The Human Cruciform-binding Protein, CBP, Is Involved in DNA Replication and Associates in Vivo with Mammalian Replication Origins*

We previously identified and purified from human (HeLa) cells a 66-kDa cruciform-binding protein, CBP, with binding specificity for cruciform DNA regardless of its sequence. DNA cruciforms have been implicated in the regulation of initiation of DNA replication. CBP is a member of the 14-3-3 family of proteins, which are conserved regulatory molecules expressed in all eukaryotes. Here, the in vivo association of CBP/14-3-3 with mammalian origins of DNA replication was analyzed by studying its association with the monkey replication origins ors8 andors12, as assayed by a chromatin immunoprecipitation assay and quantitative PCR analysis. The association of the 14-3-3β, -ε, -γ, and -ζ isoforms with these origins was found to be ∼9-fold higher, compared with other portions of the genome, in logarithmically growing cells. In addition, the association of these isoforms with ors8 and ors12 was also analyzed as a function of the cell cycle. Higher binding of 14-3-3β, -ε, -γ, and -ζ isoforms with ors8 and ors12 was found at the G1/S border, by comparison with other stages of the cell cycle. The CBP/14-3-3 cruciform binding activity was also found to be maximal at the G1/S boundary. The involvement of 14-3-3 in mammalian DNA replication was analyzed by studying the effect of anti-14-3-3β, -ε, -γ, and -ζ antibodies in thein vitro replication of p186, a plasmid containing the minimal replication origin of ors8. Anti-14-3-3ε, -γ, and -ζ antibodies alone or in combination inhibited p186 replication by ∼50–80%, while anti-14-3-3β antibodies had a lesser effect (∼25–50%). All of the antibodies tested were also able to interfere with CBP binding to cruciform DNA. The results indicate that CBP/14-3-3 is an origin-binding protein, acting at the initiation step of DNA replication by binding to cruciform-containing molecules, and dissociates after origin firing.

Inverted repeat sequences (IRs) 1 are a common feature of prokaryotic and eukaryotic regulatory regions, and their dis-tribution is nonrandom and clustered (1). Promoters (2,3), terminators (4), amplified genes (5), and origins of DNA replication from prokaryotes (6 -8), viruses (9), and eukaryotes (10) all contain IRs. Origins of DNA replication from higher eukaryotes, such as monkeys and humans (11)(12)(13)(14)(15), are also enriched in IRs. IRs have been implicated in the regulation of initiation of DNA replication in plasmids, bacteria, eukaryotic viruses, and mammals (reviewed in Ref. 16). IRs have the potential to form cruciform structures (stem loop or hairpin) through intrastrand base pairing and under conditions of torsional strain on the DNA (reviewed in Refs. 16 -18). Such cruciform structures have been shown to form in vivo (reviewed in Ref. 16) in prokaryotes and lower eukaryotes (19 -22), and in mammalian cells (23)(24)(25). Such structures are thought to serve as recognition signals for regulatory proteins involved in critical cellular processes such as transcription, recombination, and DNA replication (reviewed in Refs. 16 and 18).
We and others have previously demonstrated the involvement of cruciform structures in the initiation of mammalian DNA replication (20,(23)(24)(25) (reviewed in Ref. 16) We previously identified and isolated from human cells (HeLa) a cruciform-specific binding protein, CBP (26), of 66 kDa with binding specificity for cruciform-containing DNA (26). CBP binds at the base of four-way junctions (27), interacting with them in a different manner from other proteins known to bind such junctions (26,27).
CBP is a member of the 14-3-3 protein family, a highly conserved family (28) of proteins through plants, invertebrates, and higher eukaryotes (reviewed in Ref. 29). 14-3-3 proteins are multifunctional, involved in diverse cellular processes, such as neurotransmitter biosynthesis, signal transduction pathways, and cell cycle control (reviewed in Ref. 30). 14-3-3-associated proteins include receptors (31), kinases (32), phosphatases (33), docking molecules (34), death regulators (35), and oncogene products (36). The interaction of CBP/14-3-3 with cruciform structures is functionally important for the regulation of DNA replication. Immunofluorescence studies using anti-cruciform DNA monoclonal antibodies showed that cruciforms are at a maximum number at the G 1 /S boundary (24,25). The same antibodies had an enhancing effect on DNA replication in mammalian cells, presumably by stabilizing cruciform structures (23) and increasing DNA synthesis through continuing protein-protein interactions by signaling pathways essential to DNA replication.
