Mechanistic Insights into Replication Termination as Revealed by Investigations of the Reb1-Ter3 Complex of Schizosaccharomyces pombe

Relatively little is known about the interaction of eukaryoticreplication terminator proteins with the cognate termini andthe replication termination mechanism. Here, we report a biochemicalanalysis of the interaction of the Reb1 terminator protein ofSchizosaccharomyces pombe, which binds to the Ter3 site presentin the nontranscribed spacers of ribosomal DNA, located in chromosomeIII. We show that Reb1 is a dimeric protein and that the N-terminaldimerization domain of the protein is dispensable for replicationtermination. Unlike its mammalian counterpart Ttf1, Reb1 didnot need an accessory protein to bind to Ter3. The two myb/SANTdomains and an adjacent, N-terminal 154-amino-acid-long segment(called the myb-associated domain) were both necessary and sufficientfor optimal DNA binding in vitro and fork arrest in vivo. Theprotein and its binding site Ter3 were unable to arrest forksinitiated in vivo from ars of Saccharomyces cerevisiae in thecell milieu of the latter despite the facts that the proteinretained the proper affinity of binding, was located in vivoat the Ter site, and apparently was not displaced by the "sweepase"Rrm3. These observations suggest that replication fork arrestis not an intrinsic property of the Reb1-Ter3 complex.

INTRODUCTION

Top

INTRODUCTION

In many prokaryotic systems, DNA replication is terminated ina polar mode with respect to the origin by the interaction ofa replication terminator protein with specific sequences calledreplication termini (5, 6). In contrast, not all eukaryoticreplicons have specific replication termini, but such sequences(called Ter or replication fork barriers) are present at specializedlocations, such as the nontranscribed spacer sequences of ribosomalDNA (rDNA) (11, 28-30, 34, 51, 52) and at the mating type locusof Schizosaccharomyces pombe (18, 19). Previous work has identifiedthree different replication terminator proteins in fission yeastcalled Sap1, Reb1, and Rtf1. Sap1 binds to Ter1 (28, 29, 34)and to SAS1, a site needed for optimal mating type switching(3). Binding of Sap1 to Ter1 but not to SAS1 causes polar forkarrest (29, 34), and the architectures of the protein-DNA complexat these two sites are quite different (28). Reb1 binds to theTer2 and Ter3 sites to cause polar fork arrest, and the magnitudeof fork arrest at Ter3 is consistently greater than that atTer2 (30, 52). Rtf1 binds to the RTS1 site located near themating type locus and also causes polar fork arrest (56). Reb1and Rtf1 belong to the myb family of proteins, which containtwin canonical myb/SANT (SANT is an acronym for the transcriptionfactors Swi1 Ada2, NCOR, and TFIIIB) domains that are knownto be involved in DNA binding (1, 39) and/or interaction withhistone tails (9, 35). Hitherto, a detailed analysis of a eukaryoticterminator protein-Ter DNA interaction was available only forthe Sap1-Ter1 complex (28).

All of the replication terminator proteins of fission yeastrequire the activity of the intra-S-phase checkpoint proteinsSwi1 and Swi3 (44) for sustaining stable fork arrest (19, 30).Although the mechanism of action of Swi1 and Swi3 at the replicationtermini is presently unknown, some information has been derivedfrom investigations of the corresponding proteins, called Tof1and Csm3, of Saccharomyces cerevisiae (14, 36, 55). We havepreviously reported that Tof1/Csm3 promote stable fork arrestat Ter sites by protecting the Fob1 terminator protein fromtransient displacement by the activity of the Rrm3 helicaseduring replication fork passage (36). The Rrm3 helicase/"sweepase"acts with a 5'-to-3' polarity to promote fork passage past avariety of DNA-bound nonhistone proteins (4, 23). It is thereforepossible that Swi1/Swi3 perform a similar function in fissionyeast by protecting the replication termination complex froman unidentified "sweepase."

The Reb1 protein of S. pombe, like its mammalian counterpartcalled Ttf1 (31), also arrests transcription catalyzed by RNApolymerase I at the Ter2 and Ter3 sites. The transcription isarrested when approaching the Ter site in a direction oppositethat of replication (58). Similarly, the Rtf1 protein of S.pombe also arrests transcription catalyzed by RNA polymeraseII (16). This is in contrast with prokaryotic replication terminatorproteins Tus of Escherichia coli and RTP of Bacillus subtilis,which arrest not only the cognate replicative helicase (e.g.,DnaB of E. coli) by binding to the Ter sites and with the replicativehelicase (43) but also a variety of other helicases, such assimian virus 40 tumor antigen (SV40 Tag) (2, 8) and PriA, butnot helicases involved in rolling circle replication, chromosometransfer (helicase I), or DNA repair (e.g., UvrD and helicaseII) (25, 26, 50). These terminator proteins also arrest RNApolymerase of E. coli with the same polarity as that for DnaB(38). Mutant forms of the terminator protein that interact withTer DNA with normal or near-normal affinity but fail to interactwith the helicases also fail to arrest forks in vivo and invitro, thereby supporting the conclusion that in addition toterminator protein-Ter interaction, protein-protein interactionwith the helicase is critical to the mechanism (33, 43). Thecrystal structures of the terminator proteins and protein-DNAcomplexes have been determined, and these provide some insightsinto the fork arrest mechanisms (5, 6, 13). The DNA-bindingand fork-arresting domains of prokaryotic terminator proteinsmostly overlap each other; consequently, noninteracting mutantsisolated on the basis of the crystal structure are very fewin number (43). The Tus protein of E. coli was reported to causepolar fork/helicase arrest strictly by enhanced DNA-proteininteraction between the Tus protein and Ter DNA promoted byhelicase-catalyzed DNA melting near the blocking end of Tus,base flipping, and capture of the flipped base by Tus (42).However, our recent work shows that DNA melting and base flippingare not necessary for the polar arrest of helicase progressionand that Tus-DnaB contact(s) plays a critical role in the mechanismof helicase arrest (7).

Since a detailed knowledge of the interaction of a terminatorprotein with a replication terminus would be useful for molecularanalysis of the mechanism of fork arrest (28, 48, 49), we wishedto identify the minimal domains of the myb-like Reb1 proteinthat are needed for DNA binding and fork arrest and to determinethe critical residues of the sequence of Ter3 that are necessaryfor the process. The mammalian Ttf1 protein is a monomer witha self-inhibitory, N-terminal flap that requires interactionwith a chromatin remodeling factor for DNA binding. Withoutthe chromatin remodeling factor, the protein requires an N-terminaltruncation of the flap to bind to its cognate binding site (53).In the present work, we show that Reb1 of fission yeast is ahomodimer that apparently lacks a self-inhibitory flap and consequentlydispenses with a chromatin remodeling factor for DNA binding.The two myb/SANT domains of Reb1 and approximately 155 aminoacid residues N terminal to the SANT domains are necessary andsufficient for Ter3 binding and fork arrest. We show furtherthat the dimerization domain of Reb1 is located in the N-terminal145 amino acids. This domain is dispensable for fork arrest.Reb1 causes DNA bending, and mutations that reduced DNA bindingalso caused a reduction in DNA bending and fork arrest. Finally,expression of Reb1 in an S. cerevisiae strain that containeda chimeric plasmid containing an ars site of budding yeast andTer3 of fission yeast did not cause fork arrest at Ter3. Byfollowing replication fork movement in this reporter plasmid,we discovered that although Reb1 readily bound to Ter3 withapparently undiminished binding affinity, it could not causefork arrest. The failure of fork arrest was not caused by theRrm3 sweepase. The data suggest that fork arrest is not an intrinsicfunction of the Ter3-Reb1 complex and that additional interactions(possibly with one or more replisomal proteins) are probablyneeded to promote polar fork arrest.

