Binding to four-way junction DNA: a common property of architectural proteins?

^a Department of Biochemistry and Biophysics, Oregon State University, Corvallis, Oregon 97331–7305, USA
^b Institute of Genetics, Bulgarian Academy of Sciences, 1113 Sofia, Bulgaria

	ABSTRACT

TOP ABSTRACT INTRODUCTION STRUCTURE OF THE FOUR-WAY... THE HMG1 BOX PROTEINS THE HMGI-Y PROTEIN FAMILY THE WINGED HELIX PROTEINS SWI/SNF: A CHROMATIN REMODELING... PROKARYOTIC 4WJ BINDING PROTEINS MODES OF PROTEIN BINDING... EVIDENCE FOR AN ARCHITECTURAL... POSSIBLE IN VIVO FUNCTION... REFERENCES

 
Proteins that can be shown to strongly bind in vitro to the four-way (Holliday) junction DNA include not only the obvious candidates such as enzymes involved in recombination, but also a remarkably diverse group of seemingly unrelated proteins. These include the HMG1 box proteins, members of the HMGI-Y family, winged helix proteins (including linker histones), the SWI/SNF complex, and some totally unrelated prokaryotic proteins. What these proteins seem to share is a propensity to bind to bent DNA, to bend DNA upon binding, and/or to preferentially interact with DNA crossings. Thus, they appear to be, in the main, architectural proteins, although some (like the SWI/SNF complex) have very specific functional roles as well. Perhaps because they bind to or promote the formation of particular DNA structures, the four-way junction binding proteins are frequently interchangeable in cellular function. Furthermore, since a given kind of structure can be recognized by many different protein motifs, it is not surprising that apparently unrelated proteins can fall into such a single functional class.—Zlatanova, J., van Holde, K. Binding to four-way junction DNA: a common property of architectural proteins? FASEB J. 12, 421–431 (1998)

Key Words: HMG box • HU • IHF • linker histones • SWI/SNF

	INTRODUCTION

TOP ABSTRACT INTRODUCTION STRUCTURE OF THE FOUR-WAY... THE HMG1 BOX PROTEINS THE HMGI-Y PROTEIN FAMILY THE WINGED HELIX PROTEINS SWI/SNF: A CHROMATIN REMODELING... PROKARYOTIC 4WJ BINDING PROTEINS MODES OF PROTEIN BINDING... EVIDENCE FOR AN ARCHITECTURAL... POSSIBLE IN VIVO FUNCTION... REFERENCES

 
IN RECENT YEARS it has been observed that a remarkable variety of structurally and functionally unrelated DNA binding proteins bind with strong preference to the particular DNA structure called a four-way junction (4WJ).² The four-way (Holliday) junction DNA was initially proposed as the central intermediate in homologous genetic recombination; it is also involved in replication-related events (1). Both the in vivo formation of 4WJs and their resolution into two duplexes in recombination are the result of the action of specific enzymes (2). These enzymes will not be considered here. We will, instead, focus on proteins of nonenzymatic character that have recently been shown to preferentially bind 4WJ in vitro. These proteins have no relationship to either the formation of 4WJs (as in recombination events) or the known biological functions of these structures. Why, then, should they share this unusual propensity? We would like to present the thesis that these proteins, diverse as they may be, all perform a particular kind of architectural role in vivo. Their in vitro preference to 4WJ structures may, in fact, reflect their common ability to bend, wrap, cross, and supercoil DNA as part of their involvement in a wide variety of otherwise unrelated in vivo functions. Remarkably, although the normal functions of these proteins are so varied, it is often found that they may substitute for one another in most unexpected ways. In what follows, we will first classify the multitude of 4WJ binding proteins into groups based on structural similarities of their DNA binding domains and common features of their interaction with DNA. Then, we will briefly review evidence for the architectural roles of many of these proteins. Finally, we will describe several possible mechanisms of action of such architectural proteins and show how each accomplishes the same purpose.

	STRUCTURE OF THE FOUR-WAY JUNCTION

TOP ABSTRACT INTRODUCTION STRUCTURE OF THE FOUR-WAY... THE HMG1 BOX PROTEINS THE HMGI-Y PROTEIN FAMILY THE WINGED HELIX PROTEINS SWI/SNF: A CHROMATIN REMODELING... PROKARYOTIC 4WJ BINDING PROTEINS MODES OF PROTEIN BINDING... EVIDENCE FOR AN ARCHITECTURAL... POSSIBLE IN VIVO FUNCTION... REFERENCES

 
Although its detailed structure has not yet been reported from X-ray or nuclear magnetic resonance (NMR) studies, a variety of other experimental approaches and molecular modeling studies have provided a generally accepted picture of the 4WJ. Elegant electrophoretic studies and fluorescence energy transfer experiments, when carried out in solutions containing at least 0.1 mM divalent cations, both argue for the kind of crossed structure diagrammed in Fig. 1 (1). In such structures, two of the strands form continuous B-helices; the other two make sharp bends at the point of pair interchange. Srinivasan and Olson (3) have carried out computer modeling of 4WJs that predicts a range of acceptable structures, including the antiparallel stacked X-forms described above. They demonstrate that a smooth conversion from parallel to antiparallel structures is possible via a 90° crossing intermediate.

View larger version (48K): [in this window] [in a new window]   Figure 1. A ribbon model of the high-salt, stacked X structure of the four-way junction DNA (face view). Figure courtesy of D. M. J. Lilley, reproduced with permission.

The structure of the 4WJ at low ionic strength is not fully understood. Under these conditions the arms of the junction are maximally separated, probably forming a flat, unstacked conformation (1). In any event, since virtually all biological processes occur in the presence of significant concentrations of divalent cations, it is the stacked X structure that should be of major interest to molecular biologists. A very clear description of this form and its conformers is provided in a recent review by Lilley (4). The ease with which the angle between the stacked X arms may change in its interactions with proteins may explain why it can accommodate such a wide variety of interactions (see below).

