Functional Importance of the DNA Binding Activity of Candida albicans Czf1p

The human opportunistic pathogen Candida albicans undergoes a reversible morphological transition between the yeast and hyphal states in response to a variety of signals. One such environmental trigger is growth within a semisolid matrix such as agar medium. This growth condition is of interest because it may mimic the growth of C. albicans in contact with host tissue during infection. During growth within a semisolid matrix, hyphal growth is positively regulated by the transcriptional regulator Czf1p and negatively by a second key transcriptional regulator, Efg1p. Genetic studies indicate that Czf1p, a member of the zinc-cluster family of transcriptional regulators, exerts its function by opposing the inhibitory influence of Efg1p on matrix-induced filamentous growth. We examined the importance of the two known activities of Czf1p, DNA-binding and interaction with Efg1p. We found that the two activities were separable by mutation allowing us to demonstrate that the DNA-binding activity of Czf1p was essential for its role as a positive regulator of morphogenesis. Surprisingly, however, interactions with Efg1p appeared to be largely dispensable. Our studies provide the first evidence of a key role for the DNA-binding activity of Czf1p in the morphological yeast-to-hyphal transition triggered by matrix-embedded growth.


Introduction
Candida albicans is a diploid fungus that has adapted primarily for survival as a commensal of mammalian hosts. It is a common member of the human microbiota of the gastrointestinal tract, mucus membranes and the skin. In instances when the host immune system is weakened (for example, during immunosuppression therapy or in neonates) or the competing microflora is eliminated (for example, during antibiotic treatment) the organism can cause a range of diseases [1]. Oral thrush is an example of a superficial mucosal infection caused by C. albicans. If the organism gains access to the bloodstream, a condition known as candidaemia, it may cause a life-threatening systemic disease. Such Candida infections are the fourth-leading cause of nosocomial disease in the United States [2].
The organism's adaptability for survival in a range of niches within the host is in great part due to its highly plastic behavior. For example, growth of C. albicans can be in the yeast form or either of two filamentous states: pseudohyphae or true hyphae [3]. The transition between the different morphological forms is reversible. Filamentous growth is thought to help C. albicans invade and penetrate tissue, escaping the original site of infection and reaching distal sites, while growth in the yeast form may be better suited for dissemination in the bloodstream. As a result, polymorphic growth is considered to be an important virulence trait of C. albicans [4]. This is supported by studies showing that mutants locked in one morphological state or another are attenuated for virulence [5,6]. A mutant of hgc1, a hypha specific cyclin, that is defective in filamentous growth but expresses several hyphaassociated genes, is also attenuated for virulence in the mouse model of systemic infection [7].
A variety of signals trigger the yeast-to-hyphal transition. In a laboratory setting, several of these are growth at 37uC in the presence of serum, alkaline pH or a nitrogen limiting environment [3]. A large network of connections is involved in the control of morphogenesis and is responsive to a range of hyphal-inducing signals. Two highly conserved pathways, the cAMP dependent protein kinase A (PKA) and the mitogen activated protein kinase (MAPK) pathway, are at the heart of the network and have a positive effect on the morphological switch [5]. Mutations in components that lie within the two pathways (for recent review, see [8]) and in particular, the downstream transcriptional regulators Efg1p of the cAMP-PKA pathway and Cph1p of the MAPK pathway, lead to defects in filamentous growth [5,9,10,11]. A strain lacking both EFG1 and CPH1 exhibits a dramatic defect in filament formation during most hyphal-inducing conditions, and is avirulent in a model of systemic infection in the mouse [5]. However, interestingly, in an alternative model of C. albicans infection, an efg1D cph1D mutant is able to filament and invade tissue [12].
