 1996 Oxford University Press

1780–1786 Nucleic Acids Research, 1996, Vol. 24, No. 9

Examining the contribution of a dA+dT element to the conformation of Escherichia coli integration host factor–DNA complexes Laura M. Hales+, Richard I. Gumport1,2 and Jeffrey F. Gardner* Departments of Microbiology, 1Biochemistry and 2College of Medicine, University of Illinois, Urbana, IL 61801, USA Received November 1, 1995; Revised and Accepted March 11, 1996

ABSTRACT DNA binding proteins that induce structural changes in DNA are common in both prokaryotes and eukaryotes. Integration host factor (IHF) is a multi-functional DNA binding and bending protein of Escherichia coli that can mediate protein–protein and protein–DNA interactions by bending DNA. Previously we have shown that the presence of a dA+dT element 5′-proximal to an IHF consensus sequence can affect the binding of IHF to a particular site. In this study the contribution of various sequence elements to the formation of IHF– DNA complexes was examined. We show that IHF bends DNA more when it binds to a site containing a dA+dT element upstream of its core consensus element than to a site lacking a dA+dT element. We demonstrate that IHF can be specifically crosslinked to DNA with binding sites either containing or lacking this dA+dT element. These results indicate the importance of flanking DNA and a dA+dT element in the binding and bending of a site by IHF. INTRODUCTION Protein-induced DNA bending plays an important role in numerous cellular processes, including regulation and activation of transcription, DNA replication, transposition and site-specific recombination (1–9). The formation of nucleoprotein structures that mediate these processes often involves the interaction of one or more DNA binding or bending proteins and intrinsic structural elements, such as sequence-directed bends, to direct the DNA path in a manner that promotes efficient complex formation. A major DNA binding and bending protein of Escherichia coli and other Gram negative bacteria is integration host factor (IHF). DNA bends induced by IHF can function to create loops to bring activator and repressor proteins into proximity to RNA polymerase and to facilitate assembly of higher order protein–DNA structures (10–14). Site-specific recombination of bacteriophage λ is an excellent model for studying IHF action: IHF binds to three sites in attP and aids in the condensation of a 250 bp region of the λ chromosome to form a nucleoprotein structure, called an intasome, which is active in recombination (15–19). Evidence that IHF acts * To

primarily as an architectural element comes from bend–swap experiments (20–25). Binding sites for IHF usually consist of a common core consensus sequence, WATCAANNNNTTR (10). Additionally, flanking DNA sequences are important (26), as demonstrated by mutational analyses (27–29) and by the large region of DNA protected from nuclease attack upon binding by IHF (30,31). Expanded consensus sequences have been proposed that define a dA+dT element 5′ of the core consensus as important for IHF binding (32,33). The role of this element in the binding of IHF to sites in the λ attP region has been analyzed (28,34,35). The H′ site in attP contains a dA+dT element, while the H1 site does not. This element is required for IHF binding to the H′ site and significantly enhances the binding of IHF to the H1 site when placed at the appropriate distance from the core consensus element (34). It was proposed that the dA+dT element provides an additional protein–DNA contact or structural element that adds to contacts defined in the core consensus. Bending of DNA by IHF would account for the requirement for upstream sequences and the large footprint (31), but little is known about the precise role of flanking sequences in DNA bending by this protein. In this study we investigated the role of flanking DNA in the bending of a site by IHF. The mobility of protein–DNA complexes of IHF with various binding sites were examined in vitro using circular permutation analysis (36). The results suggest that IHF bends DNA more when bound to a site containing a dA+dT element than when associated with a site lacking it. A chimeric attP of bacteriophage λ was constructed to analyze the effects of placing a dA+dT element upstream of the H1 site on recombination. The results show that the ability of this chimeric substrate to undergo recombination is lower than that of the wild-type substrate, indicating that the intasome formed by the chimeric attP may not be optimal for recombination. Protein–DNA interactions between IHF and DNA flanking the consensus element were investigated using phenylazide-mediated photo-crosslinking. We demonstrate that the DNA upstream of the core consensus element, regardless of whether it contains a dA+dT element or not, is in proximity to IHF, thereby allowing protein–DNA crosslinks to occur. These results indicate that the flanking DNA of an IHF site is important in the specific binding and bending of DNA by this protein and suggest that IHF may bend the DNA in a different manner when a dA+dT element is present.

whom correspondence should be addressed

+Present

address: Department of Microbiology, Columbia University, College of Physicians and Surgeons, 701 West 168th Street, New York, NY 10032, USA

1781 Nucleic Acids Acids Research, Research,1994, 1996,Vol. Vol.22, 24,No. No.19 Nucleic

1781

Figure 1. Sequences (5′→3′) of the top strand of orientation (I) of the oligonucleotides used in the circular permutation studies. Orientation (II) contains the sites cloned into the HincII site in the opposite orientation. An IHF consensus sequence (W, dA or dT; R, dA or dG; N, any nucleotide) is shown at the top (10,33). The core consensus and 5′-proximal DNA elements are underlined in each of the sites. (A) Sequences of the H′ and H1 site derivatives (34) used in this study. (B) Sequences of mutant H′ sites (28,49) used in this study. Mutant sites are named for the position of the nucleotide change and the mutated base (i.e. H′22T). The numbering is based on the position of the bases with respect to position 0 of the core sequence of attP (30). All mutations occur within the dA+dT element of the H′ site.