Our laboratory has also previously isolated origin-enriched sequences (ors) from early replicating monkey kidney (CV-1) cells (37)(38)(39), which are capable of conferring autonomous replication to plasmids in vivo (14,40) and in vitro (41). In vivo mapping of ors12 by competitive PCR demonstrated that it acts as a chromosomal replication origin (42). Among the ors, ors8 and ors12 have been characterized in detail. They both contain inverted repeats that give rise to a cruciform structure (43) (reviewed in Ref. 39). Deletion analysis of ors8 and ors12 revealed that the minimal sequences required for origin function (186 bp for ors8 (44) and 215 bp for ors12 (45)) retain an intact IR. The IR present in ors8 is capable of extruding into a cruciform both in vivo (23) and in vitro (16,46).
In the present study, we analyzed the cruciform binding activity of CBP/14-3-3 as a function of the cell cycle, its association in vivo with replication origin-containing sequences (exemplified by ors8 and ors12), and its involvement in mammalian DNA replication. The in vivo binding of CBP/14-3-3 to ors8 and ors12 was analyzed using the formaldehyde cross-linking technique (47), followed by immunoprecipitation of protein-DNA cross-links with antibodies against the 14-3-3␤, -⑀, -␥, and -isoforms. PCRs were then performed using the immunoprecipitated material as template. CBP/14-3-3 was found to associate specifically with ors8 and ors12, since DNA fragments containing these ors were enriched in the immunoprecipitate compared with other portions of the genome. Furthermore, higher binding of 14-3-3␤, -⑀, -, and -␥ isoforms to ors8 and ors12 was found at the G 1 /S border by comparison with other stages of the cell cycle. The cruciform binding activity of CBP was also found to be maximal at the G 1 /S interphase. In addition, the anti-14-3-3␤, -⑀, -␥, and -antibodies were able to interfere with CBP-cruciform complex formation and inhibit the in vitro DNA replication of p186. The anti-14-3-3⑀, -␥, and -antibodies had a greater inhibitory effect on the in vitro DNA replication of p186 DNA than did the anti-14-3-3␤ antibody. The data suggest an involvement of CBP in mammalian DNA replication as an origin-binding protein.

EXPERIMENTAL PROCEDURES
Cell Culture and Synchronization-HeLa (S3) cells growing in suspension were synchronized by serum starvation and treatment with aphidicolin 2 g/ml (23) and mimosine 400 M (48). CV-1 cells growing in monolayers were cultured in minimal essential medium ␣ (Invitrogen) supplemented with 10% fetal bovine serum (Invitrogen) (termed "regular medium") at 37°C, as previously described (49). For synchronization to the G 0 /G 1 phase, 80% confluent CV-1 cells were placed in serum-free medium for 48 h. For synchronization to G 1 /S, S (50), and M (51) phases, the procedure was modified as follows: 40% confluent CV-1 cells were treated with 2 mM thymidine (Sigma) for 12 h and then released for 9 h in regular medium without thymidine and subsequently incubated for 12 h with 2 mM thymidine and 400 M mimosine (Sigma). For S phase synchronization, the cells were released from the thymidine/mimosine block for 4 h in regular medium. For synchronization to M phase, the cells were released from the thymidine/mimosine block in regular medium supplemented with 1 g/ml of nocodazole (Sigma), for 14 h. Cell synchronization was monitored by flow cytometry.
In Vivo Cross-linking-In vivo cross-linking was performed as described by Ritzi et al. (52) with some modifications. In brief, CV-1 cells, grown as previously described, were washed twice with phosphatebuffered saline to remove all traces of serum, and then formaldehyde (1%) in warm minimal essential medium ␣ without serum was added for 10 min. Cells were subsequently lysed (at 4°C) in lysis buffer (50 mM HEPES/KOH, pH 7.5, 140 mM NaCl, 1% Triton X-100, 1 mM EDTA, 1 mM phenylmethylsulfonyl fluoride, 1 capsule of protease inhibitors (Roche Molecular Biochemicals) by being drawn into and out of a 21-gauge hypodermic needle three times to effect lysis and dispersion of nuclei. Cell lysates were then layered over 4 ml of 12.5% glycerol in lysis buffer, and nuclei were pelleted by spinning at 750 ϫ g for 5 min in a benchtop centrifuge. The nuclear pellet was resuspended in 1 ml of lysis buffer.
