The 14-3-3 Protein Homologues from Saccharomyces cerevisiae , Bmh1p and Bmh2p, Have Cruciform DNA-binding Activity and Associate in Vivo with ARS307*

We have previously shown that, in human cells, cruciform DNA-binding activity is due to 14-3-3 proteins (Todd, A., Cossons, N., Aitken, A., Price, G. B., and Zannis-Hadjopoulos, M. (1998) Biochemistry 37, 14317-14325). Here, wild-type and single- and double-knockout nuclear extracts from the 14-3-3 Saccharomyces cerevisiae homologues Bmh1p and Bmh2p were analyzed for similar cruciform-binding activities in relation to these proteins. The Bmh1p-Bmh2p heterodimer, present in the wild-type strain, bound efficiently to cruciform-containing DNA in a structure-specific manner because cruciform DNA efficiently competed with the formation of the complex, whereas linear DNA did not. In contrast, the band-shift ability of the Bmh1p-Bmh1p and Bmh2p-Bmh2p homodimers present in the bmh2 (cid:1) and bmh1 (cid:1) single-knockout cells, respectively, was reduced by (cid:1) 93 and 82%, respectively. The 14-3-3 plant homologue GF14 was also able to bind to cruciform DNA, suggesting that cruciform-binding activity is a common feature of the family of 14-3-3 proteins across species. Bmh1p

Although 14-3-3 protein was primarily isolated and characterized as an abundant brain protein (1), during the last decade, it has become evident that members of the 14-3-3 protein family are present in all types of eukaryotic cells. The family is represented by a group of highly homologous proteins (isoforms) encoded by separate genes. The number of isoforms range from 12 in Arabidopsis thaliana (2) to 9 in mammalian cells (3)(4)(5) and 2 in unicellular eukaryotes such as Saccharomyces cerevisiae (6,7) and Schizosaccharomyces pombe (8), among others. In eukaryotic cells, 14-3-3 proteins are largely found in the cytoplasmic compartment, but some isoforms have been detected at the plasma membrane and in intracellular organelles such as the Golgi apparatus and the nucleus (9 -15). 14-3-3 proteins function as homo-or heterodimers, and each monomer (30)(31)(32)(33)(34)(35) is able to bind a phosphorylated target protein to its amphipathic binding groove (16). Involvement of 14-3-3 proteins in important biological processes such as apoptosis, signal transduction, and the cell cycle has been well documented (for review, see Ref. 17). Although there is a large list of binding partners (currently Ͼ50 signaling proteins) (17), functional studies of 14-3-3 proteins are incomplete.
Previously, we identified and characterized a cruciform DNA-binding protein (CBP) 1 in HeLa cells (18,19). Microsequence analysis of three tryptic peptides of CBP revealed 100% homology to the 14-3-3 family of proteins (15), which in turn revealed a novel activity associated with 14-3-3 proteins, namely the binding to cruciform-containing DNA. Because cruciforms have been implicated in the initiation of DNA replication in prokaryotic plasmids (20 -22), eukaryotic viruses (23), and mammalian cells (24), the cruciform-binding activity of 14-3-3 proteins represents a new direction in the study of these multifunctional proteins. Furthermore, in recent studies, we have found that several 14-3-3 isoforms (⑀, ␤, ␥, and ) associate in vivo with origins of DNA replication in a cell cycle-dependent manner and are involved in DNA replication (25).
In view of the strong conservation of the 14-3-3 proteins not only with respect to sequence, but also to function, data obtained in yeast would have great relevance for the higher eukaryotic systems. The use of the powerful genetic and biochemical techniques available for S. cerevisiae should advance our knowledge of the role of 14-3-3 proteins in cruciform binding and eukaryotic DNA replication. To investigate whether the cruciform-binding activity of 14-3-3 found in mammalian cells is also present in yeast, we analyzed the 14-3-3 S. cerevisiae homologues Bmh1p and Bmh2p by electrophoretic mobility shift assays or band-shift assays. S. cerevisiae was used as a model yeast system because DNA replication is better characterized in it by comparison with the fission yeast S. pombe. Bmh1p and Bmh2p, like their mammalian counterparts, are involved in interactions with a large number of proteins that are important in intracellular regulatory processes (for review, see Ref. 26). These two proteins are Ͼ60% identical to the mammalian 14-3-3⑀ isoform, and except for the C-terminal amino acids, Bmh1p (30.1 kDa) and Bmh2p (31.1 kDa) are 91 and 97% identical, respectively, to it, with the highest homology residing in the conserved domains of these proteins (26).