	THE HMG1 BOX PROTEINS

TOP ABSTRACT INTRODUCTION STRUCTURE OF THE FOUR-WAY... THE HMG1 BOX PROTEINS THE HMGI-Y PROTEIN FAMILY THE WINGED HELIX PROTEINS SWI/SNF: A CHROMATIN REMODELING... PROKARYOTIC 4WJ BINDING PROTEINS MODES OF PROTEIN BINDING... EVIDENCE FOR AN ARCHITECTURAL... POSSIBLE IN VIVO FUNCTION... REFERENCES

 
The HMG1 DNA binding domain was first identified in the abundant nonhistone chromatin protein HMG1, where it is present as two closely related structured motifs, called HMG1 boxes A and B. Each box contains about 80 amino acid residues, which fold into a characteristic 3-dimensional, L-shaped structure (reviewed in ref 5) ( Fig. 2A). The longer arm of the `L' is formed by the extended amino-terminal peptide packed against the helical carboxyl-terminal region, while the shorter arm comprises the two central -helices. The two arms form an angle of about 70–80°, which is maintained by a hydrophobic core of conserved amino acid residues situated at the apex of the `L'.

View larger version (40K): [in this window] [in a new window]   Figure 2. A) NMR solution structure of the HMG1 box A of rat HMG1 (9). Figure courtesy of J. O. Thomas, reproduced with permission. B) Solution structure of a complex of the LEF-1 HMG1 box and an adjacent basic region with its cognate DNA (14). The HMG1 box is represented by a ribbon. Figure courtesy of P. E. Wright.

Regions homologous to the HMG1 boxes have been found in a multitude of other proteins in a wide variety of eukaryotic organisms. Most of these proteins are either known or suspected transcription factors (6, 7; for subclassification of proteins belonging to the HMG1 superfamily, see ref 8). A loose `signature' sequence has been defined for the HMG1 box, characterized mainly by conservation in the position and spacing of the aromatic amino acid residues tyrosine, tryptophan, and phenylalanine. The boxes whose structures have been studied to date have 3-dimensional conformations remarkably similar to those of the boxes of HMG1. Proteins containing the box display a very broad range of sequence specificity in DNA binding, ranging from the nonspecific HMG1 and 2, through those that can be footprinted on certain DNA sequences but whose protected sites do not have a recognizable consensus sequence (e.g., UBF, mtTF, HM), to those that possess a well-defined consensus sequence (e.g., SRY, LEF-1). Even in the latter group, the sequence specificity is fairly low compared with most transcription factors. This lower specificity is not surprising since transcription factors typically bind in the major groove of DNA, whereas interaction of the HMG1 box with DNA is through contacts with the minor groove on one side of the duplex (9). The minor groove provides more limited opportunity for sequence-specific recognition (10). The TATA binding protein TBP also interacts with the minor groove of the TATA box and shows strong affinity but limited sequence specificity, about 1000-fold lower than that of a typical sequence-specific major groove binding protein (11).

How can DNA binding domains of almost identical 3-dimensional fold display such enormous variability in sequence specificity? The main reason could be differences in particular amino acid residues in the subdomains of the box. Arm-swap experiments have identified the importance of the long arm of the box in sequence recognition (12). The presence of proline in position 7 and serine in position 12 in the non-sequence-specific boxes can be contrasted to a hydrophobic (valine or isoleucine) residue in position 7 and an asparagine residue in position 12 in the sequence-specific boxes. Indeed, Asn¹² has been shown by NMR to make direct contacts with DNA in the hSRY–DNA complex (13).

Recent studies have demonstrated contributions from carboxy-terminal basic extensions to the minimal HMG1 box both in enhancing DNA binding and bending (see below) and to sequence selectivity. In accordance with biochemical results, the solution structure of a complex between a fragment of LEF-1 comprising the HMG1 domain and an adjacent basic region with its cognate DNA (14) ( Fig. 2B) has indicated that a kink in helix 3 (the helix involved in forming the longer arm of the `L') allows the carboxy-terminal region to make extensive contacts with DNA, including insertion of Tyr⁷⁵ deep into the minor groove and interactions of the adjacent basic amino acids with the sugar-phosphate backbone. The kink in helix 3 is caused by Pro⁶⁷, which is invariant in sequence-specific binders but absent from the subfamily lacking sequence specificity in DNA binding. The basic region itself is also missing in many members of this subfamily.

The non-sequence-specific members of the HMG1 box superfamily and the sequence-specific binders both appear to recognize structural features of DNA, for they all bind to distorted DNA structures (4WJs, cis-platin-modified DNA, etc.) and/or bend the duplex DNA upon binding ( Table 1). In most cases studied, the nonspecific binders also induce DNA looping and change the superhelicity of topologically constrained DNA molecules ( Table 1). To our knowledge, looping has not been directly demonstrated for any sequence-specific members of the family. It is intriguing that when the SRY HMG1 box was tested in a supercoiling assay, it failed (in contrast to boxes from HMG1) to induce supercoiling (15). The reason for this result is unclear.

We have quite detailed information as to how the HMG1 box interacts with DNA. This comes from many mutational studies, domain-swap experiments, and, most important, NMR analyses. The proteins induce a kink in the DNA and bind via the concave face of the L-shaped protein structure on the outside of the DNA kink ( Fig. 2B).

How does the HMG1 box interact with prebent DNA, in particular with 4WJ DNA? Some information has come from mutational analysis of HMG1 boxes and the use of natural SRY mutants causing sex reversal (SRY is the genetic `master switch' for testis development in mammals). Mutations of a highly conserved tryptophan residue W48 in the A domain of HMG1, which is believed to stabilize the 3-dimensional fold of the L-shaped molecule, largely disrupted the tertiary structure but did not abolish preferential binding to 4WJ (16). This suggested that 4WJ-specific binding might reside in the amino-terminal extension of the long arm. However, mutations of two proline residues, P5 and P8, thought to contribute to the extended conformation of the amino terminus did not abolish 4WJ binding either. Studies of the Ile to Thr substitution at position 68 (close to the amino-terminal end of helix I) that occurs in spontaneous mutants of SRY indicate that the isoleucine side chain in the native protein inserts between specific base pairs in duplex DNA, inducing a bend in straight DNA. Analogous side chain insertion also occurs upon binding to 4WJ DNA, establishing a shared mechanism for sequence- and structure-specific binding (17), in accordance with views expressed by Thomas and colleagues (9). Although this specific contact occurs in both types of DNA binding, it is not required for binding to the intrinsically bent 4WJ DNA. Since the L-shaped HMG1 box must bend around the DNA, we suggest that it may bind to 4WJ as shown in Fig. 3C.

View larger version (35K): [in this window] [in a new window]   Figure 3. Exterior and interior modes of binding of proteins to DNA. A) Exterior mode, with the protein reaching over the outside of the DNA bend. B) Interior mode, with the protein interacting on the concave surface of the DNA bend. C) A possible conformation for the exterior binding mode to 4WJ DNA. D) A possible conformation for the interior binding mode to 4WJ DNA.

