Mid1p-dependent regulation of the M–G1 transcription wave in fission yeast

The control of gene expression at certain times during the mitotic cell division cycle is a common feature in eukaryotes. In fission yeast, at least five waves of gene expression have been described, with one transcribed at the M–G1 interval under the control of the PBF transcription factor complex. PBF consists of at least three transcription factors, two forkhead-like proteins Sep1p and Fkh2p, and a MADS box-like protein Mbx1p, and binds to PCB motifs found in the gene promoters. Mbx1p is under the direct control of the polo-like kinase Plo1p and the Cdc14p-like phosphatase Clp1p (Flp1p). Here, we show that M–G1 gene expression in fission yeast is also regulated by the anillin-like protein, Mid1p (Dmf1p). Mid1p binds in vivo to both Fkh2p and Sep1p, and to the promoter regions of M–G1 transcribed genes. Mid1p promoter binding is dependent on Fkh2p, Plo1p and Clp1p. The absence of mid1+ in cells results in partial loss of M–G1 specific gene expression, suggesting that it has a negative role in controlling gene expression. This phenotype is exacerbated by also removing clp1+, suggesting that Mid1p and Clp1p have overlapping functions in controlling transcription. As mid1+ is itself expressed at M–G1, these observations offer a new mechanism whereby Mid1p contributes to controlling cell cycle-specific gene expression as part of a feedback loop.


Introduction
The process by which a cell duplicates and divides to produce two identical daughter cells is controlled by many different mechanisms. Prominent amongst these is the specific regulation of gene expression, so that proteins required at particular cell cycle times are produced only when they are needed. A simple paradigm that exists amongst eukaryotes to achieve this goal is groups of genes transcribed at different times in the cell cycle. Cell cycle-specific expression of each group is controlled by a combination of short, repeated DNA motifs found in the gene promoters to which a transcription factor complex binds. It is the combination of the promoter sequences and transcription factor complex, particular to each group of genes, that ensures that gene expression occurs only at a certain cell cycle time (Spellman et al., 1998;Rustici et al., 2004;McInerny, 2004;Bähler, 2005).
In fission yeast, an important wave of gene expression occurs at the end of the cell cycle during late M phase and cytokinesis, the M-G1 interval (Anderson et al., 2002;Rustici et al., 2004). The genes are co-ordinately regulated by DNA sequence motifs called pombe cell cycle boxes (PCBs) found in the gene promoters, which are bound by a transcription complex, PCB binding factor (PBF). Three components of PBF have been identified with their functions partially characterised: two forkhead-like transcription factors, Sep1p and Fkh2p, and a MADS box-like protein, Mbx1p (Buck et al., 2004;Bulmer et al., 2004;Papadopoulou et al., 2008). PBF regulation of gene expression appears to be controlled by a combination of cell cycle-specific binding of Sep1p and Fkh1p, with Sep1p stimulating, and Fkh2p inhibiting M-G1 specific transcription. Furthermore, Mbx1p phosphorylation status is controlled by the polo-like kinase, Plo1p, and the Cdc14p-like phosphatase Clp1p (Flp1p) (Papadopoulou et al., 2008;Papadopoulou et al., 2010). As plo1 + is itself under PBF-PCB control, this potentially forms a feedback loop to control gene expression at M-G1. Related mechanisms appear to be operating in budding yeast and humans, suggesting that such processes are conserved and important (Darieva et al., 2006;Fu et al., 2008).
The M-G1 wave of expression controls many genes functioning during cytokinesis (Anderson et al., 2002;Rustici et al., 2004). This process involves the assembly of an actomyosin contractile ring in the cell middle, whose contraction produces two daughter cells of equal size. Medial positioning of the contractile ring is controlled by anillin-like protein Mid1p (Dmf1p) (Chang et al., 1996;Sohrmann et al., 1996), whose expression is controlled by PBF-PCB (Anderson et al., 2002). During interphase, Mid1p binds to cortical nodes organized by Cdr2p at the medial cortex (Almonacid et al., 2009), where it recruits components of the contractile ring at the onset of mitosis to initiate contractile ring assembly in the cell middle (Pollard and Wu, 2010). Mid1p also spends part of the cell cycle in the nucleus, and this nuclear function appears to be important because the export of Mid1p from the nucleus couples the position of the division plane to the position of the nucleus (Paoletti and Chang, 2000;Almonacid et al., 2009). It is noteworthy that Mid1p interacts with both Plo1p kinase and Clp1p phosphatase to anchor them to the contractile ring in mitosis (Bähler et al., 1998;Clifford et al., 2008), whereas Plo1p triggers Mid1p export from the nucleus (Bähler et al., 1998).
Here, we explore further the nuclear function of Mid1p. Because the mid1 + gene is under PBF-PCB control, we examined whether Mid1p might have a role in regulating M-G1 gene expression. We found that Mid1p binds to both Fkh2p and Sep1p and to the promoters of genes containing PCB motifs expressed at the M-G1 boundary. Furthermore, Mid1p binding to PCBs requires Fkh2p, Plo1p and Clp1p, and is necessary for correct gene expression at this cell cycle time. These experiments reveal a novel role for an anillin-like protein in controlling cell cycle-specific gene expression.