Chromatin Fragmentation-Cross-linked or non-cross-linked cell nuclei were sonicated 10 times for 30 s each time, and the chromatin size was monitored by electrophoresis (53). This treatment generated fragments of ϳ20 kb. To further reduce the chromatin size into smaller fragments of 1.5-3.5 kb, DNA was digested with SphI, HindIII, PstI, and EcoRI restriction endonucleases in NEB2 buffer (100 units of each; New England Biolabs) at 37°C for 6 h. These enzymes were chosen because they did not cut in either the origin or non-origin-containing sequences chosen.
Real Time PCR Quantification Analysis of Immunoprecipitated DNA-PCRs were carried out in 20 l with one-two hundredth of the immunoprecipitated material using LightCycler capillaries (Roche Molecular Biochemicals) and the LightCycler-FastStart DNA Master SYBR Green I from Roche Molecular Biochemicals. The PCR contained 3 mM Mg 2ϩ and a 1 M concentration of each primer of the appropriate primer set used: ors8 150, ors12 D2, EEЈ, or CD4 intron. Primer set ors8 150 and ors12 D2 were used to amplify a 150-or 251-bp corresponding genomic fragment of ors8 or ors12 (see Fig. 2A). Primer set EEЈ was used to amplify a 250-bp genomic fragment that was mapped ϳ5 kb downstream of the origin of replication ors12 (see Fig. 2A). A control set of primers from the CV-1 CD4 gene (accession number AB052204 (54)) was also used. The primer set CD4 intron amplifies a fragment of 258 bp from genomic CV-1 DNA. Primers were designed as 20 -22-mers with ϳ50% GC content. The sequences for the primers used are shown in Table I.
Genomic CV-1 DNA (9.3, 18.6, 27.9, 37.2, and 55.8 ng), used for the standard curve reactions (necessary for quantification of the PCR products) (see Fig. 2B), were obtained from total cell lysates of noncrosslinked logarithmic 80% confluent cells. The quantification of the PCR products was assessed by the LightCycler (Roche Molecular Biochemicals) instrument, using SYBR Green I dye as the detection format (55). The quantification program used a single fluorescence reading at the end of each elongation step. Arithmetic background subtraction was used, and the fluorescence channel was set to F1. No primer-dimers were detected that interfered with the quantification of the PCR products. Typically, an initial denaturation for 10 min at 95°C was followed by 35 cycles with denaturation for 15 s at 95°C; annealing for 10 s at 45°C (primer set ors8 150), 50°C for 15 s (primer sets EEЈ and CD4 intron), or 55°C (primer set ors12 D2); and polymerization for 15 s at 72°C. The specificity of the amplified PCR products was assessed by performing a melting curve analysis after the PCR amplification. The fluorescence of the SYBR Green I dye bound to double-stranded amplified product declines sharply as the fragment is denatured. The melting temperature of this fragment was visualized by plotting the first negative derivative (ϪdF/dT) of the melting curve on the y axis and temperature (°C) on the x axis (see Fig. 2C). The melting curve analysis cycle has a first segment set at 95°C for 0 s and a temperature transition of 20°C/s; a second segment set at 45°C, 50°C, or 55°C (depending on the annealing temperature of the primer set used) with a temperature transition rate of 20°C/s; and, finally, a third segment set at 95°C with a temperature transition rate set at 0.2°C/s. PCR products were also separated on 2% agarose gels, visualized with ethidium bromide, and photographed with an Eagle Eye apparatus (Speed Light/BT Sciencetech-LT1000) (see Fig. 3B).
