JOURNAL OF BACrERIOLOGY, June 1993, p. 3443-34510021-9193/93/113443-09$02.00/0Copyright © 1993, American Society for Microbiology

Vol. 175, No. 11

Fine-Structure Analysis of the P7 Plasmid Partition SiteFINBARR HAYES, MICHAEL A. DAVIS, AND STUART J. AUSTIN*

Laboratory of Chromosome Biology, ABL-Basic Research Program, NCI-Frederick CancerResearch and Development Center, Frederick, Maryland 21702

Received 19 January 1993/Accepted 18 March 1993

Thepar region ofbacteriophage P7 is responsible for active partition ofthe P7 plasmid prophage into daughtercells. The cis-acting partition site was defined precisely as a 75-bp sequence that was necessary and sufficient topromote correct segregation of an unstable vector plasmid when the two P7 partition proteins, ParA and ParB,were supplied in trans. Roughly the same region was necessary to exert partition-mediated incompatibility. Theminimal site contains an integration host factor (IEHF) protein binding site bracketed by regions containingheptamer repeat sequences that individually bind ParB. An additional sequence forms the left boundary of thesite. Site-directed mutations in the latter sequence, as well as the IHF motif and the rightmost ParB box, blockedsite function. Although the P7 site shares 55% sequence identity with its counterpart in bacteriophage P1,functional interactions between the partition sites and the Par proteins of the two plasmids were entirely speciesspecific in vivo. The P1 sequence has similar IHF and ParB binding motifs, but the left boundary sequence differsradically and may define a point of species-specific contact with the Par proteins. No evidence was found for theexistence of a functional P7 analog of the P1 parS core, a small subregion of the P1 site that, in isolation, acts asan enfeebled partition site with modified incompatibility properties.

Bacterial plasmids encode a variety of distinct mecha-nisms designed to ensure their stable inheritance at celldivision (30). In addition to site-specific plasmid resolutionand cell-killing systems (30), loci implicated in active parti-tion have been identified on a number of low-copy-numberplasmids (23, 30, 37). Such systems are critical for plasmidswhose copy number is too low to depend wholly on passivecytoplasmic diffusion to guarantee their faithful segregation.

Bacteriophage P1 lysogenizes its Escherichia coli host as ahighly stable unit-copy plasmid (6). This remarkable stabilityis primarily due to an active partition system composed oftwo essential trans-acting genes, parA and parB, and adownstream cis-acting site, parS, which appears to be func-tionally analogous to a eukaryotic centromere (3, 4, 18, 28).TheparB gene encodes a site-specific DNA binding protein(ParB) that has been postulated to recognize a set of hep-tamer repeats inparS (14). The host protein, integration hostfactor (IHF), binds cooperatively with ParB atparS (14, 15,19), at which it induces DNA bending in vitro (20). Althoughboth the ParA and ParB proteins are essential for partition(18), no direct binding of ParA to parS has been detected.ParA is a member of a diverse group of ATPases thatfunction in plasmid maintenance (16, 29). TheparA andparBgenes form an autoregulated operon. Although ParA is theonly protein known to bind directly to the operon promoterregion (16), both ParA and ParB are required for completepar operon transcriptional autoregulation (18).

Plasmids that share the P1 partition region express mutualincompatibility (3). This characteristic is most convenientlyexplained by considering plasmids a pool from which indi-viduals are selected randomly for pairwise partition (31).Two plasmids that share the same partition site cannot bedistinguished from each other during the partition process,and random assortment ultimately gives rise to populationsthat have one or the other plasmid but not both. The P1 parincompatibility effect has been termed IncB+ (2, 15). As thecrucial recognition locus for P1 partition is parS, the parS

* Corresponding author.

site and the determinant for IncB+ should be the same locus.This proves to be the case (4). The limits of the sequence thatconfers the IncB+ phenotype have been mapped to an 84-bpregion (15; Fig. 1). This 84-bp site (parS) is necessary andsufficient to stabilize low-copy-number plasmid constructswhen P1 partition proteins are supplied in trans. However,in specific contexts, a 22-bp sequence internal to the 84-bpregion retains significant parS activity (27). This parS coresequence appears to be an inefficient site, as plasmids thatrely on it for partition are less stable than their wild-typecounterparts and are incapable of competing in incompati-bility tests with par plasmids containing wild-type sites (15,28). Although the parS core site is enfeebled and functionsonly in specific contexts (22), it must contain key informa-tion for partition. The core sequence binds ParB and consistsof a pair of inversely repeated heptamer boxes and someadjacent sequences (27; Fig. 1). Deletion and substitutionmutations in this 22-bp sequence fully abolishparS-mediatedpartition activity (27).P7 is a bacteriophage closely related to P1, with which it

shares an active partition system that likely originates froma common ancestral source (26). However, while the P1 andP7 plasmid partition operons are homologous and collinear,mutations in PlparA andparB are not complemented by theequivalent P7 genes and vice versa (26). Similarly, while theP1 and P7 cis-acting partition sites are unequivocal homologsand possess identically spaced IHF and ParB binding motifs(26; Fig. 1), they function as partition sites only when theircognate proteins are supplied (this study). An examination ofthe molecular basis for the species specificity of thesesystems should allow a definition of the key macromolecularinteractions in plasmid partition. This report describes anassay procedure suitable for a comparison of the two sys-tems and examines the partition and incompatibility proper-ties of the P7 cis-acting partition site.

MATERIALS AND METHODS

Media, enzymes, and materials. Cultures were grown inLuria (L) medium at 370C unless otherwise stated. Antibiot-

3443

on February 9, 2018 by guest

http://jb.asm.org/

Dow

nloaded from

3444 HAYES ET AL.

P1

P7

T

parS(incB+)d

r ." parScore (incB )

D S HI I

oiim o3O--*-

F Hp X D HL I I I

D3LS DOID

50 bp

FIG. 1. Partial restriction map alignments of the P1 and P7 parSregions. The open boxes indicate the positions of the heptamerrepeat sequences recognized in vitro by ParB. IHF motifs are

represented by the hatched boxes. Perfect (P1) and imperfect (P7)palindrome sequences are indicated by the horizontal invertedarrows. The limits of the P1 minimal parS (incB+) and parS core

(incBd) sequences are shown above the maps. D, DraI; F, Fnu4HI;H, HincII; Hp, HinPI; S, StyI; T, TaqI; X, XmnI.

ics were added to media at the following concentrations (inmicrograms per milliliter): ampicillin, 50; chloramphenicol,10; and spectinomycin, 25. Restriction endonucleases wereobtained from Bethesda Research Laboratories, Inc. (Gaith-ersburg, Md.) or New England BioLabs (Beverly, Mass.)and used as suggested by the manufacturers. T4 DNA ligase,the Klenow fragment of DNA polymerase I, and calf intes-tinal alkaline phosphatase were supplied by BoehringerMannheim Biochemicals (Indianapolis, Ind.). T4 polynucle-otide kinase was provided by Pharmacia LKB Biotechnol-ogy (Piscataway, N.J.). Reagent-grade chemicals were sup-plied by Sigma Chemical Co. (St. Louis, Mo.).