Single deletion of the BMH1 and BMH2 genes had a modest effect on the physiology of the cells, but the bmh1/bmh2 double knockout resulted in lethality for the budding yeast (6,7). This lethal disruption has been complemented by at least four different Arabidopsis isoforms (27).
This study demonstrates the interaction of Bmh1p and Bmh2p with cruciform-containing DNA in vitro and their in vivo association with the S. cerevisiae autonomous replication sequence ARS307. A cruciform-specific association of these yeast homologues of 14-3-3 is also illustrated, akin to their human counterpart. Furthermore, the presence of cruciform structure at ARS307 is revealed by anti-cruciform antibody immunoprecipitation and conventional PCR.

Cell Cycle Synchronization
S. cerevisiae wild-type cells (strain GG582-5D) were cultured in YEPD medium at 30°C until they reached exponential phase (A 595 ϳ 0.5). For synchronization to the G 0 phase, the cells were transferred to (NH 4 ) 2 SO 4 -free minimal medium for 48 h. For synchronization to G 1 /S, the cells were placed in YEPD medium containing 10 M hydroxyurea (Sigma) for 36 h. Synchronization to G 2 /M was achieved by releasing the cells from the hydroxyurea block into YEPD medium containing 10 g/ml nocodazole (Sigma) for 24 h. Cell synchronization was monitored by flow cytometry.

Isolation of Yeast Nuclei
Preparation of nuclei from S. cerevisiae was carried out by differential centrifugation as previously described (28).

Electrophoretic Mobility Shift Analysis
Band-shift Assays-Cruciform-containing DNA (pRGM21ϫpRGM29) was constructed and end-labeled as described previously (18). 5 g of GG582-5D, GG583-24A, GG583-24D, or GG1259 nuclear extract as well as 3 g of human CBP/14-3-3 (15,18) were used as a positive control and were incubated with ϳ3 ng of labeled cruciform DNA for 20 min on ice in binding buffer as previously described (18). The mixtures were subjected to 4% polyacrylamide gel electrophoresis at 180 V for 2 h. The gel was dried and exposed for autoradiography. The same protocol was followed with increasing amounts (3, 6, and 12 g) of wild-type, bmh1 Ϫ , or bmh2 Ϫ S. cerevisiae nuclear extracts.
Competition Binding Assays-0.1 g of wild-type S. cerevisiae nuclear extract was incubated with 1 ng of labeled cruciform DNA (ϳ8 ϫ 10 Ϫ3 pmol) and increasing molar amounts (100-, 500-, 1000-, 1500-, and 2000-fold) of either unlabeled cruciform DNA or unlabeled linear DNA (18), used as a nonspecific competitor, for 30 min on ice in binding buffer (18). The samples were subjected to 4% polyacrylamide gel electrophoresis at 180 V for 2 h. The gel was dried and exposed for autoradiography.

Band-shift Elution of Bmh1p, Bmh2p, and GF14 and Immunoblotting
Bmh1p, Bmh2p, and GF14 were purified as previously described (15) with some modifications. In brief, 300 ng of labeled cruciform DNA (18) was used for binding either Bmh1p or Bmh2p from 100 g of wild-type S. cerevisiae nuclear extract or GF14 from the same amount of bmh1 Ϫ bmh2 Ϫ double-knockout (grown in the presence of galactose) nuclear extract. Cruciform-Bmh1p/Bmh2p and cruciform-GF14 complexes were loaded onto a 4% polyacrylamide gel and subjected to electrophoresis at 180 V for 1.5 h. Each complex was eluted from the gel by isotachophoresis as previously described (15). The eluates were concentrated to ϳ3 g/l using a Microcon YM-10 concentrator (Millipore Corp.). Then, 30 g of each band shift-eluted preparation and the corresponding nuclear extracts (total of 30 g) were mixed with 1ϫ SDS sample buffer containing 100 mM dithiothreitol, and the samples were loaded onto a 12% SDSpolyacrylamide gel. Electrophoresis was carried out at 200 V for 40 min, and the gel contents were subsequently transferred to Immobilon TM P transfer membrane at 100 V for 1 h at 4°C. Membranes were probed with rabbit anti-Bmh1p/Bmh2p antibody (7) or with anti-GF14 monoclonal antibody (kindly supplied by Dr. R. J. Ferl) (29).