CBP Cruciform Binding Activity as a Function of the Cell Cycle-HeLa cells growing in suspension were synchronized by serum starvation and treatment with aphidicolin or mimosine (as described above). Extracts were prepared as before (41)  In Vitro DNA Replication Assays-In vitro DNA replication assays were performed as previously described (41) with some modifications. Approximately 100 g of total HeLa cell extracts were preincubated with either 20 g of anti-14-3-3␤, anti-14-3-3⑀, anti-14-3-3␥, anti-14-3-3 antibodies (Santa Cruz Biotechnology), normal rabbit serum (NRS) (Santa Cruz Biotechnology), or hypotonic solution (20 mM Hepes, pH 7.8, 5 mM KAc, 0.5 mM MgCl 2 , 0.5 mM dithiothreitol) or with a combination of three (7 g each of anti-14-3-3⑀, anti-14-3-3␥, and anti-14-3-3) or four (5 g each of anti-14-3-3␤, anti-14-3-3⑀, anti-14-3-3␥, and anti-14-3-3) anti-14-3-3 antibodies for 20 min on ice. This mixture was used to replicate in vitro (41) 150 ng of p186 plasmid DNA (44). Unmethylated pBluescript KSϩ was included in each of the reactions as an internal control for differences in DNA recovery and completeness of the DpnI digestion. One-third of the replication products were digested with 1.5 units of DpnI for 60 min. Both undigested and digested products were resolved by electrophoresis in a 1% agarose gel in 1ϫ TAE buffer at 50 V for 15 h; then the dried gel was exposed to an imaging plate for 6 h, and the DpnI-resistant bands (forms II and III) were quantified by densitometric measurements using Image Gauche (Fuji Photo Film Co., Ltd.). The results were normalized for the amount of DNA recovered from the in vitro replication assay and for the amount loaded on the agarose gel by quantitative analysis of the amount of radionucleotide incorporated in the unmethylated pBluescript KSϩ propagated in dam(Ϫ) bacteria cells. This incorporation was due to DNA repair, since this plasmid did not contain a mammalian origin of DNA replication. Also, the unmethylated plasmid cannot be digested by DpnI, since DpnI cleaves only fully methylated DNA. In addition, a reaction with methylated pBR322, a plasmid that does not contain a mammalian origin of DNA replication, was also performed to show that the observed DpnI-resistant bands (forms II and III) were origin-dependent. The total amount of radionucleotide incorporated was expressed as a percentage of the control reaction with hypotonic buffer.
Molecular Biochemicals) was of the expected size, reaction products were separated on a 2% agarose gel and then visualized by EtBr staining under UV light. All primer sets generated the expected corresponding 150-, 251-, 250-, or 258-bp DNA fragments (Fig. 3B), respectively. In addition, melting curve analysis of the amplification products verified that only the specific product was generated by the respective primer set and that no primer-dimers interfered with the quantification of the products in the LightCycler instrument (Fig. 2C). Primer set EEЈ generated a primer-dimer peak (ϳ74°C) when water was present instead of template DNA (Fig. 2C, EEЈ), which, however, did not interfere with the quantification of the specific product when template DNA is used instead of water. The melting temperatures of the PCR products generated with their respective primer sets were ϳ78°C (amplified by primer set ors8 150), ϳ87°C (amplified by primer set ors12 D2), ϳ90°C (amplified by primer set CD4 intron), or ϳ81°C (amplified by primer set EEЈ), respectively (Fig. 2C).
CBP Cruciform Binding Activity Is Maximal at the G 1 /S Boundary-HeLa cells growing in suspension were synchronized by serum starvation and treatment with aphidicolin (23) or mimosine (48). Extracts were prepared as before (41) from arrested and released cells as well as cells in G 0 and log phases (see FACS analysis in Fig. 5B). Bandshift unique to the cruciform (D complexes) molecules were obtained in all reactions from the indicated treatment, except for the serum-starved (G 0 ) cells (Fig. 5A, see D complexes). The cruciform binding activity of CBP was maximal at G 1 /S cells (arrested with either mimosine or aphidicolin) and in early S phase cells that had been released from the mimosine or aphidicolin block for 45 min (Fig. 5A).