Growth within semi-solid agar matrix is an environmental condition that triggers morphogenesis, and may mimic the encounter and growth of C. albicans in contact with tissue [13]. Here, a control mechanism distinct from that seen during other hypha-inducing conditions at 37uC is active. Efg1p, required for filamentous growth at 37uC in a range of conditions, has a negative effect on this process during matrix-embedded growth at 25uC [14]. A second pathway mediated by the transcriptional regulator Czf1p relieves the inhibitory effect of Efg1p and promotes morphogenesis [13,14]. Cph1p promotes filamentation, but has a more minor role during this process [14]. Ectopic expression or deletion of CZF1 do not enhance or reduce filamentation during embedded growth in the absence of Efg1p, making the efg1D mutant phenotype epistatic to that of a czf1D mutant. The epistatic interactions during matrix-embedded growth at 25uC point to a functional relationship between Czf1p and Efg1p where Czf1p relieves Efg1p-mediated inhibition of filamentation.
EFG1 encodes a DNA-binding protein of the APSES family of fungal transcriptional regulators with a multitude of functions in C. albicans [11,15,16,17]. In the absence of EFG1, a number of genes are either up or downregulated [15,16,18,19]. The precise timing and magnitude of EFG1 expression is essential for its function in morphogenesis [11,20]. Efg1p, characterized by a conserved helixloop-helix (bHLH) domain involved in DNA-binding, binds to EGR elements found in promoter regions of target genes during yeast growth [21]. Induction of hyphal growth is associated with a shift of Efg1p binding away from these elements to several other sequences that have similarities with other regulatory sequences [21]. Efg1p is thought to direct downstream effects both directly and indirectly, through intermediate transcriptional regulators.
CZF1 encodes a DNA-binding protein of the Cys 6 Zn 2 class of transcriptional regulators [13,22,23]. CZF1 expression is regulated by a variety of environmental conditions such as carbon source, growth phase, temperature and the physical environment, and the CZF1 promoter region appears to be under complex regulatory control [17,21,24]. Interestingly, CZF1 expression is Efg1p dependent under all conditions tested [24]. It is also subject to negative autoregulation [24]. A czf1D mutant has a defect in filamentation during embedded growth but not during other liquid based hypha-inducing conditions pointing to the specific role of Czf1p in the control of matrix-induced filamentation [13].
How Czf1p and Efg1p work together to control morphogenesis is not understood. Czf1p binds to DNA [23] and physically interacts with Efg1p [14]. The biological function of the protein's DNA-binding activity and the significance of the Czf1p-Efg1p interaction in the control of matrix-induced filamentation have not been previously studied. Here, we show that the DNA-binding activity of Czf1p and the ability to interact with Efg1p could be separated by mutation. We also show that DNA-binding of Czf1p is essential for its regulatory function and role as a positive regulator of filamentation, whereas interaction with Efg1p is dispensable for this function.

Results
czf1 point mutants defective in the yeast-to-hyphal morphological transition during matrix-embedded growth Czf1p belongs to a class of transcriptional regulators that possess a well-conserved Cys 6 Zn 2 motif, with six cysteines binding to two zinc atoms and coordinating the fold of the domain [22]. While the majority of zinc-finger proteins bind to DNA, functionally they are very diverse. For example, the Cys 6 Zn 2 motif has been implicated in physiological roles such as zinc sensing, chromatin remodeling and protein interactions [25,26]. Whether the Cys 6 Zn 2 motif of Czf1p is important for its role as a positive regulator of matrix-induced filamentation and is involved in DNAbinding had not been explored. We examined the possible contribution of residues within this domain to the function of Czf1p as a positive regulator of morphogenesis during embedded growth.
Using the structure of Saccharomyces cerevisiae Gal4p, a DNAbinding protein and activator of GAL genes for comparison, we identified conserved residues in the putative zinc-finger domain of Czf1p that were predicted to contact DNA, Arg321 and Lys322 (FIG 1). In order to test their contribution to DNA-binding and Czf1p activity, we substituted each residue with an alanine. We also constructed an alanine substitution mutation of Thr328 because an analogous residue in Leu3p, His50, contributes to the function of the protein as a repressor [27].
Constructs carrying CZF1-myc, czf1 T328A -myc, czf1 R321A -myc or czf1 K322A -myc were introduced into a czf1D cph1D null strain in order to test their effects on filamentation during matrixembedded growth at 25uC. The double null mutant is very defective in filamentation under embedded conditions facilitating analysis of mutant phenotypes.
