JOURNAL OF BACTERIOLOGY, Aug. 1990, p. 4603-4609

Vol. 172, No. 8

0021-9193/90/084603-07$02.00/0 Copyright © 1990, American Society for Microbiology

Intrinsic Bends and Integration Host Factor Binding at F Plasmid oriT MEI-MEI TSAI,t Y.-H. FLORENCE FU, AND RICHARD C. DEONIER* Molecular Biology, Department of Biological Sciences, SHS 172, University of Southern California, Los Angeles, California 90089-1340 Received 22 February 1990/Accepted 3 June 1990

F plasmid oriT DNA extending from the F kilobase coordinate 66.7 (base pair [bp] 1 on the oriT sequence map) rightward to bp 527 was analyzed for intrinsic bends (by permutation assays) and for binding of integration host factor (IHF) (by gel retardation and DNase footprinting). Intrinsic bending of the 527-bp fragment (bend center approximately at bp 240) was represented as a composite of at least two components located near bp 170 and near bp 260. IHF bound primarily to a site extending from bp 165 to 195 and with lower affinity to a site extending from bp 287 to 319. The intrinsic curvature and sequences to which IHF binds (IHF is known to bend DNA) may play a structural role in oriT function.

The F plasmid of Escherichia coli is the paradigm for a class of conjugative plasmids (for reviews, see references 9 and 24). The oriT locus is the site at which one strand of F DNA is nicked as a prelude to transfer, and its function is therefore crucial for the conjugation process (21, 23). oriT has been functionally localized to a 375-base-pair (bp) BglIIHaeII fragment (4). The 527-bp BglII-SalI fragment containing this DNA migrated more slowly than expected on polyacrylamide gels (20). This reduced mobility, often diagnostic of bent DNA (14), suggests the presence of a sequence-directed bend. Integration host factor (IHF) is involved in a diverse set of processes (e.g., integration of bacteriophage lambda, transcriptional regulation of bacteriophage Mu, and replication of pSC101; see reference 7 for a review) including F conjugation. The latter process is significantly reduced in E. coli mutants lacking IHF (8), and examination of the F oriT region revealed three sets of sequences that resemble the IHF-binding site consensus sequence (15). Since IHF causes sharp bends in DNA (10, 16-19), it seemed possible that oriT DNA might contain IHF sites that would allow it to be organized into a complex three-dimensional structure in the presence of appropriate proteins, possibly including tra proteins required for nicking at oriT. In this study, we describe the mapping of intrinsic bends in the oriT locus and identification of at least two binding sites for IHF. The locations of these sequence features suggest possible roles for IHF in oriT function.

in Fig. 1B. If required for plasmid construction, fill-in of cohesive ends was performed with the Klenow fragment of DNA polymerase I (Bethesda Research Laboratories, Inc.). Restriction enzymes BamHI, EcoRI, HindIII, Sail, SmaI, DraI, RsaI, HaeII, and TaqI were purchased from Bethesda Research Laboratories; Mnil and HphI were purchased from New England BioLabs, Inc.; DdeI was purchased from Boehringer Mannheim Biochemicals; and PstI was purchased from Pharmacia, Inc. Plasmid pXRD625 was constructed by insertion of the 0.55-kilobase (kb) EcoRI-HincII fragment containing oriT (bp 1 to 529) from pXRD608 into pXRD620 linearized at the EcoRI site. The remaining recessed 3' terminus at the EcoRI site was filled in prior to circularization. The resulting plasmid, pXRD625, contained two directly repeated copies of oriT sequence bp 1 to 527. Plasmid pXRD626 was constructed by ligating the 0.26-kb EcoRI-HindIII fragment containing oriT bp 1 to 237 from pXRD620A104 into pXRD62OA104 linearized at the unique HindIII site. Plasmid pXRD62OA104 will be described in more detail elsewhere (Y.-H. F. Fu, M.-M. Tsai, Y. Luo, and R. C. Deonier, unpublished data). The ends of the bimolecular products were filled in, and intramolecular ligation was conducted. The correct clone was identified by digestion with EcoRI and HindlIl. Plasmid pXRD627 was constructed by ligation of the 0.32-kb DraI-HindIII fragment containing oriT (bp 229 to 527) from pXRD608 to a terminus generated by HindIII digestion of pXRD608. The remaining HindlIl end was filled in, and the fragments were circularized by blunt-end ligation. Combinations of EcoRI, HindIII, and DraI digestions allowed the identification of the correct recombinant, in which two copies of the oriT sequence bp 229 to 527 were present as direct repeats. The HindIII sites in this construct were destroyed by unknown processes. Permutation assays. Permuted fragments were generated from pXRD625, pXRD626, or pXRD627 by using restriction enzymes that cut once in each of the directly repeated elements (25). The restriction enzymes used to generate permuted fragments were SmaI, MnlI, DraI, RsaI, HaeII, and TaqI for pXRD625; SmaI, MnlI, HphI (fragment cut at bp 113 purified after incomplete digestion), and DraI for pXRD626; and RsaI, DdeI, HaeII, HphI, Sall, and PstI for pXRD627. Distances migrated by purified, permuted frag-