Effect of Anti-14-3-3 Antibodies on the in Vitro DNA Replication of p186 -In view of the association of 14-3-3␤, -⑀, -␥, and -isoforms with ors8 and ors12 and the observed interference of their respective antibodies with CBP-cruciform complex formation, the effect of the same antibodies on the in vitro DNA replication of p186 was analyzed. A representative autoradiogram of the in vitro replication experiment, before and after digestion with DpnI, is shown in Fig. 7A (left panel (ϪDpnI) and right panel (ϩDpnI)). The in vitro replication reaction products included relaxed circular (form II) and linear (form III) forms of p186 as well as replicative intermediates and topoisomeric forms, as previously observed (41), whereas the supercoiled (form I) forms overlapped with the unmethylated pBluescript plasmid DNA (Fig. 7A). The unmethylated pBluescript plasmid was included in each in vitro replication reactions to control for DNA recovery and completeness of the DpnI digestion. As expected, this plasmid was not digested by DpnI (Fig. 7A, right panel, lanes 1-9), whereas the methylated pBR322 control was fully digested by DpnI (Fig. 7A, right  panel, lane 9), indicating that the presence of a mammalian replication origin is required for replication of the plasmid DNA. Quantitation of the DpnI-resistant DNA showed that preincubation of the HeLa cell extracts with 20 g of anti-14-3-3⑀, anti-14-3-3␥, or anti-14-3-3 antibodies decreased the level of p186 in vitro DNA replication to ϳ30 -40% of control reactions, in which the HeLa cell extracts were preincubated with the same amount of either hypotonic buffers (since the nuclear extracts are resuspended in hypotonic buffer) or NRS (Fig. 7B), while preincubation with the anti-14-3-3␤ antibody had a lesser effect (50 -75% of control) on p186 replication (Fig.  7B). The combination of either anti-14-3-3⑀, -␥, and -or anti-14-3-3␤, -⑀, -␥, and -isoforms decreased the level of replication to ϳ25 and 20%, respectively, of control reactions (Fig. 7B). These combinations where chosen because of their profound effect on the CBP-14-3-3-cruciform complex formation (Fig. 6,  A and B). DISCUSSION The multifunctional 14-3-3 family of proteins consists of seven known mammalian isoforms (␤, ⑀, ␥, , , , and ) that are highly conserved molecules and are expressed in a broad range of tissues and cell types (59). In eukaryotes, 14-3-3 proteins are largely found in the cytoplasm, but they can also be detected at the plasma membrane and in intracellular organelles, including the nucleus (34, 60). 14-3-3 proteins participate in the regulation of essential cellular processes, including cell cycle control, survival signaling, cell adhesion, neuronal plasticity, and DNA replication (reviewed in Refs. 29 and 30).
Finally, the cell cycle studies indicated that the association of 14-3-3␤, -⑀, -␥, and -with ors8 and ors12 was the highest at the onset of S phase, being ϳ10-fold (for the 14-3-3␤ isoform) or ϳ20-fold higher (for 14-3-3⑀, -, and -␥), in cells synchronized at the G 1 /S boundary, compared with that in cells that were blocked at G 0 phase by serum starvation (Fig. 4B). The higher association of 14-3-3 isoforms with ors8 and ors12 at the onset of S phase was specific and occurred at a time when the origin becomes activated (37). In addition, the cruciform binding activity of CBP/14-3-3 was also maximal at the G 1 /S phase of the cell cycle (Fig. 5A).
The same anti-14-3-3 antibodies that were used in the chromatin immunoprecipitation assays were also used in an in vitro replication assay to investigate their effect on the in vitro replication of the minimal origin of ors8, p186. The anti-14-3-3⑀, -, and -␥ were each capable of reducing replication activity of p186 to 30 -45% of control levels, while the anti-14-3-3␤ antibody gave a lesser decrease in replication activity (50 -80% of control levels) (Fig. 7B). Thus, 14-3-3␤ appears not to be involved in DNA replication in the same way as 14-3-3⑀, -␥, and -. A possible explanation might be that the ␤ isoform is phosphorylated and hence unable to bind DNA. Phosphorylation analyses are ongoing. Alternatively, the heterodimers and homodimers of 14-3-3 involved in DNA replication are composed of different combinations of 14-3-3␤, -⑀, -␥, and -isoforms, where the 14-3-3␤ is represented in lesser amounts.
Furthermore, in bandshift/supershift experiments, the same anti-14-3-3 antibodies interfered with the formation of the CBP-cruciform complex. Antibodies to all of the four 14-3-3␤, -⑀, -␥, and -isoforms tested were able to interfere with the ability of CBP/14-3-3 to bind to cruciforms, suggesting that the epitopes of these antibodies may overlap with sites within 14-3-3 that are important for cruciform binding or that there is steric interference of 14-3-3 binding due to the large immunoglobulin molecule. The combinations of anti-14-3-3⑀, -␥, andand 14-3-3␤, -⑀, -␥, and -showed a greater interference with DNA binding than did each one of these antibodies alone. However, none of the antibodies alone or in various combinations with the other antibodies (Fig. 6B) were able to abolish completely the CBP-cruciform complex formation. This might possibly be due to the presence of additional 14-3-3 isoforms in CBP and/or the problems of stoichiometry of the binding of antibodies to their target, such as bivalency, steric hindrance, etc.
