’Q sites on its own it fails to activate transcription in the absence of 1 (2). This finding suggests the function of 1 is more complex than simply the recruitment of Mcm1 to P’Q promoter elements. In this study we provide molecular insight into the requirement of 1 for transcriptional activation.
It has been shown that Mcm1 causes a large bend in the DNA when it binds a P-element and there is evidence that supports a model that this bending may be important for transcriptional activation (13,16,17). However, due to the degeneracy on one side of the site, Mcm1 does not bend a P’Q site to the same degree as a P-PAL site. This decrease in bending may explain why Mcm1 fails to activate transcription from a P’Q promoter on its own. Bending of a P’Q element is significantly increased with the Mcm1–1 complex compared with Mcm1 alone, indicating that one function of 1 might be to increase the DNA bend to an angle that promotes transcriptional activation. This model is further supported by our finding that the presence of 1 suppresses the defect in bending by the V34A mutant at a P’Q site. This result may explain why we observed previously that the V34A mutant had a less severe defect in transcriptional activation in complex with 1, compared with activation of a P-PAL element on its own (19).
The 1 protein likely increases the DNA-bend through interaction with Mcm1 by substituting for the lack of optimal Mcm1–DNA contacts at bases 5–8 on the degenerate side of the P’ element, which is directly flanked by the binding site for 1. However, the DNA-bending assays of the P’Q site suggest that Mcm1 alone and in complex with 1 uses different mechanisms to bend the DNA. The mechanism of DNA-bending by Mcm1 on its own at a P’Q element is similar to bending of a P-PAL site. For example, the mcm1 mutants that are defective for bending at a P-PAL, such as V34A, S37A, K40A and T66A, show similar defects when bound on their own to a P’Q element (17). In addition, a T–7G mutation in the context of a P’Q element shows a decrease in DNA-bending by Mcm1 on its own, similar to that seen for P-PAL (16). These data suggest that many of the Mcm1–DNA contacts and the mechanism of Mcm1-mediated DNA bending at a P’Q element are similar to a P-PAL element. In contrast, the mcm1 bending mutants and the T–7G mutation have markedly altered effects when Mcm1 is bound in complex with 1. The mutants do not show significant decreases in DNA-bending in complex with 1. In fact, the T–7G mutation caused an increase in the apparent DNA bending angle mediated by the Mcm1–1 complex. The crystal structure of the Mcm1–2 proteins bound to DNA show a complex DNA bend that does not lie in a single plane (13). It is possible the Mcm1–1 proteins also bend the DNA in multiple planes. One model for the altered bending phenotypes seen with Mcm1–1 compared with Mcm1 alone is that the 1 protein induces a DNA bend in a distinct plane relative to the Mcm1 protein on its own. The bending data provided does not differentiate multiple planes within a DNA bend. However, these data suggest the Mcm1–1 mediated DNA bend has a distinct architecture and that cofactor interaction can markedly alter the DNA bend induced by a MADS-box protein.
Our data suggest the degree of the DNA-bend mediated by Mcm1–1 is important for full transcriptional activation. The P’Q-T–7G site showed greater than a 4-fold decrease in the level of transcriptional activation compared with a wild-type site, which may in part be due to the defect in DNA bending. It is unlikely the decrease in activation was due to binding defects because the relative binding affinities of Mcm1 for these sites differed <1.5-fold (data not shown). Interestingly, the mcm1-V34A mutant, which showed roughly wild-type levels of DNA bending at the P’Q-T–7G site (98°), also activated the P’Q-T–7G reporter significantly better than the wild-type or S37A proteins, which both caused hyper-bending of the mutant site. These data suggest that the restoration of the appropriate DNA-bend angle by V34A at a P’Q-T–7G site results in partial suppression of the transcriptional defect, suggesting there may be a correlation between the proper bend angle with transcriptional activation.
To understand requirements for cofactor interaction we wanted to determine if bending by the Mcm1–DNA binary complex would alter the ability of 1 to bind (18). Our data suggest there is no significant correlation between the degree of bending of the Mcm1–DNA complex and the ability to interact with 1. The V34A, S37A, K40A and T66A mutants, along with the V34A, S37A double mutant, which all showed significant decreases in Mcm1-mediated DNA bending when bound alone, did not show a correlating defect in their ability to interact with 1 in vitro. One exception is the T66E mutation, which caused both a decrease in Mcm1-mediated DNA bending as well as a defect in interaction with 1 (18). The crystal structure of Mcm1 bound to DNA shows the T66 residue is the sole DNA contact from the ?-sheet layer in the protein and is positioned directly above V34 (Fig. 8) (13). The T66E mutation introduces a negative charge that would likely repel the DNA in a fully bent form, resulting in a bending defect (18). However, the S73 residue, which is positioned in the same ?-sheet layer as T66 and is important for interaction with 1, has approximately wild-type levels of DNA bending. The mutation S73R also causes a defect in interaction with 1, suggesting the ?-strand may be involved in a direct interaction with the cofactor 1 (26). These findings suggest that proper DNA bending is not required for Mcm1 to interact with the 1 cofactor.
Figure 8. The Mcm1 residue T66 contributes to both DNA-bending and cofactor interaction (18). Model shown of the Mcm1 dimer is based on the coordinates from the Mcm1–2 DNA ternary complex (13). Side chains of residues required for DNA bending, V34, S37, K40 and T66 are highlighted in the right panel. Side chains shown to be important for interaction with 1, T66 and S73 are highlighted in the left panel.
Protein-mediated DNA bending has been implicated in several cases to be involved in transcriptional regulation: including, Sox2, the estrogen receptor, and Reb1 (28–30). Therefore, establishment of the correct DNA architecture may be critical for proper transcriptional regulation of many genes. The data presented in this work reveals a novel mechanism by which a MADS-box transcription factor can achieve transcriptional activation. Our results demonstrate that interactions by 1 with Mcm1 facilitate the protein-mediated DNA bend and the bend directly correlates with an increase in transcriptional activation. The results also demonstrate that Mcm1 can establish DNA-bends through different mechanisms, one when the Mcm1 protein is bound to the DNA alone and the other when in complex with the cofactor 1. The proper DNA bend by Mcm1 on its own or in complex with 1 may be necessary for the transcriptional co-activators to associate with Mcm1 or serve as scaffolding for the basal transcription machinery.