mid1 + shows genetic interactions with genes encoding components of PBF
To explore the reasons why Mid1p should spend part of the cell cycle in the nucleus, and as the mid1 + gene is expressed at M-G1 under PBF-PCB control (Paoletti and Chang, 2000;Anderson et al., 2002), we searched for synthetic phenotypes in double mutants of mid1 and mutants in genes encoding components of PBF. No obvious synthetic phenotypes were seen in mid1 mbx1 and mid1 sep1 double deletion mutants (data not shown). However, we found that a mid1 fkh2 double deletion mutant is synthetically lethal (supplementary material Fig. S1A). We extended this observation by creating double mutants of fkh2 with various nuclear localisation signal mutants in mid1 (Paoletti and Chang, 2000;Almonacid et al., 2009), and found that these also revealed synthetic phenotypes, being considerably sicker than the parents, showing very slow growth, along with septation defects similar to those observed in fkh2 strains (supplementary material Fig. S1A,B; data not shown). As Mid1p nuclear localization mutants do not exhibit septation defects unless the nuclear position is artificially displaced by centrifugation (Almonacid et al., 2009), these data suggest that the synthetic phenotype with fkh2 might be linked to a novel nuclear function of Mid1p.

Mid1p binds to both Fkh2p and to Sep1p in vivo
One possible explanation for genetic interactions between genes is that the encoded proteins interact in vivo. We tested this by completing co-immunoprecipitation experiments using tagged versions of Mid1p, Fkh2p and Sep1p, expressed from their native promoters (Fig. 1A). Immunoprecipitation of Mid1p resulted in limited but reproducible co-immunoprecipitation of both Fkh1p and Sep1p. Relatively smaller amounts of Sep1p were detected compared with Fkh2p, but these experiments confirmed that Mid1p interacts with the two forkhead transcription factors in cells. Fkh2p and Sep1p interact with each other in vivo (Papadopoulou et al., 2008), so it is possible that Sep1p co-immunoprecipitates with Mid1p indirectly through Fkh2p, which would account for the low levels detected. Alternatively, this difference might reflect the fact that Sep1p is present at lower levels than Fkh2p, at least when detected by western blot analysis (Fig. 1A).

Mid1p binds to PCB promoters in vivo
The observation that Mid1p binds to both Fkh2p and Sep1p suggests that Mid1p might have a role in controlling M-G1 gene expression. One way it could do this is through binding to the promoters of genes containing PCB motifs that are expressed at this cell cycle time, and with which both Fkh2p and Sep1p interact (Papadopoulou et al., 2008). To examine this, we completed chromatin immunoprecipitation (ChIP) experiments using a tagged version of Mid1p under the control of its native promoter, and found that it bound in vivo to PCB promoters, including its own (Fig. 1B). This is the first time an anillin-like protein has been shown to contact chromatin in any organism. Interestingly, however, Mid1p did not contact all of the PCB promoters tested. We detected binding to PCB motifs from the cdc15 + , plo1 + and mid1 + promoters, but not from fkh2 + .

4367
Mid1p regulation of M-G1 gene expression Components and regulators of PBF have been observed to bind to PCB promoters at different times of the cell cycle, offering clues to how they control gene expression (Papadopoulou et al., 2008;Papadopoulou et al., 2010), so we examined Mid1p chromatin association in synchronised cells. Such analysis revealed that Mid1p is bound in similar amounts to PCB promoters throughout the mitotic cell cycle, with no apparent changes (Fig. 1C). This observation is surprising because we also used ChIP to examine the effect of mutations in Mid1p nuclear localization sequences on the ability of Mid1p to bind to PCB promoters and found that the mutations prevented binding from occurring, as no enrichment of Mid1p at PCB promoters was observed (supplementary material Fig. S3). These latter observations suggest that nuclear entry of Mid1p is required for it to bind to promoter DNAs.