MATERIALS AND METHODS

Top

MATERIALS AND METHODS

Expression and purification of Reb1-His₆ protein from E. coli. Full-length and various truncated forms of Reb1 were clonedinto pET15b and pQE31 vectors, respectively, and expressed inE. coli strain M15 or BL21. The cells were grown at to an opticaldensity at 600 nm of $~$ 0.55 to 0.65, induced by 1 mM IPTG (isopropyl-β-D-thiogalactopyranoside)at 18°C for 5 to 6 h or 25°C for 1 to 2 h, and harvested.Lysis was carried out on ice with 0.75 mg/ml lysozyme in bufferA (10 mM Tris-Cl, pH 8.0, 10% sucrose, 200 mM KCl), followedby three cycles of freeze-thawing. Cell extracts were brieflysonicated and cleared by ultracentrifugation at 30,000 rpm for30 min at 4°C. Supernatants were bound in batches for 1h at 4°C to metal ion affinity resin beads prewashed inbuffer B (10 mM Tris-Cl, pH 7.9, 200 to 400 mM KCl, 0.01% TritonX-100, 5% glycerol, 2 mM β-mercaptoethanol) containing20 mM imidazole. The beads were then packed into a disposablecolumn and washed again with 30 to 50 column volumes with thesame buffer. Protein was eluted in buffer B containing 200 mMimidazole. Fractions were immediately adjusted to 1 mM dithiothreitoland 5 mM MgCl₂ and analyzed by sodium dodecyl sulfate-polyacrylamidegel electrophoresis (SDS-PAGE) followed by Coomassie blue staining.Positive fractions were pooled. The protein was >98% pureas judged by SDS-PAGE. The protein concentration was estimatedusing the Bradford assay (9a).

Model building. With the structure of the mammalian C-myb DNA binding domainas a model (47), the structure of the DNA binding domain ofReb1 was modeled using Swiss-Model/Swiss PDB (http://swissmodel.expasy.org/SWISS-MODEL.html).

Expression of Reb1 in S. pombe and complementation. The various Reb1 open reading frames (ORFs) were expressed inthe expression vector pREP or pREP-NTAP from the inducible nmt1promoter in a reb1 ${Delta}$ strain. The ability of the mutant form toarrest replication was investigated by two-dimensional (2D)gel electrophoresis and Southern blot analyses of replicationintermediates with labeled probes for the Ter2-3 region of therDNA, as described previously (10, 37). The presence of a distincttermination spot corresponding to Ter3 (Ter2 gives a weak spot)was taken as evidence for positive complementation. A faintTer3 spot seen in replicate gels in comparison with that fora wild-type (WT) control was indicative of partial complementation.

Preparation of labeled substrates for footprinting. Ter3 fragments were PCR amplified with Vent polymerase frompReb1-IRT2 (30) to generate Ter3 fragments of $~$ 169 bp in length.The fragment was cloned into pUC19 between EcoRI and HindIIIsites, and the plasmid DNA was digested with either EcoRI orHindIII and 5' end labeled with Optikinase (USB) and [³²P] ${gamma}$ -ATP(3,000 Ci/mmol) according to the manufacturer's instructions.After labeling, the fragment was digested again with eitherHindIII or EcoRI so that the 169-bp DNA fragments were labeledat only one end.

Assays. Gel retardation was carried out as described previously (29).Dimethyl sulfate (DMS) protection, interferences, and ethylationinterference were carried out as described previously (28).Potassium permanganate (KMnO₄) probing was carried out as describedpreviously (15). Helicase assays were performed as describedpreviously (21).

Missing base contact. Missing base contact interference assays (12) were performedusing either formic acid for depurination or hydrazine for depyrimidation,as described previously (28). The binding reactions were performedusing 250 fmol modified DNA and 2.5 to 25 pmol His₆-Reb1 inReb1 binding buffer. Binding was allowed to proceed for 30 minat room temperature.

Point mutations at Ter3. Mutant forms were generated synthetically by annealing complementaryoligonucleotides containing different single and double basepair point mutations in the Reb1 binding site and were end labeledusing [³²P] ${gamma}$ -ATP and T4 polynucleotide kinase. Briefly, 2 pmolof each oligonucleotide was labeled and mixed with 10 pmol ofa complementary oligonucleotide in the presence of 70 mM Tris-Cl,pH 7.6, 10 mM MgCl₂, and 5 mM dithiothreitol or 10 mM Tris-Cl,pH 8.0, 1 mM EDTA, pH 8.0, and 600 mM NaCl. The mixture washeated in boiling water for 5 min and cooled overnight. Thelabeled DNA fragment was passed through a CL-4B spin columnbefore use. To study the role of tryptophan residues in theSANT domain of the Reb1 protein, two derivatives of N146Reb1,F414KReb1 and W353KReb1, were generated by site-directed mutagenesisin which a phenylalanine residue and a tryptophan residue atpositions 414 and 353, respectively, were each replaced witha lysine residue, as described previously (37).

N-terminal and C-terminal deletions of Reb1. To construct N-terminal deletions (N156Reb1, N166Reb1, N186Reb1,and N196Reb1) and C-terminal deletions (146-484Reb1, 146-418Reb1,and 146-364Reb1), a series of PCR amplifications utilizing differentoligonucleotides were used. After purification, the PCR productswere cut with NdeI and XhoI and ligated into a pET15b vector.The constructs were sequenced for verification. The resultingplasmids were transformed into E. coli BL21(DE3). The expressionsof recombinant proteins were induced by incubation of mid-log-phasecultures with 1 mM IPTG for 1 to 5 h.

DNA bending. DNA-bending experiments were performed essentially as describedpreviously (27). Restriction fragments of pBend-Ter3 were elutedfrom agarose gels and end labeled with Optikinase (USB) and[³²P] ${gamma}$ -ATP (3,000 Ci/mmol) according to the manufacturers' instructions.Excess free [³²P] ${gamma}$ -ATP was removed by Sephadex G-25 spin columns.Binding was performed under conditions identical to those usedfor gel shift assays, and the complexes were resolved by 8%PAGE in 1x Tris-borate-EDTA buffer. The relative mobility ofeach band was calculated for the free and bound probes by measuringthe distance each migrated from the wells.