Bacterial strains and bacteriophages. E. coli HB101 (F-mcrB mrr hsdS20 [rB- mB-] leu supE44 ara-14 galK2 lacYlproA2 rpsL20 [Strr] xyl-5 mtl-i recA13 A) (10) was used forplasmid propagation and in cloning experiments. StrainN100 (recA13 galK pro Strr) (13) was used in plasmidincompatibility tests. The effect of an IHF mutation onplasmid incompatibility was determined with strain RW1840(supE supF metB r- m- galAJ82 himA::TnlO), which wassupplied by R. Weisberg. CC1572 is an N100 derivative intowhich chromosomally located bacteriophage XW82 (xis6ind-) was introduced. The presence of XW82 in this strainblocks A lytic growth during partition tests by providing astable source of the X repressor (8). CC1531, CC1532,CC1577, CC1579, and CC1596 are CC1572 transformantsthat harbor pGB2, pALA480, pBR322, pALA271, andpALA1023, respectively. Bacteriophage X stocks were main-tained on strain YMC (supF) (17). Phage Ximm21-P7:79Cm isa derivative of imm21-P7: 79 (25). The cat gene was intro-duced into X-P7:79 by homologous recombination as de-scribed previously for the construction of XcI857-P1:5RCm(15). X-P7:79Cm lysogenizes E. coli as a stable unit-copyminiplasmid directed by the P7 plasmid maintenance system.Similarly, XcI857-P1:5RCm replicates as a low-copy-numberP1 miniplasmid driven by the P1 replication and partitionregions (15). Bacteriophage XcI857-P1:5RA1005 contains theP1 replication region but no partition sequences (5).

Plasmids. Plasmid pALA271 (2) consists of the P1 parA

and parB genes cloned in pBR322 (9). Plasmids pALA407(28) and pALA1693 contain a 109-bp TaqI-StyI parS se-quence of P1 cloned in pBR322 and the partition pickupassay vector, pALA1626, respectively.

Plasmid pALA480 contains the P7 parA and parB genessubcloned in pGB2 (12) as an EcoRI-HindIII fragment frompALA1023 (26). Plasmids pALA1602 (P7 bp 2382 to 2476),pALA1609 (P7 bp 2301 to 2476), pALA1611 (P7 bp 2335 to2476), pALA1613 (P7 bp 2301 to 2451), and pALA1614 (P7bp 2335 to 2450) (see Fig. 2) contain overlapping restrictionfragments from the vicinity of P7 parS and cloned asblunt-end segments in EcoRV-cleaved pBR322. PlasmidpALA481 consists of a BglII-DraI fragment (P7 bp 1946 to2450) spanning P7 parS and cloned into BamHI-EcoRV-cleaved pBR322. Plasmid pALA1601 (see Fig. 2) was con-structed by restricting pALA481 with XmnI and religating it,thereby generating a deletion derivative of pBR322 contain-ing P7 bp 2382 to 2451. Similarly, pALA1616 (P7 bp 2301 to2382) is a deletion derivative generated by digestingpALA1609 with XmnI and religating it.

Plasmid constructs for mapping of the right boundary ofP7 parS were made as follows. Insertion of a double-stranded blunt-end synthetic 30-bp linker containing uniqueStuI, BglII, and SmaI sites in XmnI-digested pALA1616generated pALA1617. The latter plasmid therefore containsP7 bp 2301 to 2391 immediately followed by restriction sitesinto which synthetic oligonucleotides composed of differentlengths ofparS sequence to the right of position 2390 wereconveniently cloned in a manner that facilitated the exactreconstruction of P7 sequences at the insertion point. Plas-mids pALA1621 to pALA1625, pALA1630, pALA1638,pALA1639, pALA1645, and pALA1672 (see Table 3, set A)were generated in this fashion by use of double-strandedoligonucleotides with ends complementary to StuI and BglIIsites.

Plasmid constructs for determining the left boundary ofparS were made as follows. A 66-bp sequence (P7 bp 2335 to2400) containing a defined right partition site boundary waspurified as a HinPI-SmaI fragment from pALA1638 andligated simultaneously to (i) a double-stranded 37-bp syn-thetic oligonucleotide containing bp 2301 to 2334 (but with aC-to-G mutation at position 2332) and EcoRV and HinPIcomplementary ends and (ii) EcoRV-digested pBR322. Thisprocedure produced plasmid pALA1648, which contains P7bp 2301 to 2400 but with a C-to-G substitution at position2332 that results in the formation of a unique BssHII site atthis position. In addition, the P7 material in this construct isflanked by EcoRV and BglII sites at the left and rightboundaries, respectively. Double-stranded synthetic oligonu-cleotides composed of different lengths of partition site se-quence to the left of position 2333 were introduced intoEcoRV-BssHII-digested pALA1648 to generate pALA1649 topALA1653, pALA1673, and pALA1675 (see Table 3, set B).The pBR322- and pALA1626-based plasmids (pALA1753

to pALA1756 and pALA1757 to pALA1760, respectively)containing site-specific mutations in the minimalparS regionwere generated by inserting oligonucleotides with the appro-priate mutations and EcoRV and BamHI compatible ends inEcoRV-BamHI-cleaved pBR322 and pALA1626 (see Fig. 4).Plasmids containing wild-type parS regions and similarflanking vector sequences for use as controls were thepALA1626-based plasmid pALA1659 (see Table 3, set B)and the pBR322-based plasmid pALA1770. The latter plas-mid was generated by digesting pALA1652 (see Table 3, setB) withBamHI and BglII and religating it, thereby producinga plasmid in which the vector-derived sequences immedi-

J. BACTERIOL.

on February 9, 2018 by guest

http://jb.asm.org/

Dow

nloaded from

P7 PLASMID PARTITION SITE 3445

ately adjoining the rightmost boundary of the parS sitediffered at only one position for plasmids harboring eitherwild-type or mutated sequences.

Nucleotide sequence determinations. Minipreparations ofplasmid DNA for nucleotide sequencing were purified by thealkaline lysis method essentially as described elsewhere (34)and further treated with the GeneClean kit (Bio 101, Inc., LaJolla, Calif.) as suggested by Tiesman and Rizzino (36).Plasmid-specific sequencing primers were annealed toNaOH-denatured double-stranded DNA under conditionsdescribed previously (36). Sequencing was performed by thedideoxynucleotide chain termination method (35) with theSequenase sequencing kit from U.S. Biochemical Corp.(Cleveland, Ohio) and [32P]dATP from Amersham Corp.(Arlington Heights, Ill.). Oligonucleotides for sequencingand cloning were synthesized on an Applied Biosystems(Foster City, Calif.) DNA synthesizer (model 380B) andused without further purification.