MATERIALS AND METHODS

Plasmid constructions. In the constructions described below, direct repetitions of different portions of the oriT locus were generated by using DNA from plasmids pXRD608 and pXRD620. All relevant plasmids are described in Fig. 1A. Plasmid pXRD608 contains the 527-bp BglII-SalI fragment of oriT inserted into pUC8 between the BamHI and Sall sites (B. Horowitz and R. Deonier, unpublished data). Plasmid pXRD620 contains the 1,078-bp BgllI fragment of F (includes oriT) cloned in the viable orientation into the BamHI site of pUC8. Part of the oriT DNA sequence (20) is shown * Corresponding author. t Present address: Amgen, Inc., Thousand Oaks, CA 91320-1789.

4603

J. BACTERIOL.

TSAI ET AL.

4604

:

A. pXRD620

S 'too * ^VW#

E

Il PXFD620A104

--

-

894'S P H

l

II-

pXRD60B

e0bd

pXRD625

_

N

_Wonc"%

pXAUZb

Ei_

pXRD627

E

m

L

_

_

_

bcd

HI

H?D' I

_

'

-1

I

m

a

S.

150 140 120 130 110 BCTCAACAOSTTSBTSTTCTCACCACAASACAACCACGCAM 200 190 AT*MATTTATSAAAAATTA 270

280

1g0 170 10 ----TTT ---ATTTTCTTTATAM

210 220 CCTIC TCT

240 250 240 230 AMlA TS C TACAACACG

290

310

300

320

330

340

mGTAgMnTA CTATTTTATAAAAAACATTATTTTATATTBA&TgCTBTACA0C FIG. 1. (A) Cloned DNA segments used for permutation, gel retardation, or footprinting analysis. Inserts in pXRD625, pXRD626, and pXRD627 contain direct repetitions of all or part of the oriT region. The vector is pUC8 in all cases, and the inserts are located between the EcoRI and HindITI sites in the polylinker. Base-pair coordinates for restriction sites within the BglIT fragment from F are indicated above the representation of the pXRD620 insert. DNA derived from the polylinker is shown as small boxes (a, b, c, and d), and DNA derived from F is shown as open, filled, stippled, or hatched ribbons or arrows. The hatched region to the right of the SalI site at bp 527 was retained in some constructs but was not duplicated. The location of the oriT nick is indicated by a solid triangle in the insert for pXRD620. Restriction sites are abbreviated as follows: E, EcoRI; D, DraI; Dd, DdeI; H, HindIII; Ha, HaeII; Hp, Hphl; M, Mnll; P, Pstl; R, RsaI; S, Sall; Sm, SmaI; and T, TaqI. Junctions formed from joining termini generated by different restriction enzymes are indicated with primes. Details of the constructions are given in the text. (B) DNA sequence of oriT region (20). The strand shown is defined as the top strand. ecsC

ments were measured by densitometry from photographs of gels stained with ethidium bromide or from autoradiograms if fiagments were labeled. Labeling at 5' ends was introduced with [y-32P]ATP (Dupont NEN Research Products, 3,000 Ci/mmol) and T4 DNA kinase (Bethesda Research Laboratories). A 1-kb size standard ladder (Bethesda Research Laboratories) was labeled similarly for use as size standards. Polyacrylamide gels (7.5 or 10%6, acrylamide:bisacrylamide ratio = 29:1, 1-mm thickness) in Tns-borate-EDTA buffer (13) were prerun at 4°C and 10 V/cm for at least 2 h or overnight. The same voltage gradient and temperature were applied during actual runs. Gels loaded with labeled samples were dried at 80°C for 2 h under vacuum. Autoradiography was at room temperature with Kodak XRP-5 ifim. Retardation gels. DNA fragments containing selected regions of oriT were 3' end labeled with [a-32P]dATP (Dupont NEN Research Products, 3,00 Ci/mmol) and Klenow fragment of DNA polymerase I. Except for experiments with mixed sites, fragments were purified from agarose gels. The concentration for each binding site (A or B) was either 1 or