Reb1 expression in S. cerevisiae. To express S. pombe Reb1 in S. cerevisiae, we cloned N146Reb1as an XhoI-KpnI fragment with a calmodulin-binding protein (CBP)tag at an N-terminal site into a pESC-TRP vector. The resultingconstruct (pESC-Reb1) has an additional myc tag in front ofCBP. S. pombe Reb1 binding site Ter3 (169 bp) was cloned intoanother S. cerevisiae vector, pBB3NTS1 (pBB3-Ter3), at the HpaIsite. Both of the plasmids were cotransformed into both WT andfob1 ${Delta}$ strains, and the proteins were expressed under the GAL1promoter. Reb1 protein binding was shown by an in vitro bindingassay using the CBP tag or by in vivo binding by chromatin immunoprecipitation(ChIP) using the myc tag.

In vitro binding of Myc-CBP-Reb1 to Ter3. In vitro binding of Myc-CBP-Reb1 to Ter3 was carried out asdescribed previously (37).

ChIP analysis. The BY4741 fob1 ${Delta}$ strain, expressing myc-CBP-Reb1 (induced by2% galactose) containing pESC-Reb1 and pBB3-Ter3, was used forall ChIP assays, as reported previously (30). PCR was performedusing a Ter3-specific primer along with a 35S-specific primer(used as a control), as described previously (36).

RESULTS

Top

RESULTS

Reb1 contains an N-terminal dimerization domain. Nucleotide sequence data revealed that the ORF of Reb1 (withoutthe introns) encodes a polypeptide of 504 amino acids (Fig.1A) with an estimated molecular mass of 58 kDa, a pI of 8.7,and two predicted canonical myb/SANT domains. Figure 1B showsa molecular model of the segment of Reb1 from residues 300 to418, which includes the myb I domain and part of the myb IIdomain. The model suggests that the region from amino acid residues330 to 418 of Reb1 contains the two sets of canonical R1-loop-R2-loop-R3motifs (R1, R2, and R3 being ${alpha}$ -helices) of the myb I and mybII domains. R3 is predicted to be the DNA recognition helixthat contacts the major groove of the DNA (Fig. 1B). The molecularmodel was computed using the primary sequence homology betweenReb1 and the myb family of proteins and the known solution structureof the DNA binding domain of human c-myb (46).

View larger version (51K):
[in this window]
[in a new window]

FIG. 1. Expression and characterization of fork-arresting activities of full-length and N-terminally truncated Reb1. (A) Domain structure of Reb1, showing the N-terminal dimerization (dimeriz.) domain (in green), the DNA-binding and fork-arresting domain, and the transcription termination domain (58). The myb/SANT domains are shown in blue. The numbers correspond to amino acid residues. (B) Molecular model of the myb/SANT domain of Reb1. The numbers refer to amino acid residues. (C) Gel filtration profiles of the full-length and N-terminally truncated proteins. (D) SDS-PAGE profiles of the purified proteins. (E) Gel mobility shift experiments using a labeled Ter3 probe and an increasing range of concentrations of full-length Reb1 (FLReb1) and N146Reb1. (F) Diagram showing the locations of the terminators Ter1 to Ter3 and the fork pausing site RFP4. The arrows with vertical bars show the relative locations of the arrest sites of replication forks and transcription catalyzed by RNA polymerase I. (G to K) Autoradiograms of 2D gel electrophoresis of replication intermediates of (G) an S. pombe reb1 ${Delta}$ strain complemented with His-tagged N146Reb1, (H) WT Reb1 complemented with the blank vector, (I) a reb1 ${Delta}$ strain complemented with the blank vector, (J) a reb1 ${Delta}$ strain complemented with full-length Reb1, and (K) a reb1 ${Delta}$ strain complemented with N146Reb1. The termination spots for Ter1 to Ter3 and the pause site RFP4 are identified with arrows.

We expressed both the full-length ORF of Reb1 and a truncatedform lacking the first 145 amino acids that was tagged witha His₆ affinity tag in E. coli and purified the proteins (Fig.1D). Previously, this truncated form of the tagged protein wasused to show that it terminated transcription catalyzed by RNApolymerase I at Ter2/Ter3 in vitro (58) (Fig. 1A). This observationindicated that the truncated form of Reb1 did not require anadditional protein to bind to DNA in vitro and to terminatetranscription (58). Comparative analytical gel filtration ofthe full-length and the truncated forms revealed that the formerexisted mostly (>90%) as a dimer of an estimated monomericmass of 58 kDa whereas the truncated form eluted as a monomerwith an apparent molecular mass of $~$ 38 kDa (Fig. 1C and D). Thedata supported the conclusion that the dimerization domain ofReb1 was located within the first 145 N-terminal amino acidresidues (Fig. 1A). Both forms of the protein were further investigatedfor their binding affinities for Ter3 and for their abilitiesto promote replication fork arrest in vivo, as described below.

Both full-length and truncated forms of Reb1 bind to Ter3 in vitro and terminate DNA replication in vivo. A diagram of the nontranscribed spacer region of S. pombe locatedon either ends of chromosome III is shown in Fig. 1F. The locationof ars, the directions of replication and transcription approachingthe Ter sites, and the locations at which these are arrestedare shown (Fig. 1F). First, we investigated whether the His-taggedfull-length Reb1 and the N-terminally truncated form (retainingresidues 146 to 504) had equivalent DNA binding activities byperforming gel mobility shift analysis using a labeled 169-bpintergenic fragment of rDNA containing the Ter3 site. We discoveredthat both forms of the protein bound to the Ter3 DNA with comparableaffinities ( $~$ 50 nM), as estimated by the concentration of theprotein needed for half-maximal binding (Fig. 1E; quantificationis shown in Fig. 5C). We then compared the abilities of bothforms of the protein to arrest forks at Ter2 and Ter3 in vivo(compare Fig. 1G, H, J, and K). The N-terminally truncated formwas no more efficient than the full-length protein in promotingfork arrest despite the fact that, in the example given, thespots for the N-terminally truncated form (Fig. 1K) were somewhatmore intense than those for the full-length protein (Fig. 1J).We believe that these differences are due to random experimentalvariations.

View larger version (26K):
[in this window]
[in a new window]

FIG. 5. DNA binding affinities and fork-arresting functions of various truncated forms of Reb1 and their CD spectra. (A) Schematic representation of the various N-terminal and C-terminal deletions of Reb1 and their DNA binding and in vivo fork-arresting activities. Binding column symbols: ++, strong binding; +, weak binding; –, no binding. Termination column symbols: +, fork arrest; –, no fork arrest (termination). (B) The two SANT domains, labeled DI and DII, are shown, with the amino acid residues that were mutated therein given at right. (C and D) DNA binding curves of the N-terminal and C-terminal deletions of Reb1, respectively. (E and F) 2D gel patterns of Ter3 cloned into pReb1-URA (derivative of pReb1-IRT2) and allowed to replicate in a reb1 ${Delta}$ strain. Replication intermediates were complemented with representative deletions from the N and C termini. (G and H) Comparative CD spectra of the various truncated forms of Reb1 that were expressed from an nmt1 promoter in the pREP vector.

