Intrahelical pseudoknots and interhelical associations mediated by mispaired human minisatellite DNA sequences in vitro

Gene, 121 (1992) 279-285 0 1992 Elsevier Science Publishers

GENE

B.V. All rights reserved.

279

0378-l 119/92/$05.00

06767

Intrahelical pseudoknots and interhelical associations mediated human minisatellite DNA sequences in vitro

by mispaired

(Repetitive DNA; DNA bend; theta-shaped structures; Holliday junctions; branch migration; Z-DNA)

electron

Lesley W. Coggins, Margaret

bow-shaped

O’Prey and Shahnaz

structures;

DNA conformation;

microscopy;

Akhter *

Beatson Institute for Cancer Research, CRC Beatson Laboratories. Bearsden, Glasgow G61 1BD. UK Received by D.M. Skinner:

6 April 1992; Accepted:

17 June 1992; Received

at publishers:

30 July 1992

SUMMARY

The human minisatellite arrays, 33.6 and 33.15, consist of tandem reiterations of a 37-nucleotide (nt) and a 16-nt repeat unit sequence, respectively, both of which contain a majority of purine bases on one strand. Knot-like tertiary structures, which mapped to the cloned arrays, were observed by electron microscopy (EM) in homoduplex molecules produced by denaturation and reannealing in vitro. They result from a primary hybridization between misaligned repeat units of the array, forming a slipped-strand structure with staggered single-stranded DNA loops, followed by a secondary hybridization between repeat units in the two loops. Depending on the relative alignment of the loops when they hybridize, a particular form of intrahelical pseudoknot is produced. Theta-shaped, figure-of-eight, and bow-shaped structures were the most common conformational isomers observed in homoduplexes flattened into two dimensions during EM preparation. At the site of a bow-shaped structure, a conformation-dependent bend of approximately 60” between the flanking DNA segments is induced; the other conformations generally do not deflect the line of the main DNA axis. Paired loops, similar to the bowshaped structure, were apically situated in some supercoiled plasmids containing the 33.6 array. Both plasmids formed intermolecular associations, consisting of two (or more) homoduplex molecules held together at or immediately adjacent to a nexus which mapped to the minisatellite sequences. These associations might arise either by interhelical hybridization between arrays or by knot-like structures interfering with branch migration of X-form Holliday junctions.

INTRODUCTION

Human minisatellite arrays consist of tandemly reiterated DNA repeat units which are generally lo-50 bp long, G+C-rich, with tracts of purine bases on one strand and

Correspondence to: Dr. L.W. Coggins,

Beatson

Institute

for Cancer

Re-

search, Garscube Estate, Bearsden, Glasgow G61 IBD, UK. Tel. (44-41) 339-8855, ext. 6897; Fax (44-041) 942-6521. * Present

address:

St., Glasgow Abbreviations:

Gil

Institute

of Genetics,

University

5JS, UK. Tel. (44-041)

bp, base pair(s); c-H-ras,

of Glasgow,

339-8855,

c-Harvey

Church

ext. 6231.

ras gene; EM, electron

microscopy; HSV-1, herpes simplex virus type 1; kb, kilobase or 1000 bp; nt, nucleotide(s); ss, single strand(ed); TE, 0.1 M Tris/ 0.01 M EDTA pH 8.5; VTR, variable

tandem

repeat.

pyrimidine bases on the other (Jeffreys et al., 1985). Minisatellite arrays 33.6 and 33.15 are typical examples of the class. The consensus repeat unit sequence of 33.6, 5’TGGAGGAAGGGCTGGAGGAGGGCTCCGGAGGAAGGGC, contains an imperfect triple repeat of the purine-biased sequence GGAGGAAGGGCT; the 16-nt consensus sequence of 33.15 is 5’-GAGGTGGGCAGGTGGA (Jeffreys et al., 1985). Minisatellite array length shows a high frequency of heterozygosity between unrelated individuals, and this reflects a high rate of mutation of the repeat copy number. The 33.6 and 33.15 arrays show 67% and 6% heterozygosity in length at their cognate loci on chromosomes 1 (D 1S 111) and 7 (D7S437), respectively (Jeffreys et al., 1990). Many minisatellite arrays have a consensus core