Chromatin Immunoprecipitation Assay
In Vivo Cross-linking-In vivo cross-linking was carried out essentially as previously described (30). In brief, 100 ml of strain GG582-5D in YEPD medium containing 2% glucose was grown to A 595 ϭ 0.8 -1. 0.1 volume of 11% formaldehyde solution was added to the culture (final concentration of 1%) and incubated for 10 min at 26°C with gentle shaking. The cell culture was then chilled in an ice-water bath for 50 min with occasional shaking. Cells were harvested by centrifugation at 1000 ϫ g at 4°C for 15 min. The pellet was washed three times with ice-cold buffer I (50 mM Hepes/KOH (pH 7.5), 140 mM NaCl, 1 mM EDTA (pH 7.5), 1% (v/v) Triton X-100, and 0.1% (w/v) sodium deoxycholate).
Preparation of Whole Cell Extracts-Pellets from either cross-linked or uncross-linked cells were resuspended in 500 l of ice-cold buffer I containing protease inhibitors (1 mM phenylmethylsulfonyl fluoride and one capsule of protease inhibitors (Roche Molecular Biochemicals)). The same volume (500 l) of glass beads was added to the cell suspension, and the mixture was vigorously vortexed for 30 s and then chilled on ice for another 30 s for eight consecutive cycles until Ͼ80% of the cells were lysed. After spinning the samples, the supernatants were recovered and subjected to sonication to shear the chromatin DNA to sizes ranging between 0.5 and 1.0 kb. The suspensions were sonicated for 30 s four times on ice, leaving 1-min intervals on ice between each pulse to chill the samples. After centrifugation at 14,000 rpm for 15 min at 4°C, the supernatants were transferred to fresh Eppendorf tubes on ice. The protein concentration of each supernatant was adjusted to 20 g/l by dilution with ice-cold buffer I. An aliquot of this supernatant served as the whole cell extract (WCE).
Immunoprecipitation and DNA Isolation-200 l of the WCE was incubated with 50 l of protein G-agarose (Roche Molecular Biochemicals) on a rotating wheel for 1 h at 4°C to reduce the background caused by nonspecific adsorption of irrelevant proteins/DNAs to protein Gagarose beads. The cleared chromatin lysates were then incubated at 4°C for 12 h with 10 l of either preimmune serum or rabbit anti-Bmh1p/Bmh2p polyclonal antibody (7). 50 l of protein G-agarose was then added, and the incubation was continued for 2 h. The precipitates were successively washed twice for 5 min at 4°C with 1 ml of each of the following buffers: ice-cold buffer I, ice-cold buffer II (50 mM Hepes/KOH (pH 7.5), 500 mM NaCl, 1 mM EDTA (pH 7.5), 1% (v/v) Triton X-100, and 1% (w/v) sodium deoxycholate); ice-cold buffer III (10 mM Tris-Cl (pH 8.0), 250 mM LiCl, 1 mM EDTA (pH 7.5), 0.5% (v/v) Nonidet P-40, and 0.5% (w/v) sodium deoxycholate); and ice-cold Tris/EDTA buffer (pH 7.6). Finally, the pellets were resuspended in 200 l of extraction buffer (1% SDS/Tris/EDTA buffer). Half of the sample was then incubated at 65°C overnight to reverse the protein-DNA cross-links, followed by a 2-h incubation at 37°C with 50 g of proteinase K (Roche Molecular Biochemicals). The other half (non-reversed cross-link) was only incubated with the proteinase K. At the end, samples were processed to purify the DNA by passing them through QIAquick PCR purification columns (QIAGEN Inc., Valencia, CA).

PCR Amplification of the Co-immunoprecipitated DNA
The immunoprecipitated material (WCE and genomic DNA) was used as template in conventional PCRs with Ready-To-Go PCR beads (Amersham Biosciences). Primers ARS307A (5Ј-ATATTGCAATTACT-TCTTCTCATGCAC-3Ј) and ARS307B (5Ј-GGTAGGGATAATAATCTG-TAATGAGGA-3Ј) (1 M each; GENSET Corp.) were used to amplify a 370-bp DNA fragment from the yeast autonomous replication sequence ARS307 (GenBank TM /EBI accession number AF087952). An initial denaturation for 5 min at 94°C was followed by 35 cycles of denaturing for 30 s at 94°C, annealing for 30 s at 50°C, polymerization for 1 min at 72°C, and a final extension for 10 min at 72°C. PCR products were separated on 1.5% agarose gels, visualized with ethidium bromide, and photographed with an Eagle Eye apparatus (Speed Light/BT Sciencetech-LT1000).