Replication intermediates were extracted from a reb1 ${Delta}$ strainand the same strain complemented with the full-length and thetruncated Reb1 ORFs present in the expression vector pREP-NTAP,and fork progression and arrest were monitored by 2D agarosegel electrophoresis (10). The autoradiograms of the 2D gelsshowed that whereas the control reb1 ${Delta}$ strain showed a singletermination spot corresponding to Ter1 (caused by the bindingof Sap1) (29, 34), the spots corresponding to Ter2 and Ter3,which required Reb1 binding, were absent (Fig. 1H and I). Thehost strain complemented with either the full-length or thetruncated form of Reb1 showed fork arrest at both Ter2 and Ter3(Fig. 1J and K). In some cases, the RFP4 spot (presumably generatedby replication and transcription collision) was also resolved(Fig. 1J) (30). In all of the experiments reported here, weused His-tagged Reb1, and therefore it was necessary to determinewhether the tagged protein was equivalent to the nontagged proteinin fork-arresting activity in vivo; the result was in the affirmative(Fig. 1G). The data are consistent with the conclusions that(i) as contrasted with mouse Ttf1 (53), full-length Reb1 wascapable of binding in vitro to Ter3 without the need for anaccessory factor, and (ii) the N-terminal dimerization domainof Reb1 was dispensable for DNA binding and polar fork arrest.This truncated form of Reb1 is also known to terminate transcriptioncatalyzed by RNA polymerase I in vitro without the need foran accessory protein for binding to Ter3 (58).

Base-specific recognition of Ter3 by Reb1. We wished to investigate further which bases of the Ter3 sequencewere contacted by Reb1. Although previous DNase I footprintinghad approximately established the general region of the DNAsequence protected by Reb1 (see Fig. 3C) (58), the finer andimportant details of the DNA-protein interactions were not known.Consistent with our previous work (30), 2D gel electrophoresis(Fig. 1J and K) showed that Ter3 was more efficient than Ter2in arresting replication forks, suggesting that Reb1 probablybound more efficiently to Ter3 than to Ter2. We also found thatReb1 was not a single-strand DNA binding protein and did notbind to the two separated complementary single strands of Ter3(data not shown). Therefore, in all subsequent experiments wefocused on the interaction of Reb1 with double-stranded Ter3DNA.

View larger version (28K):
[in this window]
[in a new window]

FIG. 3. Chemical footprinting patterns of 146-364Reb1, which contains a single SANT domain. (A) Autoradiograms showing the methylation protection and interference patterns. The open arrowhead shows attenuation whereas the filled arrowhead shows enhancement of bases, as revealed in the methylation protection analysis. F, free; B, bound. (B) Summary of the methylation protection and interference data. (C) Summary of the base and phosphate contact and KMnO₄ sensitivity data shown in Fig. 2. The blue arrows show the directions from which either RNA polymerase I or the replication forks approach the Ter3-Reb1 complex.

First, we mapped the purine residues of Ter3 contacted by thefunctionally competent truncated form N146-504Reb1 by performingmethylation protection and interference experiments using DMSas a chemical probe. DMS methylates the N3 group of adenineand the N7 group of guanine, which are accessible in the minorand the major groove, respectively. The purines that were contactedby the protein were expected to be protected from methylationby DMS. The methylated bases were detected by depurination,followed by cleavage of the sugar-phosphate bond at the missingbase by β-elimination with piperidine, followed by resolutionof the products in denaturing DNA sequencing gels (Fig. 2A).Occasionally, a hydrophobic pocket of the protein that accumulatesDMS and overmethylates a purine residue located close to itwas revealed by enhanced breakage at that residue (Fig. 3A)(22).

View larger version (49K):
[in this window]
[in a new window]

FIG. 2. Autoradiograms showing the interaction of N146Reb1 with Ter3. (A) Methylation protection patterns. (B) Methylation interference patterns. (C) Summary of the methylation protection and interference patterns. (D) Missing base contact patterns of purines of Ter3. (E) Missing base contact patterns of pyrimidines of Ter3. (F) Summary of the patterns of purines and pyrimidines that, when missing, block interaction of N146Reb1 with Ter3. (G) Ethylation interference patterns showing the critical phosphate contacts. (H) KMnO₄ sensitivity patterns showing the locations of departure from Watson-Crick base pairing. (I) Summary of the ethylation interference and KMnO₄ sensitivity patterns. F, free; B, bound.

Methylation interference was detected by first methylating theprotein-free, 5'-end-labeled Ter3 DNA and then binding the randomlymethylated DNA to Reb1 and separating the bound from the freeDNA, followed by depurination and strand breakage by β-elimination.The results showed that the G residues at positions 2, 3, 7to 9, and 14 of the top strand were protected from methylationby Reb1. The single G residue on the bottom strand of Ter3 wassimilarly protected (Fig. 2A, B, and C). Methylation interferenceexperiments showed that the same G residues that were protectedfrom methylation by the protein were also the ones that, uponmethylation, prevented Reb1 from binding to the DNA. This conclusionis in accord with site-directed mutagenesis of Ter3, which isdiscussed below. In summary, the above-mentioned G residuesthat were physically contacted by the protein were also criticalfor Reb1-Ter3 interaction.

We also performed missing base contact analyses that detectedpurines and pyrimidines that, after modification with formicacid and hydrazine, respectively, prevented binding of the proteinto the chemically modified Ter3 sequence (12). Formic acid-generatedmissing purine contacts revealed that residues 5 to 9, 11, 12,and 14 of the top strand and residues 4, 10, 12, and 15 of thebottom strand were critical for Reb1 binding (Fig. 2D and F).Hydrazine-generated missing pyrimidine contacts revealed thatpyrimidines at residues 2, 5, 6, and 16 of the bottom strandand residues 4, 10, and 13 of the top strand were essentialfor Reb1-Ter3 interaction (Fig. 2E and F). Taken together, themissing base contact data showed the residues that were essentialfor interaction of Reb1 with Ter3.

Sugar-phosphate backbone contacts with Reb1. Apart from the base-specific contacts, ionic interactions betweenthe DNA sugar-phosphate backbone and the binding protein alsocontribute to the energy required for protein-DNA interaction.In addition, sequence-specific interaction is enhanced by contactsbetween protein and the DNA backbone. Phosphate contacts madeby Reb1 were determined by ethylnitrosourea treatment of theDNA phosphates. Ethylnitrosourea randomly ethylated the phosphategroups. Following cleavage of the backbone at the ethylatedresidues and during electrophoretic separation, each cleavageproduct migrated as a doublet consisting of a mixture of 3'-hydroxyland 3'-ethyl phosphates. The phosphates critical for bindingwere identified by comparing the cleavage patterns of the protein-boundfractions with those of the free DNA fractions. As shown (Fig.2G and I), the strongest inhibition to binding occurred uponethylation of the phosphate at the 2G, 5A, 6A, 8G, 11A, and12A positions of the top strand and positions 5T, 7C, and 14Cof the bottom strand of Ter3. We conclude from the experimentsthat the aforementioned phosphate groups were necessary forReb1-Ter3 interaction.