1.5 nM in a reaction volume of 10 or 25 ,ul. Purified IHF (gift from H. Nash via D. Galas) was diluted to the proper concentration with 2x binding buffer (O.lM Tris [pH 7.4], 2 mM EDTA, 2 mM 1-mercaptoethanol, 20% glycerol, 400 Lg of bovine serum albumin, 140 mM KCl, 14 mM MgCl2, 6 mM CaCl2 [16]). Poly(dIdC) poly(dIdC) (Pharmacia) at concentrations up to 50 ,ug/ml was used as the competitor when required. The IHF-binding reaction was performed as described by Prentki et al. (16). End-labeled DNA samples with or without the competitor were suspended into a 0.5 volume of distilled water, and the appropriate amount of 2x binding buffer and the required amount of IHF in 2x binding buffer were added. Samples were incubated at room temperature for 25 min. Loading dye (0.20 volume; 1 mg of bovine serum albumin, 50o glycerol, 0.01% xylenecyanol) was added, and samples were immediately subjected to electrophoresis. Sample loading was performed with voltage gradients (2 V/cm) applied. Electrophoresis conditions were as described for the permutation assays. Association equilibrium constants were determined after densitometry of autoradiograms and integration of peak areas as described below. DNase footprintng. For footprinting site A, the EcoRTHindTII fragment of pXRD62OA104 containing oriT sequence bp 1 to 237 was labeled either at the EcoRI end (bottom strand) or at the HindIll end (top strand) by using [k32P]dATP (Dupont NEN Research Products, 3,000 Ci! mmol) and Klenow fragment. For footprinting site B, the HindIII-Sall fragment containing oriT bp 229 to 527 from a pXRD608 derivative containing a decameric HindIII linker at the Dral site (bp 228) was labeled either at the Hindlll end (bottom strand) or the SalT end (top strand). IHF-binding reactions were performed as described for the retardation gel procedures, except for omission of the competitor. After 25 min of incubation at room temperature, a 0.10 volume of bovine pancreatic DNase I (Bethesda Research Laboratories) was added to the reaction mixtures at the desired concentration (0.1 or 0.01 ag/ml), and mixtures were incubated at room temperature for 30 s. Reactions were terminated by the addition of an equal volume of DNase stop solution (2.1 M sodium acetate, 50 mM EDTA). Lambda DNA was added as carrier prior to precipitation with 2.5 volumes of ethanol. Samples were electrophoretically resolved in 8% denaturing polyacrylamide gels. Calculations. Association equilibrium constants (Ka) were usually determined in the presence of poly(dIdC) poly (dIdC). This suppressed formation of uncharacterized, lowmobility species that formed at large IHF concentrations in the absence of poly(dIdC) poly(dIdC) competitor. The concentration of the unbound IHF, [I], is affected by nonspecific binding to competitor, but this effect is less severe at large initial concentrations of IHF, [IJ. To deterne Ka under conditions for which nonspecific binding can be neglected, we performed the following extrapolation. The quantity Ka' is defined as Ka' = [AI]/([A] [Io]), in which [A] and [AI] are the concentrations of unbound site A and of site A bound to IHF, respectively. The ratio [AI]/[A] is calculated from the ratio of peak areas for complex and unbound fragment, determined by densitometry of the autoradiogram. [T0] is known. It can be shown that -

Ka' Ka + Ka([Ao] + [So])[!J-'1 + ([A.] + KaK-1[So])[o]-2 +... where [S.] is the initial competitor concentration and K. is to =

the association equilibrium constant for IHF binding

VOL. 172, 1990 A.

BENDING AND IHF BINDING AT oriT B.

100 200 300 400 500 bose poirs

4605

C.

460 500 545

284 324 364

bose poirs

100

200

bose pairs

0.005

0.009

0.013

1/l

FIG. 2. Analysis of intrinsic DNA bends by gel permutation assay. (A) Analysis of the oriT region from bp 1 to 529 by using pXRD625. (B) Analysis of the region from bp 229 to 527 by using pXRD627. (C) Analysis of the region from bp 1 to 237 by using pXRD626. In panels A to C, average values of R (the ratio of the observed relative electrophoretic mobility to the expected relative mobility for an unbent fragment of the given size) are plotted against base-pair coordinates for the restriction sites employed. Restriction enzymes used to generate each permuted fragment are labeled as in Fig. 1. Average values for positions at which R is maximized are stated in each panel. (D) Location of the bend in the bp 1 to 228 fragment determined by using deletion mutants derived from pXRD620. R is plotted against 1/1, where I is the fragment length. For panel D, all fragments contain 21 bp of polylinker DNA in addition to oriT DNA.

competitor (unknown). This equation is obtained by recognizing that Ka' = Ka([I]/[Io]) and employing the relationships [I] = [I] - [AI] - [SI], [A] = [Ao](1 + Ka [I])-1, and [SI = [SO](1 + Ks [I])-1. After use of these relations and appropriate rearrangements, the equation above is obtained with the additional approximations (1 + x)-1 1 - x and [I] [Io] when [Ij] is sufficiently large. To use this equation, Ka' was determined at different values of [I], and Ka' was plotted versus [Io]-1. If terms in [10f-2 and beyond can be neglected, the plot will produce a straight line with Ka as the intercept on the Ka' axis (i.e., for [10f-1 = 0). The same value of Ka should be obtained for different values of [SO]. Permutation analyses for determining intrinsic bend locations (no bound protein) were based on the quantity R = The actual relative mobility (XI/X.) is the (X/X1)/(X/X1)U. migration distance for the fragment (Xi) divided by the migration distance for an arbitrary standard fragment (Xe). (X/IX.)u is the relative mobility of an unbent fragment of identical size (X, is calculated from a calibration curve in this case). The coordinate at which R is maximized (R = 1.00) represents the bend location. Linear regression was used to extrapolate R values from each side of the maximum, with data from at least three independent experiments. Each branch of the curve, when extrapolated to R = 1.00, yielded coordinates that agreed within 10 bp, except for the largest fragment, which contains a composite bend. -