Acknowledgements
We are grateful to George Sprague for the supply of plasmids and antisera used in this study. This work was funded through the NIH grant (GM49265).
REFERENCES
Shore,P. and Sharrocks,A.D. (1995) The MADS-box family of transcription factors. Eur. J. Biochem., 229, 1–13.
Herskowitz,I., Rine,J. and Strathern,J. (1992) Mating-type Determination and Mating-type Interconversion in Saccharomyces cerevisiae. Cold Spring Harbor Laboratory Press, Cold Spring Harbor, NY, USA.
Kuo,M.H., Nadeau,E.T. and Grayhack,E.J. (1997) Multiple phosphorylated forms of the Saccharomyces cerevisiae Mcm1 protein include an isoform induced in response to high salt concentrations. Mol. Cell. Biol., 17, 819–832.
Messenguy,F. and Dubois,E. (1993) Genetic evidence for a role for MCM1 in the regulation of arginine metabolism in Saccharomyces cerevisiae. Mol. Cell. Biol., 13, 2586–2592.
Lydall,D., Ammerer,G. and Nasmyth,K. (1991) A new role for MCM1 in yeast: cell cycle regulation of SW15 transcription. Genes Dev., 5, 2405–2419.
Ammerer,G. (1990) Identification, purification and cloning of a polypeptide (PRTF/GRM) that binds to mating-specific promoter elements in yeast. Genes Dev., 4, 299–312.
Passmore,S., Maine,G.T., Elble,R., Christ,C. and Tye,B.K. (1988) Saccharomyces cerevisiae protein involved in plasmid maintenance is necessary for mating of MAT alpha cells. J. Mol. Biol., 204, 593–606.
Keleher,C.A., Passmore,S. and Johnson,A.D. (1989) Yeast repressor alpha 2 binds to its operator cooperatively with yeast protein Mcm1. Mol. Cell. Biol., 9, 5228–5230.
Bender,A. and Sprague,G.F.,Jr (1987) MAT alpha 1 protein, a yeast transcription activator, binds synergistically with a second protein to a set of cell-type-specific genes. Cell, 50, 681–691.
Jarvis,E.E., Clark,K.L. and Sprague,G.F. (1989) The yeast transcription activator PRTF, a homolog of the mammalian serum response factor, is encoded by the MCM1 gene. Genes Dev., 3, 936–945.
Pellegrini,L., Tan,S. and Richmond,T.J. (1995) Structure of serum response factor core bound to DNA. Nature, 376, 490–498.
Huang,K., Louis,J.M., Donaldson,L., Lim,F.L., Sharrocks,A.D. and Clore,G.M. (2000) Solution structure of the MEF2A-DNA complex: structural basis for the modulation of DNA bending and specificity by MADS-box transcription factors. EMBO J., 19, 2615–2628.
Tan,S. and Richmond,T.J. (1998) Crystal structure of the yeast MATalpha2/MCM1/DNA ternary complex. Nature, 391, 660–666.
West,A.G., Shore,P. and Sharrocks,A.D. (1997) DNA binding by MADS-box transcription factors: a molecular mechanism for differential DNA bending. Mol. Cell. Biol., 17, 2876–2887.
West,A.G. and Sharrocks,A.D. (1999) MADS-box transcription factors adopt alternative mechanisms for bending DNA. J. Mol. Biol., 286, 1311–1323.
Acton,T.B., Zhong,H. and Vershon,A.K. (1997) DNA-binding specificity of Mcm1: operator mutations that alter DNA-bending and transcriptional activities by a MADS box protein. Mol. Cell. Biol., 17, 1881–1889.
Acton,T.B., Mead,J., Steiner,A.M. and Vershon,A.K. (2000) Scanning mutagenesis of Mcm1: residues required for DNA binding, DNA bending and transcriptional activation by a MADS-box protein. Mol. Cell. Biol., 20, 1–11.
Lim,F.L., Hayes,A., West,A.G., Pic-Taylor,A., Darieva,Z., Morgan,B.A., Oliver,S.G. and Sharrocks,A.D. (2003) Mcm1p-induced DNA bending regulates the formation of ternary transcription factor complexes. Mol. Cell. Biol., 23, 450–461.
Mead,J., Bruning,A.R., Gill,M.K., Steiner,A.M., Acton,T.B. and Vershon,A.K. (2002) Interactions of the Mcm1 MADS box protein with cofactors that regulate mating in yeast. Mol. Cell. Biol., 22, 4607–4621.
Mead,J., Zhong,H., Acton,T.B. and Vershon,A.K. (1996) The yeast alpha2 and Mcm1 proteins interact through a region similar to a motif found in homeodomain proteins of higher eukaryotes. Mol. Cell. Biol., 16, 2135–2143.
Koranda,M., Schleiffer,A., Endler,L. and Ammerer,G. (2000) Forkhead-like transcription factors recruit Ndd1 to the chromatin of G2/M-specific promoters. Nature, 406, 94–98.
Kuo,M.H. and Allis,C.D. (1999) In vivo cross-linking and immunoprecipitation for studying dynamic protein:DNA associations in a chromatin environment. Methods, 19, 425–433.
Hagen,D.C., Bruhn,L., Westby,C.A. and Sprague,G.F.,Jr (1993) Transcription of alpha-specific genes in Saccharomyces cerevisiae: DNA sequence requirements for activity of the coregulator alpha 1. Mol. Cell. Biol., 13, 6866–6875.
Primig,M., Winkler,H. and Ammerer,G. (1991) The DNA binding and oligomerization domain of MCM1 is sufficient for its interaction with other regulatory proteins. EMBO J., 10, 4209–4218.
Tan,S. and Richmond,T.J. (1990) DNA binding-induced conformational change of the yeast transcriptional activator PRTF. Cell, 62, 367–377.
Bruhn,L. and Sprague,G.F.,Jr (1994) MCM1 point mutants deficient in expression of alpha-specific genes: residues important for interaction with alpha 1. Mol. Cell. Biol., 14, 2534–2544.
Boros,J., Lim,F.L., Darieva,Z., Pic-Taylor,A., Harman,R., Morgan,B.A. and Sharrocks,A.D. (2003) Molecular determinants of the cell-cycle regulated Mcm1p-Fkh2p transcription factor complex. Nucleic Acids Res., 31, 2279–2288.