Among the group of binders possessing an intermediate degree of sequence specificity, UBF (upstream binding factor) deserves special attention in view of its known function and suggested mechanism of action. UBF is a transcription factor involved in Pol I-mediated transcription of eukaryotic ribosomal genes. Depending on the species of origin, UBF can contain from four to six HMG1 boxes. UBF binds to two precisely spaced sequence elements—the upstream control element and the core promoter element—so as to allow binding of the selectivity factor 1 (SL1) and recruit Pol I to the preinitiation complex (for a review, see ref 18).

UBF has been footprinted to various probes, but no consensus sequence could be deduced either from comparisons of the footprinted regions or by random selection strategies. In vitro DNA binding assays showed that it interacts preferentially with 4WJ DNA, bends linear DNA to mediate circularization of small DNA fragments, and loops DNA in the rDNA promoter region to form an `enhancesome'. In this structure, an UBF dimer bends the DNA duplex to form a near 360° loop; two such UBF-created loops have been hypothesized to exist in the promoter region (19), thus bringing the upstream control region and the core element into a close juxtaposition to allow cooperative binding of two SL1 complexes (see Fig. 4C). The exact mechanism whereby UBF forms the enhancesome is still under debate. Although the ligation-mediated circularization and supercoiling could be induced by truncated versions of UBF containing only the dimerization domain and the first HMG1 box, interaction of HMG1 boxes 1, 2, and 3 of the UBF dimer with DNA might be necessary to form the full 360° loops present in the enhancesome. Each box may introduce an independent kink by binding to the minor groove on the outside of the kink, as is the case for the A domain of HMG1 or the HMG1 boxes of LEF-1 and SRY (see above). Appropriate phasing of these kinks may induce the overall DNA looping in the enhancesome (19).

View larger version (70K): [in this window] [in a new window]   Figure 4. Possible mechanisms of action of architectural proteins. A) Formation of a specific protein–DNA complex, illustrated by the example of phage intasome. Integrase binds to its two nonadjacent binding sites by virtue of the DNA bend caused by the binding of IHF to its specific binding site at the attL locus of phage genome. Redrawn after Grosschedl (40). B) Formation of a specific protein–protein complex on the enhancer region of the TCR- gene. Interaction among the various transcription factors on the enhancer is made possible by the DNA bending caused by the binding of the site-specific architectural protein LEF-1. In panels A and B, binding of the two elements should be cooperative. C) Juxtaposition of defined DNA promoter elements. In the structure of the Pol I promoter, the upstream control element is brought into proximity with the core element in order to provide a surface for binding of SL1 complexes to these two promoter elements (19). This structure is created by binding of two UBF dimers to the DNA, the first three HMG1 boxes of each UBF monomer each inducing an independent kink in the DNA; the kinks are phased so as to induce in-phase bending. Figure courtesy of T. Moss, reproduced with permission. D) Straightening out of intrinsic DNA curvature to facilitate protein–protein interaction in a functional nucleoprotein complex (47). The intrinsic curvatures of the positive regulatory domains (PRD) II and IV of the INF-ß enhancer are illustrated on the left. These curvatures are counteracted by the binding of NF-B p50/p65, ATF-2/c-Jun, and HMGI-Y to these sites: the curvature at PRD IV is straightened whereas that at PRD II is actually reversed. This change in enhancer architecture is suggested to orient transcription factors bound to the enhancer in a way favoring their interaction. The activation domains of NF-B p65 and ATF-2/c-Jun are represented as ovals tethered to the DNA binding domains. The role of HMGI-Y is to further modulate DNA structure and interact with the transcription factors. Redrawn from Falvo et al. (47).

	THE HMGI-Y PROTEIN FAMILY

TOP ABSTRACT INTRODUCTION STRUCTURE OF THE FOUR-WAY... THE HMG1 BOX PROTEINS THE HMGI-Y PROTEIN FAMILY THE WINGED HELIX PROTEINS SWI/SNF: A CHROMATIN REMODELING... PROKARYOTIC 4WJ BINDING PROTEINS MODES OF PROTEIN BINDING... EVIDENCE FOR AN ARCHITECTURAL... POSSIBLE IN VIVO FUNCTION... REFERENCES

 
The HMGI-Y protein family proper (reviewed in ref 5) consists of three presently known members: HMGI and HMGY are identical in sequence except for a small internal deletion in the latter produced by alternative splicing, and HMGI-C is a closely related protein (~50% homology) encoded by a separate gene. The members of this family are not related to HMG1 and 2, as one might expect on the basis of the similarity in their names, which came from early observations that these proteins all had high mobilities on electrophoretic gels.

The DNA binding domains, also known as `A·T-hook' motifs, have been shown to preferentially bind to A·T-DNA via minor groove contacts. Proteins containing sequences similar to the A·T-hook are found in numerous other proteins from a wide variety of eukaryotic organisms, most of which are suspected of being transcription factors. The mode of interaction of these proteins with 4WJ DNA has been clarified in recent footprinting studies by Hill and Reeves (20). It is demonstrated that HMGI-Y binds at the junction crossing. From the contacts found on the arms, the protein must wrap around the cross in much the manner suggested in Fig. 3C.

	THE WINGED HELIX PROTEINS

TOP ABSTRACT INTRODUCTION STRUCTURE OF THE FOUR-WAY... THE HMG1 BOX PROTEINS THE HMGI-Y PROTEIN FAMILY THE WINGED HELIX PROTEINS SWI/SNF: A CHROMATIN REMODELING... PROKARYOTIC 4WJ BINDING PROTEINS MODES OF PROTEIN BINDING... EVIDENCE FOR AN ARCHITECTURAL... POSSIBLE IN VIVO FUNCTION... REFERENCES

 
The lysine-rich, or linker histones (so named because of their binding to linker DNA connecting successive nucleosomes in the chromatin fiber), are members of another class of 4WJ binding proteins (21, 22). The preferential binding is determined by the structured globular domain, which occupies the central portion of the polypeptide chain and is highly conserved in evolution. Both the solution (NMR) and crystal structure of the 80 amino acid-long globular domain have recently been determined (reviewed in ref 23). The globular domain contains three -helices and a ß-hairpin at the carboxyl terminus, with a short ß-strand situated between helices I and II ( Fig. 5). An almost identical structure has been identified in several transcription factors from both prokaryotic (CAP) and eukaryotic (HNF-3, engrail, Ets-1) origin, all belonging to a large evolutionarily conserved family, named the HNF-3/fork head or `winged helix' protein family (24). The term `wings' refers to the large loops that connect structured portions of these proteins.