280

Fig. 1. Electron

micrographs

of (a-c) knot-like

g) knot-like

structures

supercoiled

~6.3. All micrographs

made by conventional with ScaI (Gibco reannealed

tertiary

and (h) an intermolecular methods

are shown at the same magnification

BRL) or BglI (Pharmacia) x

in buffers provided

TE, 0.1 mg per ml cytochrome

Philips EM 300. Quantitative

and (d) an intermolecular

in a ScaI-cut

~15.1 homoduplex

and the bar represents

and purified by CsCI density centrifugation

at 37°C for 20 min, using the formamide

40% formamide/l

structures

association

analysis

tific); data in the text are expressed

of electron

technique

micrographs

as mean k standard

by the manufacturer,

originally

c, were spread

(Sambrook described

on a hypophase

was carried

deviation;

in a BglI-cut

preparation;

0.25 pm. Methods: supercoiled

denatured

by Westmoreland

in 50% formamide,

of molecules

plasmid

molecules,

preparation;

(e-

at apices of

preparations plasmids

were

were cut

1 x TE at 80°C for 1 min, and

et al. (1969). Homoduplexes

of 10% formamide,

number

~6.3 homoduplex

(i-l) paired loops (arrowheads)

et al., 1989). To form homoduplex

out with a Summagraphics

n indicates

sequence containing the motif 5 ’ -GCAGGAXG, which resembles 5’-GCTGGTGG, the chi recombination signal of Escherichiu coli (Jeffreys et al., 1985). Part of the 33.15 consensus sequence, 5’-GCAGGTGG, differs at one position from this chi signal. Each 37-nt 33.6 repeat unit contains two copies of another &-like sequence, 5’GCTGGAGG, single copies of which are associated with

association

0.1 x TE (Coggins,

or native plasmids

1987) and examined

digitizer and SigmaScan

program

(Jandel

in

with a Scien-

scored.

some oncogene translocation sites (Krowczynska et al., 1990). The sequence 5’-GGCAGG, also present in the 33.15 repeat unit, is reported to affect the germline stability of tandem sequences (Mitani et al., 1990). Wahls et al. (1990) found that plasmids carrying tandem copies of a minisatellite sequence show increased intermolecular recombination on transfection into cultured human cells. In

281 situ hybridization mosomes

of the 33.15 sequence

showed that it is localized

to meiotic

at chiasmata

chro-

(Chand-

ley and Mitchell, 1988). We have shown that the VTR array near the human c-H-rus oncogene (Capon et al., 1983) forms knot-like tertiary structures in homoduplexes produced by reannealing in vitro (Coggins and O’Prey, 1989). Using EM to examine homoduplexes containing the 33.6 and 33.15 arrays, we have now investigated whether the ability to form these novel DNA conformations, which result from misaligned hybridization, is a more general property of minisatellite arrays. We demonstrate that the 33.6 and 33.15 arrays can form intrahelical pseudoknot structures, and show that intermolecular associations are also mediated in their vicinity.

RESULTS

AND DISCUSSION

(a) Intrahelical tertiary structures in homoduplex molecules Linearized ~6.3 or ~15.1 plasmids, carrying the 33.6 and 33.15 arrays, respectively, were denatured and reannealed to form homoduplexes. Knot-like tertiary structures were observed by EM in 22.2% of homoduplexes derived from BglI-cut ~6.3 (n = 500) and 13.4% of homoduplexes from ScaI-cut ~15.1 (n = 500). O-shaped (Fig. la,d), figure-ofeight (Fig. lb,e) and bow-shaped structures (Fig. lc,f) were most common. In ~6.3 homoduplexes, their position was 0.73 ? 0.06 kb to 1.25 ? 0.07 kb from the proximal ScaI-cut end (n = 35) and 1.03 + 0.06 kb to 1.60 & 0.08 kb from the BglI-cut end (n = 35). In ~15.1 homoduplexes, knot-like structures were 0.68 k 0.05 kb to 1.14 + 0.05 kb from the proximal ScaI-cut end (n = 35), and were more centrally situated in BglI-cut molecules. These data map the structures to the minisatellite arrays (Fig. 2). The duplex DNA comprising the structures was 540 + 61 bp (n = 70) in ~6.3