MATERIALS AND METHODS Media, chemicals, enzymes and proteins The media and buffers used have been described (37). Oligonucleotides were obtained from the University of Illinois Biotechnology Center (Urbana, IL) or from Operon Technologies Inc. Bovine serum albumin (BSA) and proteinase K were purchased from Sigma Chemical Co. and were used as described. Enzymes were purchased from Life Technologies Inc. Purified IHF and Int proteins were a generous gift from H. Nash (NIH). Plasmid constructions DNA manipulations have been described (34). Plasmids pBB105 (38) and pBR322::attP7 (C. Bauer) contain the E.coli attB and λ attP sequences (respectively) cloned into pBR322. Plasmid pLH130 was constructed using the Transformer site-directed mutagenesis kit (Clontech Laboratories Inc.) according to the manufacturer’s instructions. An oligonucleotide of sequence 5′-CAGACTACATAATACTGTAAAACACAACATAAAAAATCACTATGAATCAACTAC-3′, which hybridizes to λ attP DNA on the plasmid pBR322::attP7, introduces six dA residues (underlined) 8 bp upstream of the H1 core consensus sequence. Plasmids were sequenced using primer 1215 (New England Biolabs). Plasmids used in the circular permutation studies were constructed as follows. Plasmid pTN152 contains the ScaI–EcoRI fragment of pBend2 (39) cloned into pUC19 (40). Oligonucleotides specifying various IHF binding sites (Fig. 1; 34) were cloned into the unique HincII (SalI) site of pTN152 using linker tailing (41). Cloning into this site positions the IHF binding site so that it is flanked on both sides by direct repeats of 17 restriction endonuclease sites (39). The top strand of orientation (I) of the clones contains the binding site in the order 5′-W6N8WATCAANNNNTTR-3′ (28). Clones were sequenced using primer 1212 (New England Biolabs).

total DNA was diluted into a 20 µl reaction mix as described (42). The concentration of purified IHF used in each reaction was 0.5 µM. Samples were loaded onto a 5% polyacrylamide gel and electrophoresed at 30 mA for 2 h. The gels were stained in ethidium bromide solution and photographed using Polaroid Type 667 film. Recombination assays Assays for integrative recombination in vitro were performed as previously described (35). Supercoiled attP substrates (pBR322::attP7 or pLH130, 0.4 µg/ml, 0.1 pmol) and linear attB (pBB105, 0.2 µg/ml, 0.05 pmol) were incubated with purified Int and IHF proteins in a reaction mix (38) for 1 h and the products were electrophoresed on 1.2% agarose gels (35). Protein–DNA crosslinking analysis Fifty picomoles of H′3, H′7, H13, H17, H1AT3 and H1AT7 (Fig. 2) were radiolabeled with [γ-32P]ATP (NEN Research Products) and T4 polynucleotide kinase as described (43). The reaction of p-azidophenacyl bromide (APB) with DNA was performed essentially as described (44), except that the reaction mixture volume was 200 µl. The crosslinking reaction contained ∼0.2 pmol (0.22 µg) DNA and was performed as described (44). Following irradiation with UV light the samples were loaded onto a 12% SDS–polyacrylamide gel as described (44). Rainbow markers (Amersham) were loaded in the first lane of each gel. The concentration of purified IHF used in each reaction was 0.5 µM. PhosphorImager analysis (Molecular Dynamics) using values for integrated volumes of specific bands revealed the efficiencies of crosslinking for each of the DNA samples. RESULTS

Gel mobility shift assays

Gel mobility shift assays using circularly permuted fragments

Restriction endonuclease digestion of the pBend2 derivatives by EcoRV, BamHI or MluI generated a fragment of ∼150 bp that was used in gel mobility shift assays. The DNA fragment containing the cloned IHF binding site was not purified and therefore the vector DNA remained in the reaction mix. Approximately 4 pmol (4.8 µg)

To examine and compare the mobility of protein–DNA complexes of IHF bound to various sites, circular permutation analysis (36) was performed. IHF binding sites were cloned in both orientations into pTN152, a derivative of the pBend2 vector (39). Gel mobility shift assays were performed with different permutations of the