Affinity Purification of Cruciform-containing DNA
Immunoprecipitation of cruciform-containing DNA using 2D3, an anti-cruciform DNA monoclonal antibody (mAb) (31), was carried out essentially as previously described (32). Briefly, 500 l of hybridoma 2D3 or myeloma P3 cell culture supernatant containing 5-10 g/ml immunoglobulin was diluted 1:1 with phosphate-buffered saline and adjusted to pH 7.4 with HCl. Antibody reaction with DNA was accomplished by addition to the antibody solution of 10 g of high molecular mass yeast genomic DNA prepared using a QIAGEN genomic DNA kit in a polypropylene microtube. After a 1-h incubation on ice, 600 units of high concentration restriction endonuclease EcoRI (New England Biolabs Inc.) was added, and the reaction mixture was incubated at 37°C overnight (12-16 h). Another 600 units of EcoRI was added after the overnight incubation, and the reaction mixture was incubated a further 90 min at 37°C. Antibody-bound DNA was extracted from unbound DNA using rabbit anti-mouse immunoglobulin conjugated to polyacrylamide beads (Bio-Rad). The Immunobeads were prepared by washing them once with 0.15 M NaCl and 10 mM Tris (pH 7.4) and once with Tris/EDTA buffer (pH 7.4) and then resuspending them in Tris/EDTA buffer (pH 7.4) at 12 mg/ml. 3 mg of Immunobeads was incubated with the antibody-bound, EcoRI-digested DNA on ice for 1 h and mixed occasionally. The DNA-antibody-Immunobeads complex was washed free of unbound DNA three times with 1 ml of buffer containing 0.5 M NaCl, 1 mg/ml bovine serum albumin, and 10 mM Tris (pH 7.4) by centrifuging the bead complex at 12,000 ϫ g for 20 s, decanting the supernatant, and gently resuspending the pellet in the wash solution. Washing was then repeated three times with 1.0 ml of 0.15 M NaCl and 10 mM Tris (pH 7.4) containing 0.1% Nonidet-40. The affinity-purified DNA was eluted from the bound immunoglobulin by resuspending the bead complex in 50 l of Tris/EDTA buffer (pH 7.4) containing 2% SDS; the Immunobeads were removed by centrifuging the mixture at 12,000 ϫ g for 2 min. The DNA was further purified using the QIAquick PCR purification columns. The immunoprecipitated material was used as template for amplification with either primers ARS307A and ARS307B for the ARS307 region (described above) or primers YCL010C(A) (5Ј-TACCTGCCTATAACCCTACAACATCAC-3Ј) and YCL010C(B) (5Ј-TACACCTTGCCTGAGTTGCCCAATTCA-3Ј) for the YCL010C region (GenBank TM /EBI accession number NC_001135). Fig. 1A (compare lanes 1 and 3), nuclear proteins from S. cerevisiae wild-type cells (strain GG582-5D) were able to retard the migration of a cruciform-containing DNA molecule (see "Experimental Procedures"). As a positive control, human CBP/14-3-3 was used (Fig. 1A, lane 2), which produced the expected band-shift pattern (15,18). Nuclear extracts from the bmh1 Ϫ (strain GG583-24A) or bmh2 Ϫ (strain GG583-24D) single-knockout cells failed to produce large amounts of band shifts (Fig. 1A, lanes 4 and 5), although ϳ18 and 7% retardation was observed when extracts from the GG583-24A and GG583-24D strains were used, respectively (lane 4). However, the latter band shift was not always reproducible. Band-shift assays were also carried out using nuclear extracts from the bmh1/bmh2 double-knockout cells (strain GG1259) (Fig. 1A, lanes 6 -9). Viability of this S. cerevisiae strain, containing a genetic deletion of the Bmh1p and Bmh2p proteins (7), was supported by the 14-3-3 Arabidopsis homologue GF14 (7). The cDNA encoding the GF14 protein is under the control of the GAL1 promoter (7). Induction of GF14 expression occurs in the presence of galactose, whereas repression occurs in medium containing glucose. Nuclear extracts from the GG1259 cells grown in 2% galactose showed a retardation in the migration of the cruciform DNA (Fig. 1A, lane 6). However, replacement with medium containing 2% glucose showed that, after 24 h in this new condition, the band shift was reduced (Fig. 1A, lane 7) and was completely lost after 48 h of repression of GF14 in the presence of glucose (lane 8). Although the S. cerevisiae double-knockout cells did not grow on medium with glucose (7), they remained viable for at least 48 h. Viability of 48-h cultures was confirmed by dilution transfer of the cells into new medium containing 2% galactose, which led to the recovery of the cruciform-binding activity (Fig. 1A, lane and Cruciform-complex (Human) represent the complexes between the 14-3-3 yeast homologues or human CBP/14-3-3 proteins and the cruciform probe, respectively. B, the amount of 32 P-labeled cruciform DNA was the same as described for A, and it was incubated with 3 g (lanes 2-4), 6 g (lanes 5-7), and 12 g (lanes 8 -10) of wild-type and bmh1 Ϫ and bmh2 Ϫ single-knockout nuclear extracts, respectively. 9). Furthermore, when these cells were plated on solid medium, their morphology was normal, excluding the possibility of mutant selection (data not shown).