Helical distortion of Ter3 by Reb1. Did Reb1 binding cause any distortion of Ter3 from the canonicalB form of DNA? To answer this question, we carried out KMnO₄chemical probing experiments. KMnO₄ efficiently oxidizes unpairedthymine and, to a lesser extent, cytosine residues exposed byprotein-induced unwinding, base flipping, or other departuresfrom the canonical Watson-Crick base pairing. KMnO₄ oxidationleads to conversion of a thymine to a cis-thymine glycol (5,6-dihydroxy-5,6-dihydrothymine);the oxidized base undergoes further degradation, leading tocleavage of the DNA strand after piperidine treatment. The resultsshowed that a T at residue 10 consistently became more reactivewith KMnO₄ in the Reb1-Ter3 complex. KMnO₄ sensitivities werenot found elsewhere in the top or the bottom strand (Fig. 2H and I).

We plotted the contact points of Reb1 with Ter3 on a planarprojection of the DNA double helix (Fig. 3C). It was clear fromthe plot that a majority of contacts that were located in themiddle of the top strand in the major groove of DNA were essentialfor DNA binding as well as for replication termination (as shownbelow).

Site-directed mutagenesis of Ter3. In order to determine which of the bases of Ter3 contacted byReb1 were necessary for DNA binding and fork arrest, we synthesizeda series of complementary oligonucleotides that, upon annealing,had base substitutions at positions 2, 3, 4, 6, 7 to 10, 12,14, and 15 (Fig. 4A). We performed gel mobility shift analyseswith labeled double-stranded Ter3 DNA containing each of themutated bases (Fig. 4B) and plotted the binding as a functionof the concentration of Reb1 (Fig. 4C). The binding affinitieswere estimated from the concentration of protein that yielded50% binding to the DNA substrates. The results showed that themutants M2, M3, M4, M12, and M14-15 did not detectably reducebinding affinity with respect to that of the control, M0 (K_d= 50 to 60 nM). The mutants M6, M8, M9, and M10 had a partialreduction in binding affinity (2- to 3-fold reduction), whereasM7 showed a significant ( $~$ 7-fold) reduction in binding affinity(350 nM). We then performed 2D gel analysis of fork arrest invivo with a selected set of the mutant Ter3 sites that werecloned into a plasmid. We discovered that M6, which caused apartial reduction in binding affinity that did arrest forks,also arrested forks with a reduced efficiency compared to thatof the M0 control in vivo, whereas M9 and M10 did not show anyfork arrest in vivo (Fig. 4D).

View larger version (44K):
[in this window]
[in a new window]

FIG. 4. Interactions of various mutated forms of Ter3 with N146Reb1. (A) WT Ter3 (M0) and the various mutant forms (M2 to M15). The mutated base is shown in boldface font. (B) Autoradiograms showing gel mobility shifts of the WT and the mutant forms of Ter3 as a function of N146Reb1 concentration. (C) Binding curves of the Ter3 DNAs. (D) Autoradiograms of 2D gel electrophoresis patterns of replication intermediates of several representative mutant forms of Ter3 (shown in panel A) that were cloned in the plasmid pIRT2 (30).

Minimal domain of Reb1 necessary for DNA binding and fork arrest. Reb1 is a multidomain protein. In order to delineate the domainsinvolved in DNA binding and replication termination, a seriesof deletion mutants that were C-terminally tagged with the His₆affinity tag were constructed in the Reb1 ORFs. The truncatedproteins were expressed in E. coli, were soluble, and were purifiedby affinity chromatography on cobalt-nitrilotriacetic acid columns(not shown). A summary of the various truncated forms of Reb1and their DNA-binding and fork-arresting activities is shownin Fig. 5A. The data from DNA binding with a labeled Ter3 probeare summarized in Fig. 5C and D. The ability of each of themutant forms to terminate replication either at Ter3 locatedin chromosomal rDNA (see Fig. S1 in the supplemental material)or in a plasmid was determined by 2D gel electrophoresis ofreplication intermediates isolated from a reb1 ${Delta}$ strain, whichwere complemented with either the WT or the mutant forms ofReb1 (Fig. 5E and F). A summary of the DNA binding and terminationactivity is presented in Fig. 5A. The data showed that the N-terminalboundary of a functional protein was located at residue 156or between residues 156 and 166 whereas the C-terminal boundarywas either at residue 418 or between residues 418 and 364 (Fig.5A). As mentioned above, the deletion of up to 155 amino acidsfrom the N-terminal end of Reb1 reduced neither specific DNAbinding to Ter3 (K_d = 45 nM) in vitro nor its ability to arrestforks in vivo. This observation is in contrast with observationsfor mammalian homolog Ttf1, in which the full-length proteindid not bind to its cognate site unless its self-inhibitoryN-terminal flap was opened up by interaction with a chromatinremodeling protein. However, the removal of 10 more amino acids(leaving residues 166 to 504) reduced the binding affinity drastically(K_d = 400 to 500 nM) (Fig. 4C). Further, deletion of the N-terminal186 amino acids also, as expected, completely abolished binding(Fig. 5C). Thus, the N-terminal boundary of the DNA bindingdomain seems to reside near residue 156.

The deletion of 20 C-terminal amino acids (from residues 146to 484) did not affect the binding affinity (K_d = 40 nM). Therewas also no detectable effect on binding after further removalof a total of 86 amino acids from the C-terminal portion (K_d= $~$ 40 nM) (Fig. 5A and D). However, deletion of the entire Cterminus, up to and including the entire myb II domain (146-364Reb1),significantly reduced the binding affinity (K_d = 130 nM).

While working with deleted forms of a protein, a major concernis whether the deletions (or point mutations) cause loss offunction by ablating a functional domain or whether the lossof function alternatively has a more trivial origin, such asglobal misfolding of the derivative protein. For example, the146-364 deletion, which lacks the second SANT domain, showedreduced binding to Ter3 and had lost the ability to terminateforks. Two lines of evidence show that this loss of functionwas not caused by a global misfolding of the protein: (i) themethylation protection of the single SANT domain derivativewas very similar to that of the 146-418 form, which was fullyfunctional (compare Fig. 3A and B with 2A and B, respectively),and (ii) the circular dichroism (CD) spectra of the 146-364protein were almost identical to those of the 146-418 form,thereby showing that the derivatives did not have a disorderedstructure (Fig. 5F). Similarly, any N-terminal deletion thatwent past amino acid 156 was defective in both DNA binding andfork arrest. The CD spectra of the N-terminal deletions werevirtually indistinguishable from those of the 146-504 form.An example of the data is shown in Fig. 5G.