Schultz,J.R., Loven,M.A., Melvin,V.M., Edwards,D.P. and Nardulli,A.M. (2002) Differential modulation of DNA conformation by estrogen receptors alpha and beta. J. Biol. Chem., 277, 8702–8707.
Scaffidi,P. and Bianchi,M.E. (2001) Spatially precise DNA bending is an essential activity of the sox2 transcription factor. J. Biol. Chem., 276, 47296–47302.
Angermayr,M., Oechsner,U. and Bandlow,W. (2003) Reb1p-dependent DNA bending effects nucleosome positioning and constitutive transcription at the yeast profilin promoter. J. Biol. Chem., 278, 17918–17926.(Edward A. Carr, Janet Mead and Andrew K.)
*To whom correspondence should be addressed. Tel: +1 732 445 2905; Fax: +1 732 445 5735; Email: vershon@waksman.rutgers.edu
ABSTRACT
The yeast Mcm1 protein is a founding member of the MADS-box family of transcription factors that is involved in the regulation of diverse sets of genes through interactions with distinct cofactor proteins. Mcm1 interacts with the Mat1 protein to activate the expression of the -cell type-specific genes. To understand the requirement of the cofactor 1 for Mcm1–1-dependent transcriptional activation we analyzed the recruitment of Mcm1 to the promoters of -specific genes in vivo and found that Mcm1 is able to bind to the promoters of -specific genes in the absence of 1. This suggests the function of 1 is more complex than simply recruiting Mcm1. Several MADS-box transcription factors, including Mcm1, induce DNA bending and there is evidence the proper bend may be required for transcriptional activation. We analyzed Mcm1-dependent bending of a Mcm1–1 binding site in the presence and absence of 1 and found that Mcm1 alone shows a reduced DNA-bend at this site compared with other Mcm1 binding sites. However, the addition of 1 markedly increases the DNA-bend and we present evidence this bend is required for full transcriptional activation. These results support a model in which proper DNA-bending by the Mcm1–1 complex is required for transcriptional activation.
INTRODUCTION
The MADS-box family of DNA-binding proteins are involved in transcriptional regulation in a wide array of eukaryotes, including yeast, plants and mammals (1). Mcm1, a MADS-box transcription factor in yeast, is an essential protein involved in regulating diverse sets of genes; including, cell mating-type, cell cycle, osmotic regulation and arginine metabolism (2–5). In order to carry out these regulatory roles Mcm1 interacts with several different cofactors that help determine its function. For example, in cell-type determination, Mcm1 binds P-elements in the promoters of a-specific genes to activate their transcription in the a-cell type (6). However, in the -cell type, Mcm1 interacts with the 2 protein at these promoter elements to repress this same set of genes (7,8). Mcm1 also interacts with the 1 protein to bind P’Q promoter elements to activate the expression of -specific genes (9,10). Therefore, cofactor interaction markedly alters the function and specificity of the Mcm1 MADS-box protein.
A comparison of P’Q elements found upstream of -specific genes to the consensus P-element derived from the promoters of a-specific genes shows they are similar (Fig. 1). However, the P’Q significantly differs from the consensus P-element on the side that is directly flanked by Q, the binding site for 1 (9). Since Mcm1 is unable to activate transcription on its own from a P’Q site, the 1 protein must help Mcm1 by either increasing its ability to bind to the site or to recruit the transcriptional machinery .
Figure 1. A comparison of the consensus P-element to STE3 P’Q. The STE3 P’Q shows strong conservation with the consensus P-element except for the degenerate side, bases 5–8 (bold), which is flanked by the Q site, the binding site for 1. The predicted DNA contacts by Mcm1 with the outer regions of the site based on the crystal structure of Mcm1 are indicated (13). Base-specific contacts are indicated with a solid line, phosphate contacts are indicated with dashed lines. The asterisk indicates the axis of symmetry in the Mcm1 binding site. Positions –7 and 7 are important for DNA bending of the consensus P-element (16).
The crystal structures of several MADS-box proteins show that many produce a significant bend in the DNA when they bind their sites (11–13). Biochemical and genetic analysis suggest that this bending can play an important role in transcriptional activation (14–18). For example, the mcm1-V34A mutant, which reduces the apparent bend angle of a P-element from 95° to 82°, causes a >10-fold decrease in transcriptional activation, yet has only a minor effect on DNA-binding affinity (17,19). Other mutants with larger defects in binding affinity have less of an effect on activation than the V34A mutant, suggesting that bending may play a role in activation. Interestingly, when the mcm1-V34A mutant was assayed for the ability to activate a P’Q element in complex with 1, it showed a less severe defect in transcriptional activation than many other mutants, demonstrating that either activation by Mcm1–1 does not require DNA bending or that 1 helps mediate the bending (19).
To more fully understand the relationship of DNA-bending and cofactor interaction for transcriptional activation, we analyzed bending of a P’Q element with wild-type and mutant Mcm1 proteins alone and in complex with 1. Our results demonstrate the 1 cofactor contributes to the DNA bend mediated by the complex and that a change in the bend angle results in a significant decrease in transcriptional activation. These results support a model of the requirement for a proper DNA-bend by the 1–Mcm1 complex for full transcriptional activation.
MATERIALS AND METHODS
Yeast and bacterial strains and plasmids
Chromatin immunoprecipitations were performed in either the yeast strain JMY041 (MATa ade2-1 trp1-1 his3-11,15 can1-100 ura3-1 leu2-3 mcm1::kanr/pSL1574-MCM1 CEN URA3) or EC7 (MAT ade2-1 trp1-1 his3-11,15 can1-100 ura3-1 leu2-3,112 mat1::TRP1 mcm1::kanr/pSL1574-MCM1 CEN URA3) derived from W303-1A and W303-1B, respectively. The strains were transformed with plasmids that contained either wild-type or mutant derivatives of untagged Mcm1 (pJM231, MCM1 CEN HIS3) or a V5-tagged derivative (pJM421, MCM1-V5 CEN HIS3), both of which complement a mcm1 strain for viability (17). Cells that had lost the plasmid pSL1574 with a wild-type copy of MCM1 were selected by growth in the presence of 5-FOA. In addition, these strains were transformed with pRS415, as an empty vector control, or a derivative of pEC43 (MAT1-Myc CEN LEU2) that expresses MAT1 from its native promoter with a 13-myc tag at the C-terminus. The ORF of MAT1 in pEC43 was engineered with silent restriction sites at the following positions, PstI (bp 61), NheI (bp 95), StuI (bp111), EagI (bp134), XbaI (bp175), NruI (bp 210), SpeI (bp250), EcoR1 (bp 280), SphI (bp 369), XmaI (bp 435), BamHI (bp 530) by recursive PCR as described previously (20). The engineered MAT1 construct complements a mat1 strain for mating and activates transcription of a heterologous STE3-lacZ promoter at the same level as an unmodified gene.