It was previously found that binding of CBP to cruciforms is not DNA sequence-specific but rather depends on the presence of the cruciform structure (27). In this study, the association of at least four of the mammalian 14-3-3 isoforms with mammalian origins of DNA replication in vivo and their cruciform binding activity was demonstrated. CBP can be a plausible combination of any two 14-3-3␤, -⑀, -␥, and -isoforms, since 14-3-3 proteins function as homo-or heterodimers (71,72). The N-terminal part of each monomer interacts extensively with the N-terminal part of the opposing monomer, forming a central channel suitable for binding to protein ligands and DNA structures, such as cruciforms (73,74). The amino acids lining the channel show extensive sequence conservation among all seven 14-3-3 isoforms found in mammalian cells (29,73). Hence, CBP may likely be a combination of any two 14-3-3 isoforms.
The relative order of isoform importance in each of the assays in this study (i.e. G 1 /S-specific abundance as detected by immunoprecipitation, in vivo origin association by chromatin immunoprecipitation, inhibition of CBP-cruciform complex formation, and inhibition of DNA replication) was analyzed. The chromatin immunoprecipitation assay showed that the association of 14-3-3 isoforms ⑀, ␥, and in vivo with mammalian origins of DNA replication, at the G 1 /S phase of the cell cycle, was approximately equivalent. The abundance of 14-3-3␤ isoform association with origins was approximately half that of the other three 14-3-3 isoforms (Fig. 4B). The abundance of total 14-3-3 isoforms immunoprecipitated, in G 1 /S phase indicates that the 14-3-3⑀ isoform is present in the nucleus in higher amounts than isoforms ␤, ␥, and , which were immunoprecipitated in approximately equal amounts (Fig. 1, A-D). This immunoprecipitation assay indicates the total amount present in the nucleus, suggesting additional roles for the 14-3-3⑀ isoform in the nucleus at the G 1 /S phase other than origin binding. The inhibition of CBP-cruciform complex formation by anti-14-3-3 antibodies indicates that the 14-3-3⑀, -␥, andisoforms interfered with the CBP-cruciform complex formation to approximately the same extent (Fig. 6B, lanes 4, 10, and 13), while the 14-3-3␤ isoform interfered to a lesser extent with it (Fig. 6B, lane 7). The in vitro replication assay also showed that anti-14-3-3⑀, -␥, and -antibodies inhibited DNA replication to approximately the same extent (ϳ60 -70%), whereas the anti-14-3-3␤ antibody again had a lesser effect, inhibiting replication to a level of ϳ25-50% of the control reaction (Fig. 7B). The combination of all four anti-14-3-3␤, -⑀, -␥, and -antibodies reduced DNA replication to ϳ20% of control levels, perhaps suggesting the presence of additional 14-3-3 isoforms involved in DNA replication.
A large number of 14-3-3 isoforms and a large number of their target proteins have been identified (29) (reviewed in Ref. 30) in many organisms. A common hypothesis is that all 14-3-3 isoforms bind with more or less specificity to a defined target and that the isoform-specific ligand interactions observed by others are due to either particular subcellular localization or transcriptional regulation of isoforms rather than to fundamental differences in their ability to bind to specific ligands. Furthermore, structural studies have not supported isoformspecific interactions with different targets (see Ref. 75 and references therein) (29), and there is little indication that such isoform specificity exists.
The data presented in this study suggest that all four 14-3-3 isoforms studied (␤, ⑀, ␥, and ) are associated with mammalian origins of replication and are involved in the initiation of DNA replication, 14-3-3␤ being less important for both origin binding and in vitro replication.
Taken together, the data suggest that CBP/14-3-3 is, as an origin-binding protein, acting at the initiation step of DNA replication by binding to cruciform-containing (activated) origins and that it dissociates after origin firing. The higher association of 14-3-3⑀, ␤, , and ␥ with origins at the G 1 /S phase of the cell cycle suggests that these isoforms act at the level of initiation of replication, presumably by binding and stabilizing the cruciform structure.