1782 Nucleic Acids Research, 1996, Vol. 24, No. 9

Figure 2. Duplex sequences of the derivatized strand of the oligonucleotides used in the protein–DNA crosslinking studies. An IHF consensus sequence (5′→3′) is shown at the top (10,33). The core consensus sequence is separated from the 5′-proximal DNA (dA+dT) by a spacer region (N8). The core consensus and the flanking DNA elements are underlined in each of the sites. The arrowhead indicates positions where phosphorothioates are incorporated into the DNA backbone. (A) The 3 series of binding sites derivatized in the dA+dT element. (B) The 7 series of binding sites derivatized in the N8 element.

fragment containing the binding site (Fig. 3). Restriction of the plasmid with EcoRV placed the IHF binding sites in the middle of the DNA fragment (67 bp from the left end and 54 bp from the right end). Restriction with BamHI placed the IHF binding sites at the left end of the fragment (13 bp away from the end) and restriction with MluI placed the IHF binding sites at the right end of the fragment (6 bp away from the end) (Fig. 3). Because IHF bends DNA upon binding (45,46), a fragment containing an IHF site at its direct center will migrate more slowly through polyacrylamide gels when complexed with IHF as compared with a fragment containing an IHF site at either end of the DNA fragment, because migration is dependent upon the end-to-end distance (36,47,48). To compare the mobility of complexes of IHF with binding sites containing or lacking a dA+dT element gel mobility shift assays were performed with permuted DNA fragments containing the H′, H1 and H1AT sites (Fig. 1A). The H′ site naturally contains a dA+dT sequence upstream of the core consensus element, while the H1 site does not. The H1AT site is a hybrid H′/H1 site that contains the sequence of the core consensus element of the H1 site and the DNA 5′ of the core consensus element (including the dA+dT element) of the H′ site (34). Derivatives of the pBend2 vector containing these sites in orientation (I) were digested with EcoRV, BamHI and MluI, incubated with IHF and subjected to gel electrophoresis (Fig. 4A). The mobility of the shifted DNA of each of the EcoRV-H′(I), EcoRV-H1(I) and EcoRV-H1AT(I) sites is approximately the same. However, the mobilities of the BamHI-H′(I) and BamHI-H1AT(I) fragments are approximately the same and both migrate faster than the BamHI-H1(I) fragment. Additionally, the mobilities of MluIH′(I) and MluI-H1AT(I) are approximately the same and these fragments migrate faster than the fragment containing the MluIH1(I) site.

The same analysis was performed with DNA fragments containing the H′, H1 and H1AT sites in orientation (II) (Fig. 4B). Again, the mobility of the shifted DNA of the EcoRV-H′(II), EcoRV-H1(II) and EcoRV-H1AT(II) sites is approximately the same. The mobilities of the BamHI-H′(II) and BamHI-H1AT(II) fragments are approximately the same and are faster than that of the BamHI-H1(II) fragment. The mobilities of MluI-H′(II) and MluI-H1AT(II) are approximately the same and migration is faster than the fragment containing the MluI-H1(II) site. Electrophoresis at 4
C, instead of at room temperature, had no effect on the mobilities of the protein–DNA complexes (data not shown). It is interesting to note that almost all of the DNA containing the H1AT site is bound by IHF when compared with the amount of H′ or H1 DNA bound. This indicates that IHF has a higher relative affinity for the H1AT site, which is in agreement with in vivo binding studies using challenge phages (34). In summary, a more pronounced difference in the mobilities between the H′ and H1AT sites relative to the H1 site is observed when the flanking DNA of the sites is at the extreme ends of the DNA fragment (the BamHI fragments, Fig. 4A, or the MluI fragments, Fig. 4B). To examine the effects of mutations in the dA+dT element on the mobility of IHF–H′ site complexes three single base pair mutations were made in the H′ binding site (Fig. 1B). The choice of mutations was based on previous studies of the wild-type H′ site (28) and the H′ site in the context of the λ attL arm (49). The mutations H′22T, H′22G and H′25G disrupted the binding of IHF to the H′ site alone (H′22T and H′25G; 28) or in attL (H′22G; 49) as measured in the challenge phage assay in vivo (50). The results of gel mobility shift assays with BamHI-generated fragments of the three mutant sites in orientation (II) are shown in Figure 5. Mutants H′22T and H′25G bound IHF and migration of the protein–DNA complex was like that of the complex of IHF with

1783 Nucleic Acids Acids Research, Research,1994, 1996,Vol. Vol.22, 24,No. No.19 Nucleic

1783

Figure 4. Circular permutation and gel mobility shift analysis of the H′, H1 and H1AT sites in orientation (I) (A) and orientation (II) (B). The positions of the DNA and the DNA complexed with IHF are indicated at the right of the figure. Lane 1 shows migration of the EcoRV-H′ site. Lanes 2–4 show the mobility of the EcoRV-H′, EcoRV-H1 and EcoRV-H1AT sites respectively when incubated with purified IHF protein. Lanes 5–7 show the mobility of the BamHI-H′, BamHI-H1 and BamHI-H1AT sites respectively when incubated with purified IHF protein. Lanes 8–10 show the mobility of the MluI-H′, MluI-H1 and MluI-H1AT sites respectively when incubated with purified IHF protein.