Transcription assays
Mcm1–1-dependent transcriptional activation was monitored in the yeast strain EC7 by monitoring lacZ expression from reporter constructs derived from pTBA23 (2 μ URA3) containing a CYC1-lacZ reporter promoter with the CYC1 UAS deleted (16). The transcription reporter constructs contained either a single P’Q wild-type site from STE3 (pEC25) or a P’Q site with a T–7 to G mutation (P’Q-T–7G) (pEC111) (9). ?-Galactosidase assays were performed in triplicate with independent isolates and the standard deviations of the calculated ?-galactosidase values were <10% (17).
Chromatin immunoprecipitation assays
Chromatin immunoprecipitation assays of Mcm1 binding were performed based on the procedures described previously (21). Cultures (50 ml) were grown to an OD600 of 0.8 and were cross-linked with 1% formaldehyde for 30 min. The cells were washed twice in ice-cold TBS and frozen at –80°C. The cells were then suspended in 500 μl FA lysis buffer and 0.5 g acid washed glass beads and were vortexed 40 min at 4°C. The lysates were cleared by centrifugation for 1 min at 14 000 r.p.m. at 4°C. The supernatants were sonicated 6x for 5 s to shear the chromatin to 500 bp fragments and then centrifuged for 5 min at 14 000 r.p.m. at 4°C. A total of 50 μl of the supernatant was removed for total chromatin (TC) control. The remaining supernatant was pre-cleared with protein G–agarose beads and then incubated overnight with 1 μl anti-V5 antibody (Invitrogen). Protein G–agarose beads (50 μl) were incubated for 1 h and the beads were washed 1x in low salt wash (0.1% SDS, 1% Triton X-100, 150 mM NaCl 2x TE); 1x in high salt wash (0.1% SDS, 1% Triton X-100, 500 mM NaCl 2x TE); 1x in LiCl wash (0.25 M LiCl, 1% IGEPAL, 1x TE, 1% sodium deoxycholate); and 2x in 1x TE. Protein–DNA complexes were eluted from the beads by washing 2x with 250 μl elution buffer (1% SDS, 0.1 M NaHCO3). Cross-links were reversed by the addition of 20 μl 5 M NaCl and incubating at 65°C for 4 h. Protease digestion was performed by adding 10 μl 0.5 M EDTA, 20 μl Tris pH 7.4, 2 μl proteinase K (10 μg/μl) and incubating at 42°C for 45 min. The samples were then phenol–chloroform extracted and the DNA precipitated with ethanol.
Quantitative PCR was performed as described previously (22) using 1/50 of immunoprecipitated DNA fragments and 1/1000 of the TC material in a 50 μl total reaction volume. PCR amplification of STE6 or STE2 promoter region served as a positive control. Primers to amplify the promoters of ACT1 or YDL223C served as negative controls.
Protein purification
The wild-type and mutant Mcm1 proteins used in the in vitro assays were purified as described previously except that the proteins were expressed in Escherichia coli BL21 (codon +) and the maltose-binding protein (MBP) was not removed after thrombin cleavage (17). The Mcm1 proteins isolated were >95% homogeneous. The 1 protein was purified as an MBP–1 fusion from BL21 (codon +) bacteria cells expressing pSL2187 (23) grown in LB media containing ampicillin and chloramphenicol to an OD600 of 0.6 followed by induction with 0.3 mM isopropyl-?-D-thiogalactopyranoside for 20 h at 14°C. Cells were lysed and protein was purified to >95% homogeneity as described previously exclusive of cleavage of the MBP (23). Protein concentrations were normalized by Bradford assays and verified on Coomassie stained SDS–PAGE gels.
Electrophoretic mobility shift assays
Oligonucleotides containing STE3 P’Q wild type or P’Q-T–7G were end-labeled with ATP using polynucleotide kinase and purified by Qiagen nucleotide removal columns (Qiagen) according to the manufacturer. The oligonucleotides were made double-stranded by mixing with a 3-fold excess of the matching strand, incubating at 90°C for 20 min and slowly cooling to 25°C overnight in a water bath. The circular permutation assays were performed with BamHI, NheI, HindIII or EcoRI-generated end-labeled fragments as described previously (16) containing either a P-PAL site, STE3-P’Q or P’Q-T–7G site. All binding reactions with purified Mcm11–97 or MBP–1 were carried out in 10 mM Tris–HCl (pH 7.5), 40 mM NaCl, 4 mM MgCl2, 6% (w/v) glycerol, 10 mg/ml BSA, 10 μg/ml of sonicated salmon sperm DNA and 32P-labeled DNA probes (4500 c.p.m.) in a total volume of 30 μl at room temperature for 60 min. All protein dilutions were made in 20 mM Tris–HCl (pH 8), 50 mM NaCl, 1 mM EDTA, 1 mg/ml BSA, 5 mM ?-mercaptoethanol and 1 mM PMSF. Samples were analyzed on a 6% polyacrylamide gel . Gels were dried after electrophoresis, exposed to a phosphor screen and scanned on a Model Storm 840 Molecular Dynamics phosphorimager. Gels were quantified using IPlab Gel imaging software.