I

4

I

Fig. 2. Maps

I

of .&I-cut

boxes), the flanking

human

s

B I B I

~6.3 and ~15.1, showing DNA

sequences

I p6.3

S I

p15.1

the arrays

(blackened

(hatched

boxes) and the

vector plasmid T7/T3-18 (Gibco BRL) (open boxes). The 2.95-kb plasmid ~6.3 carries the 33.6 array (GenBank accession No. M30548), consisting of 18 copies of a 37-nt repeat unit sequence a deletion

in one repeat);

~15.1) carries

the 33.15 array

copies of a 16-nt sequence. are marked 500 bp.

the 2.75kb

plasmid

(GenBank

accession

The mean positions

above the respective

(about 640 bp due to

p15.1.11.14

(here termed

No. M30551),

of the knot-like

29

structures

maps. B, BglI; S, ScaI. Bar represents

and 460 k 52 bp (n = 35) in p 15.1, in proportion

to the array

lengths of 33.6 (640 bp) and 33.15 (460 bp). In both homoduplex preparations, the structural diversity of the knot-like forms was similar to that previously found for the c-H-ras-associated VTR (Coggins and O’Prey, 1989). We proposed that their formation is initiated by a nucleation event between two arrays which are out-of-register when they anneal. This is supported by the correlation between the observed frequency of the structures (in 10% of homoduplexes containing the VTR, in 13% for 33.15 and 22% for 33.6) and the array length as a percentage of total plasmid length (9%) 16% and 21%, respectively). Hybridization of misaligned repeat units produces a ‘slipped structure’ with two staggered ssDNA loops (McKeon et al., 1984). This type of structure was never found in our preparations, and we surmise that secondary hybridization of the complementary repeat sequences in the loops normally ensues, thus producing the various structures observed. A O-shaped structure is produced when repeat unit sequences in the base of one loop hybridize to homologous sequences in the apex of the other loop. Hybridization when the bases of both loops are aligned produces a pair of duplex DNA loops. Two types of doubleloop structures were distinguished, in which the loops were on the same side (bow-shaped structure) or on opposite sides (figure-of-eight) of the main DNA axis. The loops were sometimes of unequal sizes (Fig. lb,c), possibly due to different degrees of slippage and/or translocation of the ss loops by branch migration (Lee et al., 1970) during their formation. All these structures are interhelical pseudoknots, rather than topologically true knots, as the two DNA strands are not interlinked. (b) A bend is associated with bow-shaped structures The angle between the plasmid vector arms at their junction with tertiary structures, extrapolated assuming a 150-bp persistence length (Muzard et al., 1990) was determined. The main duplex DNA axis exhibited little deviation in the vicinity of figure-of-eight and O-shaped structures in either ~6.3 or ~15.1. In contrast, the arms emerging from bow-shaped structures invariably formed an acute angle (Fig. lc,f) of 56.4 k 17.3’ (n = 30) in ~6.3 homoduplexes and 67.3 + 11.9” (n = 30) in ~15.1 homoduplexes. This novel DNA bend, resulting from the presence of tandemly reiterated sequences, brings sequences flanking the arrays into closer proximity than in homoduplexes containing figure-of-eight and O-shaped structures. It is a conformation-induced bend, analogous to bending at the site of cruciforms formed by inverted repeat sequences (Gough and Lilley, 1985) or of intramolecular triplexes formed by purine tract mirror repeats (Wells et al., 1988) but unlike the curvature resulting from the intrinsic properties of the kinetoplast DNA sequence (Griffith et al., 1986).

282 An angle of 60” conforms with a planar six-way helical junction (360”/6) at the centre of both two-dimensional models that fit the observed bow-shaped structure. The complementary DNA strands behave differently in the first model (Fig. 3C), as only one strand bends at the centre and boundaries of the structure. In the other (Fig. 3D), one strand bends in the centre and the other strand at the boundaries. The figure-of-eight model (Fig. 3B) also contains a six-way planar junction, but the vector arms are 180” apart and both DNA strands have similar topologies. A rope model of any double-loop structure (Fig. 3B, C or D) can be manipulated into the other double-loop forms without separating the two strands. All the observed structures were derived from threedimensional structures in free solution, but were flattened in two dimensions on a liquid surface for EM preparation. For a bow-shaped structure, the 60” angle of the planar form is increased to 90” between evenly spaced arms in three dimensions. However, the actual structure(s) in solution could differ from this simplistic model, particularly since molecules were spread for EM in the absence of divalent cations. Helix-helix stacking, influenced by the local DNA sequence (see section c), alfects the conforma-

secondaryhybridisation of single-strandedloops

Fig. 3. Formation

of knot-like

structures

by a schematic

array.

of the strands (shown parallel not antiparallel) are drawn without the strands intersecting.