Transformation. Cells were transformed by the single-stepmethod of Chung et al. (11), except that dimethyl sulfoxidewas omitted from the transformation and storage solution.

Partition tests. Partition tests were performed by a pickupassay with a high-copy-number P1 miniplasmid vector (1).Plasmid pSP102 (32) is a P1 miniplasmid that contains nopartition sequences, has a deletion of the copy numbercontrol locus, incA, and replicates at a moderate copynumber. The introduction of a synthetic multiple cloningregion in the BamHI site of pSP102 generated plasmidpALA1626, which contains adjacent unique AvaI, BamHl,BglII, BssHII, ClaI, EcoRV, SphI, XbaI, and XhoI restric-tion sites, allowing the more convenient manipulation ofsequences. P7 partition sites to be tested for activity weresubcloned into this multiple cloning region from the set ofpBR322 derivatives described above and transformed intorecombination-proficient strain YMC. Bacteriophage X-P1:5RA&1005 was grown on these YMC transformants, and thephage lysate containing in vivo composite pALA1626::X-P1:5RA1005 phage was used to lysogenize CC1572 deriv-atives, with selection for pALA1626-encoded chloramphen-icol resistance at 32°C. Because X-P1:5RA1005 lacks se-quences essential for chromosomal integration (5), therecombinant prophage replicates as a plasmid directed bythe P1 replication machinery. In addition, because the re-combinant phage possess the copy number control locusincA, replication proceeds at a low copy number (7). Themaintenance stability of the composite prophage plasmidfollowing approximately 25 generations of unselected growthwhen partition proteins were supplied in trans on a compat-ible plasmid was used as an indicator of whether an activepartition site was present on the cloned insert as follows.Eight chloramphenicol-resistant lysogens were streaked forsingle colonies on chloramphenicol-containing L agar at32°C for 24 h. Eight of the resulting colonies from this initialselection were restreaked on nonselective L agar at 32°C for24 h. This step was repeated, and eight individual coloniesfrom each of the final eight streaks were tested for retentionof the chloramphenicol resistance marker by being pickedand stabbed to L agar containing chloramphenicol at 32°C.

Partition-mediated incompatibility tests. Incompatibilitytests with X-P7:79Cm were performed at 32°C essentially asdescribed previously for X-P1:5RCm (27). In brief, strainN100 or RW1840 was lysogenized with X-P7:79Cm, withselection for chloramphenicol resistance. Lysogens weretransformed to ampicillin resistance with the pBR322-basedtest plasmid and restreaked once on ampicillin-containingmedium. Following growth under nonselective conditions

for approximately 25 generations as described above, trans-formants were assayed for X-P7:79Cm retention.

Protein purification and DNA footprinting assays. Purifica-tion of IHF and P7 ParB proteins and conditions for DNaseI protection assays (21) were described by Davis et al. (14,16). DNA fragments for footprinting analyses were endlabelled and amplified from pBR322-based plasmids by thepolymerase chain reaction with vector-derived primersflanking the cloned P7 sequences essentially by the methodof Hooft van Huijsduijnen (24). End-labelled fragments werepurified by polyacrylamide gel electrophoresis.

RESULTS

Assay system for P1 and P7 partition sites. Successfulpartition site assays use conditional vector systems that canbe maintained at either a high or a low copy number (1). Thehigh-copy-number state is convenient for the manipulationof DNA and is necessary for long-term maintenance ofpartition-defective derivatives. However, plasmid mainte-nance stability due to par must be tested in the low-copy-number mode, in which the partition system has an easilymeasurable phenotype. Previously, the dual replicon vectorpALA136 was used for mapping the P1 partition site (27, 28).The high-copy-number ColEl-derived replication origin ofpALA136 is suppressed in apolA host, in which replicationproceeds at a low copy number solely via the P1 replicationregion. This system allows an estimate of plasmid mainte-nance stability due to cloned partition sites when Par pro-teins are provided in trans on a compatible plasmid (28).While this protocol works with modest efficiency with Plparsites, P7 par site activity has proven weak and inconsistent(22). The sequence context and/or small size of this vectormay impose on the inserted par sites restrictions that arepartially tolerated by the P1 site but less well by its P7 analog(7). An alternative assay suitable for the comparative anal-ysis of both par sites was therefore developed (1; seeMaterials and Methods). Potential partition sites are clonedinto a moderate-copy-number mutant (incA) P1 miniplasmid.The sites are tested after recombinational pickup onto aphage X vector that supplies the missing P1 incA locus. Theresulting composite replicates at a low copy number (7), andthe efficacy of the cloned par sites can be estimated bymeasuring the maintenance stability of the hybrid when theappropriate Par proteins are supplied in trans on a compat-ible plasmid. A fragment encompassing the entire P1 parSregion confers stability in response to the P1 Par proteins,and the level of stabilization achieved (>80% retention in 25generations) is substantially higher than that in the previ-ously used assay (approximately 40% retention; 28). Thelimits of the functional P1 parS site in the new assay areapproximately equivalent to those of the previously definedP1 IncB+ region (Fig. 1). However, shorter sequences,including the core 22-bp site, are nonfunctional (22).

Identification and specificity of the P7 plasmid incompati-bility and partition locus. The P1 parS region determinesboth partition site and incompatibility properties. It containsa number of distinctive protein binding motifs, including aset of heptamer repeat sequences recognized in vitro byParB and an IHF binding site (14, 15; Fig. 1). Specificmutations in the IHF binding site or in the ParB heptamerrepeat boxes affect partition ability and/or partition-medi-ated incompatibility (15, 20, 27). Ludtke et al. (26) alignedthe region immediately downstream of the P7parB sequencewith the P1 parS sequence, thereby localizing a highlyrelated group of identically spaced protein binding motifs

VOL. i75, 1993

on February 9, 2018 by guest

http://jb.asm.org/

Dow

nloaded from

3446 HAYES ET AL.

TABLE 1. Incompatibility specificities of the P1 andP7 parS regionsa

% Retention of the followingIncoming IncB+ sequence on resident miniplasmid in 25plasmid the incoming generations:plasmid

X-P1:5RCm X-P7:79Cm

pBR322 None 99 1 60 + 9pALA1609 P7 176-bp IncB+ 90 ± 7 <2pALA407 P1 109-bp IncB+ <2 77 ± 14

a All values are the averages of at least three individual experiments + 1standard deviation.