Mid1p requires Fkh2p to bind to PCB promoters
Because Mid1p binds to PCB promoters, we were interested to see whether any of the known transcription factors that bind to these DNAs to control gene expression are required for this interaction. To this end, we used ChIP to assay Mid1p binding to PCB promoters in cells lacking Mbx1p (mbx1), Fkh2p (fkh2), or Sep1p (sep1). Mid1p binding to PCB motifs was unaffected in cells lacking Mbx1p ( Fig. 2A). Similar observations were seen in cells lacking Sep1p although, interestingly, Mid1p was detected binding to the fkh2 + promoter (Fig. 2C), which was not observed in wild-type cells (Fig. 1B). By contrast, in cells lacking Fkh2p, Mid1p was not detected binding to any PCB promoters (Fig. 2B), indicating that Fkh2p is required for Mid1p to bind to these DNA motifs. Reciprocal ChIP experiments revealed that Mid1p was not required for either Sep1p or Fkh2p to bind to PCB promoters (data not shown).

Co-requirement for Mid1p, Plo1p and Clp1p for binding to PCB promoters
Two known regulators of the PBF complex, the polo-like kinase Plo1p and the Cdc14p-like phosphatase Clp1p, which control the phosphorylation status of Mbx1p and themselves contact PCB promoters through Mbx1p (Papadopoulou et al., 2008;Papadopoulou et al., 2010), have been shown to interact directly with Mid1p at the contractile ring (Bähler et al., 1998;Clifford et al., 2008). This raises the possibility that they also interact on PCB promoters, and so we tested the requirement of Plo1p and Clp1p for Mid1p binding. ChIP analysis in cells expressing a temperature-sensitive mutant (plo1-ts35), a phosphatase-dead mutant (clp1-D257A) or no Clp1p (clp1) revealed no Mid1p binding in vivo ( Fig. 3A-C), indicating that both Plo1p and Clp1p are required for Mid1p to contact PCB promoters. However, a gain-of-function mutant of clp1, clp1-3A, which was not inhibited in phosphatase activity as effectively as the wild type (Wolfe et al., 2006), did not inhibit Mid1p binding to PCB promoters (Fig. 3D). Importantly, we also found that the reciprocal observation was true: both Clp1p and Plo1p require Mid1p to bind to PCB promoter DNAs (Fig. 4A,B). Because both Plo1p and Clp1p themselves bind to PCB promoters, it is possible that this kinase and phosphatase bind directly to Mid1p on PCB promoters as well. Unfortunately, this hypothesis is not testable because the methods used to detect Mid1p binding to both Plo1p and Clp1p do not distinguish between contractile ring and DNA binding populations.

Mid1p is required for correct M-G1 specific gene expression
The observation that Mid1p binds to the promoter regions of genes transcribed at the M-G1 interval, including its own, suggests that it might have a role in controlling gene expression. We tested this by assaying mRNA levels of genes usually transcribed at M-G1 in cells containing a chromosomal deletion of mid1 + (mid1). cdc25-22 mid1 cells were synchronised by transient temperature arrest 4368 Journal of Cell Science 123 (24) and cell samples taken for northern blot analysis of mRNA levels. In parallel, a culture of cdc25-22 mid1 + cells were also processed to allow direct comparison with a wild-type control. Such analysis revealed that cdc15 + and plo1 + mRNAs were no longer expressed just at M-G1 as seen in wild-type cells but, instead, were present in a much broader peak (Fig. 5A). The effect of mid1 was specific to M-G1 transcribed genes because cdc22 + , a gene expressed at G1-S under MCB-MBF control (Maqbool et al., 2003), was unaffected, with similar narrow peaks of expression in both mid1 and wild-type cells. The fact that the absence of Mid1p causes gene expression outside the M-G1 interval suggests that this protein has a repressive role in controlling transcription.
We have also shown that the Cdc14p-like phosphatase Clp1p regulates M-G1 gene expression with an apparently repressive role in this process (Papadopoulou et al., 2010). In both cases, chromosomal deletions of either clp1 + (clp1) or mid1 + resulted in partial loss of cell cycle-regulated gene expression, and it is possible that these phenotypes reflect redundancy in these controls. To address this, we examined M-G1 gene expression in a clp1 mid1 double deletion mutant. These cells show a very strong misshaped phenotype, more pronounced than either single mutant, which might in part reflect de-regulated gene expression. We examined mRNA levels of cdc15 + and plo1 + in a clp1 mid1 double deletion mutant through the cell cycle and found that these genes were no longer expressed just at M-G1, but instead with higher levels of mRNA throughout the cell cycle (Fig. 5A).
To directly compare levels of mRNA in the single and double mutants we completed northern blots with cells from asynchronous cultures (Fig. 5B). This revealed that these M-G1 genes were overexpressed in clp1 mid1 double deletion mutant cells compared with the wild type and with single deletion mutants, supporting the idea that both genes have a redundant, repressive role in controlling M-G1 cell cycle-specific gene expression.