View larger version (44K): [in this window] [in a new window]   Figure 5. A ribbon model of the 3-dimensional structure of the globular domain of histone H5, solved by X-ray crystallography (52). Side chains of the amino acids that form the two DNA binding sites on opposite sides of the domain are indicated. Figure courtesy of V. Ramakrishnan.

The structure of the winged helix motif bound to its cognate sequence has been determined for some sequence-specific members of the family (CAP, HNF-3, Ets-1) (reviewed in refs 25 and 26), none of which has been studied with respect to 4WJ binding. In general, all these proteins interact with DNA via principle contacts of a `recognition' -helix in the major groove with additional, mainly minor groove contacts by the wings. Notwithstanding the overall similarity in the 3-dimensional fold of the winged helix binding domains and in their presumed contacts with the DNA binding site, there are considerable differences in the effect of protein binding on DNA conformation. In the case of CAP, the DNA binding site is bent by ~90°, the bend resulting mainly from two major 40° kinks that occur on each side of the dyad axis of the complex (CAP interacts with its cognate sequence as a dimer). HNF-3 binding induces a 13° bend, narrowing the major groove in which the recognition helix is located. The proteins belonging to the Ets family of transcription factors induce a slight (8°) bend, also achieved by uniform curving of the DNA without distinct kinks.

In view of the structural similarities between the globular domain of linker histones and the sequence-specific binders of the HNF-3/fork head family, as well as the fact that the linker histones are 4WJ binding proteins, one might expect members of the HNF-3/fork head family to also preferentially bind to 4WJ DNA. Unfortunately, this still has not been tested.

Binding of the globular domain of the linker histones to DNA has not yet been resolved by structural studies. Despite their similarities, it must not necessarily be assumed that the nonspecific linker histones bind in the same manner as do the related specific transcription factors. However, topological assays have defined linker histones as DNA unwinding proteins, formally unwinding DNA by ~10° per histone molecule bound (27). This unwinding could be due to bending, similar to that observed with HNF-3 and Ets-1. That linker histones may induce bending might be expected on the basis of the recent observation that they show a strong preference to bind to DNA that has been bent by modification with the antitumor drug cis-platinum (28).

Almost nothing is known about the molecular details of 4WJ binding of linker histones apart from the observation that it is probably dependent on the geometry of the 4WJ (22) and requires the intactness of the two DNA binding sites present on opposite sides of the globular domain (29). Surprisingly, our knowledge of exactly how linker histones bind to any duplex DNA is also very limited, despite numerous studies.

	SWI/SNF: A CHROMATIN REMODELING ACTIVITY

TOP ABSTRACT INTRODUCTION STRUCTURE OF THE FOUR-WAY... THE HMG1 BOX PROTEINS THE HMGI-Y PROTEIN FAMILY THE WINGED HELIX PROTEINS SWI/SNF: A CHROMATIN REMODELING... PROKARYOTIC 4WJ BINDING PROTEINS MODES OF PROTEIN BINDING... EVIDENCE FOR AN ARCHITECTURAL... POSSIBLE IN VIVO FUNCTION... REFERENCES

 
SWI/SNF is a multicomponent protein complex, identified as an activity necessary for enhancing the action of many sequence-specific transcriptional activators in yeast (for a recent review, see ref 30). It is believed to function by remodeling the chromatin structure of the promoters of transcribing genes. Although the molecular mechanism of action is still far from being resolved, it is believed to somehow modify the structure or stability of the nucleosomal particle so as to make it accessible to the transcriptional machinery. Indeed, in vitro studies have demonstrated binding of SWI/SNF to isolated nucleosome cores, leading to disruption of the 10 bp DNase I repeat of the DNA, which is diagnostic of its wrapping around the histone octamer in the core particle. It has recently been shown that SWI/SNF binds with high affinity and preference to synthetic 4WJs (31), which may resemble the entry/exit point of the DNA in the nucleosome (see below). On the basis of this and other in vitro observations, it has been suggested that SWI/SNF might initially bind at this point and cause changes in the helical twist of the DNA; these, in turn, may lead to destabilization of the histone–DNA contacts in the core particle, leading either to loss of a H2A-H2B dimer or to eviction of the whole octamer.

Studies of SWI/SNF binding to DNA or nucleosomes are still at an early stage, mainly because of the enormous complexity of the system. The SWI/SNF complex itself contains 11 polypeptide chains, and obviously only some small portion of it can bind to a structure as small as a 4WJ. A more detailed structural analysis will have to await acquisition of more knowledge of the functional roles of each subunit and the availability of recombinant peptides.

	PROKARYOTIC 4WJ BINDING PROTEINS

TOP ABSTRACT INTRODUCTION STRUCTURE OF THE FOUR-WAY... THE HMG1 BOX PROTEINS THE HMGI-Y PROTEIN FAMILY THE WINGED HELIX PROTEINS SWI/SNF: A CHROMATIN REMODELING... PROKARYOTIC 4WJ BINDING PROTEINS MODES OF PROTEIN BINDING... EVIDENCE FOR AN ARCHITECTURAL... POSSIBLE IN VIVO FUNCTION... REFERENCES

 
Prokaryotes contain two abundant, well-studied basic proteins that bind preferentially to 4WJs. These are IHF (integration host factor) and HU, proteins that share significant sequence homology. Whereas IHF is a sequence-specific binder, HU binds DNA without much sequence specificity. IHF is a heterodimer of two basic peptides and is involved in a wide range of processes, including coliphage site-specific recombination, DNA replication, and transcription regulation from certain promoters (32). It is believed that its major function is to generate bends at certain defined DNA sequences, regardless of the final physiological effect (33). HU is also a multifunctional protein, believed to play a role in the structure of the bacterial nucleoid and to be involved in a variety of DNA transactions requiring the formation of higher order nucleoprotein complexes (34, 35).

IHF and HU both bend and loop DNA; HU has also been shown to constrain negative supercoils in circular DNA ( Table 1). HU is among the very few proteins that have been footprinted on the 4WJ. Hydroxyl-radical footprinting shows protection by HU of residues located at and near the junction point, the main sites of protection being situated on the unpaired oligonucleotides opposite each other (36). These results were interpreted to support a model in which two HU protein dimers specifically bind to two equivalent angles opposite each other in the 4WJ DNA, with almost no dimer–dimer interactions.