(c) Repeat unit sequence and tertiary structure formation All the conformational isomers are proposed to result from misaligned hybridization within the array. Tandem repeats are evidently necessary for this, but are they sufficient or does the sequence of the repeat unit also play a role? Some sequence divergence between repeat units occurs in the VTR (Capon et al., 1983) and in both minisatellite arrays (Jeffreys et al., 1985), and is evidently tolerated in the structures. All three arrays are G+C-rich, and are purine-biased (33.6 and 33.15) or contain tracts of purines and pyrimidines (VTR). We have not determined whether arrays with other sequence characteristics can form knotlike structures. The repeat unit sequence might influence the formation or stability of the structures, for example by affecting helix-helix stacking at the helical junctions. The purine-rich and pyrimidine-rich strands might preferentially adopt particular configurations in these junctions (Duckett et al., 1988). Polypurine sequences can adopt a number of unusual DNA structures (Wells et al., 1988) including interhelical interactions (Sen and Gilbert, 1988). Unlike the primary hybridization between linear strands, secondary hybridization between ss loops (Fig. 3) should be topologically constrained. Pairing of complementary ss DNA circles usually generates underwound DNA, containing left-handed and right-handed helical duplex segments alternating with unpaired DNA (Malamy et al., 1972; Broker et al., 1977; Brahmachari et al., 1987). However, the tertiary structures appeared to consist of duplex DNA with no extensive underwound regions (Fig. la-c, e-g). Although the EM technique might not have revealed short interspersed unpaired regions in the small loops, our results suggested that the minisatellite sequences may form duplex DNA with no net interlinking of the strands. Such paranemic pairing is usually attributed to facile formation of lefthanded DNA (Haniford and Pulleyblank, 1983; Holliday, 1989). Unfortunately, our EM observations cannot distinguish between left-handed and right-handed duplex DNA regions. In principle, none of the arrays examined contains sequences known to form left-handed Z-DNA readily, such as tracts of alternating purine and pyrimidine bases, although Cantor and Efstratiadis (1984) have suggested that polypurine tracts may assume a non-Z-form left-handed DNA structure.

Arrow-

heads represent repeat units and the thick lines represent nonrepetitive flanking DNA. The resulting structures are displayed as two-dimensional conformations. For clarity in structures A-D, arrows indicate only the direction junctions

tion of four-way helical junctions in the presence of cations (Duckett et al., 1988; 1990). O-shaped structures apparently contain two square-planar four-way junctions (Holliday, 1964).

and the helical

(d) Interhelical associations In addition to the intrahelical pseudoknots observed in individual homoduplex molecules, associations of two (occasionally more) unit-length linear molecules were found at

283 a low frequency

in reannealed

~6.3 and p 15.1 preparations

(Fig. Id, h). A knot-like structure, often more complex than those described in section a, was observed where the X-

frequency, but supercoils with a pair of loops similar in size and form to a bow-shaped structure were sporadically observed (Fig. li-1).

They usually

occurred

apically,

where

shaped molecules intersected. The position of this nexus was mapped in the BgZI-cut ~6.3 homoduplex preparation. The lengths of the flanking arms were similar to those in unassociated homoduplexes. There were two shorter and two longer arms, each pair of approximately equal length. The summed lengths of the smaller pair was 43.61 + 5.72%

the DNA duplex turns back on itself, a position expected to be favoured by the bend associated with bow-shapes (section b). Curved kinetoplast DNA preferentially adopts an apical position in supercoils (Laundon and Griffith, 1988) and the bend associated with triplex DNA is pro-

of the total length of all four arms (n = 20). This agrees with

the map position of the paired loops could not be determined in uncut supercoiled plasmids, so we were unable to