Mid1p regulation of M-G1 gene expression
In this paper, we have examined the role of Mid1p in fission yeast and discovered a new and unexpected function for this anillin-like 4369 Mid1p regulation of M-G1 gene expression protein in controlling gene expression through binding to transcription factors and gene promoters. The gene expression it helps to control is cell cycle-specific, with transcripts being maximally present at the end of the cell cycle, during late M phase and cytokinesis, with many of the encoded proteins having functions at this cell cycle time. ChIP experiments with a tagged version of Mid1p-13myc on the cdc15 + , fkh2 + , plo1 + and mid1 + promoters. WCE (whole cell extracts; non-immunoprecipitated sample) and IP (immunoprecipitates). As negative controls, beads alone and mouse IgG were used. A negative control ChIP with anti-Myc antibody was also completed with a wild-type untagged strain (Papadopoulou et al., 2008). As a positive control, histone H3 was analysed by ChIP with an anti-H3 antibody. PCR conditions were selected to ensure amplification within a linear range, because PCR amplification with fivefold serially diluted WCE DNA resulted in a corresponding reduction in the observed signal (data not shown). Promoter regions amplified during the ChIP procedure are shown in supplementary material Fig. S2. (C)Cell cycle binding of Mid1p to PCB promoter DNAs. A population of cdc25-22 clp1-13myc cells was synchronised by transient temperature arrest and samples were taken every 20 minutes after return to the permissive temperature for ChIP and northern blot analysis. Crosslinked DNA was prepared from each sample and Mid1p-13myc analysed by ChIP using anti-Myc antibodies. Binding of Mid1p-13myc to the cdc15 + , fkh2 + , plo1 + and mid1 + promoters was detected by PCR. As a loading control, PCR was performed with 10% WCE containing input DNA and cdc15 + oligos. Quantification of Clp1p binding against input DNA is shown. RNA was prepared from duplicate samples and cdc15 + , plo1 + and mid1 + mRNA levels detected by northern blot analysis; 'asy' indicates control RNA sample from asynchronous cells prior to synchronisation. mRNA levels of cdc22 + , a known G1-S expressed transcript independent of PBF-PCB control, were also detected to confirm synchrony of the experiment.

Fig. 2. Requirement of components of PBF for Mid1p promoter binding in vivo.
ChIP experiments with Mid1p-13myc from extracts of cells containing chromosome deletions of either mbx1 (A), fkh2 (B) or sep1 (C). Crosslinked DNA was prepared from each cell type, and Mid1p-13myc analysed by ChIP using anti-Myc antibodies. Binding to PCB promoter fragments from cdc15 + , fkh2 + , plo1 + and mid1 + was detected by PCR. WCE (whole cell extracts; non-immunoprecipitated sample) and IP (immunoprecipitates). As negative controls, beads alone and mouse IgG were used for precipitations. As a positive control, histone H3 was analysed by ChIP with an anti-H3 antibody. Mid1p promoter binding in vivo could be restored by placing plasmid-borne wild-type fkh2 + in fkh2 cells. Equal levels of Mid1p-13myc protein in the strains used for ChIP were confirmed by western blotting (supplementary material Fig. S4).
It is striking that one of the genes expressed at M-G1 under PBF-PCB control is mid1 + . This suggests that Mid1p controls its own transcription, and this is supported by the observation that Mid1p is detected binding to its own promoter.