Cruciform-binding Activity in S. cerevisiae Nuclear Extracts-As shown in
The Bmh1p-Bmh2p Heterodimer Is More Efficient in Binding to the Cruciform than the Homodimers-14-3-3 proteins can associate with their partners either as monomers or as protein dimers. However, the cruciform-binding activity is found solely in 14-3-3 dimers (15). The data in Fig. 1A show that the Bmh1p-Bmh2p heterodimer (lane 3, nuclear extract from GG582-5D cells) is more efficient in binding to cruciform DNA than either of the homodimers, Bmh2p-Bmh2p (lane 4, nuclear extracts from GG583-24A cells) and Bmh1p-Bmh1p (lane 5, nuclear extract from GG583-24D cells). This conclusion was confirmed in a set of electrophoretic mobility shift assays in which increasing amounts (3, 6, and 12 g) of nuclear extract from the three strains were used (Fig. 1B). A comparison of the band shift for each of the nuclear extract amounts indicated that Bmh2p-Bmh2p (strain GG583-24A) was able to retard the migration of the cruciform DNA (Fig. 1B, lanes 3, 6, and 9), although, even at the highest concentration (lane 9), its band shift was less efficient and weaker than the one obtained with the wild-type nuclear extract (lanes 2, 5, and 8), even at the lowest amount used (lane 2). No band shift was detectable when nuclear extracts from strain GG583-24D (bmh2 Ϫ ) were used (Fig. 1B, lanes 4, 7, and 10), even at the highest concentration (lane 10), indicating that the Bmh1p-Bmh1p homodimer is not able to bind to the cruciform DNA.
Structure-specific Binding of Bmh1p and Bmh2p to Cruciform-containing DNA-Human CBP/14-3-3 has been shown to recognize and to bind to cruciform DNA specifically by structure rather than sequence (18). To analyze whether the binding of the 14-3-3 S. cerevisiae homologues Bmh1p and Bmh2p to cruciform DNA is sequence-or structure-specific, competition binding assays were carried out as previously described for human CBP/14-3-3 (18). As shown in Fig. 2A, the cruciform complex-specific band decreased at a 100-fold molar excess of unlabeled cruciform competitor and was completely abolished at a 500-fold molar excess of unlabeled cruciform competitor, as has also been observed for human CBP/14-3-3 (18). Use of a Ͻ100-fold molar excess of unlabeled cruciform did not result in competition (data not shown), as also previously observed with human CBP/14-3-3 (18). However, the same complex was unaffected by as much as a 2000-fold molar excess of unlabeled linear competitor (Fig. 2B).