Strains carrying these CZF1 alleles were grown embedded in YPS agar at 25uC and representative colony images are shown in Figure 2A. Filamentous colonies were counted at the times indicated, as described in Materials and Methods (FIG 2B). In strains carrying either wild-type CZF1-myc or CZF1 T328A -myc, filamentation initiated at around the 72-hour time-point and about fifty percent of colonies were filamentous at 96 hours. Strains carrying czf1 R321A -myc or czf1 K322A -myc were highly defective in filamentation and resembled the czf1D cph1D double null mutant during the first 96 hours of growth under embedded conditions. After this time point, the czf1 R321A -myc strain remained primarily non-filamentous like the czf1D cph1D double null strain, whereas the czf1 K322A -myc strain showed some increase in filamentation. At 96 hours, differences in the percent of filamentous colonies between CZF1-myc and either czf1 R321A -myc or czf1 K322A -myc were statistically significant (Table 1).
When untagged alleles were analyzed in a Ura + strain background, similar relative effects were observed (Table 1), demonstrating a statistically significant defect in formation of filamentous colonies for czf1 R321A and czf1 K322A strains in comparison to the CZF1 + strain. Furthermore, expression of    a czf1 mutant in a strain where both copies of CZF1 and CPH1 were present on the chromosome did not cause a defect in filamentation suggesting that the czf1 mutants were loss of function mutants (data not shown).
To show that the defective mutant proteins were produced normally, extracts of cells carrying myc-tagged alleles were analyzed by immunoblotting analysis with anti-myc antibody. Results showed that wild-type and mutant Czf1 proteins were present at similar levels (FIG 2C).
These results demonstrate that, as in Gal4p and other members of this family, two residues in region A of Czf1p, the highly conserved Arg321 and Lys322, contribute to the function of the protein in invasive filamentation. In contrast, Thr328 in region B does not appear to be important. czf1 mutants are defective in DNA-binding to a target DNA sequence Czf1p binds to specific fragments of the CZF1 promoter region, but not to promoter sequences of the ACT1 and SOD3 genes [23]. To test the DNA-binding activities of the Czf1p R321A and Czf1p K322A mutants, GST fusions of wild-type and mutant Czf1p were expressed in Escherichia coli and purified by standard methods (FIG 3A).
To test the abilities of these proteins to bind DNA we used a known Czf1p target DNA sequence (fragment E) located at positions 23381 to 22816 relative to the start of the CZF1 ORF. This region of the promoter is immediately upstream of the transcriptional start site [23]. Figure 3B shows that DNA-binding by wild-type GST-Czf1p was detectable at a concentration as low as 25 nM and became more robust as protein concentration increased. At lower protein amounts, a single protein-DNA complex was observed, while at higher protein concentrations multiple complexes were detected. In previous studies, we showed that the two complexes formed by binding of Czf1p to fragment E were not observed when another fragment of the CZF1 promoter was used as the probe, demonstrating binding specificity [23].
Binding of GST-Czf1 K322A to DNA was detectable at protein concentrations of 25-50 nM but the amount was low and the lower molecular weight species was most predominant. At 0.8 mM the pattern of protein-DNA complexes was different from that of wild-type GST-Czf1p. Thus, to achieve the level of binding seen with a given concentration of WT GST-Czf1p, higher concentrations of GST-Czf1 K322A were required.
Binding of GST-Czf1 R321A had a more dramatic defect and no DNA-binding activity was observed at any protein concentration tested. Therefore, one highly conserved residue within the Cys 6 Zn 2 motif was essential for Czf1p target DNA-binding and important for Czf1p function as a regulator of filamentation.
Czf1p DNA-binding and interaction with Efg1p are independent of one another Epistasis studies point to a Czf1p and Efg1p genetic interaction during matrix-embedded growth [14]. Yeast-two-hybrid analysis further supports the notion of an in vivo physical interaction between the two transcriptional regulators. The significance of this physical interaction is unclear.