Role of conserved tryptophan residues. The structure of the DNA binding domain of c-myb has been shownto consist of three imperfect tandem repeats of 51 or 52 aminoacids which form three well-defined helices (45) (Fig. 1B).Each helix contains three conserved tryptophan residues spaced18 or 19 amino acids apart. This structure, referred to as the"tryptophan cluster," represents a characteristic property ofthe myb family of proteins (24). The homology of the DNA bindingdomain of TtfI to those of c-myb and Reb1 prompted us to focuson conserved tryptophan residues that are known to be part ofthe hydrophobic core of the myb/SANT DNA binding motif. To examinethe functional significance of this sequence conservation, weconstructed an N146Reb1 protein derivative in which the phenylalanineresidue at position 414 (equivalent position of Trp688 of TtfI)in DNA binding domain II was replaced by a lysine residue (F414K)by site-directed mutagenesis. The binding affinity of this mutantwas compared with that of N146Reb1 (Fig. 5B and D). Equal amountsof fully functional protein and the F414K mutant form were examinedby gel mobility shift assays, and the results showed no significantchange in binding affinity (K_d = 38 to 39 nM) between the twoforms. In contrast, the W353K mutant form of myb I caused completeloss of Ter binding (Fig. 5D). The mutation of conserved phenylalanineresidues in SANT domain II (F414KReb1) caused no effect on replicationtermination (see Fig. S1G in the supplemental material), butthe W335K mutant form of myb I could not arrest forks at eitherTer2 or Ter3 (see Fig. S1H in the supplemental material). Notsurprisingly, all mutations that significantly reduced DNA bindingactivity were also impaired in replication termination.

Alternate sequence recognition by a Reb1 derivative containing a single myb/SANT domain. We wished to determine the base contacts made individually bythe canonical myb I and myb II domains on the Ter3 sequenceby chemical footprinting techniques. We performed methylationprotection and interference studies with two truncated proteins,146-418Reb1 (without the C-terminal region including a portionof myb II) and 146-364Reb1 (containing the single myb I domain).It should be noted that the first deletion was expected to retainthe helix-loop-helix-loop motif of myb II, according to thestructural model presented in Fig. 1B. The data showed minordifferences in methylation protection and interference patternsbetween the protein having just myb I (Fig. 3A) and the controlhaving both myb I and myb II motifs (not shown). The methylationinterference patterns, in contrast, did not change upon removalof myb II. Methylation protection showed that G8 was not protectedbut G9 showed significantly higher enhancement when myb II wasdeleted, thereby partially revealing the myb II contacts versusthose contributed by myb I. The data are consistent with theinterpretation that although myb I contributed most of the contactswith Ter3, the myb II contacts, although fewer, were necessaryfor optimal DNA binding and for fork arrest (Fig. 3A and B).

Reb1-induced bending of Ter3. Several DNA binding proteins, including the replication terminatorprotein RTP of Bacillus subtilis, are known to bend DNA (27).The bent DNA, along with other structural changes by the DNAligand, induces asymmetry in the otherwise symmetrical RTP andprovides the structural basis for accomplishing polar fork arrestfrom an otherwise symmetrical apoprotein (57). These considerationsled us to determine whether Reb1 caused bending of Ter3 andwhether binding mutants showed reduced bending.

Bent DNA migrates anomalously in a nondenaturing polyacrylamidegel because of its reduced ability to reptate through the poresof the gel matrix in comparison with straight, more flexibleDNA of identical length and nucleotide sequence (17, 41). Themobility of the protein-DNA complex depends on the locationof the bending locus within DNA fragments with respect to thetwo ends of the DNA. Minimal migration occurred when the bendwas located near the middle of the fragment, and the mobilityproportionately increased by progressive placement of the bindinglocus closer to the ends of the DNA. The pBend2 vector (27),which provides permuted linear DNA fragments of identical lengthsbut with the bending locus at different locations with respectto the ends, was utilized to study Reb1-mediated DNA bending.The vector contains tandem direct repeats of identical DNA segmentscontaining 17 restriction sites surrounding the cloning sites.Restriction by various enzymes thus produces fragments withcircularly permuted binding sites. A 29-bp DNA fragment containingthe 17-bp Ter3 was cloned into this vector to produce pBend-Ter3.Gel shift assays were performed with fully functional N146Reb1as well as truncated 146-418Reb1 and 146-364Reb1 proteins (Fig.6). Figure 6A shows the position of Ter3 in the $~$ 153-bp fragments.The relative mobility of the Reb1-Ter3 complex was inverselyproportional to the distance of Ter3 from the end of the fragment.Figure 6B shows that the mobility was most retarded when Ter3was located at the center of the fragment and least affectedwhen Ter3 resided at the ends. The migration patterns were similarin all three cases, but, interestingly, in the case of 146-364Reb1,the bending was reduced in comparison with that of the 146-504protein (Fig. 6C). The retardations of mobility for N146Reb1and 146-418Reb1 were almost similar (Fig. 6C). This appearsto be consistent with the observation that the C-terminal regionis not necessary for DNA binding and bending. We have shownabove that the myb I domain was important for replication termination.The truncated 146-364 protein, which contained the single mybI domain and showed reduced binding affinity for Ter3, alsocaused less bending. The mutant forms M9 and M10, which showedpartial reduction in affinity for Reb1, showed slightly reducedbending (Fig. 6D). There appeared to be an approximate correspondencebetween the DNA binding affinity and the degree of DNA bending,and a reduction in bending appeared to cause a loss of the abilityto arrest forks.

View larger version (41K):
[in this window]
[in a new window]

FIG. 6. DNA-bending and helicase-arresting activities of N146Reb1. (A) Various circular permutations of the Ter3 sequence present in DNA fragments of identical lengths generated by cutting the pBend2 vector. (B) Gel mobility profiles of the DNA fragments shown in panel A in a 5% nondenaturing polyacrylamide gel. (C) Restriction site location versus mobility of the fragments that were bound to various truncated forms of Reb1. (D) Restriction site location versus mobility of mutated Ter3 fragments that were bound to N146Reb1. (E) Autoradiogram of 8% nondenaturing polyacrylamide gel showing the ability of the Tus protein of E. coli to arrest DnaB helicase in a partially duplex M13 DNA hybridized to a complementary oligonucleotide containing the Ter site. S, substrate; H, substrate plus DnaB. (F) Same as for panel E, except that the M13 DNA substrate containing Ter3 (Reb1 binding site) and the Reb1 protein were tested against DnaB activity. (G and H) Quantification of the data (from triplicate gels) shown in panels E and F, respectively. oligo, oligonucleotide.