S. cerevisiae CBP Contains Both Bmh1p and Bmh2p-The protein composition of the complex obtained with the wild-type nuclear extract (strain GG582-5D) (Fig. 3A, lane 2) was exam-ined by Western blot analyses using a polyclonal antibody that recognizes both S. cerevisiae isoforms, Bmh1p and Bmh2p. The nuclear extract from strain GG582-5D was used as positive control, and as expected, both proteins were present in it (Fig.  3A, lane 3). Both Bmh1p and Bmh2p were also detected in the sample from the band-shifted complex (Fig. 3A, lane 4). The presence of both proteins in the band-shifted complex was further supported by the fact that both isoforms were present in the total cell extract from wild-type cells (strain GG582-5D) (Fig. 3A, lane 5), whereas only Bmh2p (lane 6) or Bmh1p (lane 7) was present in the total cell extract from bmh1 Ϫ (strain GG583-24A) or bmh2 Ϫ (strain GG583-24D) cells, respectively. The same analysis was performed with the complex obtained with the nuclear extract from bmh1⁄bmh2 double-knockout cells (strain GG1259) cultured in the presence of galactose (Fig.  3B, lane 2). Using a monoclonal antibody against GF14 (Mob-19, a gift from the laboratory of Dr. R. J. Ferl), the 14-3-3 plant protein was detected in both the nuclear extract (Fig. 3B, lane  3

) and the band-shifted complex (lane 4).
Cruciform-containing DNA at ARS307-DNA replication origins have been well characterized in the budding yeast S. cerevisiae owing to the isolation and characterization of autonomous replication sequences (ARSs), which were recognized by their ability to promote the autonomous replication of plasmids into which they were cloned (33,34). To investigate the presence of cruciform-containing DNA at S. cerevisiae replication origins, anti-cruciform DNA antibody immunoprecipitations were carried out (see "Experimental Procedures") using mAb 2D3, which is specific for cruciform-containing DNA (31,35). The immunoprecipitated DNA was subsequently amplified by conventional PCR using primers specific for amplifying a 370-bp fragment of ARS307 (GenBank TM /EBI accession number AF087952) (36, 37) (see "Experimental Procedures"). ARS307 is the only one of four chromosome III ARS elements sequenced by Palzkill et al. (38) that contains a perfect (11 of 11) match with the core consensus sequence. A 370-bp fragment was specifically amplified by these primers when DNA immunoprecipitated by mAb 2D3 (Fig. 4A, lane 1) was used as template. This amplification was specific because the same primers failed to amplify the 370-bp fragment when the DNA used as template in the PCR was that immunoprecipitated by P3 (Fig. 4A, lane 2), a mAb secreted by the parental myeloma line of 2D3 (32), used here as a negative control. The identity of the 370-bp fragment was confirmed by PCR in which S. cerevisiae genomic DNA was used as template (Fig. 4A, lane 3). An additional control was carried out by amplifying the 2D3-immunoprecipitated DNA with a specific set of primers for a

14-3-3 Yeast Proteins Bind to Cruciform DNA at ARS307
500-bp fragment of the YCL010C region, which is located upstream of the ARS307 region (Fig. 4B, lane 1) (EcoRI digestion excluded the entire 500-bp fragment from the one amplified by the ARS307 primers). The results show that these primers failed to produce any amplification when the 2D3-immunoprecipitated DNA was used as template, whereas they efficiently amplified the 500-bp fragment from genomic DNA (Fig. 4B,  lane 2). In contrast and as previously observed, the ARS307 primers amplified the specific 370-bp fragment from both the 2D3-immunoprecipitated and genomic DNAs (Fig. 4B, lanes 3  and 4, respectively).
To determine whether the 370-bp fragment contains inverted repeat sequences with the potential to form stem-loop (cruciform) structures, the ARS307 sequence was subjected to analysis using Stem-loop software (Genetics Computer Group, Inc., Madison, WI). Several potential stem-loop structures were identified in this manner (data not shown), but the one included between nucleotides 143 and 182 of ARS307 (Fig. 4C, panels i and ii), containing 10 bp in the stem and 10 bp in the loop, was the best in that it contained no mismatches or gaps and only two G-T base pairings.
Using BestFit software (Genetics Computer Group, Inc.), the ARS consensus sequence (TTTTGTATTTA) in the 370-bp ARS307 fragment is located adjacent to this inverted repeat sequence, between nucleotides 182 and 192 (Fig. 4C, panel i). A similar analysis of ARS305, ARS309, and ARS310 revealed potential stem-loop structures existing within a maximum of 110 bp from the ARS elements (data not shown).