	MODES OF PROTEIN BINDING TO THE FOUR-WAY JUNCTION: RELATION TO DNA BENDING

TOP ABSTRACT INTRODUCTION STRUCTURE OF THE FOUR-WAY... THE HMG1 BOX PROTEINS THE HMGI-Y PROTEIN FAMILY THE WINGED HELIX PROTEINS SWI/SNF: A CHROMATIN REMODELING... PROKARYOTIC 4WJ BINDING PROTEINS MODES OF PROTEIN BINDING... EVIDENCE FOR AN ARCHITECTURAL... POSSIBLE IN VIVO FUNCTION... REFERENCES

 
The question of exactly how proteins bind to 4WJs has been little investigated. Most of the proteins known to bind to 4WJs also bend DNA or bind preferentially to bent DNA. This is probably not surprising, since a 4WJ is a kind of bent structure ( Fig. 1). In principle, there are at least two distinct ways in which proteins can bend DNA or bind to bent DNA. One, which we shall call the exterior mode, is exemplified by the HMG1 box, the L-shaped motif that reaches over the outside, or convex, face of the bend ( Fig. 2B and Fig. 3A). The other, which we call the interior mode, is illustrated, for example, by the binding of the HU dimer. Here the protein interacts with the concave surface of the bend. The structural requirements for proteins that bind in these two ways would be quite different. Proteins binding in the exterior mode must possess a stable bend or curvature to fit over the bent DNA, whereas those binding in the interior mode can be of virtually any shape, but must have at least two binding sites to interact simultaneously with the two arms of the bend; alternatively, the interior binders possessing one DNA binding site must interact as dimers. Linker histones and HU may be regarded as paradigms for these two subtypes of interior binders.

If we approach the question of binding to 4WJ from this viewpoint, we see that quite different kinds of interactions are possible. To interact with a 4WJ, exterior binders must extend over the crossing point, perhaps in the manner shown in Fig. 3C. From careful analysis of the footprinting results (20), this appears to be the case for HMGI-Y. Thus, the proteins will lie on one face of the crossed X structure. As Lilley and Clegg (1) point out, the faces are not equivalent in terms of DNA groove exposure; hence, some selectivity of face might be expected. In any event, no more than two molecules can bind in this way.

Interior binders could, in principle, fit between any two arms of the junction ( Fig. 3D); which pair is chosen will depend on the geometric relationship of the protein binding sites and the conformation of the junction. In principle, up to four sites could be occupied; the globular domain of H5 (but not intact H5) itself can occupy multiple sites on the 4WJ (22).

	EVIDENCE FOR AN ARCHITECTURAL ROLE OF 4WJ BINDING PROTEINS: DNA BENDING AND CROSSING

TOP ABSTRACT INTRODUCTION STRUCTURE OF THE FOUR-WAY... THE HMG1 BOX PROTEINS THE HMGI-Y PROTEIN FAMILY THE WINGED HELIX PROTEINS SWI/SNF: A CHROMATIN REMODELING... PROKARYOTIC 4WJ BINDING PROTEINS MODES OF PROTEIN BINDING... EVIDENCE FOR AN ARCHITECTURAL... POSSIBLE IN VIVO FUNCTION... REFERENCES

 
It is now clear that a wide variety of proteins bend and loop DNA in vitro; many have also been shown to bind to 4WJ DNA or other distorted DNA structures such as cis-platin adducts or bulged DNA. These in vitro findings have initiated numerous studies in an attempt to understand whether these peculiar binding properties may correlate with certain physiologically relevant features. Space limitations do not allow a detailed description of these studies, but the reader is referred to some excellent recent reviews (6, 37–40).

A strong indication that the property of certain proteins to preferentially bind to 4WJ DNA may be diagnostic for an architectural (rather than specifically functional) role in vivo comes from observations that many 4WJ binding proteins can substitute for each other in a variety of (often unrelated) functions ( Table 2). Such mutual complementation between proteins that are completely unrelated structurally and functionally but share the ability to bend DNA (or bind to bent DNA) suggests that it is this shared property that plays a crucial role in the specific reactions studied. An even stronger indication that this is the case comes from experiments in which a sequence-specific binding protein could be functionally substituted for by an intrinsically curved DNA sequence (e.g., ref 41). Spacing and/or phasing of the bend with respect to regulatory DNA elements (e.g., binding sites for transcription factors) was found to be critical to its ability to perform the corresponding reaction. Such experiments show that it is the bend, not the protein, that is functionally important.

It is important to mention two other intriguing functional substitutions that involve proteins not studied with respect to 4WJ binding. These proteins are not included in Table 2, although they will likely turn out to be 4WJ binders. CAP is a prokaryotic protein belonging to the winged helix family, with proven DNA bending ability. It can substitute for IHF in the in vitro integrative recombination of bacteriophage into the Escherichia coli chromosome (41). CAP and IHF bind DNA in entirely different ways: CAP is a major groove binder whereas IHF interacts with DNA mainly through the minor groove. What seems to matter in the recombination reaction is the presence of a specific DNA bend that would promote formation of a stable complex of the enzyme integrase with DNA ( Fig. 4A).

The second remarkable substitution is that of IHF by a dimer of histones H2A-H2B to form intasomes at one of the attachment sites (attL) used in excision recombination (42). The histone dimer, however, cannot support the excision reaction itself, most likely because it cannot assemble a sufficiently stable intasome at the other attachment site (attR). Unfortunately, this intriguing observation concerning H2A-H2B has not been investigated further; thus, the structural reason for this complementation remains obscure.

Binding to 4WJ can be symptomatic of a protein's preference for crossed as well as bent DNA. As shown in Fig. 3, a cross or bend can, in principle, be recognized by the same kind of protein conformations. A notable example: The binding of linker histones to 4WJ DNA may be a consequence of a possible structural similarity between 4WJs and the DNA at the entry/exit point of the DNA at the nucleosome (43), which is currently believed by most to be the major binding site of linker histones to the nucleosome (23, but see also ref 44; for a recent discussion, cf. ref 45). A similar observation for SWI/SNF (see above) presumably reflects the same structural similarity between 4WJ and DNA crossings.