Reb1 does not arrest several helicases in a polar mode. We wanted to investigate whether purified Reb1 would arrestany of the helicases derived from prokaryotes, archaea, or amammalian virus. Some of these helicases, such as DnaB, translocateon DNA with a 5'-to-3' polarity, whereas others do so with theopposite (3'-to-5') polarity (e.g., PriA, SV40 Tag, and archaealMCM1) (50). Helicase assays were performed using single-strandedM13 circular DNA that contained a cloned binding for Ter3 inboth orientations and a 3'- or 5'-tailed labeled oligonucleotidethat was complementary to the Ter3 sequence present in M13 DNAand formed a partial duplex with it. The ability of helicasesto melt and dislodge the oligonucleotide from the partial duplexin the presence of increasing concentrations of Reb1 was analyzedby 8% nondenaturing PAGE. We also performed a control DnaB helicaseassay in the presence of the replication termination proteinof E. coli called Tus (20). The results showed that while DnaB,as expected, was arrested in a polar mode by the Tus-Ter complex,the Ter3-Reb1 complex did not arrest DnaB with any recognizablepolarity (Fig. 6E to H). The same was true for SV40 Tag andarchaeal MCM1 (not shown). Since the putative replicative helicaseMCM2-7 of S. pombe is not available at this time in a physiologicallyactive form, we were unable to determine whether Reb1 functions,like its prokaryotic counterparts, by arresting the homologousreplicative helicase. Since, as noted above, Reb1 does not needa chromatin remodeling protein or a second protein to bind toTer3 in vitro, the failure to arrest the helicases was probablynot due to a failure to bind to Ter3 but because the helicaseswere not intrinsically targeted by the Ter3-Reb1 complex. Sincepurified Reb1 expressed in E. coli does not require accessoryproteins for robust binding to Ter3, the failure to arrest thehelicases in a polar mode (there was nonpolar arrest due tothe Ter3-Reb1 interaction [Fig. 6F and H]) was not due to afailure of the Ter3-Reb1 interaction in vitro but more likelyoccurred because these helicases were not targeted by the protein-DNAcomplex.

Fork arrest is not an intrinsic property of the Reb1-Ter3 complex. We addressed the interesting mechanistic question as to whetherReb1-Ter3 interaction was both necessary and sufficient forpolar fork arrest or whether polar fork arrest needed additionalinteractions. Although the few point mutants and several deletionsof Reb1 showed that every one that was defective in DNA bindingwas also defective in fork arrest, these results do not necessarilysupport the hypothesis that Reb1-Ter3 interaction is both necessaryand sufficient for polar fork arrest. The observation that theTer3-Reb1 complex arrests RNA polymerase I in vivo and in vitroin one orientation (58) and replication forks approaching thecomplex from the opposite direction is difficult to explainon the basis of just the Reb1-Ter3 interaction without any additionalinteractions with transcription and the replication complex.

We do not yet know whether there are separable domains for DNAbinding and for fork arrest in the protein or whether thesedomains are mostly overlapping, as in the case of bacterialterminator proteins (6). We therefore approached the questiondiscussed above by using a different method, as described below.

If polar fork arrest were an intrinsic property of the Reb1-Ter3complex, one would expect that a properly formed Ter3-Reb1 complexwould arrest replication forks in a foreign cell milieu, suchas that of the budding yeast, provided that the Reb1 proteindid not lose binding affinity for Ter3 in vivo, perhaps dueto a posttranslational modification, or that it was not displacedfrom Ter3 during fork passage by the Rrm3 "sweepase."

We proceeded to test this hypothesis by constructing a chimericreplicon, pBB3-Ter3, that included ars of budding yeast andTer3 of fission yeast placed immediately before the locationsof the Ter1 and Ter2 sites of budding yeast. We also constructedthe plasmid pBB-Ter3', which contained the Ter3 site locatedimmediately after the Ter2 site of budding yeast, further awayfrom ars (Fig. 7A). The proper expression, correct folding,and retention of sequence-specific Ter3-binding activity oftagged Reb1 in budding yeast were determined as follows. Weextracted the Reb1 protein tagged with CBP and myc tags fromthe cells during the exponential phase of growth and immobilizedthe soluble cell extract onto a calmodulin-Sepharose columnas described previously (37). A radiolabeled Ter3 fragment anda control-labeled fragment obtained from a DdeI-digested pUC18DNA were mixed and applied to the immobilized Reb1 affinitycolumn. The same DNA was also applied to a blank column (notshown). The column was washed, and the bound DNA was elutedwith EGTA and resolved by 8% nondenaturing PAGE. The autoradiogramsshowed that only the Ter3 fragment in the free form and as aReb1-Ter3 complex remained bound to the column and was elutedby EGTA. None of the control fragments generated from pUC18DNA were retained on the affinity column after the initial washingsteps (Fig. 7B). We also investigated whether Reb1 protein expressedin budding yeast was able to bind to the Ter3 site located inthe chimeric replicon in vivo by performing ChIP assays usinganti-myc antibodies and found that the protein was bound specificallyto Ter3 and not to the 35S RNA-encoding region of the rDNA (Fig.7C). However, it is possible that expression in vivo in buddingyeast could cause a decrease in binding affinity for Ter3, perhapsdue to a posttranslational modification of the protein. We thereforecompared the binding affinities of the protein expressed inbudding yeast and that from E. coli and found no detectabledifference (Fig. 7D). The protein expressed in E. coli retainsrobust DNA binding activity in vitro and is able to arrest RNApolymerase I-catalyzed transcription in vitro without any needfor an accessory protein to augment its DNA binding affinity(58).

View larger version (54K):
[in this window]
[in a new window]

FIG. 7. DNA binding and fork-arresting activities of N146Reb1 expressed in S. cerevisiae. (A) Structures of the tagged Reb1 donor plasmids and the chimeric replicons pBB3-Ter3 and pBB-Ter3'. (B) Autoradiogram of a 5% nondenaturing polyacrylamide gel showing in vitro sequence-specific binding of Reb1 expressed in S. cerevisiae to a labeled Ter3 probe. The protein was immobilized on a calmodulin-Sepharose column. (C) Photograph of a 1% agarose gel showing results from a ChIP assay of myc-tagged Reb1 binding to the Ter3 site in vivo and not to the 35S region of rDNA. ab, antibody. (D) Comparative binding affinities of Reb1 expressed in E. coli (E-Reb1) and that from S. cerevisiae (Y-Reb1), as determined by gel mobility shift assays. (E to H) Autoradiograms of 2D gel patterns of replication intermediates of pBB-Ter3 or pBB-Ter3' isolated from S. cerevisiae. (E) pBB-Ter3 from WT yeast. (F) pBB-Ter3 from the fob1 ${Delta}$ strain. (G) pBB-Ter3' from the fob1 ${Delta}$ strain. Diagonal line, region of the arc where a Ter3 spot would have been visible if fork arrest had occurred. (H) pBB-Ter3 from the rrm3 ${Delta}$ fob1 ${Delta}$ strain. 1, Ter1; 2, Ter2; 3, Ter3; ${chi}$ -arc, double-Y arc.