Figure 3. The circular permutation assay as utilized in this study. An IHF site is cloned into the unique HincII (SalI) site of a derivative of the pBend2 vector (39). Digestion of the plasmid with a particular endonuclease results in placement of the IHF binding site at different positions in the DNA fragment. The endonucleases used in this study were BamHI, EcoRV and MluI. Orientation (I) of IHF sites cloned into this vector places the dA+dT element closest to the end of the fragment generated by BamHI digestion. Orientation (II) places the dA+dT element closest to the end of the fragment generated by MluI digestion.

the wild-type H′ site. Binding of IHF to mutant H′22G was undetectable, suggesting that this base is also important for binding in vitro. These results indicate that the protein–DNA complex that IHF forms with the H′22T and H′25G sites has a conformation that is similar to that with the wild-type H′ site. Recombination of a chimeric substrate To examine the effects of placement of a dA+dT element in the H1 site on λ recombination a chimeric substrate containing dA residues 5′ of the H1 consensus element was constructed. In vitro integration assays were performed with this chimeric substrate and compared with assays performed with the wild-type attP substrate. The results show that the efficiency of integrative recombination is severely decreased with the chimeric substrate (Fig. 6, compare lane j with lane v). Using 2 U Int protein or varying the potassium chloride concentration in the reaction mix had no effect on the amount of chimeric recombinant product (data not shown). It is interesting to note that the amount of recombination of the wild-type substrate (lanes d–j) decreases with decreasing IHF concentration, while the amount of recombination from the chimeric substrate (lanes p–u) seems to be at a level that is independent of the concentrations of IHF protein tested. This may be due to the fact that the H1AT site has been shown to bind IHF with higher relative

Figure 5. Circular permutation and gel mobility shift analysis of the mutant H′ sites in orientation (II). The positions of the DNA and the DNA complexed with IHF are indicated at the right of the figure. Lanes 1–5 show the mobility of the BamHI-H′, BamHI-H1, BamHI-H′22T, BamHI-H′22G and BamHI-H′25G sites respectively when incubated with purified IHF protein.

affinity in vivo when compared with the H′ site (34). Consequently, this site may be tightly bound by IHF in the chimeric attP at an early step in the reaction and form a defective intasome intermediate that cannot progress along the recombination pathway. Photo-crosslinking analysis of the IHF–DNA complex To determine if IHF could form a specific protein–DNA crosslink with the DNA flanking the core consensus element of binding sites containing or lacking a dA+dT element azide-mediated crosslinking was performed. Phosphorothioate linkages were incorporated at specific positions in the H′, H1 and H1AT sites. In one series of oligonucleotides (the 7 series, Fig. 2B) two phosphorothioate linkages were incorporated in the N8 spacer region located between the core consensus and dA+dT elements. In another series of oligonucleotides (the 3 series, Fig. 2A) two phosphorothioate linkages were positioned in the dA+dT element. The phosphorothioate group locates the reaction site (51) of APB (52) with the DNA. IHF is allowed to bind the DNA duplex and upon exposure of the complex to long wavelength light (302 nm) a highly reactive

1784 Nucleic Acids Research, 1996, Vol. 24, No. 9

Figure 6. In vitro integration assays with wild-type and chimeric substrates. Lanes a–l show recombination of the wild-type attP substrate. Lanes m–x show recombination of the chimeric attP substrate. Lanes a and m contain the supercoiled attP substrate. Lanes b and n contain the linear attB substrate. Lanes c and o contain the reaction mix incubated with the DNA substrates. Lanes d–l and p–x contain the reaction mix incubated with the DNA substrates, with 1 U Int protein and successive dilutions of purified IHF protein as follows: 1:10 (lanes d and p); 1:20 (lanes e and q); 1:30 (lanes f and r); 1:40 (lanes g and s); 1:50 (lanes h and t); 1:75 (lanes i and u); 1:100 (lanes j and v); 1:200 (lanes k and w); 1:500 (lanes l and x). The DNA bands corresponding to attP (sc, supercoiled; nc, nicked), attB and the recombinant product (rec) are indicated on the right of the figure.