To test the effects of the czf1 R321A and czf1 K322A mutations on Czf1p-Efg1p two-hybrid interaction, we performed a yeast twohybrid assay. S. cerevisiae strain EGY40 carrying a lacZ reporter gene with lexA operator sites was transformed with LexA DNAbinding domain plasmids (either CZF1-lexA or the control fusion lexA-FOS). Gal4 activation domain fusions (either GAL4AD-Efg1 or the control fusion GAL4AD-Slk19) were introduced into these strains. As expected, strains carrying the control fusion proteins, Gal4AD-Slk19 and LexA-Fos produced low levels of b-galactosidase activity (FIG 4A), demonstrating that these fusions did not interact with Czf1p and Efg1p, respectively, nor did they interact with one another. b-galactosidase activity in transformants carrying wild-type Czf1-LexA and Gal4AD-Efg1 was higher than that of transformants carrying control fusions (Gal4AD-Slk19 and LexA-Fos) (FIG 4A), consistent with previous results [14]. Importantly, the Czf1 R321A and Czf1 K322A fusion proteins interacted with Gal4AD-Efg1 similarly to wild-type Czf1p (FIG 4A). In addition, wild-type and mutant Czf1p fusions showed similar protein levels (FIG 4B). Therefore, by individually introducing single amino-acid changes into the zinc-finger motif of Czf1p, we generated czf1 mutants that were defective in DNA-binding, but not in interaction with Efg1p. As a result, we separated the two known activities of Czf1p and showed that a DNA binding defect confers a czf1 mutant phenotype. These results provide the first evidence of a key role for the DNA-binding activity of Czf1p in regulating the morphological yeast-to-hyphal transition triggered by matrixembedded growth.
Interaction between Czf1p and Efg1p has a minor role during matrix-embedded growth To test the importance of the physical interaction between Efg1p and Czf1p we analyzed a previously characterized mutant form of Efg1p, Efg1-D3, that is compromised in the ability to bind Czf1p in the two-hybrid assay [28]. Efg1-D3 is deleted for amino acids 131-202, adjacent to the APSES domain of the protein [28]. Wild-type EFG1 and EFG1-D3 were introduced into an efg1D strain background and the ability of the strains to produce filamentous colonies during embedded growth at 23uC was measured.
As shown in Figure 5A and 5B, at a time when the wild-type strain maintained growth in the yeast state, an efg1D null strain was hyperfilamentous, consistent with previous results [14]. Introducing one copy of EFG1 + partially complemented the efg1D mutant phenotype, as observed previously. A strain carrying EFG1-D3 behaved very similarly to the wild-type complemented strain. By detecting the HA tag at the N-terminus of the protein, we showed that the wild-type and mutant Efg1p were produced in similar amounts, consistent with previous reports [28]. Because the efg1D cph1D double null mutant behaved similarly to wild-type C. albicans [14], EFG1-D3 was not tested in this strain background.
In summary, during matrix-embedded growth, the EFG1-D3 allele had wild type filamentation. Previous results showed that EFG1-D3 has no defect in hyphal induction in various other media at 37uC, but has a slight defect in the maintenance of the white state in a WO-1 strain of C. ablicans [28]. Thus, the interaction between Czf1p and Efg1p mediated by the D3 region of Efg1p is dispensable for the control of morphogenesis during embedded growth, as well as under other conditions.

Discussion
The contributions of the two known activities of Czf1p to its function as a positive regulator of filamentation had not been previously explored. Here, for the first time, we provide evidence for the biological significance of the DNA-binding activity of Czf1p. A czf1 mutant, Czf1 R321A , was defective during in vitro DNA-binding using a target sequence from the CZF1 promoter region. Importantly, this mutant was also defective in the yeast-tohyphal transition during matrix-embedded growth at 25uC. The importance of Arg321, a highly conserved residue within the Cys 6 Zn 2 motif of Czf1p, for DNA binding is consistent with the role of the corresponding residue within the same sub-region of other well-studied members of this class of transcriptional regulators [27,29,30,31,32]. For example, substitution of Glu for Lys at position 17 of S. cerevisiae Gal4 (analogous to Arg321 of Czf1p) and substitution of Ala for Lys at position 44 in Leu3 (analogous to Lys322 of Czf1p) lead to an approximately 20-fold and 15-fold reduction in DNA binding activity, respectively [31,32]. Furthermore, substitution of Lys322 led to reduced DNA binding and a defect in matrix-embedded filamentation. In contrast, mutant Czf1 T328A exhibited no defect in filamentation. This result is consistent with previous structural studies of other members of the Cys 6 Zn 2 family, which showed that residues found in this position and its proximity are facing away from DNA rather than contacting the DNA [29,30].