An intriguing example may be the participation of HMGI-Y in in vitro integration of HIV-1 cDNA (46). The requirement for the protein to be present in preintegration complexes seems beyond doubt, although the mechanism through which it may act remains a mystery. We suggest that this mechanism may involve stabilization of integration intermediates in which the target DNA and the HIV cDNA must form a structure with two DNA helices crossing each other in close spatial proximity, closely resembling the structure of 4WJ DNA. Perhaps HMGI-Y forms a protein bridge between the two DNA molecules, allowing the proper enzymatic reactions to take place.

	POSSIBLE IN VIVO FUNCTION OF PROTEIN-INDUCED DNA BENDS AND CROSSES

TOP ABSTRACT INTRODUCTION STRUCTURE OF THE FOUR-WAY... THE HMG1 BOX PROTEINS THE HMGI-Y PROTEIN FAMILY THE WINGED HELIX PROTEINS SWI/SNF: A CHROMATIN REMODELING... PROKARYOTIC 4WJ BINDING PROTEINS MODES OF PROTEIN BINDING... EVIDENCE FOR AN ARCHITECTURAL... POSSIBLE IN VIVO FUNCTION... REFERENCES

 
Some possible mechanisms have been suggested for in vivo functions of protein-induced (or stabilized) DNA bending or crossing. The most common proposal involves a role for DNA bends in promoting the formation of higher order protein–DNA or protein–protein complexes (6, 40). Thus, for instance, IHF bound to specific sequences at the attachment locus attL of phage induces a sharp bend in the DNA so as to align two distant binding sites for the enzyme integrase, facilitating its simultaneous binding to both sites; the integrase then acts as a site-specific topoisomerase, cleaving and rejoining DNA broken ends to achieve the strand exchange that occurs during integration of phage DNA into bacterial DNA (33) ( Fig. 4A). In other cases, protein-induced bends may be instrumental in bringing together transcription factors bound at widely separated DNA sites. A well-studied example is the DNA bending caused by LEF-1 binding to the center of the enhancer of the T cell receptor gene, thus allowing interactions among several transcription factors (see ref 6 for further references) ( Fig. 4B).

An unexpected alternative to using protein-induced DNA bending to promote functionally important interactions of proteins is the use of proteins to straighten out intrinsic DNA curvature at specific DNA sites. Thus, for instance, HMGI-Y has recently been suggested to recruit transcription factors by partially counteracting the curvature that exists at its binding sites within the interferon ß (INF-ß) enhancer (47) ( Fig. 4D). A clue to the mechanism that possibly underlies such an effect comes from the recent solution structure of an HMGI-Y-DNA complex (48). The DNA binding motif interacts in the minor groove, stabilizing the B-form of DNA. Thus, the principal architectural role of these proteins seems to involve reversing and preventing intrinsic distortions in DNA conformation.

The effects of protein-induced DNA bending are not necessarily confined to the bend region: there are examples in which bending seems to cause conformational change in the DNA double helix at distant sites. Thus, CAP binding to sites at different positions relative to a reference sequence in a minicircle can cause alteration in DNase I cleavage patterns, suggesting changes in the geometry of the DNA double helix at a distance (49). Insertion of intrinsically curved DNA regions upstream of the TATA box, the binding site of TBP, increases in a distance-dependent manner the affinity of TBP binding in the context of a minicircle (50). These results again suggest that structural deformations induced at one site of a topologically constrained DNA domain may influence DNA conformation at distant sites, thereby possibly affecting protein binding there. Finally, a recent example shows that IHF binding to an upstream activating sequence of an E. coli gene contained in a supercoiled DNA template alters the structure of the DNA helix ~80 bp downstream, close to the transcription initiation site (51). The allosteric change caused by IHF binding was demonstrated by enhanced reactivity to KMnO₄ (a probe detecting sharply distorted and single-stranded DNA regions) of a pair of adjacent thymine residues at positions -11 and -12 from the transcription start site. This change was also accompanied by protection of a KMnO₄ site in the IHF binding region. Furthermore, results from an abortive transcription initiation assay showed that IHF binding increased the rate of open complex formation at the promoter site.

Protein stabilization of crossed DNA structures may be of equal importance to a variety of in vivo functions. It is clear that binding of linker histones to nucleosomal fibers stabilizes the nucleosome against opening or displacement, thereby blocking transcription. The role of HMGI-Y in the integration of HIV cDNA has been mentioned above; so far, this role has been demonstrated only in vitro.

In summary, the correlation that exists for many proteins between their preferential in vitro binding to 4WJ DNA and their ability to bend, loop, cross, or supercoil DNA provides a fundamental insight into the versatility and importance of architectural proteins. It may also indicate that 4WJ binding can be a very useful tool in searching for certain classes of architectural proteins of physiological relevance. A note of caution should be added: 4WJ binding may not always be in direct correlation with the physiologically relevant mechanisms of action of certain proteins. An instructive example can be found among spontaneous mutants in the human SRY gene that cause sex reversal. Although such mutants possess near wild-type affinity for 4WJ, they display reduced affinity for their sequence-specific DNA binding sites (7, 17) and cannot perform their sex-determining function. What this probably means is that although 4WJ binding in vitro is useful in identifying proteins that perform certain architectural functions, their localization to physiologically significant sites may be determined by sequence specificity. Additional assays are needed to actually prove or disprove an in vivo role for any 4WJ binding protein.

	ACKNOWLEDGMENTS

 
This work was supported by National Institutes of Health Grant GM50276 to K.v.H. and J.Z. We thank Drs. D. M. J. Lilley, J. O. Thomas, P. E. Wright, V. Ramakrishnan, and T. Moss for providing figures. We apologize to the researchers whose work could not be cited due to space limitations.

	FOOTNOTES

 
¹ Correspondence: Department of Biochemistry and Biophysics, Oregon State University, Corvallis, OR 97331–7305, USA. E-mail: zlatanoj{at}ucs.orst.edu

² Abbreviations: 4WJ, four-way junction; NMR, nuclear magnetic resonance; UBF, upstream binding factor; IHF, integration host factor; INF, interferon; SL1, selectivity factor 1; TBP, TATA binding protein.