the shorter arm being 42% of the length of both arms in monomer BglI-cut homoduplexes containing knot-like structures. As expected, more length asymmetry was evident between the arms of ScaI-cut X-shaped molecules (Fig. lh). Thus decussate molecules were metastably associated at, or immediately adjacent to, the minisatellite arrays. They were neither fortuitously overlapping monomers nor (with one exception noted below) linked by chi junctions with a variably located point of homologous interaction (Holliday, 1964; Potter and Dressler, 1976). It was not usually possible to trace individual DNA strands through the nexus in order to determine whether exchange of partners had occurred between the flanking arms of X-shaped molecules. In theory, the incidence of exchange depends on how the associations arise in vitro. Several mechanisms can be postulated and could occur at different frequencies. (i) No exchange of flanking sequences would result from the interaction of unpaired array sequences in knot-like or slipped structures in two otherwise duplex molecules. (ii) Exchange of partners would ensue if two homoduplex molecules with knot-like structures interacted via homologous unpaired regions in the flanking arms. Hybridization to the ‘wrong’ partner might proceed as far as the knots which could act as an obstacle to further branch migration. A single example (not shown) was found of two molecules possibly in the process of forming or resolving an interhelical association. Each carried a knot-like structure, and was connected by a chi junction to the other molecule at identical locations in the arms. (iii) Exchange could also occur if two strands underwent misaligned hybridization in the array (as described in section a) and a third strand hybridized to one of the unpaired arms. Displacement by branch migration might similarly be impeded by the knot-like structure, enabling another strand to interact with the remaining unpaired DNA derived from two molecules. (e) EM of supercoiled plasmids In an attempt to investigate whether knot-like structures can be formed under torsional stress, native supercoiled ~6.3 plasmids were examined by EM. The topology of the intertwined DNA duplexes was difficult to interpret. No individual type of unusual structure was observed at a high

posed to be similarly located (Wells et al., 1988). However,

prove that they were produced by the 33.6 array. Theoretically, formation of knot-like structures could be initiated under torsional stress by local denaturation within the array, followed by misaligned hybridization as described in section a. A limited denaturation might favour slipped hybridization between shorter repeat units. The postulated initiation mechanism is similar to C-type cruciform formation by inverted repeats, which however occurs in an A+Trich environment (Lilley, 1985) whereas the minisatellite arrays are G+C-rich. The resulting structures might generally be smaller than those produced as a result of complete denaturation in homoduplex preparation. Their formation would be energetically favoured by progressive relaxation of supercoiling as the effective length of the plasmid is decreased. The cloned 33.6 and 33.15 minisatellite arrays are hypersensitive to S 1 nuclease and Pl nuclease in plasmids under torsional stress (L.W.C., L. Scobie, N.E. Bradshaw and M. O’P., unpublished), and we are investigating the nature of the unorthodox DNA structure responsible. (f) Relevance to minisatellite behaviour in vivo Evidence that minisatellite sequences promote recombination is described in the INTRODUCTION. In addition, a recombination hotspot in the murine major histocompatibility complex is associated with a minisatellite-like array (Steinmetz et al., 1986). The genome segment inversion site in HSV-1 also contains tandem arrays, including an array of 12-nt direct repeats (DR2), with a strong bias to purine bases on one strand, which is minisatellite-like (Coggins and O’Prey, 1989; Wohlrab et al., 1991). Associated with the fragile X site is a larger array than normal of the tandemly repeated trinucleotide CCG, suggesting the hypothesis that similar unstable DNA sequences could be a source of genomic instability and hence a cause of genetic disorders (Sutherland et al., 1991). New length alleles in minisatellites often arise with no exchange of flanking markers, suggesting generation by mitotic mechanisms (Wolff et al., 1989). Slipped-strand mispairing is a possible mechanism, and has been invoked as a general method for the evolution of tandem repetitive sequences (Levinson and Gutman, 1987). The formation of

284 knot-like structures such a propensity.

by minisatellite

arrays in vitro reflects

Other hotspots for recombination in vivo are associated with sequences that have the potential to form unorthodox DNA structures in vitro. These include Z-DNA (Haniford and Pulleyblank,