In Vivo Association of Bmh1p and Bmh2p with ARS307 Detected by Chromatin Immunoprecipitation Assay-The in vivo association of Bmh1p and Bmh2p with ARS307 was analyzed using formaldehyde cross-linking and immunoprecipitation with anti-Bmh1p/Bmh2p antibody as previously described (39), followed by PCR (see "Experimental Procedures"). Formaldehyde is an easily reversible cross-linking agent that efficiently produces both DNA-protein and protein-protein crosslinks in vivo by inducing covalent coupling of endogenous proteins bound to DNA or to each other. Antibodies are then used to immunoprecipitate specific proteins coupled to their target DNA (DNA fragments ranged in size from 0.5 to 1.0 kb; see "Experimental Procedures"). When the immunoprecipitated protein-DNA cross-link was reversed by incubation at 65°C overnight (39), the 370-bp fragment was successfully amplified (Fig. 5A, lane 1). In contrast, the immunoprecipitated material from either untreated (not cross-linked) cells or crosslinked but not reversed cells did not result in PCR amplification (Fig. 5A, lanes 2 and 3, respectively), indicating that formaldehyde cross-linking is necessary before Bmh1p/Bmh2p immunoprecipitation and that the PCR amplification is blocked if the cross-linking is not reversed, as previously demonstrated (39). PCR amplification using material that was immunoprecipitated by the preimmune serum did not produce the 370-bp ARS307 fragment (Fig. 5A, lane 4), confirming the specificity of the amplification, whereas the use of S. cerevisiae WCE (Fig.  5A, lane 5) or genomic DNA (lane 6) as template produced the expected 370-bp fragment amplification.
To examine whether the absence of amplification in untreated (not cross-linked) cells was due to inefficient immunoprecipitation, the same material (without proteinase K treatment) was immunoblotted using rabbit anti-Bmh1p/Bmh2p polyclonal antibody. Both Bmh1p and Bmh2p were detected with the immunoprecipitated material from either the crosslinked or untreated cells (Fig. 5B, lanes 1 and 2, respectively). In contrast, these two proteins were not detected in the sample that was immunoprecipitated by the preimmune serum (Fig.  5B, lane 3).
Cell Cycle Profile of the Association of Bmh1p and Bmh2p-Real-time PCR was used to examine the quantity of Bmh1p and Bmh2p associated with ARS307 through the cell cycle. S. cerevisiae wild-type cells (strain GG582-5D) were synchronized to G 0 , G 1 /S, and G 2 /M (see "Experimental Procedures"), and the synchronization was monitored by fluorescence-activated cell sorting analysis (Fig. 6A). After synchronization, cells at different phases of the cell cycle were subjected to in vivo formaldehyde cross-linking and immunoprecipitation with either anti-Bmh1p/Bmh2p antibody or preimmune serum. The immunoprecipitated material was then amplified by real-time PCR using the ARS307 primers. The association of the 14-3-3 yeast homologues Bmh1p and Bmh2p was found to be ϳ1.7-fold higher at the G 1 /S boundary by comparison with the G 0 and G 2 /M phases (Fig. 6B). By contrast, the materials that were immunoprecipitated by the preimmune serum were of consistently low abundance.

DISCUSSION
This study provides evidence for the binding of the 14-3-3 S. cerevisiae homologues to cruciform-containing DNA. The data show that similar to their human CBP/14-3-3 counterpart (15,25), Bmh1p and Bmh2p have cruciform-binding activity both in vitro and in vivo. Comparison of the band shifts produced with the nuclear extracts from the GG582-5D (wildtype), GG583-24A (bmh1 Ϫ ), and GG583-24D (bmh2 Ϫ ) strains indicated the function of these yeast isoforms as cruciform- binding proteins, which was further confirmed in vivo by the chromatin immunoprecipitation assays. The fact that amplification was not observed with the non-reversed immunoprecipitated material suggests that the 14-3-3 S. cerevisiae homologues Bmh1p and Bmh2p are directly associated with or near ARS307, an active origin of DNA replication in S. cerevisiae (36,37). Furthermore, the cell cycle studies indicate that the association of Bmh1p and Bmh2p with ARS307 is higher at the boundary of G 1 /S, akin to what has been observed for some mammalian 14-3-3 isoforms (␤, ⑀, ␥, and ) (25). Additional characterizations of the cell cycle-dependent association of the 14-3-3 yeast homologues with other active ARSs should allow a more generalized conclusion to be reached in this respect. The 14-3-3 plant homologue GF14 was also able to produce a band shift with cruciform DNA, supporting the notion that 14-3-3 proteins from different organisms can complement each other's function. The loss of band shift after 48 h in the presence of glucose in the GG1259 strain confirmed that cruciform-binding activity is a function of the two 14-3-3 yeast homologues.