Mid1p binds to forkhead transcription factors
Mid1p contacts the promoters of genes expressed at M-G1 and the Fkh2p and Sep1p forkhead transcription factors (Fig. 6). This has been demonstrated through ChIP and immunoprecipitation experiments that show that Mid1p binds both to PCB promoter DNA and to Fkh2p and Sep1p in cells. It is likely that Mid1p binding to promoter DNA occurs primarily through Fkh2p, because the absence of this transcription factor in fkh2 cells prevents Mid1p binding. The fact that Mid1p can still be detected contacting PCB DNAs in sep1 cells implies that Sep1p is not required for this function. The apparently contradictory results from immunoprecipitation and ChIP experiments with Mid1p and Sep1p can be reconciled if Sep1p interacts with Mid1p through Fkh2p, which is possible because the forkhead transcription factors bind to each other (Papadopoulou et al., 2008). This would explain why lower levels of Sep1p are detected in immunoprecipitation experiments than Fkh2p.
The conclusion that Mid1p control of gene expression operates through Fkh2p is also supported by genetics and mutant phenotypes. , grown at 25°C. Crosslinked DNA was prepared from each cell type, and Mid1p-13myc was analysed by ChIP using anti-Myc antibodies. Binding to PCB promoter fragments from cdc15 + , fkh2 + , plo1 + and mid1 + was detected by PCR. WCE (whole cell extracts; non-immunoprecipitated sample) and IP (immunoprecipitates). As negative controls, beads alone and mouse IgG were used for precipitations. As a positive control, histone H3 was analysed by ChIP with an anti-H3 antibody. Note that both the clp1-D257A and clp1-3A mutants were also Myctagged, but neither bound to PCB DNA in vivo in ChIP assays (Papadopoulou et al., 2010); thus any precipitated DNA is bound by Mid1p. Mid1p promoter binding in vivo could be restored by placing plasmid-borne wild-type plo1 + in plo-ts35, or clp1 + in clp1 and clp1-D257A cells. Levels of Mid1p-13myc were confirmed by western blotting (supplementary material Fig. S4).

Fig. 4. Requirement of Mid1p for Plo1p and Clp1p binding to PCB promoters in vivo.
ChIP experiments with Clp1p-13myc (A) and Plo1p-3HA (B) from extracts of mid1 cells. Crosslinked DNA was prepared, and Clp1p-13myc or Plo1p-3HA was analysed by ChIP using anti-Myc or anti-HA antibody. Binding to PCB promoter fragments from cdc15 + , fkh2 + and plo1 + was detected by PCR. WCE (whole cell extracts; non-immunoprecipitated sample) and IP (immunoprecipitates). As negative controls, beads alone and mouse IgG were used for precipitations. As a positive control, histone H3 was analysed by ChIP using an anti-H3 antibody. Clp1p and Plo1p DNA binding could be restored by placing plasmid-borne wild-type mid1 + in mid1 cells. Equal levels of Mid1p-13myc protein in the strains used for ChIP were confirmed by western blotting (supplementary material Fig. S4).
The mid1 fkh2 double mutant is synthetically lethal, whereas the mid1 sep1 double mutant is viable. Furthermore, both mid1 and fkh2 result in similar effects on M-G1 gene expression, with transcripts found throughout the cell cycle. These phenotypes suggest that both proteins have a role in repressing gene expression and might account for the synthetic lethality. Deletion of both genes results in such extreme mis-regulation of M-G1 genes that this kills cells.

Mid1p binding to DNA is controlled by Plo1p and Clp1p and vice versa
Experiments presented here suggest that both Plo1p and Clp1p have a role in Mid1p regulation of gene expression, because both are required for it to bind to DNA (Fig. 6). Similarly, Plo1p and Clp1p require Mid1p to bind to PCB promoters. This is perhaps not surprising because this kinase and phosphatase not only directly control M-G1 gene expression through modifying Mbx1p, a MADS box transcription factor, but also interact with Mid1p at the contractile ring to regulate its function (Bähler et al., 1998;Hachet and Simanis, 2008;Clifford et al., 2008;Papadopoulou et al., 2008;Papadopoulou et al., 2010). Thus, it appears that Plo1p and Clp1p have dual functions in modifying Mid1p activity: to regulate formation of the contractile ring and to control M-G1 gene expression.
The observation that deletion of both mid1 + and clp1 + results in a stronger loss of cell cycle-regulated gene expression suggests that Clp1p and Mid1p have a related, overlapping function in controlling M-G1 gene transcription. However, the fact that neither 4371 Mid1p regulation of M-G1 gene expression protein can bind to promoter DNAs in the absence of the other suggests that part of the mechanism might occur away from PCB promoters.