1983; Kmiec and Holloman,

1986; Fowler

et al., 1988; Boehm et al., 1989; Weinreb et al., 1990), triplex DNA (Fowler and Skinner, 1986; Collier et al., 1988; Wells et al., 1988; Weinreb

et al., 1990), and anisomorphic

DNA formed by the HSV-1 DR2 array (Wohlrab et al., 1987). Local changes in torsional stress associated with transcription (Tsao et al., 1989) or replication have been suggested as a mechanism that could drive such structural transitions in DNA in vivo. If minisatellite arrays in the cell could also form knot-like structures (section e), further explanations for their recombinogenic behaviour can be envisaged. Proteins have recently been identified which bind minisatellite DNA sequences (Collick and Jeffreys, 1990; Wahls et al., 1991) or cleave DR2 repeat sequences (Wohlrab et al., 1991). Knot-like structures might also interact with proteins which recognize unusual conformations such as bent DNA, resolve helical junctions, or repair mismatched sequences. Mitotic recombination between sister chromatids and meiotic recombination between homologous chromosomes are interhelical events. Jeffreys et al. (1991) have recently shown that meiotic recombination between minisatellite sequences can occur. We found that metastable interhelical associations between homoduplexes occurred in the vicinity of arrays which had formed knot-like structures. If interactions similar to the two major mechanisms postulated, namely interhelical hybridization between misaligned arrays [perhaps facilitated by paranemic pairing (section c)] or interference with branch migration of chi junctions by intrahelical pseudoknot( could occur in vivo, they might promote recombination within the minisatellite array or the sequences that flank it.

(4) Paired loops, similar to a bow-shaped structure, are found at the apices of some ~6.3 supercoils but their position could not be mapped since restriction enzyme treatment is precluded. (5) Intermolecular associations diate vicinity of complex nexuses

ellite arrays. The associations could be due either to interhelical hybridization between tandem arrays or to branch migration in mispaired molecules being obstructed by knotlike structures.

ACKNOWLEDGEMENTS

The original plasmids

~6.3 and ~15.1.11.14

were gener-

ously provided by Dr. K. Brown, University of Bristol, UK; the 33.6 and 33.15 arrays were used with kind permission of Dr. A.J. Jeffreys, University of Leicester, UK. The Beatson Institute is supported by the Cancer Research Campaign.

REFERENCES Boehm, T., Mengle-Gaw, A. and Rabbitts,

L., Kees, U.R., Spurr, N., Lavenir,

T.H.: Alternating

mote chromosomal phoid tumours. Brahmachari, probed Broker,

seen in a variety

of human

lym-

C.R. and McClelland,

M.:

Y.S., Cantor,

that adopt non-B-DNA

by enzymic

T.R.,

methylation.

Soll,

self-renatured

L.

and

conformation

sequences.

L.T.:

Underwound

of inverted

loops

duplications

in and

J. Mol. Biol. 113 (1977) 579-589.

C.R. and Efstratiadis,

homopyrimidine

in form V DNA as

J. Mol. Biol. 193 (1987) 201-21 I.

Chow,

DNA can be diagnostic

translocated Cantor,

tracts may pro-

translocations

S.K., Shouche,

Sequences

I., For&r,

purine-pyrimidine

EMBO J. 8 (1989) 2621-2631.

A.: Possible

S l-hypersensitive

structures

of homopurine-

sites. Nucleic Acids Res. 12 (1984)

8059-8072. Capon,

D.J., Chen, E.Y., Levinson,

D.V.: Complete cinoma

nucleotide

oncogene

A.D.,

sequences

and its normal

Seeburg,

P.H. and Goeddel,

of the T24 human bladder

homologue.

Nature

car-

302 (1983) 33-

37. Chandley,

A.C. and Mitchell,

are sites for crossing-over

(g) Conclusions (I) The cloned human minisatellite arrays 33.15 and 33.6 form knot-like tertiary structures in homoduplex molecules produced by in vitro reannealing. (2) It is proposed that the conformational isomers (Oshapes, figures-of-eight and bow shapes) are intrahelical pseudoknots, formed by a primary slipped-strand hybridization in the misaligned array, followed by secondary hybridization between the resulting staggered ss loops (with no evident topological constraint). (3) Bow-shaped structures induce a novel conformationdependent bend of about 60” in the DNA axis of homoduplexes constrained in two-dimensions for EM preparation.

are found in the immethat map to the minisat-

A.R.: Hypervariable

minisatellite

at meiosis in man. Cytogenet.

regions

Cell Genet.

48 (1988) 152-155. Coggins,

L.W.: Preparation

Sommerville, lecular

of nucleic acids for electron microscopy.

J. and Scheer, U. (Eds.), Electron

Biology.

A Practical

Approach.