nitrene group is generated (53) that immediately forms covalent crosslinks with relatively low chemical specificity for proteins (54). Figure 7 shows the crosslinked complexes formed by the interaction of IHF with the 7 series of binding sites (the H′7, H17 and H1AT7 duplexes, Fig. 2B). Protein–DNA crosslinks do not form when the derivatized IHF binding sites are incubated with 0.5 µg/ml BSA, thereby showing the specificity of the reaction. However, IHF forms a specific protein–DNA crosslink with the H′, H1 and H1AT sites when the DNA is derivatized in the N8 spacer region. Figure 8 shows the results of an experiment performed when the DNA is derivatized within the dA+dT element (the H′3, H13 and H1AT3 duplexes, Fig. 2A). When IHF is added to the DNA, a specific protein–DNA crosslink is formed with the H′3, H13 and H1AT3 sites. When BSA alone is incubated with the respective derivatized duplexes, no crosslinked complex is formed. Taken together these results indicate that amino acids of IHF protein are in proximity to the derivatized bases within both the N8 spacer region and the DNA 5′ of this region, regardless of their nucleotide sequence. Identical controls were performed for all sites in both series (data not shown) and the controls confirmed the following. IHF forms a specific protein–DNA crosslink that is dependent upon IHF and irradiation to generate an available reactive nitrene. Formation of the crosslinked complex can be prevented by pre-incubation of IHF with a 100-fold excess of a specific unlabeled DNA duplex containing an IHF site (the H1AT7 site; Fig. 2B), but not with a 500-fold excess of a DNA duplex that lacks an IHF site (X1X2F, oligonucleotides 1 and 2 containing binding sites for the λ Xis and E.coli FIS proteins only; 55). Also, the complex is susceptible to a 30 min digestion with 5 mg/ml proteinase K. Using APB in a similar manner, other groups reported crosslinking efficiencies of 5–20% (44,56,57). In the experiments presented here, the crosslinking efficiencies ranged from 8.5 to 19%, with an average efficiency of 13% for the six sites. DISCUSSION Previous studies of full-length, truncated and hybrid H′ and H1 sites from attP of bacteriophage λ suggest that the DNA flanking an IHF

Figure 7. Photo-crosslinking analysis of IHF with the H′7, H17 and H1AT7 sites containing APB moieties incorporated into the N8 spacer region of the binding site (Fig. 2B). Migration of molecular weight markers of the designated size are indicated by arrowheads on the left of the figure. Migration of the uncoupled DNA duplex (UC), coupled DNA duplex (C) and the IHF–DNA crosslinked complex (XL) are indicated by arrows on the right of the figure. Identity of the 32P-labeled DNA duplexes is indicated at the top of the figure. The DNA duplexes were incubated alone (–) (lanes 1, 4 and 7) or with IHF (lanes 2, 5 and 8) or BSA (lanes 3, 6 and 9) as indicated by (+) at the bottom of the figure.

core consensus element is important in the in vivo binding of this protein to its sites (28,34,35). Using the challenge phage assay (50) we found that the H1 core consensus element alone (WATCAANNNNTTR; 10) was sufficient for IHF binding to the H1 site, but that artificial placement of a dA+dT element upstream of the core consensus significantly increased binding of both wild-type and mutant IHF proteins (34,35). Binding of IHF to the H′ site, however, required intact core consensus and dA+dT elements (28,34). IHF has been reported to bend DNA at both the H′ and H1 sites at an angle greater than 140
(46). However, as described above, sequence requirements for IHF binding to these two sites differ: a dA+dT element is required for IHF to bind the H′ site. Does this flanking DNA play a role in the DNA bend induced by IHF? It is possible that the conformations of the IHF–DNA complexes with the H′ and H1 sites are different. Further examination of the contribution of the flanking DNA to the conformation of the protein–DNA complexes formed by IHF at these sites was attempted in this study using circular permutation, recombination with a chimeric substrate and protein–DNA crosslinking analyses. To investigate the influence of the dA+dT element on the mobility of an IHF–DNA complex, circular permutation analysis (36) was performed. The results showed that the complexes of IHF with the H′ and H1AT sites containing the dA+dT element migrated faster than those complexes with the H1 site lacking the dA+dT element when this DNA was at the extreme end of the DNA fragment. This suggests that IHF bends sites containing the dA+dT element more than a site that does not. A similar effect was observed with CAP protein binding to ends of wild-type and mutant DNA fragments. It was proposed that an end-bound CAP complex with mutant DNA had a slower mobility because the DNA is bent to a lesser extent than it is in a wild-type CAP–DNA complex. When CAP binds at the end of a fragment, the less bent complex with mutant DNA has more frictional drag than the more bent wild-type complex, which reduces its mobility (58). The IHF–DNA effect is not due to contacts specified in the core consensus element, because the H1AT site contains the core consensus sequence of the H1 site. Because the 5′-proximal region of the H1 site is not protected from nuclease attack in

1785 Nucleic Acids Acids Research, Research,1994, 1996,Vol. Vol.22, 24,No. No.19 Nucleic

Figure 8. Photo-crosslinking analysis of IHF with the H′3, H13 and H1AT3 sites containing APB moieties incorporated into the dA+dT region of the binding site (Fig. 2A). Migration of molecular weight markers of the designated size are indicated by arrowheads on the left of the figure. Migration of the uncoupled DNA duplex (UC), coupled DNA duplex (C) and the IHF–DNA crosslinked complex (XL) are indicated by arrows on the right of the figure. Identity of the 32P-labeled DNA duplexes is indicated at the top of the figure. The DNA duplexes were incubated alone (–) (lanes 1, 4 and 7) or with IHF (lanes 2, 5 and 8) or BSA (lanes 3, 6 and 9) as indicated by (+) at the bottom of the figure.