(Fig. 1). For investigation of whether this P7 region alsofunctions as both an incompatibility locus and a partitionlocus, a 176-bp Fnu4HI-HincII fragment completely span-ning this region (Fig. 1) was tested for these properties.When a pBR322 derivative containing the fragment(pALA1609) was introduced into cells containing P7 unit-copy miniplasmid X-P7:79Cm, the latter plasmid was dis-placed efficiently (Table 1). Furthermore, in partition assays,the same fragment promoted stability of the X pickup vector(pALA1664). This activity was dependent on both the pres-ence of the fragment in the stabilized plasmid and the P7 Parproteins that were supplied in trans (Table 2). These resultsindicate that the 176-bp Fnu4HI-HincII sequence of P7contains sufficient cis-acting information for both the activepartition process and partition-mediated incompatibility.The incompatibility specificities of the P1 and P7 partition

sites were examined by comparing the effects of pBR322plasmids containing the P7 176-bp sequence or the P1 109-bpIncB+ sequence on the stability of P1 and P7 miniplasmidsX-P1:5RCm and X-P7:79Cm (Table 1). Partition-mediatedincompatibility was expressed by these derivatives onlytowards miniplasmids carrying a site from the same species.Furthermore, partition tests with these fragments were alsospecies specific: the partition site activity of the P7 fragmentwas supported when P7 proteins were provided in trans andnot when P1 proteins were supplied, whereas the derivativecontaining the P1 109-bp IncB+ sequence was dependent onP1 proteins (Table 2). The P1 and P7 partition sites aretherefore plasmid specific, both in their abilities to supportthe active partition process and in the expression of parti-tion-mediated incompatibility towards P1 and P7 miniplas-mids.

Gross limits of the P7 parS site. By use of restriction sites

TABLE 2. Partition site specificities of the P1 andP7parS regionsa

% Retention in 25 generations withthe following Par proteins

Plasmid parS supplied in trans:sequence

Noe P1 ParA and P7 ParA andNone ParB ParB

pALA1626 None <2 <2 <2pALA1664 P7 176-bp IncB+ <2 <2 72 ± 1pALA1693 P1 109-bp IncB+ <2 83 ± 1 <2

a All values are the averages of at least three individual experiments ± 1standard deviation. The indicated plasmid was incorporated into X pickupvector X-P1:5RA1005 and tested, as described in Materials and Methods, inthe presence of pBR322 (which supplies no partition proteins) or pALA271 orpALA1023 (which supplies the P1 or P7 ParA and ParB proteins, respec-tively).

near the P7 partition region, a set of deletion variants of the176-bp Fnu4HI-HincII fragment as constructed in pBR322 asoutlined in Materials and Methods. The nucleotide se-quences of the inserts were confirmed, and their associatedincompatibility phenotypes were determined (Fig. 2). Plas-mids with deletions of the 82-bp Fnu4HI-XmnI fragment(pALA1601 and pALA1602) or of the 34-bp Fnu4HI-HinPIfragment (pALA1611 and pALA1614) no longer displaced P7miniplasmid X-P7:79Cm. The leftmost boundary of the P7IncB+ sequence therefore maps between the Fnu4HI andHinPI sites. Similarly, the rightmost boundary is locatedbetween the XmnI and DraI sites, because deletion of the68-bp XmnI-DraI sequence (pALA1616) but not of the 26-bpDraI-HincII sequence (pALA1613) abolished incompatibil-ity. Furthermore, only plasmids containing IncB+ regionswere segregated stably in partition tests when P7 ParA andParB were supplied in trans (Fig. 2). Partition site assaysshowed all IncB- fragments to be inactive (Fig. 2), althoughthe possibility that these negative results were an assay-dependent phenomenon cannot be eliminated (see Discus-sion).

Precise mapping of the P7 parS boundaries. As outlined inMaterials and Methods, the rightmost limit of P7 parS wasdetermined by constructing a set of pBR322-based plasmidscontaining P7 bp 2301 to 2391 plus additional various por-tions of the interval from bp 2392 to bp 2421 (Fig. 3A) andassaying for partition-mediated plasmid incompatibility (Ta-ble 3). In addition, these P7 sequences were subcloned in thepartition assay pickup vector, pALA1626, and tested forpartition ability (Table 3). The structures of all plasmidstested were confirmed by sequence analysis. Plasmids con-taining bp 2301 to 2400 expressed complete incompatibilitytowards P7 miniplasmid X-P7:79Cm (<2% retention) andwere also partition proficient (>70% retention) in the pickupassay system when P7 partition proteins were supplied onthe compatible plasmid, pALA480. However, deletion of bp2400 resulted in both a loss of incompatibility (pALA1624)and a marginal reduction in partition proficiency(pALA1632). Deletion of an additional nucleotide furtherdecreased partition ability (pALA1684), while a plasmidcontaining bp 2301 to 2391 (pALA1633) was fully Par- (Fig.3A and Table 3). Thus, the rightmost limits of the incompat-ibility and partition determinants roughly correspond and liein the vicinity of nucleotide 2400.

Plasmid pALA1648 is a pBR322-based plasmid containingP7 bp 2301 to 2400 but with a C-to-G substitution mutation atbp 2332, which results in the formation of a unique BssHIIsite at this position (Fig. 3B). In addition, the leftmostP7-pBR322 junction of this plasmid is flanked by a uniqueEcoRV site. The leftmost boundary of the parS region wasdefined by substituting bp 2301 to 2333 with syntheticdouble-stranded oligonucleotides composed of differentlengths of P7 material and with EcoRV- and BssHII-compat-ible ends (Fig. 3B and Table 3). Deletion of bp 2301 to 2325had no dramatic effect on either incompatibility or partitionability. However, deletion of bp 2326 (pALA1651) resultedin a sharp decrease in the degree of incompatibility exertedby the P7 sequence, while elimination of an additional 2nucleotides (pALA1657) significantly lowered the retentionfrequency in partition assays. Partition ability was elimi-nated by the loss of nucleotide 2329 (pALA1656). Thus, theleftmost boundaries of the incompatibility determinant andpartition site are roughly coincident around nucleotide 2325.The minimal parS site is therefore about 75 bp in size (bp2326 to 2400). However, because clones possessing an extra3 bp at this leftmost limit yielded somewhat more reproduc-

J. BACTERIOL.

on February 9, 2018 by guest

http://jb.asm.org/

Dow

nloaded from

P7 PLASMID PARTITION SITE 3447

T- W cmo Co) GoC') mV C')cm cm cm

F Hp Xa I I

pALA1609/ 1664

pALAI61I1 / 1665

pALA1602/ 1663

pALA1613 / 1631

pALA1614/ 1666

pALA1601 /1662

0a coto CY

D H

I I

pALA1616/ 1667 1 1

Incompatibility test

% Retention X-P7:79Cm

WO RW1 840(IHF )