RESULTS Intrinsic bends in oriT DNA. Plasmid pXRD625 (Fig. 1A) was used in a permutation assay to detect bending in the oriT fragment extending from bp 1 to bp 527. Variation of R for fragments cleaved with different enzymes indicated a bend at bp 244 (Fig. 2A). The magnitude of the minimum R value corresponded to a bend angle of 50 to 600 (19). Plasmids pXRD626 (contains direct repetition of the bp 1 to 237 region; Fig. 1A) and pXRD627 (contains direct repetition of the bp 229 to 529 region) were used to refine the estimate of the bend location. Variations in R for plasmid pXRD627 again indicated that there is a bend (Fig. 2B), but the predicted location was bp 264, not bp 244, as had been seen with the larger insert in pXRD625. The minimum value of R again indicated a bend angle of approximately 500. This

discrepancy in location might have arisen if the bp 1 to 527 segment contained two (or more) bends, one to the left of bp 244 (bend I) and one to the right of bp 244 (bend II). Permutation analysis of the insert in pXRD626 (Fig. 2C) indicated that the bp 1 to 237 region does indeed contain an intrinsic bend. The location is near bp 170. Again, the mobility ratio suggests a bend angle of approximately 500. Since the magnitudes of the bend angles are similar for the bp 1 to 237 and the bp 229 to 529 segments, the bend position of the joined fragments should be the average of the individual bend positions. This average (bp 217) differs enough from the measured average (bp 244) to indicate that additional factors are affecting the composite bend. This issue will be addressed in the Discussion. Further mapping of the bend near bp 170. A jprominent sequence feature possibly associated with a bend is (A)5TCAGC(A)5 (bp 172 to 158, bottom strand). This might be a component of bend I in the bp 1 to 237 fragment. We tested this possibility by analyzing the mobilities of EcoRIHindIII fragments derived from a set of nested deletion mutants of pXRD620 (e.g., pXRD620A104, Fig. 1A; Y.-H. F. Fu et al., unpublished data). These deletions extended into oriT from the right, terminating at bp 237, 222, 186, 177, 167, 138, 123, 88, or 61, respectively. The fragments used also contained an additional 21 bp of polylinker. Bends cause the greatest decrease in mobility when the bends are near the center of the fragment (25). For a single localized bend, we reasoned that deletions from the right would locate the bend proportionally closer to the end of the shortened fragment, so that the ratio R (the actual mobility divided by the mobility predicted for an unbent molecule of the same molecular weight) should increase as the deletion endpoints approached the bend site. After the bend site was deleted, the R value should have a constant value approximately equal to unity. Similar reasoning was used by Cobbett et al. (1) for identifying a bend in the aroF regulatory region. A plot of R versus the reciprocal of the fragment length (1/1) is expected to show a change of slope after the deletion endpoints extend beyond a localized bend. The results of such a plot are shown in Fig. 2D. The two branches of the plot intersect at 1/i = 0.00543, or 1 = 184. Because the

4606

TSAI ET AL.

J. BACTERIOL.

TABLE 1. Association equilibrium constants for binding of IHF to sites A and B in the oriT region of F: comparison of Ka values obtained at different concentrations of competitor

FIG. 3. Gel retardation assays of IHF binding to oriT DNA segments. (A) Binding to segments containing bp 1 to 237, 1 to 186, and 1 to 138 is analyzed. These DNA fragments were purified from pXRD62OA104, pXRD62OA52, and pXRD62OA90, respectively. Poly(dIdC). poly(dIdC) concentration was 20 p.g/ml. (B) Binding to the segment containing oriT bp 229 to 527 (purified from pXRD608). Concentrations (nanomolar) of IHF are indicated above each lane. The presence or absence of 20 ,ug of poly(dIdC) poly(dIdC) per ml is indicated by plus or minus. The positions of the protein-DNA complexes are labeled with a solid triangle beside each panel.