footprinting studies (30,31), it is possible that the DNA flanking the core consensus element of the H1 site is not contacting the protein. Alternatively, IHF could be making transient contacts with this flanking sequence. In either case the DNA at the end of the H1 fragment would retard the complex more in the gel matrix, resulting in decreased mobility. In contrast, when a dA+dT tract is present at the end of the fragment, IHF may make an additional, more stable protein–DNA contact with the flanking DNA. Alternatively, the dA+dT element and neighboring sequences could form a DNA structure, such as a kink, that moves that region of DNA closer to the protein. Either of these possibilities could result in an end-bound protein–DNA complex with increased mobility in polyacrylamide gels. Because the IHF bend is so substantial, the end-to-end distance may not solely govern mobility when the site is in the middle of the DNA fragment (59,60) and the frictional drag may be insignificant or undetectable when the A tract region is internal to the DNA fragment. Therefore, the protein–DNA complexes of IHF with the EcoRV fragments containing the three sites in the middle all migrate with approximately the same mobility. We performed circular permutation analysis of H′ sites containing single base pair mutations in the dA+dT element to eliminate interactions that might be occurring between IHF and the dA+dT element. If so, these mutant sites would form complexes with IHF that would have a mobility more like that of the IHF–H1 complex. The results show that the mobilities of complexes of IHF with two of the mutant sites were the same as those of complexes of IHF with the wild-type H′ site, indicating that the conformation of the complex of IHF with the mutant sites is similar to that of IHF with the site lacking the mutations. These results suggest that, in addition to the dA+dT element, other factors are involved in the deformation of DNA by IHF. To further pursue the hypothesis that IHF bends sites containing a dA+dT element more than a site lacking it we took advantage of the role of IHF in λ site-specific recombination. The presence of an altered bend at a hybrid IHF site might disrupt the structure of the intasome formed. Thus a chimeric attP substrate was constructed and this substrate was examined in integrative recombination assays.

1785

The results showed that recombination of this substrate was severely decreased in vitro when compared with the wild-type substrate. This result indicates that the nucleoprotein structure formed by the chimeric substrate may lack the precise geometry required for efficient integrative recombination. Such an altered conformation could have at least two causes. The IHF-induced bend at the H1 site in the chimeric substrate might not be of the correct rotational angle. Alternatively, even though the dA·dT residues were placed at the position that corresponds to their location in the H′ site, the phasing of the IHF-induced bend may be changed in this substrate. In either case a less proficient intasome would be formed. The importance of the preciseness of this protein–DNA interaction in recombination is further exemplified by the finding that a substrate containing an H′ site whose position has been shifted 1 bp to the right or left had a severely decreased efficiency of recombination (61). Because the simple 5′-proximal addition of a dA+dT element to the H1 site disrupts recombination, IHF sites are not simply particular combinations of sequence elements. Each IHF–DNA complex may have a unique overall structure. In an attempt to determine if IHF is contacting DNA that is 5′-proximal to the core consensus element azide-mediated crosslinking experiments were performed. Yang and Nash (1994) reported that specific crosslinks occur with IHF and DNA that is derivatized in the dA+dT element or the N8 spacer region of the H′ site. Based on the conclusions of these workers phosphorothioate groups were placed at the same positions in the DNA backbone of the H′ site and at equivalent positions in the H1 and H1AT sites. The results show that IHF forms a specific protein–DNA crosslink when the DNA of the H′, H1 or H1AT sites are derivatized in the N8 spacer region and the DNA 5′ of this region. Therefore, IHF can form a specific protein–DNA crosslink with DNA flanking the core consensus whether or not it contains a dA+dT element. This finding indicates that the DNA that flanks the core consensus element of each of the sites (regardless of its DNA sequence) is close to the IHF protein. Given the results from the circular permutation analysis and the recombination assays it was surprising that the DNA flanking the core consensus element of the H1 site crosslinked with an efficiency comparable with that of the H′ and H1AT sites. This is the first evidence that the upstream DNA of the H1 site indeed plays a role in bending of DNA by IHF. In conclusion, the experiments presented here suggest a direct and important role for the sequences flanking the core consensus element in the bending of DNA by IHF. Some binding sites show a perfect match to the consensus, yet are bound by IHF weakly, if at all. Conversely, sites that have weak core consensus elements have been found to be good IHF sites (27,29,32,62–64), suggesting that flanking DNA is important. The majority of IHF sites lack a discreet dA+dT element 5′ of the core consensus, but many IHF sites occur in AT-rich regions, such as promoters. For CAP DNA binding and bending AT-rich dinucleotides flanking the binding site are optimal (65). A similar situation may also exist for IHF. Perhaps direct hydrogen bond contacts are made with bases in the WATCAA and TTR elements and indirect contacts occur with the phosphate backbone of the flanking DNA to contribute to the overall affinity and bending. Base analog studies suggest that no base-specific contacts occur between IHF and the dA+dT element (66). However, other studies using crosslinking (44) and chemical probes (D. Sun,, L. H. Hurley and R. Harshey, personal communication) implicate this 5′-proximal region as important for DNA binding to the H′ site. It is clear that DNA structure plays a role in the formation of numerous protein–DNA complexes (67,68). The