4±2 12±8

77± 2 63± 6

91 +10 59± 9

4±2 14±7

87±5 57± 8

79 ±8 94+ 1

61 ± 8 81± 5

Partition site test

% Retention

pGB2 pALA480pGB (ParA+ParB +)<2 77±15

<2 6±1

<2 <2

<2 94± 2

<2 10±6

<2 <2

<2 <2

50 bpFIG. 2. Determination of the gross limits of P7parS. The top line shows a linear restriction map with coordinates previously assigned to

this region (26). The bars under the map indicate the extent of identicalparS sequences cloned in both pBR322, denoted by the first plasmidnumber on each line, and the pickup assay vector, pALA1626, denoted by the second plasmid number on each line, as outlined in Materialsand Methods. The first two columns show the retention frequencies after approximately 25 generations of nonselective growth of a P7miniplasmid in partition-mediated incompatibility tests in strains N100 and RW1840 (IHF-) with pBR322 plasmids harboring the indicatedparS regions in trans. The retention frequencies after 25 generations of nonselective growth in partition site pickup assays of miniplasmidscontaining the relevant parS regions when provided with pGB2 and pALA480 (P7 ParA' ParB+) in trans are shown in the third and fourthcolumns, respectively. Details of incompatibility and partition site assays are provided in Materials and Methods. All data are the averagesof at least three experiments ± 1 standard deviation. Abbreviations and symbols correspond to those described in the legend to Fig. 1.

ible results in both partition site and incompatibility assays,this 78-bp sequence (bp 2323 to 2400) was used for subse-quent mutational analyses.

Does the P7 parS region exhibit dual incompatibility speci-ficities? Although the intact P1 parS region is required forwild-type (IncB+) incompatibility, the 22-bpparS core (Fig.1) is also capable of exerting an incompatibility effect underspecific circumstances (15, 27, 28). Plasmids containing thissite cannot displace a P1 miniplasmid from a wild-type strainbut can displace one from strains lacking an IHF protein(19). This limited incompatibility property is termed IncBd(15). Several different fragments lacking portions of theparSregion but retaining the parS core share this property (27).For assessment of whether a comparable situation exists forthe P7 parS region, the group of pBR322 clones describedpreviously was tested for the ability to eliminate a P7miniplasmid in incompatibility tests with strain RW1840(IHF-). Inspection of Fig. 2 shows that, in the case of P7,the full-length parS fragment was essential for exertingincompatibility in the IHF- strain. Shorter fragments thatretain the P7 region that is homologous to the P1 parS coreregion showed, at best, only a marginal ability to expelX-P7:79Cm in the IHF-defective background, and the small-est such fragment had no effect at all. Thus, in contrast to theP1 case, there is no evidence for the existence of a P7 parScore sequence with IncBd activity.

Site-directed mutational analysis of the P7 minimal parSregion. Extensive fine-structure analysis of the P1 parSregion has defined motifs critical to parS-mediated incom-patibility specificity and partition ability (15, 27). The roles

of selected related sequences in the P7 minimal parS regionwere probed by cloning synthetic oligonucleotides withmultiple mutations in two of the three ParB heptamer repeatboxes, in the IHF binding site, and in the region adjacent tothe leftmost boundary of the minimal parS site (Fig. 4).Mutations in the P7 ParB and IHF binding motifs werechosen to match alterations known to have a deleteriouseffect on expression of the P1 IncB+ phenotype (15). Suit-able constructs containing these oligonucleotides weretested for incompatibility and partition ability (Table 4).Plasmids harboring alterations in the rightmost ParB box(pALA1756), the IHF binding site (pALA1755), or the regionadjacent to the leftmost boundary (pALA1753) failed todisplace a P7 miniplasmid in incompatibility tests (Table 4).However, a plasmid containing multiple mutations in theleftmost ParB box (pALA1754) retained some capacity toexpel this miniplasmid. Parallel results were obtained inpartition site assays. Mutations in the IHF binding site(pALA1759), the region close to the leftmost limit of theminimal site (pALA1757), and the rightmost ParB heptamerrepeat box (pALA1760) fully abolished partition activity,whereas a plasmid containing substitutions in the leftmostParB box (pALA1758) was retained at approximately 50%the level of a plasmid containing the wild-type sequence(pALA1659; Table 4).

Protein binding to wild-type and mutated P7 minimalpartition site sequences. The in vitro patterns of binding ofpurified IHF and P7 ParB proteins to wild-type and mutatedP7 minimal partition sites were examined by the DNase Iprotection (footprinting) technique (21). Sequences at the

VOL. 175, 1993

L-

on February 9, 2018 by guest

http://jb.asm.org/

Dow

nloaded from

3448 HAYES ET AL.

LOI0Ob Fnu 4HIGGAGGCGGCT

R3 Hin Pi CY

TTAGCCCCCTCCCCTGTCTAMATGTCCCGCGCCT |ATCAT GTATAMTATATGATATATATAGACATTC

NO Xmn I so o of N* Nir~ ~ ~ ~ m Y Dra I C5m

ATGAAAAMATTCAGICTGAAAT TCCCACG GTTTCAC GCCTGTTTTACTTGCCCCCCTCCCCCGCACMMAATTTAAA

4ftfttt t

G

N Fnu4HI GHinPI cm

B. GGAGGCGGCTTTAGCCCCCTCCCCTGTCTAAMTGTCCCGCGCCT ATTTCAT| GTATAAATATATGATATATATAGACATTC

0

93 8c Xmn I

ATGAAMA A1TCAG G CTGAAATITCCFIG. 3. Mapping of the precise right- and leftmost limits of P7parS as determined from the data presented in Table 3. (A) Nucleotide

sequence of the Fnu4HI-DraIparS region (26) and determination of the rightmostparS boundary. Nucleotide numbering corresponds to thatpreviously assigned to this region (26). ParB heptamer binding motifs are boxed. The positions of the IHF recognition sequence and relevantrestriction sites are indicated above the sequence. An imperfect palindrome sequence is identified by the horizontal inverted arrows. Verticalarrows indicate the final rightmost nucleotide ofparS sequences commencing at the Fnu4HI site as cloned in both pBR322 and pALA1626as described in Materials and Methods. Long closed and open vertical arrows denote plasmid clones that are Inc' Par' and Inc- Par-,respectively. Short closed vertical arrows indicate plasmid clones that display partial defects in incompatibility and/or partition abilities. (B)Determination of the leftmostparS boundary. Symbols correspond to those described in panel A, except that vertical arrows indicate the finalleftmost nucleotide ofparS-containing subclones. The C-to-G base change at position 2332 generated a novel BssHII site at this position (seeMaterials and Methods) while simultaneously abolishing incompatibility and partition abilities (data not shown).