Interestingly, the DNA-binding activity of Czf1p was separated from its ability to physically interact with Efg1p. Czf1 K322A and Czf1 R321l although defective in DNA-binding, were able to interact with Efg1p. Importantly, our data indicate that the Czf1p-Efg1p interaction detected by two-hybrid analysis does not play an essential role in the ability of Czf1p to antagonize Efg1p inhibitory function on filamentation. The EFG1-D3 mutant allele encoding a mutant Efg1p that is defective in interaction with Czf1p [28], complemented an efg1D null strain to the same extent as wild type EFG1. If the physical binding of Czf1p to Efg1p were the key mechanism to relieving Efg1p-mediated repression of filamentation, the expected phenotype would have been a defect in filamentation under embedded conditions and this phenotype was not observed. Our findings are consistent with the model that Czf1p exerts its positive influence on filamentation and relief of inhibition by Efg1p, by acting at promoters via its DNA-binding activity. For example, at the CZF1 promoter region both Czf1p and Efg1p associate with sequences located upstream of the CZF1 ORF. Importantly, in the cell, Czf1p and Efg1p bind independently to the CZF1 promoter region [24]. Recent genome-wide association studies conducted by ChIP-chip [21] confirm that Efg1p associates with the CZF1 promoter region. In addition, in yeast form cells, Efg1p binds to the promoters of several genes encoding negative (Tcc1p, Nrg1p, Rfg1p and Cpp1p) and positive (Tec1p and Def1p) regulators of filamentation [21]. We propose that the effects of Efg1p on these (or other) promoters result in repression of filamentation. In our model, during growth under matrixembedded conditions, Czf1p, via its DNA-binding activity, acts at these or other promoters resulting in reversal of Efg1p-mediated repression of filamentous growth. It appears that the Czf1p-Efg1p physical interaction is dispensable for this event. This interaction may be required for other functions of Czf1p or Efg1p.
When Czf1p levels are high enough, we propose that Czf1p modulates and shuts down its own expression [23]. Therefore, once its role as a positive regulator has been fulfilled, a negative autoregulatory loop takes place. The transient function of Czf1p in this model is analogous to the transient role of Czf1p in regulation of switching between the more common ''white'' phenotype and the mating competent ''opaque'' phenotype [17]. Temporal regulation of multiple factors involved in the control of the yeast-to-hyphal transition is thus a crucial mechanism for regulating the different stages of initiation and maintenance of filamentous growth [11,21,33].
The functional relationship between Czf1p and Efg1p in the context of the yeast-to-hyphal transition is apparent only when growth takes place within a semi-solid material at temperatures at or below 35uC. These growth conditions likely simulate a specific physical environment encountered during invasion into tissue in particular parts of the body. In this environment, a distinct Czf1p/ Efg1p-dependent genetic program controlling morphogenesis is important. To understand more broadly how morphogenesis is triggered during infection of a human host, it will be of interest to  understand the contributions of the many signaling pathways that regulate filamentation in different environments.