	REFERENCES

TOP ABSTRACT INTRODUCTION STRUCTURE OF THE FOUR-WAY... THE HMG1 BOX PROTEINS THE HMGI-Y PROTEIN FAMILY THE WINGED HELIX PROTEINS SWI/SNF: A CHROMATIN REMODELING... PROKARYOTIC 4WJ BINDING PROTEINS MODES OF PROTEIN BINDING... EVIDENCE FOR AN ARCHITECTURAL... POSSIBLE IN VIVO FUNCTION... REFERENCES

 

1. Lilley, D. M. J., and Clegg, R. M. (1993) The structure of the four-way junction in DNA. Annu. Rev. Biophys. Biomol. Struct. 22, 299–329[Medline]
2. Lilley, D. M. J. (1994) A ring for a warhead. Curr. Biol. 4, 1152–1154[Medline]
3. Srinivasan, A. R., and Olson, W. K. (1994) Computer models of DNA four-way junctions. Biochemistry 33, 9389–9404[Medline]
4. Lilley, D. M. J. (1997) All change at Holliday junction. Proc. Natl. Acad. Sci. USA 94, 9513–9515[Free Full Text]
5. Bustin, M., and Reeves, R. (1996) High-mobility-group chromosomal proteins: architectural components that facilitate chromatin function. Prog. Nucleic Acid Res. Mol. Biol. 54, 35–100[Medline]
6. Grosschedl, R., Giese, K., and Pagel, J. (1994) HMG domain proteins: architectural elements in the assembly of nucleoprotein structures. Trends Genet. 10, 94–100[Medline]
7. Bianchi, M. E. (1995) The HMG-box domain. In DNA–Protein Structural Interactions (Lilley, D. M. J., ed) pp. 177–200, IRL Press, Oxford
8. Baxevanis, A. D., and Landsman, D. (1995) The HMG-1 box protein family: classification and functional relationships. Nucleic Acids Res. 23, 1604–1613[Abstract/Free Full Text]
9. Hardman, C. H., Broadhurst, R. W., Raine, A. R. C., Grasser, K. D., Thomas, J. O., and Laue, E. D. (1995) Structure of the A-domain of HMG1 and its interaction with DNA as studied by heteronuclear three- and four-dimensional NMR spectroscopy. Biochemistry 34, 16596–16607[Medline]
10. Seeman, N. C., Rosenberg, J. M., and Rich, A. (1976) Sequence specific recognition of double helical nucleic acids by proteins. Proc. Natl. Acad. Sci. USA 73, 804–808[Abstract/Free Full Text]
11. Coleman, R. A., and Pugh, B. F. (1995) Evidence for functional binding and stable sliding of the TATA binding protein on nonspecific DNA. J. Biol. Chem. 270, 13850–13859[Abstract/Free Full Text]
12. Read, C. M., Cary, P. D., Preston, N. S., Lnenicek-Allen, M., and Crane-Robinson, C. (1994) The DNA sequence specificity of HMG boxes lies in the minor wing of the structure. EMBO J. 13, 5639–5646[Medline]
13. Werner, M. H., Huth, J. R., Gronenborn, A. M., and Clore, G. M. (1995) Molecular basis of human 46X,Y sex reversal revealed from the three-dimensional solution structure of the human SRY–DNA complex. Cell 81, 705–714[Medline]
14. Love, J. J., Li, X., Case, D. A., Giese, K., Grosschedl, R., and Wright, P. E. (1995) Structural basis for DNA bending by the architectural transcriptional factor LEF-1. Nature (London) 376, 791–795[Medline]
15. Teo, S.-H., Grasser, K. D., and Thomas, J. O. (1995) Differences in the DNA-binding properties of the HMG-box domains of HMG1 and the sex-determining factor SRY. Eur. J. Biochem. 230, 943–950[Medline]
16. Teo, S.-H., Grasser, K. D., Hardman, C. H., Broadhurst, R. W., Laue, E. D., and Thomas, J. O. (1995) Two mutations in the HMG-box with very different structural consequences provide insights into the nature of binding to four-way junction DNA. EMBO J. 14, 3844–3853[Medline]
17. Peters, R., King, C.-Y., Ukiyama, E., Falsify, S., Donate, P. C., and Weiss, M. A. (1995) An SRY mutation causing human sex reversal resolves a general mechanism of structure-specific DNA recognition: application to the four-way DNA junction. Biochemistry 34, 4569–4576[Medline]
18. Moss, T., and Stefanovsky, V. Y. (1995) Promotion and regulation of ribosomal transcription in eukaryotes by RNA polymerase I. Prog. Nucleic Acids Res. Mol. Biol. 50, 25–66[Medline]
19. Stefanovsky, V. Y., Bazett-Jones, D. P., Pelletier, G., and Moss, T. (1996) The DNA supercoiling architecture induced by the transcription factor xUBF requires three of its five HMG-boxes. Nucleic Acids Res. 24, 3208–3215[Abstract/Free Full Text]
20. Hill, D. A., and Reeves, R. (1997) Competition between HMG-I(Y), HMG-1 and histone H1 on four-way junction DNA. Nucleic Acids Res. 25, 3523–3531[Abstract/Free Full Text]
21. Varga-Weisz, P., van Holde, K., and Zlatanova, J. (1993) Preferential binding of histone H1 to four-way helical junction DNA. J. Biol. Chem. 268, 20699–20700[Abstract/Free Full Text]
22. Varga-Weisz, P., Zlatanova, J., Leuba, S. H., Schroth, G. P., and van Holde, K. (1994) The binding of histones H1 and H5 and their globular domains to four-way junction DNA. Proc. Natl. Acad. Sci. USA 91, 3525–3529[Abstract/Free Full Text]
23. Zlatanova, J., and van Holde, K. (1996) The linker histones and chromatin structure: New twists. Prog. Nucleic Acids Res. Mol. Biol. 52, 217–259[Medline]
24. Lai, E., Clark, K. L., Burley, S. K., and Darnell, J. E., Jr. (1993) Hepatocyte nuclear factor 3/fork head or "winged helix" proteins: A family of transcription factors of diverse biologic function. Proc. Natl. Acad. Sci. USA 90, 10421–10423[Abstract/Free Full Text]
25. Brennan, R. G. (1993) The winged-helix DNA binding motif: another helix-turn-helix takeoff. Cell 74, 773–776[Medline]
26. Kodandapani, R., Pio, F., Ni, C.-Z., Piccialli, G., Klemsz, M., McKercher, S., Maki, R. A., and Ely, K. R. (1996) A new pattern for helix-turn-helix recognition revealed by the PU.1 ETS-domain-DNA complex. Nature (London) 380, 456–460[Medline]