rightmost boundary of the wild-type site were protectedfrom DNase I digestion by ParB (Fig. 5, lower region of gel).This region includes two inversely oriented copies of theproposed ParB heptamer recognition element, with the se-quences ATIT'TCAG and CTGAAAT (Fig. 4). ParB alsoweakly protected a region that includes the leftmost ParBheptamer ATITCAT (Fig. 4) and extends somewhat beyondthe leftmost boundary of the partition site (Fig. 5, upperregion of gel). IHF protected a region that includes the 13-bpputative IHF binding motif and additional bases, mostly tothe right of the motif. ParB and IHF together produced apattern similar to the sum of the separate effects, but theprotection was more complete. Protection from the leftmostboundary through the leftmost heptamer was now clearlyseen.Although the rightmost ParB heptamer motif was rather

poorly cut by DNase I, inspection of Fig. 5 (and of overex-posed versions not shown) shows that ParB protected it andthat this protection was complete in the presence of IHF.The introduction of multiple mutations into this box(pALA1756) altered and improved the basic DNase I cuttingpattern of the region, making the protection experimentseasier to interpret. ParB binding to this mutated box waseliminated, but binding to the adjacent box persisted. Thus,the heptamer motif is an individual ParB binding site, as hasbeen speculated for the P1 system (14). IHF binding to thecentral region was unaffected by the mutation in pALA1756,but careful inspection of the leftmost boundary of the site(closed boxes; Fig. 5) suggests that ParB binding here wasless complete than in the wild-type case.The set of mutations introduced into the putative IHF box

(pALA1755; Fig. 4) completely eliminated the recognition ofthis site by the IHF protein (Fig. 5), a result confirming theidentity and location of the IHF site (Fig. 4). The pattern ofParB binding to this mutated sequence was unaltered incomparison with the pattern obtained when ParB alonebound to the wild-type site (Fig. 5).Although both sets of mutations introduced into the left-

most end of the P7 minimal partition site (Fig. 4) had effectson partition and incompatibility in vivo (Table 4), neither setof substitutions had an obvious impact on ParB or IHFbinding in vitro (data not shown). However, as the in vitroprotein-DNA interactions in this region of the wild-typesequence are weak (Fig. 5), alterations in the binding affin-ities due to the mutated bases may be difficult to detect.

DISCUSSION

The partition systems of the P7 and P1 prophage plasmidsare substantially homologous and presumably originate froma common ancestral source (26). However, they have di-verged sufficiently so that the cis-acting partition sites sup-port plasmid segregation only when provided with theircognate proteins (Table 2). The partition systems also dem-onstrate species specificities in partition-mediated incompat-ibility (26; Table 1) and Par operon autoregulation (33).These species specificities presumably reflect steps in apathway for partition at which macromolecular componentsrecognize each other and at which elements from differentsources fail to function correctly together. By mapping thespecificity determinants to specific proteins, protein do-mains, and DNA sites, it should be possible to define an

A.

J. BACTIERIOL.

on February 9, 2018 by guest

http://jb.asm.org/

Dow

nloaded from

P7 PLASMID PARTITION SITE 3449

TABLE 3. Incompatibility and partition properties of P7 parS constructsa

Incompatibility test Partition site test

Plasmid parS region % Retention with the followingset (bp) Plasmid % Retention of Plasmid proteins supplied:

X-P7:79CmNone P7 ParA and ParB

A None pBR322 60 9 pALA1626 <2 <22301-2451 pALA1613 4 2 pALA1631 <2 94 22301-2421 pALA1623 <2 pALA1636 <2 76 12301-2415 pALA1622 <2 pALA1635 <2 91 22301-2411 pALA1621 <2 pALA1634 <2 86 52301-2405 pALA1639 <2 pALA1642 <2 85 62301-2404 pALA1625 <2 pALA1637 <2 77 72301-2403 pALA1630 <2 pALA1640 <2 88 42301-2402 pALA1645 <2 pALA1747 <2 85 52301-2400 pALA1638 <2 pALA1641 <2 72 ± 12301-2399 pALA1624 51 ± 13 pALA1632 <2 54±12301-2398 pALA1672 64 ± 6 pALA1684 <2 30 ± 32301-2391 pALA1617 68 ± 4 pALA1633 <2 <22301-2382 pALA1616 61 ± 8 pALA1667 <2 <2

B 2315-2400 pALA1653 3 ± 2 pALA1660 <2 64 ± 82323-2400 pALA1652 <2 pALA1659 <2 71 ± 22324-2400 pALA1675 8 ± 4 pALA1683 <2 59 ± 42326-2400 pALA1673 5 ± 3 pALA1682 <2 53 ± 12327-2400 pALA1651 44 ± 17 pALA1658 <2 59 ± 62329-2400 pALA1650 58 ± 8 pALA1657 <2 27 ± 32330-2400 pALA1649 55 ± 16 pALA1656 <2 4 ± 3

a Sets A and B correspond to the sets of plasmids shown in Fig. 3A and B, respectively. All values are the averages of at least three individual experiments± 1 standard deviation. Incompatibility tests were done with incoming pBR322-based constructs containing the relevant P7parS sequences. Retention of residentP7 miniplasmid X-P7:79Cm was determined following 25 generations of nonselective growth. Partition site tests were done with pALA1626 derivatives containingan identicalparS sequence. Retention of pickup vector X-P1:5RA1005 containing the indicated plasmid was determined following 25 generations of nonselectivegrowth. Assays were carried out in the presence of pGB2 (which supplies no partition proteins) or pALA480 (which supplies the P7 ParA and ParB proteins).

important subset of interactions involved in partition. Thisprocess requires the development of a convenient assaycapable of efficient detection of both P7 and P1 partition siteactivities. The pickup assay protocol described here fulfillsthis requirement. With this technique, a detailed character-ization of the P7 cis-acting partition site,parS, whose extentand exact location were previously unknown, was per-formed.The P7 minimal partition site was defined precisely and

shown to be 75 bp in size, with a partial requirement for anadditional 3 nucleotides at the leftmost boundary. The min-imal partition site includes essential ParB and IHF bindingmotifs and novel sequences at the leftmost boundary. Threeof the P1 heptamer boxes speculated to direct ParB binding(14) have homologs in the P7 minimal site. The heptamermotifs have been shown here to be individual ParB bindingsites, as destruction of the rightmost heptamer motif bymultiple mutations (Fig. 4) blocked binding to the disruptedbox, whereas binding to its adjacent wild-type partner was

maintained (Fig. 5). The integrity of the boxes, both here andtoward the leftmost end of the site, is essential for functionof the minimal partition site, as the introduction of multiplemutations into these boxes eliminated or reduced bothpartition and incompatibility activities in vivo (Table 4).The central IHF binding motif in P7 is considerably

divergent from its P1 counterpart. However, the P7 se-quence binds IHF, which is essential for the activity of theminimal partition site in vivo (Table 4). IHF promotes moreefficient ParB binding, perhaps by cooperative interactionsof the two proteins. It seems probable from the knownproperties of IHF (38) that IHF binding induces a severebend in the DNA (20) that promotes an interaction betweenthe two ends of the site. This interaction could account forthe apparent weakening in protection of the leftmost end ofthe site when the rightmost ParB box is mutated (Fig. 5). Theextreme ends of the P7 sequence clearly contain essentialinformation of unknown function. The deletion analysisdescribed here demonstrated that 16 nucleotides at the left

pALA1753/1757 pALA1754/1758 pALA1755/1759 pALAI 756/1760

GGG TAC AG CC GG ACT GTA

++*+++ ++ *f ++ +++ +++TMAATGTCCCGCGCCT IAMCAT GTATAAMTATATGATATATATAGACATTCATGAAAA IATTTCAGIGCTGAAATITCC

| 1HF l

FIG. 4. Mutational analysis of the P7 minimalparS region. Partition site and partition-mediated incompatibility assay results are presentedin Table 4. Heptamer repeats recognized by ParB are boxed. The position of the IHF binding motif is also indicated. Substitution mutationsgenerated in the sequence (see Material and Methods for details of plasmid constructions) are shown above the parS sequence. The first andsecond figures denote the plasmid numbers assigned to these sequences when cloned in pBR322 and pALA1626, respectively.