1786 Nucleic Acids Research, 1996, Vol. 24, No. 9 narrow minor groove characteristic of A tracts (69–71) may provide an additional recognition element for IHF. Precisely how the flanking DNA influences the binding and bending of a site by IHF remains to be elucidated, but this study suggests that the sequences 5′-proximal to an IHF site play an important role in IHF action. ACKNOWLEDGEMENTS We would like to thank C. Robertson in the laboratory of H. Nash for the gifts of purified Int and IHF proteins, S.-W. Yang for the crosslinking protocol, A. Burgin, R. Ebright, M. Fried and S.-W. Yang for helpful suggestions and S. Goodman, H. Nash and M. Werner for comments on the manuscript. This work was supported by NIH grant GM28717. REFERENCES 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31

Mitchell,P.J. and Tjian,R. (1989) Science, 245, 371–378. Echols,H. (1990) J. Biol. Chem., 265, 14697–14700. Crothers,D.M. (1993) Curr. Biol., 3, 675–676. van der Vliet,P.C. and Verrijzer,C.P. (1993) BioEssays, 15, 25–32. Serrano,M., Salas,M. and Hermoso,J.M. (1993) Trends Biochem. Sci., 18, 202–206. Grosschedl,R., Giese,K. and Pagel,J. (1994) Trends Genet., 10, 94–100. Wolffe,A.P. (1994) Science, 264, 1100–1101. Travers,A.A., Ner,S.S. and Churchill,M.E.A. (1994) Cell, 77, 167–169. Grosschedl,R. (1995) Curr. Opin. Cell Biol., 7, 362–370. Friedman,D.I. (1988) Cell, 55, 545–554. Freundlich,M., Ramani,N., Mathew,E., Sirko,A. and Tsui,P. (1992) Mol. Microbiol., 6, 2557–2563. Goosen,N. and van de Putte,P. (1995) Mol. Microbiol., 16, 1–7. Nash,H.A. (1996) In Lin,E.C.C. and Lynch,A.S. (eds), Regulation of Gene Expression in Escherichia coli. R.G.Landes Co., Austin, TX, in press. Hales,L.M., Gumport,R.I. and Gardner,J.F. (1996) in preparation. Better,M., Lu,C., Williams,R.C. and Echols,H. (1982) Proc. Natl. Acad. Sci. USA, 79, 5837–5841. Better,M., Wickner,S. Auerbach,J. and Echols,H. (1983) Cell, 32, 161–168. Pollock,T.J. and Nash,H.A. (1983) J. Mol. Biol., 170, 1–18. Richet,E., Abcarian, P. and Nash,H.A. (1986) Cell, 46, 1011–1021. Richet,E., Abcarian,P. and Nash,H.A. (1988) Cell, 52, 9–17. Goodman,S.D. and Nash,H.A. (1989) Nature, 341, 251–254. Goodman,S.D., Nicholson,S.C. and Nash,H.A. (1992) Proc. Natl. Acad. Sci. USA, 89, 11910–11914. Giese,K., Cox,J. and Grosschedl,R. (1992) Cell, 69, 185–195. Perez-Martin,J., Timmis,K.N. and de Lorenzo,V. (1994) J. Biol. Chem., 269, 22657–22662. Molina-Lopez,J.A., Govantes,F. and Santero,E. (1994) J. Biol. Chem., 269, 25419-25425. Hayes,F. and Austin,S. (1994) J. Mol. Biol., 243, 190–198. Kurumizaka,H., Kanke,F., Matsumoto,U. and Shindo,H. (1992) Arch. Biochem. Biophys., 295, 297–301. Prentki,P., Chandler,M. and Galas,D.J. (1987) EMBO J., 6, 2479–2487. Lee,E.C., MacWilliams,M.P., Gumport,R.I. and Gardner,J.F. (1991) J. Bacteriol., 173, 609–617. van Rijn,P.A., van de Putte,P. and Goosen,N. (1991) Nucleic Acids Res., 19, 2825-2834. Craig,N. and Nash,H.A. (1984) Cell, 39, 707–716. Yang,C.-C. and Nash,H.A. (1989) Cell, 57, 869–880.