fragments used in this analysis all contained 21 bp of polylinker sequence from the vector, this intersection point corresponds to bp 163 in oriT, indicating that the bp 158 to 172 segment is indeed bent. The agreement with the bp 170 position determined by permutation assay is excellent. Both methods located the bend within the (A)5TCAGC(A)5 sequence. This sequence is expected to display an additive intrinsic bend since the (A)5 tracts are in phase with the helical repeat distance for DNA. Binding of IHF to oriT. Sites similar to an IHF consensus have been identified in the F oriT region (15). Since IHF sharply bends DNA to which it binds, we wished to locate IHF-binding sites for comparison with locations of the intrinsic bends. Plasmids pXRD620A104, pXRD620A52, and pXRD62OA90 contain oriT DNA from bp 1 to 237, 1 to 186, and 1 to 138, respectively (cf. Fig. 1A for structure of pXRD62OA104; others are similar; Y.-H. F. Fu et al., unpublished data). The first two of these three plasmids carried the sites identified by McIntire and Dempsey (15). Gel retardation analyses in which these mutants were used revealed a single IHF-binding site (Fig. 3A). The additional, retarded species seen only with pXRD62O104 indicated that it contains an IHF-binding site (site A) that overlaps the bp 186 to 237 region. A second IHF-binding site (site B) was located in the bp 229 to 527 oriT segment purified from pXRD608 (Fig. 3B). Its affinity for IHF is lower than the affinity of site A, and at the high IHF concentrations used, a more retarded smeared band appeared unless the competitor was added. Since the nature and composition of this smeared band was unknown, most of the analysis for binding to site B was done in the presence of competitor. Association equilibrium constants for IHF and site A or B were determined individually or in mixtures containing each site on separate fragments (data not shown). IHF binding to both sites was usually studied in the presence of competitor. The true values for Ka, calculated under different conditions, were estimated by using the extrapolation described in Materials and Methods. The calculated true equilibrium constants are shown in Table 1. We have not determined the fraction of IHF that is active, so the reported values may be

Experimental

(Ka)A

(M- l )

(Ka)B (M- ')

(Ka)A/

protocol"

1. Extrapolation, 50 ,ug of poly(dIdC) poly(dIdC) per ml 2. Extrapolation, 10 ,ug of poly(dIdC) poly(dIdC) per ml 3. Direct measurement, no poly(dIdC) poly(dIdC) 4. Mixed sites, A and B on different molecules, 50 ,ug of poly(dIdC) poly(dIdC) per ml

1.2 x 107

1.7 x 106

7

(K.)B

1.2 x 107

3.4 x 106 (8.1 x 106)b

(7.5 x 105)b

11

a Experimental details: protocol 1, (K,)A is an average of two independent extrapolations (three IHF concentrations each) and (Ka)B was determined by one extrapolation in which data from four experiments and four different IHF concentrations were used; protocol 2, one extrapolation; protocol 3, one experiment; protocol 4, average of two determinations, using the relation = [AI]/[BIl [actual (K,)AI(K&)B values: 15.4 and 7.3]. (K)AI(Kj)R b Values given are K,,' (cf. equation, Materials and Methods). These are expected to be underestimates. The (K9AI(K)B ratio will be unaffected by the presence of competitor.

underestimated; however, the values obtained from different conditions are in reasonable agreement. The Ka for site A (with IHF from the same source as used in reference 16) is 1/4 to 1/10 of the values observed for IHF sites on pBR322 and comparable to the Ka for the IHF site at the left end of ISJ (16). The Ka for site B is 1/5 to 1/10 of the value for the Ka for site A. Mapping of IHF-binding sites by DNase footprinting. Protection of sites A and B by IHF in the absence of competitor is shown in Fig. 4. Protection for site A extended from approximately bp 163 to 195 on the "top" strand and from bp 165 to 195 on the "bottom" strand, although the precise limits were difficult to determine because of regions in the DNA that were relatively insensitive to DNase I digestion. This location is consistent with mapping of the site with the deletion mutants: mutant pXRD62OA52, which lacks the left portion of the site A DNA protected by IHF, did not produce a retarded species in gel retardation assays. Because site B was bound more weakly by IHF than site A, protection at site B was more difficult to visualize. Protection was seen over bp 287 to 319. The location of this site was confirmed by gel mobility shift analysis by using a series of fragments carrying oriT DNA from bp 229 to 285, 229 to 309, 229 to 325, and 229 to 362 (data not shown). Only the latter DNA segment showed IHF binding. Failure of IHF to bind to the bp 229 to 325 region (which contains the entire region protected in footprinting and six additional base pairs) emphasizes the role of more distant sequences in IHF binding (11, 26). DISCUSSION F oriT bending is composed of at least two components. In the bp 1 to 237 segment, a localized bend (bend I) occurs in the vicinity of bp 168, near a set of poly(A) tracts like those often associated with bending. In the bp 229 to 527 region, the average bend location is near bp 264. The absence of poly(A) tracts at this location suggests either that bend II is itself composite or that some sequence features other than the poly(A) tracts cause bending. In both cases, the mini-

VOL. 172, 1990

BENDING AND IHF BINDING AT oriT

A. H

D.