Materials and Methods
Strains and media C. albicans strains used in this study are listed in Table 2. C. albicans cells were routinely grown in YP medium (1% yeast extract, 2% Bacto peptone) with either 2% glucose (YPD) or 2% sucrose (YPS). For embedding experiments, 1% agar was used. For growth of CKY154 [ura3D czf1D cph1D mutant] [13] and its derivatives carrying Myc-tagged CZF1 alleles, uridine was added to the growth medium at 162 mg/ml. For czf1 mutant analysis, plasmids encoding Myc-tagged CZF1 (wild type or mutant) under control of the maltase promoter were linearized by digestion with BsgI and integrated at the ADE2 locus of strain CKY154. Transformants were selected for Nourseothricin resistance (200 mg/mL) and integrations were confirmed using primers ADE2-For1 and ADE2-Rev2 (Table 3). For efg1 mutant analysis, plasmids pTD38-HA, pTD38-HA-Efg1 and pTD-HA-Efg1-D3 [28] encoding HA-tagged versions of EFG1 under control of the endogenous promoter were linearized by digestion with PacI, integrated at the EFG1 locus of strain Can35 [efg1D mutant] [5] and correct integrations were confirmed by PCR with primers EFG1-7a and EFG1-7b for vector alone and EFG1-10a and EFG1-10b for EFG1 containing plasmids. E. coli DH5a [34], and GM2163 (dam 2 dcm 2 ; New England Biolabs) were used to propagate plasmids. E. coli strain BL21 [35,36] was used for expression and purification of recombinant Czf1p. Bacterial cells were cultured in Luria broth or on Luria plates with ampicillin added to a concentration of 100 mg/ml. S. cerevisiae strain EGY40 [37] was used for yeast-two-hybrid analysis carrying the bait and prey fusions, and was grown routinely in complete minimal media lacking uracil, leucine and histidine.

Matrix-embedding assay
Growth and embedding of C. albicans cells in agar medium was performed as previously described [13]. Briefly, following 4 hours of growth of a diluted overnight culture, C. albicans cells were embedded in YPS containing 1% molten agar and grown at 25uC for czf1 mutant analysis and 23uC for efg1 mutant analysis. Uridine was added to 162 mg/ml for embedded growth of Ura 2 strains. A colony was considered to be filamentous if more than 20 filaments protruding from the edge of the colony were visible. 100 colonies were counted for a given plate. Each assay described was performed at least 3 times.

Plasmids
Plasmids used in this study are listed in Table 4. Czf1p-Myc and HA-Efg1p. Wild-type or mutant Czf1-Myc were expressed from the maltase promoter in plasmid pDBI-Czf1-13xMyc-SAT. First, the vector pDBI52-SAT was constructed by cloning SAT1 into plasmid pDBI52 [13]. Primers SAT1F and SAT1R were used to amplify a fragment from pSFS2 [38] containing the SAT1 marker. The PCR fragment was subcloned into vector pDBI52. Cloning of CZF1-13xMyc-3UTR into pDBI-SAT was done in two steps. First, an overlap PCR fragment of 13xMyc and the 39UTR of CZF1 was generated with pADH34 [39] amplified with primers MycF and MycR to generate the 13xMyc [AB] fragment and pTONY1 (an unpublished CZF1 + plasmid isolated as described in [13]) with primers 3UTR-F and CzfXba to generate the 39UTR [CD] fragment. The fragment was then cloned into pDBI-SAT creating plasmid pDBI-13xMyc-3UTR-SAT. Second, a PCR fragment containing the CZF1 ORF and , 200bp upstream sequence was generated by primers Czf1KpnI and CzfAge1 and cloned into pDBI-13xMyc-3UTR-SAT. For this PCR, pMV102 [24] served as a template for wildtype CZF1 and czf1 mutant alleles were generated by overlap PCR. The resulting final plasmids carrying Myc-tagged Czf1p variants and used for the embedding assay are: pDBI-Czf1 R321A -13xMyc-SAT, pDBI-Czf1 K322A -13xMyc-SAT, pDBI-Czf1 T328A -13xMyc-SAT.
Yeast-two hybrid plasmids. A list of yeast-two-hybrid plasmids used in this study is provided in Table 4. To create plexA-Czf1 R321A and plexA-Czf1 K322A plasmids pDBI-Czf1 R321A and pDBI-Czf1 K322A served as templates using primers BOPO1 and BOPO2. The CZF1 ORF-containing PCR fragment was then digested with EcoRI and BamHI and cloned into vector pNlexA [14] digested with the same enzymes. GST fusion proteins for expression in E. coli. Wild-type GST-Czf1 was published previously [23]. To create GST fusions of mutant czf1, primers POM1F and POM1R were used to amplify CZF1 alleles using templates pDBI-Czf1 R321A and pDBI-Czf1 K322A . After subsequent digestion with BamHI and EcoRI the PCR fragments were cloned into plasmid pGEX-6P-2 (GE Healthcare Life Sciences) digested with the same enzymes, giving rise to pGST-Czf1 R321A and pGST-Czf1 K322A .