32 Kur,J., Hasan,N. and Szybalski,W. (1989) Gene, 81, 1–15. 33 Goodrich,J.A., Schwartz,M.L. and McClure,W.R. (1990) Nucleic Acids Res., 18, 4993-5000. 34 Hales,L.M., Gumport,R.I. and Gardner,J.F. (1994) J. Bacteriol., 176, 2999–3006. 35 Hales,L.M., Gumport,R.I. and Gardner,J.F. (1994) Biochimie, 76, 1030–1040. 36 Wu,H.-M. and Crothers,D.M. (1984) Nature, 308, 509–513. 37 Lee,E.C., Gumport,R.I. and Gardner,J.F. (1990) J. Bacteriol., 172, 1529–1538. 38 Nash,H.A. (1983) Methods Enzymol., 100, 210–216. 39 Kim,J., Zweib,C., Wu,C. and Adhya,S. (1989) Gene, 85, 15–23. 40 Numrych,T.E. (1992) PhD Thesis, University of Illinois. 41 Bass,S., Sugiono,P., Arvidson,D.N., Gunsalus,R.P. and Youderian,P. (1987) Genes Dev., 1, 565–572. 42 Lee,E.C., Hales, L.M., Gumport,R.I. and Gardner,J.F. (1992) EMBO J., 11, 305–313. 43 Maniatis,T., Fritsch,E.F. and Sambrook,J. (1982) Molecular Cloning: A Laboratory Manual. Cold Spring Harbor Laboratory Press, Cold Spring Harbor, NY. 44 Yang,S.-W. and Nash,H.A. (1994) Proc. Natl. Acad. Sci. USA, 91, 12183–12187. 45 Robertson,C.A. and Nash,H.A. (1988) J. Biol. Chem., 263, 3554–3557. 46 Thompson,J.F. and Landy,A. (1988) Nucleic Acids Res., 16, 9687–9705. 47 Lerman,L. and Frisch,H.L. (1982) Biopolymers, 21, 995–997. 48 Lumpkin,O.J. and Zimm,B.H. (1982) Biopolymers, 21, 2315–2316. 49 MacWilliams,M.P. (1992) PhD Thesis, University of Illinois. 50 Benson,N., Sugiono,P., Bass,S., Mendelman,L.V. and Youderian,P. (1986) Genetics, 114, 1–14. 51 Fidanza,J.A., Ozaki,H. and McLaughlin,L.W. (1992) J. Am. Chem. Soc., 114, 5509-–5517. 52 Hixson,S.H. and Hixson,S.S. (1975) Biochemistry, 14, 4251–4254. 53 Knowles,J.R. (1972) Acc. Chem. Res., 5, 155–160. 54 Bayley,H. and Staros,J.V. (1984) In Scriven,E.F.V. (ed.), Azides and Nitrenes: Reactivity and Utility. Academic Press, Orlando, FL, pp. 433–490. 55 Numrych,T.E., Gumport,R.I. and Gardner,J.F. (1991) J. Bacteriol., 173, 5954–5963. 56 Baker,T.A. and Mizuuchi,K. (1992) Genes Dev., 6, 2221–2232. 57 Mayer,A.N. and Barany,F. (1995) Gene, 153, 1–8. 58 Liu-Johnson,H.-N., Gartenberg,M.R. and Crothers,D.M. (1986) Cell, 47, 995–1005. 59 Koo,H.S. and Crothers,D.M. (1988) Proc. Natl. Acad. Sci. USA, 85, 1763–1767. 60 Zinkel,S.S. and Crothers,D.M. (1990) Biopolymers, 29, 29–38. 61 Nunes-Duby,S.E., Smith-Mungo,L.I. and Landy,A. 1995. J. Mol. Biol., 253, 228–242. 62 Leong,J.M., Nunes-Duby,S., Lesser,C.F., Youderian,P., Susskind,M.M. and Landy,A. (1985) J. Biol. Chem., 260, 4468–4477. 63 Gamas,P., Chandler,M.G., Prentki,P. and Galas,D.J. (1987) J. Mol. Biol., 195, 261–272. 64 Giladi,H., Gottesman,M. and Oppenheim,A.B. (1990) J. Mol. Biol., 213, 109–121. 65 Gartenberg,M.R. and Crothers,D.M. (1988) Nature, 333, 824–829. 66 Wang,S., Cosstick,R., Gumport,R.I. and Gardner,J.F. (1995) Biochemistry, 34, 13082–13090. 67 Travers,A.A. (1989) Annu. Rev. Biochem., 58, 427–452. 68 Harrington,R.E. (1992) Mol. Microbiol., 6, 2549–2555. 69 Nelson,H.C.M., Finch,J.T., Luisi,B.F. and Klug,A. (1987) Nature, 330, 221–226. 70 Yoon,C., Prive, G.G., Goodsell,D.S. and Dickerson,R.E. (1988) Proc. Natl. Acad. Sci. USA, 85, 6332–6336. 71 Katahira,M., Sugeta,H. and Kyogotu,K. (1990) Nucleic Acids Res., 18, 613–618.