IRL

Microscopy Press,

In:

in Mo-

Oxford,

1987,

pp. l-60. Coggins,

L.W. and O’Prey,

by misaligned

M.: DNA tertiary

hybridization

structures

formed in vitro

of multiple tandem repeat sequences.

Nu-

cleic Acids Res. 17 (1989) 7417-7426. Collick, A. and Jeffreys, DNA-binding

protein.

A.J.: Detection Nucleic

of a novel minisatellite-specific

Acids Res. 18 (1990) 625-629.

Collier, D.A., Griffin, J.A. and Wells, R.D.: Non-B right-handed DNA conformations of homopurine. Homopyrimidine sequences in the murine immunoglobulin 7397-7405.

Ccc switch

region.

J. Biol. Chem.

263 (1988)

Duckett, D.R., Murchie, A.I.H., Diekmann, S., von Kitzing, E., Kemper, B. and Lilley, D.M.J.: The structure of the Holliday junction, and its resolution.

Cell 55 (1988) 79-89.

285 Duckett,

D.R., Murchie,

A.I.H. and Lilley, D.M.J.:

in the conformation

of the four-way

The role of metal ions

helical junction.

EMBO

J. 9

Fowler, R.F. and Skinner, D.M.: Eukaryotic pyrimidinepurine

tract

DNA diverges at a long and

that can adopt

altered

R.F., Stringfellow,

Gough,

of a complex

G.W.

Nature

Visualization Haniford,

that as-

a specific deletion in some cloned

DNA bending

induced

M., Rauch, C.A., Kitchin,

Biomol.

Struct.

D.E.:

Jeffreys,

P.A. and Englund,

P.T.:

DNA by electron

The in-vivo

mi-

occurrence

of Z

in fungi. Genet.

Res. 5

1 (1983) 593-609.

for gene conversion

A.J., Wilson,

B-Z DNA. Trends V. and Thein,

DNA. Nature

A.J., MacLeod,

Genet.

A., Tamaki,

R., Povey,

repeat coding

clones 33.6 and

and D7S437.

C.H.

A.M.,

W.K.:

Rudders,

consensus

and Monckton,

as a digital approach pairing

to DNA typof DNA mole-

by sequences

J.D.:

Curved

of supertwisted

of the terminally

from the coliphage

of Z-DNA.

T.G.:

of oncogene

The human

translocations.

particles

DNA.

Cell 52 (1988) 545-549.

N.: A physical

repetitious,

can uniquely

study by electron

circularly

permuted

DNA

of Escherichia coli 15. J. Mol. Biol. 48

G.A.:

for DNA sequence

Slipped evolution.

strand

mispairing:

a major

Mol. Biol. Evol. 4 (1987)

Lilley, D.M.J.: The kinetic properties of cruciform extrusion are determined by DNA base sequence. Nucleic Acids Res. 13 (1985) 1443Fiandt,

M. and Szybalski,

W.: Electron

microscopy

of

in the lac operon of Escherichia coli. Mol. Gen. Genet.

119 (1972) 207-222.

D. and Lindahl,

hotspots

for meiosis.

A

Press,

complexes Nature

334

K.F.: Gene organization

and

in the murine major histocompatibility

G.R., Haan, E.A., Kremer, E., Lynch, M., Pritchard,

S. and Richards,

RI.:

Hereditary

unstable

com-

Lancet

of DNA:

direct biochemical

evidence

C., Schmidt,

A. and de Crombrugghe,

served in both the chicken

M., Yu,

DNA: a new explanation

338 (1991) 289-292.

Tsao, Y.-P., Wu, H.-Y. and Liu, L.F.: Transcription-driven

supercoiling

from in vitro studies.

L.J. and Moore, P.D.: Hypervariable

DNA is a hotspot for homologous 60 (1990) 95-103. Swenson,

A., Collier,

recombination

G. and Moore,

satellite DNA binding proteins. 3274.

Cell 56

D.A.,

minisatellite

in human cells. Cell

P.D.: Two hypervariable

mini-

Nucleic Acids Res. 19 (1991) 3269-

Birshtein,

B.K.

handed Z-DNA and intramolecular unequal sister chromatid exchange. Wells, R.D., Collier, D.A., Hanvey, oligopurine.

and biology

and

Wells,

R.D.:

Left-

triplex formation at the site of an J. Biol. Chem. 265 (1990) 1352-

J.C., Shim&

of unusual

B.: A sequence

and mouse ct2(1) collagen

promoter

DNA

M. and Wohlrab, structures

oligopyrimidine

sequences.