RESULTS
Mcm1 binds in vivo in the absence of 1
In vivo DNA footprint analysis led to the conclusion that Mcm1 is unable to bind P’Q elements in the absence of 1, suggesting, 1 may function primarily to recruit Mcm1 to P’Q elements in vivo (24). However, Mcm1 showed only a slightly reduced binding affinity (3-fold) in vitro for a P’Q site compared with a P-element, suggesting Mcm1 may be bound on its own in vivo (25). To characterize the in vivo binding of Mcm1 to P’Q elements alone and in complex with 1 chromatin immunoprecipitations were performed (Fig. 2). As expected, in the presence of 1, Mcm1 strongly bound to the promoters of -specific genes (Fig. 2, lane 7). Interestingly, there were also significant levels of Mcm1 binding to these sites even in the absence of 1 (Fig. 2, lane 3). Since Mcm1 bound to the promoters of -specific genes in vivo but failed to activate transcription, it suggests 1 has a greater role in transcriptional activation than simply recruiting Mcm1 to these promoters.
Figure 2. Mcm1 binds P’Q elements in the absence of 1 in vivo. Chromatin immunoprecipitations were performed with lysates containing untagged (lanes 1, 2, 5 and 6) or V5-epitope tagged Mcm1 (lanes 3, 4, 7 and 8) using anti-V5 antibody. A MATa strain was transformed with either an empty vector (pRS415, lanes 1–4) or wild-type 1 (pEC43, lanes 5–8). The panel shows an ethidium bromide stained gel of PCR amplifications of the indicated promoter region. PCR product using primers to the -specific genes STE3, SAG1 and MF1 are indicated. A promoter fragment of the a-specific STE6 gene bound by Mcm1 served as a positive control. YDL223C, a gene not regulated by Mcm1, was used as a negative control. Lanes 1, 4, 5 and 8 show PCR products using TC as a template, whereas lanes 2, 3, 6 and 7 used immunoprecipitated product (IP).
The 1 protein increases the bend of Mcm1 bound to a P’Q element
There is evidence that DNA bending may play a role in transcriptional regulation by MADS-box proteins (16–18). To determine if 1 affects the degree of DNA bending at a P’Q element with Mcm1 we performed EMSAs with circularly permuted fragments that contain P’Q sites. As demonstrated previously, Mcm1 induced an apparent bend angle of roughly 99° at a P-PAL element (Fig. 3, lanes 1–4) (16). However, the STE3 P’Q site was only bent with an apparent angle of 76° (Fig. 3, lanes 9–12). The decreased bend angle is presumably due to sub-optimal contacts at bases +5 to +8 in the degenerate side of the P’Q site. DNA-bending of the P’Q element was further reduced by mcm1-V34A, which contains a mutation at a residue required for DNA-bending . In the presence of both 1 and Mcm1 there was a shifted band with significantly slower mobility (Fig. 3, lanes 17–20). The presence of 1 significantly increased the bend of the P’Q site to an apparent angle of 94°. Interestingly, in the presence of 1 the mcm1-V34A mutant had a similar level of DNA bending as the wild-type protein (Fig. 3, compare lanes 17–20 with 21–24). These results indicate that the presence of 1 increases the DNA bend at a P’Q site and is able to suppress the bending defect caused by the V34A mutation.
Figure 3. The 1–Mcm1 complex induces a greater bend in the DNA than Mcm1 alone. An EMSA utilizing position permutation of either a P-PAL (lanes 1–8) or STE3-P’Q site (lanes 9–24) is shown. Lanes 1–4, 9–12 and 17–20 show binding by the wild-type Mcm1 protein. Lanes 5–8, 13–16 and 21–24 show binding by the mcm1-V34A mutant. Lanes 1–16 indicate binding by Mcm1 alone, whereas lanes 17–24 show binding in the presence of 1. The calculated bend angle for each binding complex is shown.
In addition to V34A, several other mutations in Mcm1 cause bending defects at a P-element (17,18). On their own, each of the mcm1 mutants showed a decrease in bending of the P’Q site (Fig. 4B–F, lanes 1–4). The relative decreases in bending were comparable with that seen for these mutants binding to a P-element (17,18). For example, the T66A mutation only caused a slight decrease in the apparent bend angle (Fig. 4E, lanes 1–4), whereas the V34A, S37A double mutant caused a substantial decrease (Fig. 4F, lanes 1–4). Although these mutants showed significant decreases in bending on their own, in complex with 1 none of the mutants significantly altered the apparent bend angle in comparison with the wild-type protein (Fig. 4A–F, lanes 9–12).
Figure 4. The Mcm1–1 complex has increased bending at the P’Q-T–7G site. An EMSA utilizing position permutation of the binding sites, with each panel representing either wild-type Mcm1 (A) or mcm1-mutants V34A (B), S37A (C), K40A (D), T66A (E) and V34A/S37A (F). Lanes 1–4 and 9–12 of each panel show binding to a STE3-P’Q site. Lanes 5–8 and 13–16 in each panel show binding to a STE3-P’Q site with a T–7G substitution. Lanes 1–8 are in the absence of 1 and lanes 9–16 are in the presence of 1. The calculated bend angle for each complex is shown.
To further test if 1 affects the bending of the complex we examined binding to a mutant P’Q site that contains a base pair substitution, T–7G. In the context of a P-site this substitution significantly affects Mcm1-dependent DNA bending (17,18). The T–7G substitution also caused a decrease in bending by Mcm1 in the context of the P’Q site (Fig. 4A, lanes 5–8 versus 1–4). The difference in the apparent bend angle was comparable with the decrease observed for a P-element with a similar mutation (16). The S37A, K40A and T66A mutations in Mcm1 caused a further decrease in the apparent bend angle of the P’Q-T–7G site (Fig. 4C, D and F, lanes 1–4 versus 5–8). However, V34A and the V34A, S37A double mutant showed no further decrease in bending with the P’Q-T–7G site (Fig. 4B and F, lanes 1–8). This likely indicates a direct contact between V34 and the T–7 position, as seen previously for Mcm1 at a P-element (13,17). Interestingly, in complex with 1, the bend angle of the P’Q-T–7G site mediated by the WT, S37A, K40A or T66A mutants was not decreased, but was rather significantly increased above the levels of bending by the wild-type protein to the wild-type site (Fig. 4A, C, D and E, lanes 9–12 versus 13–16.) In contrast, V34A and the V34A, S37A double mutant did not show this further increase in bending but bent the DNA with an apparent bend angle that was similar to the one caused by the wild-type protein binding to the wild-type site (Fig. 4B and F, compare lanes 9–12 with 13–16). These data indicate that in complex with 1, the mcm1-V34A mutation suppresses the increased bending defect due to the P’Q-T–7G element.