f''V

80

3

^

Ig5

r18l85

g

165

175

-287

|

*

x,

C V

g

FIG. 4. Binding of IHF to oriT sites A and B. (A) DNase footprints of IHF bound to site A (left) and site B (right). Base pair coordinates of the footprint boundaries are shown at the sides. (B) Relationship of footprint A (hatched and solid region) to (A)5TCAGC(A)5 tract (open box) centered at bp 165. The sequence of bottom strand is shown, with the 3' end at the bottom. Two boundaries of the oligo(dA) oligo(dT) tracts (bp 162 and 172) lie 1 and 2 helical turns below regions protected against hydroxyl radical attack in the IHF consensus regions of attP (26). Solid boxes are drawn opposite deoxynucleotides that match the IHF consensus C/TAANNNNTTGATA/T (12) (read down labeled sequence, beginning at bp 188). mum values for the mobility ratios, R, are in the range expected for 500 bends (19). Other experiments in which internal deletions of oriT were used (not discussed here) were performed to determine the relationship between the two bends. If the DNA entering and exiting each bend is considered to define a plane, the two planes intersect along a line corresponding to the DNA segment connecting bend I to bend II. These other experiments suggested that the planes intersect at an angle of 80 to 1400. If the angle between the planes (dihedral angle) had been 0°, the effects of the two bends on mobility would have been completely additive. Had the angle been 1800, the effects of the bends would have tended to cancel each other. The observation of a minimum R value for the bp 1 to 527 fragment that is similar to the minimum R values for frag-

4607

ments containing either of the individual components is consistent with an intermediate value for the angle between the planes. The two intrinsic bends are each within 25 bp of an IHF-binding site (see below). We have not determined whether the intrinsic bends are in phase with IHF-induced bends. If bends I and II have similar magnitudes, then it might have been expected that the bp 1 to 527 fragment containing this composite bend would appear to be bent at bp 217, rather than at the observed position at bp 244. A possible explanation for this discrepancy is that the insert used to locate bend II (contained in plasmid pXRD627) was cloned by using a DraI site that divides a TATTT sequence from an adjacent AAAAAA sequence. The novel junction adjacent to this A tract may have created an additional bend component. Moreover, phase changes associated with the novel junction in pXRD625 might have generated an artificial dihedral angle between bends I and II for some permuted fragments. Finally, it is possible that bend II is itself composite or delocalized. McIntire and Dempsey (15) identified three regions in the R100 oriT sequence that matched closely the IHF-binding site consensus C/TAANNNNTTGATA/T (12). Corresponding matches were identified for F, although the degree of sequence similarity to the IHF consensus was less for the two sites located to the right of bp 170 on the F map. The F sequence best matching this IHF consensus spanned bp 163 to 151 (Table 2). We determined that this site was not detectably bound by IHF in vitro. The DNA sequence from bp 188 to 176 corresponds best with the observed IHFbinding site. Kur et al. (11) proposed that the IHF-binding consensus sequence is C/TAANNWNCITTGWWWWNNN NNNNWWWWWW, where N is any deoxynucleotide and W is A or T. Similarity between oriT sequences and this extended consensus was greatest in the bp 154 to 180 region and in the bp 201 to 176 region (Table 2). However, the segments from bp 191 to 165 or 193 to 167 best correspond to the footprint observed for site A. Site A partially overlaps the sequence (A)5TCAGC(A)5 (bp 172 to 158, bottom strand), which we showed has an intrinsic bend. If bending promoted by (A)5TCAGC(A)5 is in the direction of the minor groove (3) or if bends occur at junctions between A tracts and neighboring DNA (see reference 22 for further discussion), then the intrinsic bend may reinforce IHF-induced bending. This might explain why IHF binds better to site A than to regions having greater similar-

TABLE 2. Comparison of DNA sequences in IHF sites A and B and their neighborhood to consensus sequences C/TAANNNNTTGATA/T or C/TAANNWNC1TTGWWWWNNNNNNNWWWWWW' Locationb

Strand

Sequence

% Similarity'

Site A 188-176 191-165 193-167 Sites near site A

Bottom Bottom Bottom

CAcNNNNTTtATA TAANNcNCTaTTTANNNNNNNAAATcA CAtNNcNCTcTATTNNNNNNNAAAAAT

78 82 82

Bottom Top Bottom

CAANNNNTTGtTT CAANNTNTTGcTgANNNNNNNTTATAA CtANNTNTTcATAANNNNNNNTTTATA

89 88 88

Bottom Top Bottom

TAANNNNaTGtTT TgANNANTTtTATANNNNNNNTTATTT TAANNTNaTGTTTTNNNNNNNAATAgT

78 88 88

163-151 154-180 201-176 Site B 317-305 289-315 318-291

a Matches to a consensus sequences (11, 12) are indicated in uppercase letters. Nucleotides that do not match are shown in lowercase type. b Base-pair number 1 is the fourth in the BglII site at 66.7 F (137 bp to the left of the oriT nick site). c Based on fraction of nucleotides matching specified nucleotides in the consensus.

4608

TSAI ET AL.