The cell cycle role of Mid1p in the nucleus
The observations presented here help to explain why Mid1p is observed in the nucleus during the cell cycle. Its presence in this organelle allows it to regulate cell cycle-specific gene expression through contacting promoter DNAs. The experiments suggest a repressive function for Mid1p, implying that its activity is changed during M phase to cause the alleviation of inhibition of transcription and allow the activation of PCB-regulated gene transcription at M-G1. The ChIP experiments suggest that Mid1p binds to chromatin throughout the cell cycle, which suggests that a change in its interaction with Fkh2p is likely to cause differences in gene expression. This might occur through Mid1p shuttling on and off the chromatin throughout the cell cycle, but with modifications during mitosis resulting in changes in its interaction with PBF to influence gene expression.
Although the experiments suggest that a pool of Mid1p remains bound to chromatin during M phase, Plo1p is known to trigger Mid1p export from the nucleus at G2-M (Bähler et al., 1998). We have shown that Plo1p is required for Mid1p binding to chromatin, therefore, Plo1p could potentially control M-G1 gene expression through three mechanisms: alleviation of the inhibition of Mid1p on gene promoters, export of Mid1p from the nucleus, and through activation of Mbx1p (a component of PBF) by direct phosphorylation (Fig. 6). RNA was prepared from samples, and cdc15 + and plo1 + mRNA levels detected by northern blot analysis. mRNA levels of cdc22 + , a known G1-S expressed transcript independent of PBF-PCB control (Maqbool et al., 2003), were also detected to confirm synchrony in each experiment. (B)Separate asynchronous populations of mid1 + (wild-type), mid1 and mid1 clp1 (two isolates) cells were grown and samples taken for northern blot analysis. RNA was prepared from samples, and cdc15 + and plo1 + mRNA levels detected by northern blot analysis.

Similarities between Mid1p and Ndd1p, and conservation of controls
A protein with related transcription functions to Mid1p is Ndd1p from budding yeast (Koranda et al., 2000). Though these two proteins show no amino acid sequence overlap, both bind to forkhead-like transcription factors to modify cell cycle-specific gene expression, and both are influenced by polo-kinases, although Mid1p has a repressive and Ndd1p an activating function (Darieva et al., 2006). Ndd1p has been so far been identified only in budding yeast, and it possible that anillin-like proteins replace its function in other eukaryotic organisms. Because anillin-like proteins, and genes whose expression is regulated by forkhead and polo-like kinases, are found in many eukaryotes (Ng et al., 2006;Pollard and Wu, 2010), it is possible that such mechanisms might be conserved throughout evolution.

Media and general techniques
General molecular procedures were performed (Sambrook et al., 1989), and the standard methodology and media used for the manipulation of Schizosaccharomyces pombe (Moreno et al., 1991). The strains used in this study are shown in supplementary material Table S1. Cells were routinely grown in complete yeast extract (YE) medium at 25°C or 30°C. Synchronous cultures of cdc25-22 cells were prepared by transient temperature shifts to 36°C for 4 hours. Septation indices were counted microscopically and plotted to indicate the synchrony of each culture. In the case of mid1 cell cycles, septation indices were not plotted because the cells showed malformed and incorrectly placed septa, precluding the use of septa to quantify cell synchrony.

Immunoprecipitations and western blots
Immunoprecipitations and co-immunoprecipitations were completed as described (Almonacid et al., 2009), except that cells were grown in 2ϫYE5S (double concentration of all components as compared to YE5S) to allow exponential growth at higher cell concentrations. Extracts were made from 100 ml of cells grown at 30°C to 4ϫ10 7 cells/ml. Mid1p was detected on western-blots with a rabbit anti-Mid1p affinity purified antibody (Celton-Morizur et al., 2004). Myc-tagged proteins were detected using the mouse anti-Myc monoclonal 9E10, and HA-tagged proteins with the mouse anti-HA monoclonal 12CA5 (Santa Cruz Biotechnology, Roche).

Chromatin immunoprecipitation
Chromatin immunoprecipitations were performed as previously described (Papadopoulou et al., 2008). Promoter regions amplified during the ChIP procedure are shown in supplementary material Fig. S2.