1 interaction with the Mcm1–DNA complex is not altered by changes in DNA bending
We next compared the relative binding affinities of the Mcm1 mutants, alone and in complex with 1 at a wild-type P’Q and the P’Q-T–7G to determine if any of these amino acid substitutions affected the formation of the complex. The relative binding affinities of the mcm1 mutants for a P’Q site were all comparable with the affinities for a P-element (17). With the exception of K40A, the Mcm1 mutants bound to a P’Q with less than a 2.5-fold reduction compared with wild-type protein (data not shown). In addition, the P’Q-T–7G mutation did not greatly alter the binding affinity of wild-type or mutant Mcm1 proteins (data not shown).
To determine if 1 interaction with the Mcm1–DNA complex was affected by the degree of DNA bending we analyzed this interaction by EMSA. The formation of the binary Mcm1–DNA complex was normalized to account for differences in mcm1 DNA binding at P’Q and at P’Q-T–7G. The ability of 1 to interact with the Mcm1–DNA complex was unaffected, even by severe mcm1 DNA-bending mutants (Fig. 5). In addition, the P’Q-T–7G mutation did not cause a decrease in 1-binding affinity. The K40A mutant showed the opposite effect with 1 interaction, partially suppressing the binding defect. There was a decrease at a P’Q-T–7G with V34A and the V34A-S37A double mutant in 1 interaction showing an approximate 2-fold decrease. However, overall there was no correlation between the degree of bending of the binary Mcm1–DNA complex and the ability of 1 to interact with the complex.
Figure 5. Mutations that affect Mcm1-dependent bending do not alter the interaction with 1. Wild-type and mutant Mcm1 proteins were normalized for DNA-binding affinity on their own. Serial dilutions (3-fold) of 1 were added to each of the Mcm1 mutants. The fold decrease for 1 interaction with the Mcm1–DNA complex is indicated relative to the Mcm1 wild-type protein.
Two mutants shown previously to be defective in interaction with 1, mcm1-T66E and mcm1-S73R, were analyzed for bending in the presence or absence of 1 to determine if defects in interaction would result in a bending deficiency (26). As reported previously, both mutants showed a significant defect in interaction with 1 (Fig. 6A) (26,27). In the absence of 1 the relative bending of the T66E mutant was significantly reduced compared with wild-type Mcm1, whereas the S73R mutant showed levels comparable with wild type (Fig. 6B). However, in complex with 1 both mutants showed apparent bend angles comparable with the wild-type protein. These data indicate that defects in the interaction between 1 and Mcm1 do not necessarily result in bending defects.
Figure 6. Defects in interaction with 1 do not result in bending defects in the Mcm1–1 complex. (A) An EMSA with WT-Mcm1, T66E or S73R bound to a P’Q site (lanes 2, 6 or 10, respectively) or in complex with 1 (lanes 3–5, 7–9 or 11–13, respectively). The Mcm1 protein concentration is constant in each lane with 3-fold serial dilutions 1. (B) A circular permutation assay is shown in the absence of 1 (lanes 1–12) or in the presence of 1 (lanes 13–24) with Mcm1-WT (lanes 1–4, 13–16), T66E (lanes 5–8, 17–20) and S73R (lanes 9–12, 21–24). The calculated bend angle for each complex is shown.
The 1–Mcm1 DNA bend must be at a proper level for full transcriptional activation
To determine if changes in the apparent bend angle caused by the T–7G mutation altered transcriptional activation we constructed heterologous reporter promoters driving lacZ transcription that contained either a single P’Q or P’Q-T–7G site. In a wild-type strain activation from the T–7G reporter was decreased greater than 4-fold compared with the wild-type site (Fig. 7). Since the DNA-binding affinity of the 1–Mcm1 complex for P’Q-T–7G was comparable with the wild-type site it suggests that the increase in the bend angle affects activation. We also analyzed the ability of mcm1-mutants to drive expression from a P’Q and P’Q-T–7G site. The S37A mutant showed levels comparable with wild-type protein at both the P’Q and P’Q-T–7G site (Fig. 7). Since both the wild-type and S37A proteins showed increased bending of the P’Q-T–7G, it is possible the change in the apparent bending angle causes a decrease in activation at the mutant site. Although the V34A mutant showed a small reduction in activation from a wild-type P’Q, this mutant was able to partially suppress the decrease in activation by the P’Q-T–7G site, which correlates with the ability to suppress the bending defect (Fig. 7). The T66E and S73R mutants both showed significant decreases in transcriptional activation at both sites, which is likely due to defects in interaction with 1 (Fig. 7) (18,26).
Figure 7. In vivo assays to determine the effect of DNA-bending on transcriptional activation. Reporter promoters containing either a P’Q (solid bars) or P’Q-T–7G (open bars) were used to drive lacZ expression in a WT or mutant Mcm1 background. Activation of the reporter in MAT cells was determined from the average of three independent assays that differed by <10%. Percentages of activation are shown relative to the WT Mcm1 protein at a WT-P’Q site.
DISCUSSION
The 1 and Mcm1 proteins combine together to form a complex that activates -specific genes by binding to promoter elements termed P’Q (2). To understand the requirements of both proteins for transcriptional activation we characterized the in vivo binding of Mcm1 to P’Q promoter elements found upstream of -specific genes with and without 1. Interestingly, 1 is not required for Mcm1 binding to the promoters of -specific genes in vivo. However, although Mcm1 binds to P’Q sites on its own it fails to activate transcription in the absence of 1 (2). This finding suggests the function of 1 is more complex than simply the recruitment of Mcm1 to P’Q promoter elements. In this study we provide molecular insight into the requirement of 1 for transcriptional activation.
It has been shown that Mcm1 causes a large bend in the DNA when it binds a P-element and there is evidence that supports a model that this bending may be important for transcriptional activation (13,16,17). However, due to the degeneracy on one side of the site, Mcm1 does not bend a P’Q site to the same degree as a P-PAL site. This decrease in bending may explain why Mcm1 fails to activate transcription from a P’Q promoter on its own. Bending of a P’Q element is significantly increased with the Mcm1–1 complex compared with Mcm1 alone, indicating that one function of 1 might be to increase the DNA bend to an angle that promotes transcriptional activation. This model is further supported by our finding that the presence of 1 suppresses the defect in bending by the V34A mutant at a P’Q site. This result may explain why we observed previously that the V34A mutant had a less severe defect in transcriptional activation in complex with 1, compared with activation of a P-PAL element on its own (19).