We performed 2D gel analyses, following digestion with SspI,to monitor replication fork progression in the chimeric repliconpBB-Ter3 present in both the WT and the fob1 ${Delta}$ strains. The datashowed that in the Fob1 cells, as expected, termination spotscorresponding to Ter1 and Ter2 of budding yeast were readilyvisible not only on the Y arc but also, more prominently, onthe double-Y arc (Fig. 7E). The spots on the double-Y arc weregenerated by the arrest of the first fork initiated and propagatedbidirectionally from ars followed by the arrest of the secondfork, which proceeded in the opposite direction around the circleand met the first one, at the two Ter sites, generating (afterSspI digestion) an X-shaped intermediate. In these DNA preparations,there seemed to be more late intermediates (double-Y or X forms)than early ones (Y forms). As expected, in fob1 ${Delta}$ cells, the terminationspots corresponding to budding yeast Ter1 and Ter2 were abolished.However, despite this reduction in the background, no terminationspot corresponding to fork arrest at Ter3 was detected in threeindependent experiments, on either the Y or the X arc (Fig.7E and F).

To ensure that a termination spot corresponding to Ter3 wasnot obscured by the prominent monomeric spot, we moved the Ter3site further away from ori by cloning it 50 bp after the Ter2site and repeated the 2D gel analysis. The data from multipleruns (which were prolonged to stretch out the Y arc) did notshow any termination spot (Fig. 7G). We conclude from theseexperiments that even though Reb1 was expressed in a functionalform and was able to recognize and bind to the Ter3 site invivo in the cell milieu of budding yeast, it failed to arrestreplication forks in budding yeast.

These results could be explained by an alternative hypothesisthat the failure of Reb1-Ter3 to arrest forks was caused bythe displacement of the protein from Ter3 by the Rrm3 sweepaseduring fork passage. We investigated this question by transformingthe pBB-Ter3 plasmid into rrm3 ${Delta}$ and fob1 ${Delta}$ rrm3 ${Delta}$ strains of S. cerevisiaeand repeating the 2D gel analyses. Neither in the rrm3 ${Delta}$ strain(not shown) nor in the fob1 ${Delta}$ rrm3 ${Delta}$ strain were there any visibletermination spots corresponding to the location of Ter3 withrespect to ori, which was detectable in the blocking orientationof the terminus (Fig. 7H). Control experiments with Ter3 placedin the nonblocking orientations were also performed and, asexpected, revealed no termination spot (not shown). We concludefrom these experiments that Reb1-Ter3 failed to arrest forksin the cell milieu of S. cerevisiae, showing that fork arrestwas not an intrinsic property of the terminator complex.

DISCUSSION

Top

DISCUSSION

The present work was undertaken to identify the minimal domainsof the interesting multidomain and multifunctional Reb1 proteinof S. pombe that were necessary for binding to the replicationterminus Ter3 in vitro and causing polar fork arrest in vivoand to define the specific bases that interacted with Reb1 tocause polar fork arrest and DNA bending. Previous work withthe replication terminator protein of B. subtilis has shownthat DNA ligand-mediated structural changes in the protein-DNAcomplex appear to be involved in polar fork arrest (57). Mutationalanalysis showed that either base changes in Ter3 or mutationalalterations in Reb1 that reduced the binding affinity of Reb1with Ter3 also caused reduction of DNA bending in vitro andelimination of the ability to arrest replication forks in vivo,thereby suggesting the possible involvement of the structuralalterations in Ter3 in fork arrest.

Although the dimerization domain of Reb1 located in the N-terminal145 amino acid residues was not necessary for causing fork arrest,our recent work shows that this domain is critical for loopingDNA (S. Singh, S. Biswas, and D. Bastia, unpublished data).What might be the biological contributions of DNA looping tofork arrest and transcription termination catalyzed by Reb1?We hypothesize that DNA looping promotes long-range interactionsbetween the two clusters of rDNA located at the opposite endsof chromosome III and Reb1 sites located in the other two chromosomesand might bring the sites to a "termination factory" locatedin the nucleolus. The looping also might contribute to more-efficientfork arrest by promoting cooperativity at a distance.

Is the fork arrest activity of Reb1 intrinsic to the Ter3-Reb1complex? In the absence of detailed knowledge of the biochemistryof the process that might have enabled a direct test of theproposition, we attempted to address this question by expressingthe Reb1 protein in the cell milieu of S. cerevisiae and askedthe question as to whether the protein-Ter3 complex could arresta generic fork initiated from the origin of replication of buddingyeast at the fission yeast replication terminus. We found thatthe forks progressed past the Ter3 site unimpeded, althoughthe protein was produced in vivo, retained its sequence-specificDNA binding activity, and was bound in vivo to the Ter3 site.

The interpretation of the aforementioned experiment has takeninto consideration the following points. Our previously publishedwork shows that stable fork arrest at Ter3 and at the otherTer sites in the rDNA of S. pombe also requires the checkpointproteins Swi1 and Swi3 (30). However, these proteins do notappear to be specific to the Reb1-Ter3 complex and tend to workat any nonhistone protein that is tightly bound to DNA (e.g.,Sap1 binding at Ter1 and Rtf1 binding to the RTS1 site). Thecorresponding system in S. cerevisiae consists of the homologousTof1/Csm3 protein complex, which also does not act in a terminatorprotein-specific fashion and promotes fork arrest by counteractingrrm3 at a variety of sites that bind tightly to nonhistone proteins(36). If Rrm3 helicase removed the Reb1 protein from Ter3 invivo despite the presence of a functional Tof1/Csm3 complex,such a polar arrest should be detectable in an rrm3 ${Delta}$ or an rrm3 ${Delta}$ fob1 ${Delta}$ strain. The data presented in this work show that the failureof the Ter3-Reb1 complex to arrest forks in S. cerevisiae couldnot be attributed to the removal of Reb1 from Ter3 by the Rrm3helicase.

Although purified Reb1 did not arrest several prokaryotic, archaeal,and eukaryotic helicases in vitro, it is possible that it mightspecifically arrest the putative fission yeast helicase MCM2-7(32). Additional work will be necessary to define the replicativehelicase of fission yeast and its interaction with proteinssuch as the GINS complex (40, 54), and isolation of the helicasein a physiologically correct and active form will be necessaryto gain further insight into the fork arrest mechanism. Determinationof the crystal structure of the Reb1-Ter3 complex should alsoprovide valuable clues to understanding the mechanism, and suchwork is in progress in our laboratory.

ACKNOWLEDGMENTS

This work was supported by grants from the NIAID and NIGMS ofthe National Institutes of Health to D.B.

We thank Tony Carr, Susan Forsburg, Bidyut Mohanty, Gregor Krings,and the anonymous reviewers of the manuscript for many helpfulsuggestions and constructive criticisms. We thank Louise Pape,Kathy Gould, Shankar Adhya, Tom Kelly, Jacob Dalgaard, and BenoitArcangioli for the gifts of many useful strains and plasmidsand Starr Hazard for building the Reb1 model.

FOOTNOTES

* Corresponding author. Mailing address: Department of Biochemistry and Molecular Biology, Medical University of South Carolina, 174 Ashley Ave., Charleston, SC 29425. Phone: (843) 792-0491. Fax: (843) 792-8568. E-mail: bastia{at}musc.edu

Published ahead of print on 15 September 2008.

Supplemental material for this article may be found at http://mcb.asm.org/.

REFERENCES

Top