CBP/14-3-3 binds to the cruciform-containing DNA only as a dimer (15). The dimer can be composed of either the same isoforms (homodimer) or different ones (heterodimer). Our group has recently demonstrated that the mammalian CBP/14-3-3 isoforms have a differential effect on DNA replication in vitro (25). Here, we have shown by the cruciform DNA retardation assay and using different S. cerevisiae genetic backgrounds that the Bmh1p-Bmh2p heterodimer was more efficient in binding to cruciform DNA than either of the corresponding homodimers. Although high concentrations of nuclear extract from strain GG583-24A were able to produce a clear band shift, no retardation of cruciform migration was observed with the nuclear extract from the GG583-24D cells, in which a Bmh1p-Bmh1p homodimer is active. However, the cruciform-binding activity of this isoform cannot be ruled out because proteins that eluted from the band-shifted complex obtained with wild-type nuclear extracts did contain Bmh1p. Moreover, because the Bmh1p-Bmh2p heterodimer was more efficient in binding to cruciform-containing DNA than the Bmh2p-Bmh2p homodimer, there must be some contribution from Bmh1p. It is also possible that the association with Bmh2p confers cruciform-binding activity to the Bmh1p isoform. The association of Bmh1p and Bmh2p with cruciformcontaining DNA was structure-specific because only unlabeled cruciform competitor was able to dissociate the protein-DNA complex, revealing a binding behavior similar to the one observed with human CBP/14-3-3 (18).
The implications of cruciform structures in the initiation of prokaryotic and eukaryotic DNA replication have been supported by numerous publications (20 -25). In this work, the issue of the presence of cruciform structure at yeast origins of DNA replication has been addressed. Using anti-cruciform DNA antibody immunoprecipitation and conventional PCR, we demonstrated, for the first time, the presence of cruciform structure at the yeast origin of DNA replication ARS307. Mutational analysis of the ARS307 consensus sequence and its flaking regions demonstrated that substitutions of A to C and G to A at positions 176 and 173, respectively, did not affect the stability (efficiency) of this ARS, but that substitution of G to T at position 181 did affect the ARS function, causing a 2-fold increase in plasmid loss rate (40). The most likely explanation for this differential effect of the mutations on ARS stability relates to the process of cruciform formation. The kinetics of cruciform extrusion require melting and nucleation to occur at the center of symmetry before cruciform extrusion can occur (41). The A at position 176 and the G at position 173, which are located near the end of the potential stem structure, would not FIG. 6. Profile of association of Bmh1p and Bmh2p with ARS307 through the cell cycle. A, shown are the results from fluorescenceactivated cell sorting analysis of DNA contained in logarithmically growing or synchronized wild-type cells (strain GG582-5D) at the G 0 , G 1 /S, or G 2 /M phase of the cell cycle. B, the amount of Bmh1p or Bmh2p on ARS307 was measured by real-time PCR using, as template, the anti-Bmh1p/Bmh2p antibody-immunoprecipitated material from synchronized (G 0 , G 1 /S, or G 2 /M) and formaldehyde-cross-linked cells. be expected to greatly affect the nucleation and therefore stability of a cruciform forming at that location, whereas the G at position 181, which is located at the base of the stem, would. In fact, an A at position 173 would substitute the existing G-T pair with an A-T pair, thus contributing to greater stability of the stem (see Fig. 4C, panel ii). Furthermore, characterization of the binding specificity of two anti-cruciform DNA monoclonal antibodies demonstrated that these antibodies specifically recognize conformational determinants at the base of the stemloop structure (31,35,42). Those mutations altering the structure at the base of the cruciform (as the one at 181 position) would affect the binding of these monoclonal antibodies to the cruciform. The finding of a potential cruciform structure adjacent to the ARS consensus sequence suggests that DNA cruciform structures may be elements of initiation within the minimal core elements of yeast origins.
Footprinting analyses revealed a novel mode of interaction between CBP/14-3-3 and the cruciform DNA whereby CBP interacts with the elbows of the cruciform junction in an asymmetric fashion (19). However, the regions of CBP/14-3-3 involved in cruciform association remain to be determined. Alignment of human, yeast, and Drosophila 14-3-3 isoforms illustrated five sequence blocks that are evolutionarily conserved across all species (43). The X-ray structures of mammalian 14-3-3 bound to two different phosphoserine peptides revealed that all peptide-interacting residues observed in the two complexes lie within four of the five conserved blocks (44). We are currently studying some mutants of the 14-3-3 yeast homologues that will improve the knowledge of this specific association.