Western analysis
Strains producing Myc-tagged Czf1p variants under control of the maltase promoter were grown in YPS at 30uC to OD 600 of 0.8-1. Strains producing HA-tagged Efg1p variants under transcriptional control of the endogenous promoter were grown similarly and as in [28]. Crude cell extracts were prepared by breakage with glass beads, in the presence of protease inhibitors (2 mM PMSF, one tablet Complete mini protease inhibitor cocktail (Roche) per 10 ml lysis buffer). Lysis buffer, RIPA, was: 50 mM TrisCl, pH 7.5, 150 mM NaCl, 0.1% SDS, 0.5% Sodium Deoxycholate, 1% NP40. Total protein was quantified with the BCA assay (Thermo Scientific). 30 mg of total protein for Czf1-Myc and 45 mg of total protein for HA-Efg1 were used. Proteins were separated by 8% sodium dodecyl sulfate-polyacrylamide gel electrophoresis (SDS-PAGE), and immunoblottings were done using mouse anti-Myc (1:2,000, Upstate) for Czf1-Myc and mouse anti-HA high-affinity antibody (1:1,000; Roche) for HA-Efg1. For visualization of Czf1-lexA protein fusion in the yeast-two-hybrid assay, cells were grown to exponential phase, harvested, adjusted to 10 OD 600 units and crude extracts were obtained using M2 lysis buffer pH 7.0 (20 mM HEPES, 150 mM NaCl) and protease inhibitors (Complete mini protease inhibitor cocktail, Roche). Immunoblottings were done using rabbit anti-LexA antibody (1:1, 5000, Sigma). Purification of GST-Czf1 For purification of GST-Czf1p, 1.2 L of cells were grown to OD 600 #0.8 at 25uC and induced with 0.5 mM IPTG at 30uC for 2.5 hrs. Cells were harvested and resuspended in ice cold 1xPBS (8 mM sodium phosphate, 2 mM potassium phosphate, 140 mM NaCl, 10 mM KCl, ph 7.4) containing a cocktail of protease inhibitors (Roche, EDTA free) and 2 mM PMSF, and passed twice through a French pressure cell at 14,000 psi. The lysate was cleared by ultracentrifugation at 90,000 r.p.m. for 20 min in a Beckman TLA 100.3 rotor. The cleared lysate was loaded onto a pre-equilibrated glutathione Sepharose column. The resin was washed with three times the bed volume with 1xPBS. Protein was first eluted with 50 mM Tris, pH 8.0, 10 mM reduced glutathione and then with 125 mM Tris, pH 8.0, 150 mM NaCl, 50 mM reduced glutathione. Protein from the second elution was used for gel shift analysis.

Mobility shift assay
Primers CZF1AF/CZF1AR (Table 3) were used to amplify a 565 bp fragment from the CZF1 promoter, which was denoted the fragment E [23]. DNA probes were gel-purified and labeled with [c-32 P] ATP using T4 polynucleotide kinase (New England Biolabs) and the manufacturer's protocol. Mobility shift experiments were modified from previously published procedures [23]. Binding reactions were performed in a 20 mL reaction that contained the following components: 12% (v/v) glycerol, 12 mM HEPES-NaOH (pH 7.9), 4 mM Tris-Cl (pH 7.9), 90 mM KCl, 2 mg poly(dI-dC), 5 mg BSA, 0.625 ng 32 P-labelled oligonucleotide probe, and varying amounts of protein. After incubation for 30 min at 30uC, the binding mixture was separated on 8% polyacrylamide, Tris-Glycine non-denaturing gel (50 mM Tris pH 8.0, 383.6 M Glycine), at room temperature in the absence of chelating agents. Running buffer was 35 mM HEPES, 43 mM Immidazole, pH 7.4. Dried gels were visualized by phosphorimaging.