J. BACTERIOL. 3

4

6A

S -N

at-

A-T rich

68

-

-

A-T rich

I

50

1OO

150

200

250

300

350

400

base pairs

FIG. 5. Relationships between IHF-binding sites, bends, and sequence features of the F oriT region. The filled region is homologous among IncF plasmids (6). The hatched portion is unique to F. One of the AT-rich regions (open boxes) extends into the homologous region (overlap: bp 150 to 170). The location of the strand-specific nick (21) is denoted by the triangle. Base-pair coordinates are measured clockwise of the BglTI site at the 66.7-kb F coordinate. Inverted or direct repetitions are labeled as in reference 9. Carets denote the two average intrinsic bend locations, solid boxes stand for the repeated sequence TTTnTAAAnA not previously recognized, and stippled boxes represent IHF-binding sites determined by footprinting. For the actual DNA sequence, see Fig. 1B.

ity to either of the two IHF consensus sequences. A tracts are also located 3' to the C/TAANNNNTTGATA/T consensus sequences in the H' and H2 sites of attP (26). Site B is located at the third site identified by McIntire and Dempsey (15). The best match to the shorter IHF consensus (two mismatches) is bp 317 to 305 (bottom strand), which includes the right half of the footprint for site B. The leftward extension of the footprint (to bp 287) overlaps an (A)5 tract (bp 299 to 295, bottom strand). Site B (bp 287 to 319) includes most of the region from bp 293 to 320, which is very rich in A+T sequences (97%). The best matches to the extended consensus are bp 289 to 315 and 318 to 291. Both of these correspond well with site B. The lesser affinity of site B for IHF compared to site A and the reduced affinity for site B in a deletant that retains DNA 6 bp to the right of the footprint again indicate that sequence determinants several base pairs away from the IHF consensus can affect binding (11, 16, 26). It is not known whether the IHF-binding sites in oriT are functionally significant, although retention of these sites in similar positions in F and the related IncFT plasmid R100 is consistent with their being functional. IHF is required for efficient tra operon expression and production of pili for F and R100 (2, 8). The R100-1 sequence revealed a potential IHF-binding site at the beginning of traJ (2), which may be one of the points at which the IHF effect is exerted (traJ is required for tra operon expression for F and R100). To determine whether IHF also acts directly at oriT, it will be necessary to construct tra expression plasmids that are independent of IHF and to assay nicking and transfer in IHF- strains. IHF is known to assist formation of higherorder structures and binding of other proteins to plasmid or phage loci. In the case of the pSC101 and R6K -y replication origins (5, 18), IHF-binding sites have been located within A+T-rich sequence tracts that are similar in length to the central A+T-rich tract of F oriT, which contains IHFbinding site A at its left margin (Fig. 5). One consequence of conjugal transfer is production of two copies of the transferred plasmid (a form of replication). The resemblence between oriT and replication origins may also extend to higher-order DNA-protein structures whose formation is assisted by IHF binding. IHF produces large bends upon binding to DNA (>1400 [19]). If IHF binds to oriT sires A and B in vivo, the resulting bends would divide oriT into three structural domains: bp 100 to 180, 180 to 300, and 300 to 350 (Fig. 5). Depending upon phase relationships and binding of other proteins, regions might be folded into a looped structure in which

inverted repetitions 1 and 5 (and proteins bound to them) are brought into close proximity. This possibility was also suggested by other data (cf. Fig. 5 and reference 21). Removal of portions of feature 5 causes a severe transfer defect (Y.-H. F. Fu et al., unpublished data), indicating that this feature is functionally important. The placement of feature 5 approximately 90 bp from the nick site (bp 137) independently suggests that oriT and F transfer proteins that bind to it may be organized into a higher-order DNA-protein structure. Because 25% of the 250 bp required for full oriT function (- bp 110 to 360) can be bound specifically by IHF, direct and indirect interactions between IHF and tra proteins presumed to bind oriT may also be important for forming a functional oriT complex in vivo. ACKNOWLEDGMENTS We thank Howard Nash for supplying the IHF and David Galas for advice on DNase footprinting and for thoughtful comments on DNA bending. This research was supported by National Science Foundation grant DMB-8717057 and by the University of Southern California Faculty Research and Innovation Fund.

ADDENDUM IN PROOF IHF-binding sites in oriT of plasmid R100 have recenLly been located (W. B. Dempsey and B. E. Fee, Mol. Microbiol., in press). IHF site 1 in R100 lies several base pairs to the left of the position for IHF site A in oriT of plasmid F. The sequences of F and R100 are not identical in this region. The location of IHF site 2 in R100 corresponds closely with the location of IHF site B in F, despite differences in 25% of the nucleotides spanned by the two footprints. LITERATURE CITED 1. Cobbett, C., B. Dickson, and L. Farmer. 1989. The role of a static bend in the DNA of the aroF regulatory region of Escherichia coli. Gene 75:185-191. 2. Dempsey, W. B. 1987. Integration host factor and conjugative transfer of the antibiotic resistance plasmid R100. J. Bacteriol.