VOL. 175, 1993

on February 9, 2018 by guest

http://jb.asm.org/

Dow

nloaded from

3450 HAYES ET AL.

TABLE 4. Incompatibility and partition properties of P7 minimal parS sites containing multiple mutationsa

Incompatibility test Partition site test

parS region % Retention with the followingPlasmid Retention of Plasmid proteins supplied:

X\-P7:79CmNone P7 ParA and ParB

None pBR322 60 ± 9 pALA1626 <2 <2Wild type pALA1770 <2 pALA1659 <2 74 ± 3Left boundary mutant pALA1753 66 ± 7 pALA1757 <2 <2Leftmost ParB box mutant pALA1754 30 ± 22 pALA1758 <2 33 ± 10IHF motif mutant pALA1755 58 ± 12 pALA1759 <2 <2Rightmost ParB box mutant pALA1756 62 ± 11 pALA1760 <2 <2

a All values represent plasmid retention during 25 generations of unselected growth and are expressed as the averages of at least three individual experiments+ 1 standard deviation. Incompatibility and partition site tests were carried out as described in Table 3, footnote a. Plasmid pBR322 or pALA1626 derivativescontained P7 bp 2323 to 2400 with or without multiple mutations in the indicated regions.

WTboundary and 3 nucleotides at the right boundary are neces-sary for incompatibility and partition site activities but donot form part of any obvious ParB heptamer or IHF bindingbox. The role of these sequences is under investigation. Thefact that they differ radically from their P1 counterpartssuggests that these sequences may play an important role inspecies specificity.Like its homolog in the P1 plasmid (3), the P7 partition site

also functions as a partition-mediated plasmid incompatibil-ity determinant. This function is thought to reflect a criticalpairing of plasmids via their partition sites (31). An incomingreplicon carrying the relevant partition site competes with aresident plasmid for partition by forming inappropriate pairsprior to segregation (15, 31). As the entire P7 75-bp minimalpartition site is required to exert incompatibility, it appearsthat pairing with the wild-type region requires the integrity ofthe complete site. However, a number of partition sitedeletion constructs that retained significant partition abilitybut exhibited little or no incompatibility with a P7 miniplas-mid were identified (Table 3). This result may reflect sitedamage that weakens the pairing reaction, such that daugh-ter plasmids are still able to pair and segregate in the pickupassay, but competition for pairing with wild-type residentplasmids in the incompatibility test is poor.The P7 and P1 partition sites share 55% identity over the

P7 75-bp minimal partition site. The relative spacings of theParB and IHF protein binding motifs are conserved perfectly(Fig. 1), and there is no clearly defined consensus differencebetween the P7 and P1 heptamer boxes. The overall config-urations of these sites, when present in an active plasmidpartition complex, are probably similar. The structure mayconsist of partition site sequences wrapped about a centralParB-IHF protein core with its ends in proximity to eachother (15). However, the relative importance of differentelements in maintaining an active structure seems to havediverged considerably between the two species. A P1 hep-tamer box at the right end of the P1 locus is essential forwild-type P1 incompatibility (15), but its counterpart in P7 isredundant (Fig. 3A). In contrast, the leftmost P7 heptamerbox is required for full activity of the minimal partition site(Table 4), whereas the properties of similar, although notstrictly comparable, P1 constructs suggest that the equivalentbox is dispensable in P1 (15). Another major differenceconcerns the function of the 22-bp region referred to, in P1, asthe parS core (Fig. 1). The isolated P1 sequence showsmodest partition site activity (27) and exerts modified (IncBd)incompatibility (15, 28). Although the equivalent region is avital component of the P7 site, there is no evidence that the

1 234

pALA17561 234

U

ElUOl

0

0

pALA17551 234

UEl1-I

0El

FIG. 5. DNase I footprinting patterns of wild-type and mutatedP7 minimal partition site sequences as tested with IHF and P7 ParBproteins. P7 DNA consisted of the 78-bp partition site (P7 bp 2323 to2400) inserted into pBR322. Arrows mark the boundaries of the P7and vector bases. Open and hatched boxes indicate the positions ofthe putative ParB binding heptamer and IHF binding motifs, respec-tively. Closed boxes mark the 3 bases mutated in pALA1753 (Fig.4). Asterisks mark the boxes containing multiple mutations (Fig. 4).WT, wild-type sequence; pALA1755, multiple mutations introducedinto the IHF box (Fig. 4); pALA1756, multiple mutations introducedinto the rightmost ParB box (Fig. 4). Lanes: 1, no protein added; 2,6.0 pLg of P7 ParB per ml; 3, 0.2 pLg of IHF protein per ml; 4, 2.0 ,ugof ParB and 0.2 ,ug of IHF protein per ml.

J. BACTERIOL.

on February 9, 2018 by guest

http://jb.asm.org/

Dow

nloaded from

P7 PLASMID PARTITION SITE 3451

relevant P7 subregion has any partition site activity in isola-tion (Fig. 2), and no IncBd-like activity was seen for this P7subregion when either P7 (Fig. 2) or P1 (22) miniplasmidswere used as targets. One or more of just 4 base differencesbetween the P1 and P7 22-bp core regions are presumablyresponsible for this important difference. The P1 core regionmay contain the critical information for partition, and theadjacent regions may form a structure in which this informa-tion is optimally presented (15). It seems probable that asimilar situation occurs in P7 but that, in this case, the criticalcore sequences are unable to function at all in the absence ofthe supporting sequences. The construction and in vivo and invitro characterization of chimeric partition sites should assistfurther in defining the crucial determinants of P7 and P1partition site specificities.

ACKNOWLEDGMENTS

This research was sponsored by NCI under contract NO1-CO-74101 with ABL.

REFERENCES1. Abeles, A. L., and S. J. Austin. 1989. The molecular genetics of

plasmid partition: special vector systems for the analysis ofplasmid partition, p. 111-125. In J. K. Setlow (ed.), Geneticengineering, vol. 11. Plenum Publishing Corp., New York.