27. Ivanchenko, M., Hassan, A., van Holde, K., and Zlatanova, J. (1996) Histone H1 binding affects superhelicity in DNA: evidence from topological assays. J. Biol. Chem. 271, 32580–32585[Abstract/Free Full Text]
28. Yaneva, J., Leuba, S. H., van Holde, K., and Zlatanova, J. (1997) The major chromatin protein histone H1 binds preferentially to cis-platinum-damaged DNA. Proc. Natl. Acad. Sci. USA 94, 13448–13451.[Abstract/Free Full Text]
29. Goytisolo, F. A., Gerchman, S. E., Yu, X., Rees, C., Graziano, V., Ramakrishnan, V., and Thomas, J. O. (1996) Identification of two DNA-binding sites on the globular domain of histone H5. EMBO J. 15, 3421–3429[Medline]
30. Peterson, C. L. (1996) Multiple SWItches to turn on chromatin? Curr. Opin. Genet. Dev. 6, 171–175[Medline]
31. Quinn, J., Fyrberg, A. M., Ganster, R. W., Schmidt, M. C., and Peterson, C. L. (1996) DNA-binding properties of the yeast SWI/SNF complex. Nature (London) 379, 844–847[Medline]
32. Friedman, D. I. (1988) Integration host factor: a protein for all reasons. Cell 55, 545–554[Medline]
33. Nash, H. A. (1990) Bending and supercoiling of DNA at the attachment site of bacteriophage . Trends Biochem. Sci. 15, 222–227[Medline]
34. Drlica, K., and Rouviere-Yanif, J. (1987) Histonelike proteins of bacteria. Microbiol. Rev. 51, 301–319[Free Full Text]
35. Pettijohn, D. E. (1990) Bacterial chromosome structure. In Nucleic Acids and Molecular Biology (Eckstein, F., and Lilley, D. M. J., eds) pp. 153–162, Springer Verlag, Berlin
36. Bonnefoy, E., Takahashi, M., and Rouviere-Yanif, J. (1994) DNA-binding parameters of the HU protein of Escherichia coli to cruciform DNA. J. Mol. Biol. 242, 116–129[Medline]
37. van der Vliet, P. C., and Verrijzer, C. P. (1993) Bending of DNA by transcription factors. BioEssays 15, 25–32[Medline]
38. Pérez-Martín, J., Rogue, F., and de Loner, V. (1994) Promoters responsive to DNA bending: a common theme in prokaryotic gene expression. Microbiol. Rev. 58, 268–290[Abstract/Free Full Text]
39. Ner, S. S., Travers, A. A., and Churchill, M. E. A. (1994) Harnessing the writhe: A role for DNA chaperones in nucleoprotein complex formation. Trends Biochem. Sci. 19, 185–187[Medline]
40. Grosschedl, R. (1995) Higher-order nucleoprotein complexes in transcription: analogies with site-specific recombination. Curr. Opin. Cell Biol. 7, 362–370[Medline]
41. Goodman, S. D., and Nash, H. A. (1989) Functional replacement of a protein-induced bend in a DNA recombination site. Nature (London) 341, 251–254[Medline]
42. Segall, A. M., Goodman, S. D., and Nash, H. A. (1994) Architectural elements in nucleoprotein complexes: interchangeability of specific and non-specific DNA binding proteins. EMBO J. 13, 4536–4648[Medline]
43. Lilley, D. M. J. (1992) DNA wraps up? Nature (London) 357, 282–283[Medline]
44. Pruss, D., Bartholomew, B., Persinger, J., Hayes, J., Arents, G., Moudrianakis, E. N., and Wolffe, A. P. (1996) An asymmetric model for the nucleosome: a binding site for linker histones inside the DNA gyres. Science 274, 614–617[Abstract/Free Full Text]
45. Crane-Robinson, C. (1997) Where is the globular domain of linker histone located on the nucleosome? Trends Biochem. Sci. 22, 75–77[Medline]
46. Farnet, C. M., and Bushman, F. D. (1997) HIV-1 cDNA integration: Requirement of HMG I(Y) protein for function of preintegration complexes in vitro. Cell 88, 483–492[Medline]
47. Falvo, J. V., Thanos, D., and Maniatis, T. (1995)( )Reversal of intrinsic DNA bends in the IFNß gene enhancer by transcription factors and the architectural protein HMGI(Y). Cell 83, 1101–1111[Medline]
48. Huth, J. R., Bewley, C. A., Nissen, M. S., Evans, J. N. S., Reeves, R., Gronenborn, A. M., and Clore, G. M. (1997) The solution structure of an HMG-I(Y)–DNA complex defines a new architectural minor groove binding motif. Nature Struct. Biol. 4, 657–665[Medline]
49. Lavigne, M., Kolb, A., Yeramian, E., and Buc, H. (1994) CRP fixes the rotational orientation of covalently closed DNA molecules. EMBO J. 13, 4983–4990[Medline]
50. Parvin, J. D., McCormick, R. J., Sharp, P. A., and Fisher, D. E. (1995) Pre-bending of a promoter sequence enhances affinity for the TATA-binding factor. Nature (London) 373, 724–727[Medline]
51. Parekh, B., and Hatfield, G. W. (1996) Transcriptional activation by protein-induced DNA bending: evidence for a DNA structural transmission model. Proc. Natl. Acad. Sci. USA 93, 1173–1177[Abstract/Free Full Text]
52. Ramakrishnan, V., Finch, J. T., Graziano, V., Lee, P. L., and Sweet, R. M. (1993) Crystal structure of globular domain of histone H5 and its implications for nucleosome binding. Nature (London) 362, 219–223[Medline]
53. Bazett-Jones, D. P., Leblanc, B., Herfort, M., and Moss, T. (1994) Short-range DNA looping by the Xenopus HMG-box transcription factor, xUBF. Science 264, 1134–1137[Abstract/Free Full Text]
54. Putnam, C. D., Copenhaver, G. P., Denton, M. L., and Pikaard, C. S. (1994) The RNA polymerase I transactivator upstream binding factor requires its dimerization domain and high-mobility group (HMG) box 1 to bend, wrap, and positively supercoil enhancer DNA. Mol. Cell. Biol. 14, 6476–6488[Abstract/Free Full Text]
55. Bor, Y.-C., Bushman, F. D., and Orgel, L. E. (1995) In vitro integration of human immunodeficiency virus type 1 cDNA into targets containing protein-induced bends. Proc. Natl. Acad. Sci. USA 92, 10334–10338[Abstract/Free Full Text]