The 1 protein likely increases the DNA-bend through interaction with Mcm1 by substituting for the lack of optimal Mcm1–DNA contacts at bases 5–8 on the degenerate side of the P’ element, which is directly flanked by the binding site for 1. However, the DNA-bending assays of the P’Q site suggest that Mcm1 alone and in complex with 1 uses different mechanisms to bend the DNA. The mechanism of DNA-bending by Mcm1 on its own at a P’Q element is similar to bending of a P-PAL site. For example, the mcm1 mutants that are defective for bending at a P-PAL, such as V34A, S37A, K40A and T66A, show similar defects when bound on their own to a P’Q element (17). In addition, a T–7G mutation in the context of a P’Q element shows a decrease in DNA-bending by Mcm1 on its own, similar to that seen for P-PAL (16). These data suggest that many of the Mcm1–DNA contacts and the mechanism of Mcm1-mediated DNA bending at a P’Q element are similar to a P-PAL element. In contrast, the mcm1 bending mutants and the T–7G mutation have markedly altered effects when Mcm1 is bound in complex with 1. The mutants do not show significant decreases in DNA-bending in complex with 1. In fact, the T–7G mutation caused an increase in the apparent DNA bending angle mediated by the Mcm1–1 complex. The crystal structure of the Mcm1–2 proteins bound to DNA show a complex DNA bend that does not lie in a single plane (13). It is possible the Mcm1–1 proteins also bend the DNA in multiple planes. One model for the altered bending phenotypes seen with Mcm1–1 compared with Mcm1 alone is that the 1 protein induces a DNA bend in a distinct plane relative to the Mcm1 protein on its own. The bending data provided does not differentiate multiple planes within a DNA bend. However, these data suggest the Mcm1–1 mediated DNA bend has a distinct architecture and that cofactor interaction can markedly alter the DNA bend induced by a MADS-box protein.
Our data suggest the degree of the DNA-bend mediated by Mcm1–1 is important for full transcriptional activation. The P’Q-T–7G site showed greater than a 4-fold decrease in the level of transcriptional activation compared with a wild-type site, which may in part be due to the defect in DNA bending. It is unlikely the decrease in activation was due to binding defects because the relative binding affinities of Mcm1 for these sites differed <1.5-fold (data not shown). Interestingly, the mcm1-V34A mutant, which showed roughly wild-type levels of DNA bending at the P’Q-T–7G site (98°), also activated the P’Q-T–7G reporter significantly better than the wild-type or S37A proteins, which both caused hyper-bending of the mutant site. These data suggest that the restoration of the appropriate DNA-bend angle by V34A at a P’Q-T–7G site results in partial suppression of the transcriptional defect, suggesting there may be a correlation between the proper bend angle with transcriptional activation.
To understand requirements for cofactor interaction we wanted to determine if bending by the Mcm1–DNA binary complex would alter the ability of 1 to bind (18). Our data suggest there is no significant correlation between the degree of bending of the Mcm1–DNA complex and the ability to interact with 1. The V34A, S37A, K40A and T66A mutants, along with the V34A, S37A double mutant, which all showed significant decreases in Mcm1-mediated DNA bending when bound alone, did not show a correlating defect in their ability to interact with 1 in vitro. One exception is the T66E mutation, which caused both a decrease in Mcm1-mediated DNA bending as well as a defect in interaction with 1 (18). The crystal structure of Mcm1 bound to DNA shows the T66 residue is the sole DNA contact from the ?-sheet layer in the protein and is positioned directly above V34 (Fig. 8) (13). The T66E mutation introduces a negative charge that would likely repel the DNA in a fully bent form, resulting in a bending defect (18). However, the S73 residue, which is positioned in the same ?-sheet layer as T66 and is important for interaction with 1, has approximately wild-type levels of DNA bending. The mutation S73R also causes a defect in interaction with 1, suggesting the ?-strand may be involved in a direct interaction with the cofactor 1 (26). These findings suggest that proper DNA bending is not required for Mcm1 to interact with the 1 cofactor.
Figure 8. The Mcm1 residue T66 contributes to both DNA-bending and cofactor interaction (18). Model shown of the Mcm1 dimer is based on the coordinates from the Mcm1–2 DNA ternary complex (13). Side chains of residues required for DNA bending, V34, S37, K40 and T66 are highlighted in the right panel. Side chains shown to be important for interaction with 1, T66 and S73 are highlighted in the left panel.
Protein-mediated DNA bending has been implicated in several cases to be involved in transcriptional regulation: including, Sox2, the estrogen receptor, and Reb1 (28–30). Therefore, establishment of the correct DNA architecture may be critical for proper transcriptional regulation of many genes. The data presented in this work reveals a novel mechanism by which a MADS-box transcription factor can achieve transcriptional activation. Our results demonstrate that interactions by 1 with Mcm1 facilitate the protein-mediated DNA bend and the bend directly correlates with an increase in transcriptional activation. The results also demonstrate that Mcm1 can establish DNA-bends through different mechanisms, one when the Mcm1 protein is bound to the DNA alone and the other when in complex with the cofactor 1. The proper DNA bend by Mcm1 on its own or in complex with 1 may be necessary for the transcriptional co-activators to associate with Mcm1 or serve as scaffolding for the basal transcription machinery.
Acknowledgements
We are grateful to George Sprague for the supply of plasmids and antisera used in this study. This work was funded through the NIH grant (GM49265).
REFERENCES
Shore,P. and Sharrocks,A.D. (1995) The MADS-box family of transcription factors. Eur. J. Biochem., 229, 1–13.
Herskowitz,I., Rine,J. and Strathern,J. (1992) Mating-type Determination and Mating-type Interconversion in Saccharomyces cerevisiae. Cold Spring Harbor Laboratory Press, Cold Spring Harbor, NY, USA.
Kuo,M.H., Nadeau,E.T. and Grayhack,E.J. (1997) Multiple phosphorylated forms of the Saccharomyces cerevisiae Mcm1 protein include an isoform induced in response to high salt concentrations. Mol. Cell. Biol., 17, 819–832.