FASEB

B.C., Szybalski,

W. and Ris, H.: Mapping

by

J. 2 (1988) 2939-

lambda Wohlrab,

in heteroduplex

by electron

microscopy.

F., McLean,

DNA molecules Science

of deletions

of bacteriophage

163 (1969) 1343-1348.

M.J. and Wells, R.D.: The segment inversion a novel DNA structure.

site

J. Biol.

Chem. 262 (1987) 6407-6416. Wohlrab,

F., Chatterjee, inversion

clear endonuclease.

S. and Wells, R.D.: The herpes simplex virus 1 site is specifically Proc.

Natl.

cleaved by a virus-induced

Acad.

Sci. USA

nu-

88 (1991) 6432-

6436. R., Jeffreys, A.J. and White, R.: Unequal

crossing

con-

over between homologous

chromosomes

con-

involved in the generation (1989) 382-384.

of new alleles at VNTR loci. Genomics

J. Biol. Chem. 259 (1984) 6636-

F.:

adopted

2949. Westmoreland,

Wolff, R.K., Plaetke,

tains sites sensitive to Sl nuclease. 6640.

M., Stephan,

recombinational

segment

1465.

McKeon,

Cloning.

Laboratory

of parallel four-stranded

motifs and its implications

of herpes simplex virus type 1 adopts

polar insertions

T.: Molecular

NY, 1989.

W.: Formation

and substitutions

G. and Gutman,

M.H.,

and Maniatis,

(1988) 364-366.

The chemistry

203-221.

Malamy,

intermediates.

1359. helix segments

(1970) l-22. mechanism

of genetic recombination:

of recombination

2nd ed. Cold Spring Harbor

Cold Spring Harbor, Sen, D. and Gilbert,

Weinreb,

R.A. and Krontiris,

Lee, C.S., Davis, R.W. and Davidson,

Levinson,

E.F.

Manual,

Wahls, W.P., Homologous

at breakpoints

and Griffith,

orient the topology microscopy

mapping

models. EMBO J.

(1989) 111-118.

Genomics

Nucleic Acids Res. 18 (1990) 1121-1127. Laundon,

265

Sci. USA 73 (1976) 3000-3004.

Wahls, W.P., Wallace,

K., Neil, D.L.

cules by Ustilago reel protein is promoted Cell 44 (1986) 545-554. minisatellite

observation

for some old genetic questions?

S. and Royle, N.J.:

354 (1991) 204-209.

Kmiec, E.B. and Holloman,

Krowczynska,

microscopy

with theoretical

D.: On the mechanism

J., Fritsch,

Sutherland,

‘minisatellite’

loci’ detected by the minisatellite

MacLeod,

D.G.: Minisatellite ing. Nature

5 (1989) 355-356.

314 (1985) 67-73.

33.15 correspond to the cognate loci DlSlll 7 (1990) 449-452. A.J.,

motif in

J. Biol. Chem.

B.: Electron

Comparison

microscopic

Steinmetz,

S.L.: Hypervariable

A., Neumann,

‘Major minisatellite

Jeffreys,

R.: A GGCAGG

instability.

plex. Cell 44 (1986) 895-904.

R.: Untwisting

regions in human Jeffreys,

electron Sambrook,

by cruciform

(1964) 282-304. Holliday,

B. and R&et,

Potter, H. and Dressier,

by guanine-rich

Dynam.

R.: A mechanism

G., Theveny,

Laboratory

of the bent helix in kinetoplast and Pulleyblank,

Muzard,

Proc. Natl. Acad.

satellite. Gene 71 (1988) 165-176.

Cell 46 (1986) 717-724.

D.B.

DNA.J. Holliday,

D.M.: A domain

313 (1985) 154-156.

Griffith, J., Bleyman, croscopy.

includes

and Lilley, D.M.J.:

formation.

their germline

9 (1990) 1289-1298.

L.A. and Skinner,

sumes a Z-conformation variants

Y. and Kominami,

affecting

of pBR322 curvature.

conforma-

tions. J. Biol. Chem. 261 (1986) 8994-9001. Fowler,

K., Takahashi,

minisatellites

(1990) 15203-15210.

(1990) 583-590. complex

Mitani,

is not the major mechanism 5