169:4391-4392. 3. Drew, H. R., and A. A. Travers. 1985. DNA bending and its relation to nucleosome positioning. J. Mol. Biol. 186:773-790. 4. Everett, R., and N. Wiletts. 1982. Cloning, mutation and location of the F origin of conjugal DNA transfer. J. Mol. Biol.

136:129-150. 5.

Fllutow,cz, M., and K. Appelt. 1988. The integration host factor

of Escherichia coli binds to multiple sites at plasmid R6K y

VOL. 172, 1990

6.

7. 8. 9. 10.

11.

12.

13.

14.

15. 16.

origin and is essential for replication. Nucleic Acids Res. 16:3829-3843. Finlay, B. B., L. S. Frost, and W. Paranchych. 1986. Origin of transfer of IncF plasmids and nucleotide sequences of the type II oriT, traM, and tra Y alleles from ColB4-K98 and the type IV tra Y allele from R100-1. J. Bacteriol. 168:132-139. Friedman, D. I. 1988. Integration host factor: a protein for all reasons. Cell 55:545-554. Gamas, P., L. Caro, D. Galas, and M. Chandler. 1987. Expression F transfer functions depends on the Escherichia coli integration host factor. Mol. Gen. Genet. 207:302-305. Ippen-Ihler, K. A., and E. G. Minkley, Jr. 1986. The conjugation system of F, the fertility factor of Escherichia coli. Annu. Rev. Genet. 20:593-624. Kosturko, L. D., E. Daub, and H. Murialdo. 1989. The interaction of E. coli integration host factor and X cos DNA: multiple complex formation and protein-induced bending. Nucleic Acids Res. 17:317-334. Kur, J., N. Hasan, and W. Szybalski. 1989. Physical and biological consequence of interactions between integration host factor (IHF) and coliphage lambda late PR' promoter and its mutants. Gene 81:1-15. Leong, J. M., S. Nunes-Duby, C. F. Lesser, P. Youderian, M. M. Susskind, and A. Landy. 1985. The 4)80 and P22 attachment sites: primary structure and interaction with Escherichia coli integration host factor. J. Biol. Chem. 260:4468 4477. Maniatis, T., A. Jeffrey, and H. Van de Sande. 1975. Chain length determination of small double- and single-stranded DNA molecules by polyacrylamide gel electrophoresis. Biochemistry 14:3787-3794. Marili, J. C., S. D. Levene, D. M. Crothers, and P. T. Englund. 1982. Bent helical structure in kinetoplast DNA. Proc. Natl. Acad. Sci. USA 79:7664-7668. McIntire, S. A., and W. B. Dempsey. 1987. oriT sequence of the antibiotic resistance plasmid R100. J. Bacteriol. 169:3829-3832. Prentki, P., M. Chandler, and D. J. Galas. 1987. Escherichia coli

BENDING AND IHF BINDING AT oriT

4609

integration host factor bends the DNA at the ends of IS1 and in an insertion hot spot with multiple IHF binding sites. EMBO J. 6:2479-2487. 17. Robertson, C. A., and H. A. Nash. 1988. Bending of the bacteriophage X attachment site by Escherichia coli integration host factor. J. Biol. Chem. 263:3554-3557. 18. Stenzel, T. T., P. Patel, and D. Bastia. 1987. The integration host factor of Escherichia coli binds to bent DNA at the origin of replication of the plasmid pSC101. Cell 49:709-717. 19. Thompson, J. F., and A. Landy. 1988. Empirical estimation of protein-induced DNA bending angles: applications to X sitespecific recombination complexes. Nucleic Acids Res. 16:96879705. 20. Thompson, R., L. Taylor, K. Kelly, R. Everett, and N. Willetts. 1984. The F plasmid origin of transfer: DNA sequence of wild-type and mutant origins and location of origin-specific nicks. EMBO J. 3:1175-1180. 21. Thompson, T. L., M. B. Centola, and R. C. Deonier. 1989. Location of the nick at oriT of the F plasmid. J. Mol. Biol. 207:505-512. 22. Travers, A. A. 1989. DNA conformation and protein binding. Annu. Rev. Biochem. 58:427-452. 23. Willetts, N. S. 1972. Location of the origin of transfer of the sex factor F. J. Bacteriol. 112:773-778. 24. Willetts, N., and R. Skurray. 1987. Structure and function of the F factor and mechanism of conjugation, p. 1110-1133. In F. C. Neidhardt, J. L. Ingraham, K. B. Low, B. Magasanik, M. Schaechter, and H. E. Umbarger (ed.), Escherichia coli and Salmonella typhimurium: cellular and molecular biology. American Society for Microbiology, Washington, D.C. 25. Wu, H.-M., and D. M. Crothers. 1984. The locus of sequencedirected and protein-induced DNA bending. Nature (London) 308:509-513. 26. Yang, C.-C., and H. A. Nash. 1989. The interaction of E. coli IHF protein with its specific binding sites. Cell 57:869-880.