Messenguy,F. and Dubois,E. (1993) Genetic evidence for a role for MCM1 in the regulation of arginine metabolism in Saccharomyces cerevisiae. Mol. Cell. Biol., 13, 2586–2592.
Lydall,D., Ammerer,G. and Nasmyth,K. (1991) A new role for MCM1 in yeast: cell cycle regulation of SW15 transcription. Genes Dev., 5, 2405–2419.
Ammerer,G. (1990) Identification, purification and cloning of a polypeptide (PRTF/GRM) that binds to mating-specific promoter elements in yeast. Genes Dev., 4, 299–312.
Passmore,S., Maine,G.T., Elble,R., Christ,C. and Tye,B.K. (1988) Saccharomyces cerevisiae protein involved in plasmid maintenance is necessary for mating of MAT alpha cells. J. Mol. Biol., 204, 593–606.
Keleher,C.A., Passmore,S. and Johnson,A.D. (1989) Yeast repressor alpha 2 binds to its operator cooperatively with yeast protein Mcm1. Mol. Cell. Biol., 9, 5228–5230.
Bender,A. and Sprague,G.F.,Jr (1987) MAT alpha 1 protein, a yeast transcription activator, binds synergistically with a second protein to a set of cell-type-specific genes. Cell, 50, 681–691.
Jarvis,E.E., Clark,K.L. and Sprague,G.F. (1989) The yeast transcription activator PRTF, a homolog of the mammalian serum response factor, is encoded by the MCM1 gene. Genes Dev., 3, 936–945.
Pellegrini,L., Tan,S. and Richmond,T.J. (1995) Structure of serum response factor core bound to DNA. Nature, 376, 490–498.
Huang,K., Louis,J.M., Donaldson,L., Lim,F.L., Sharrocks,A.D. and Clore,G.M. (2000) Solution structure of the MEF2A-DNA complex: structural basis for the modulation of DNA bending and specificity by MADS-box transcription factors. EMBO J., 19, 2615–2628.
Tan,S. and Richmond,T.J. (1998) Crystal structure of the yeast MATalpha2/MCM1/DNA ternary complex. Nature, 391, 660–666.
West,A.G., Shore,P. and Sharrocks,A.D. (1997) DNA binding by MADS-box transcription factors: a molecular mechanism for differential DNA bending. Mol. Cell. Biol., 17, 2876–2887.
West,A.G. and Sharrocks,A.D. (1999) MADS-box transcription factors adopt alternative mechanisms for bending DNA. J. Mol. Biol., 286, 1311–1323.
Acton,T.B., Zhong,H. and Vershon,A.K. (1997) DNA-binding specificity of Mcm1: operator mutations that alter DNA-bending and transcriptional activities by a MADS box protein. Mol. Cell. Biol., 17, 1881–1889.
Acton,T.B., Mead,J., Steiner,A.M. and Vershon,A.K. (2000) Scanning mutagenesis of Mcm1: residues required for DNA binding, DNA bending and transcriptional activation by a MADS-box protein. Mol. Cell. Biol., 20, 1–11.
Lim,F.L., Hayes,A., West,A.G., Pic-Taylor,A., Darieva,Z., Morgan,B.A., Oliver,S.G. and Sharrocks,A.D. (2003) Mcm1p-induced DNA bending regulates the formation of ternary transcription factor complexes. Mol. Cell. Biol., 23, 450–461.
Mead,J., Bruning,A.R., Gill,M.K., Steiner,A.M., Acton,T.B. and Vershon,A.K. (2002) Interactions of the Mcm1 MADS box protein with cofactors that regulate mating in yeast. Mol. Cell. Biol., 22, 4607–4621.
Mead,J., Zhong,H., Acton,T.B. and Vershon,A.K. (1996) The yeast alpha2 and Mcm1 proteins interact through a region similar to a motif found in homeodomain proteins of higher eukaryotes. Mol. Cell. Biol., 16, 2135–2143.
Koranda,M., Schleiffer,A., Endler,L. and Ammerer,G. (2000) Forkhead-like transcription factors recruit Ndd1 to the chromatin of G2/M-specific promoters. Nature, 406, 94–98.
Kuo,M.H. and Allis,C.D. (1999) In vivo cross-linking and immunoprecipitation for studying dynamic protein:DNA associations in a chromatin environment. Methods, 19, 425–433.
Hagen,D.C., Bruhn,L., Westby,C.A. and Sprague,G.F.,Jr (1993) Transcription of alpha-specific genes in Saccharomyces cerevisiae: DNA sequence requirements for activity of the coregulator alpha 1. Mol. Cell. Biol., 13, 6866–6875.
Primig,M., Winkler,H. and Ammerer,G. (1991) The DNA binding and oligomerization domain of MCM1 is sufficient for its interaction with other regulatory proteins. EMBO J., 10, 4209–4218.
Tan,S. and Richmond,T.J. (1990) DNA binding-induced conformational change of the yeast transcriptional activator PRTF. Cell, 62, 367–377.
Bruhn,L. and Sprague,G.F.,Jr (1994) MCM1 point mutants deficient in expression of alpha-specific genes: residues important for interaction with alpha 1. Mol. Cell. Biol., 14, 2534–2544.
Boros,J., Lim,F.L., Darieva,Z., Pic-Taylor,A., Harman,R., Morgan,B.A. and Sharrocks,A.D. (2003) Molecular determinants of the cell-cycle regulated Mcm1p-Fkh2p transcription factor complex. Nucleic Acids Res., 31, 2279–2288.
Schultz,J.R., Loven,M.A., Melvin,V.M., Edwards,D.P. and Nardulli,A.M. (2002) Differential modulation of DNA conformation by estrogen receptors alpha and beta. J. Biol. Chem., 277, 8702–8707.
Scaffidi,P. and Bianchi,M.E. (2001) Spatially precise DNA bending is an essential activity of the sox2 transcription factor. J. Biol. Chem., 276, 47296–47302.
Angermayr,M., Oechsner,U. and Bandlow,W. (2003) Reb1p-dependent DNA bending effects nucleosome positioning and constitutive transcription at the yeast profilin promoter. J. Biol. Chem., 278, 17918–17926.(Edward A. Carr, Janet Mead and Andrew K.)