2. Abeles, A. L., S. A. Friedman, and S. J. Austin. 1985. Partitionof unit-copy miniplasmids to daughter cells. III. The DNAsequence and functional organization of the P1 partition region.J. Mol. Biol. 185:261-272.

3. Austin, S., and A. Abeles. 1983. Partition of unit-copy miniplas-mids to daughter cells. I. P1 and F miniplasmids containdiscrete, interchangeable sequences sufficient to promote equi-partition. J. Mol. Biol. 169:353-372.

4. Austin, S., and A. Abeles. 1983. Partition of unit-copy miniplas-mids to daughter cells. II. The partition region of miniplasmidP1 encodes an essential protein and a centromere-like site atwhich it acts. J. Mol. Biol. 169:373-387.

5. Austin, S., F. Hart, A. Abeles, and N. Sternberg. 1982. Geneticand physical map of a P1 miniplasmid. J. Bacteriol. 152:63-71.

6. Austin, S., M. Ziese, and N. Sternberg. 1981. A novel role forsite-specific recombination in maintenance of bacterial repli-cons. Cell 25:729-736.

7. Austin, S. J. Unpublished data.8. Austin, S. J., and B. G. Eichorn. 1992. Random diffusion can

account for topA-dependent suppression of partition defects inlow-copy-number plasmids. J. Bacteriol. 174:5190-5195.

9. Bolivar, F., R. L. Rodriguez, P. J. Green, M. D. Betlach, H. W.Boyer, J. H. Crosa, and S. Falkow. 1977. Construction andcharacterization of new cloning vehicles. II. A multipurposecloning system. Gene 2:95-113.

10. Boyer, H. W., and D. Roulland-Dussoix. 1969. A complementa-tion analysis of the restriction and modification of DNA inEscherichia coli. J. Mol. Biol. 41:459-472.

11. Chung, C. T., S. L. Niemela, and R. H. Miller. 1989. One-steppreparation of competent Escherichia coli: transformation andstorage of bacterial cells in the same solution. Proc. Natl. Acad.Sci. USA 86:2172-2175.

12. Churchward, G., D. Belin, and Y. Nagamine. 1984. A pSC101-derived plasmid which shows no sequence homology to othercommonly used cloning vectors. Gene 31:165-171.

13. Das, A., D. Court, and S. Adhya. 1976. Isolation and character-ization of conditional lethal mutants of Escherichia coli defec-tive in transcription termination factor rho. Proc. Natl. Acad.

Sci. USA 73:1959-1963.14. Davis, M. A., and S. J. Austin. 1988. Recognition of the P1

plasmid centromere analog involves binding of the ParB proteinand is modified by a specific host factor. EMBO J. 7:1881-1888.

15. Davis, M. A., K. A. Martin, and S. J. Austin. 1990. Specificityswitching of the P1 plasmid centromere-like site. EMBO J.9:991-998.

16. Davis, M. A., K. A. Martin, and S. J. Austin. 1992. Biochemicalactivities of the ParA partition protein of the P1 plasmid. Mol.Microbiol. 6:1141-1147.

17. Dennert, A., and U. Henning. 1968. Tyrosine-incorporatingamber suppressor in Escherichia coli. J. Mol. Biol. 33:322-329.

18. Friedman, S. A., and S. J. Austin. 1988. The P1 plasmid-partition system synthesizes two essential proteins from anautoregulated operon. Plasmid 19:103-112.

19. Funnell, B. E. 1988. Participation of Escherichia coli integrationhost factor in the P1 plasmid partition system. Proc. Natl. Acad.Sci. USA 85:6657-6661.

20. Funnell, B. E. 1991. The P1 plasmid partition complex at parS.The influence of Eschenichia coli integration host factor and ofsubstrate topology. J. Biol. Chem. 266:14328-14337.

21. Galas, D. J., and A. Schmitz. 1978. DNAase footprinting: asimple method for the detection of protein-DNA interactions.Nucleic Acids Res. 5:3157-3170.

22. Hayes, F., and S. J. Austin. Unpublished data.23. Hiraga, S. 1992. Chromosome and plasmid partition in Esche-

nchia coli. Annu. Rev. Biochem. 61:283-306.24. Hooft van Huijsduinen, R. A. M. 1992. PCR-generated probes

for the study of DNA-protein interaction. BioTechniques 12:830-832.

25. Ludtke, D. L., and S. J. Austin. 1987. The plasmid-maintenancefunctions of the P7 prophage. Plasmid 18:93-98.

26. Ludtke, D. L., B. G. Eichorn, and S. J. Austin. 1989. Plasmid-partition functions of the P7 prophage. J. Mol. Biol. 209:393-406.

27. Martin, K. A., M. A. Davis, and S. Austin. 1991. Fine-structureanalysis of the P1 plasmid partition site. J. Bacteriol. 173:3630-3634.

28. Martin, K. A., S. A. Friedman, and S. J. Austin. 1987. Partitionsite of the P1 plasmid. Proc. Natl. Acad. Sci. USA 84:8544-8547.

29. Motallebi-Veshareh, M., D. Rouch, and C. M. Thomas. 1990. Afamily of ATPases involved in active partitioning of diversebacterial plasmids. Mol. Microbiol. 4:1455-1463.

30. Nordstr6m, K., and S. J. Austin. 1989. Mechanisms that con-tribute to the stable segregation of plasmids. Annu. Rev. Genet.23:37-69.

31. Novick, R. P., and F. C. Hoppensteadt. 1978. On plasmidincompatibility. Plasmid 1:421-434.

32. Pal, S. K., R. J. Mason, and D. K. Chattoraj. 1986. P1 plasmidreplication. Role of initiator titration in copy number control. J.Mol. Biol. 192:275-285.

33. Radnedge, L., F. Hayes, and S. J. Austin. Unpublished data.34. Sambrook, J., E. F. Fritsch, and T. Maniatis. 1989. Molecular

cloning: a laboratory manual, 2nd ed. Cold Spring HarborLaboratory, Cold Spring Harbor, N.Y.

35. Sanger, F., S. Nicklen, and A. R. Coulson. 1977. DNA sequenc-ing with chain-terminating inhibitors. Proc. Natl. Acad. Sci.USA 74:5463-5467.

36. Tiesman, J., and A. Rizzino. 1991. A rapid and reliable methodfor the purification of high-quality plasmid DNA for double-stranded sequencing. BioTechniques 10:319, 326-328.

37. Williams, D. R., and C. M. Thomas. 1992. Active partitioning ofbacterial plasmids. J. Gen. Microbiol. 138:1-16.

38. Yang, C. C., and H. A. Nash. 1989. The interaction of E. coliIHF protein with its specific binding sites. Cell 57:869-880.

VOL. 175, 1993

on February 9, 2018 by guest

http://jb.asm.org/

Dow

nloaded from