insertion of l1 elements into sites that can form non-b dna

THE JOURNAL OF BIOLOGICAL CHEMISTRY Vol. 264, No. 34, Issue of December 5, pp. 20736-20743,1989 Printed in U.S.A.

Insertion of L1 Elements into Sites That Can Form Non-B DNA INTERACTIONS OF NON-B DNA-FORMING SEQUENCES*

(Received for publication, July 14, 1989)

Karen Usdin$ and Anthony V. Furano From the Section on Genomic Structure and Function, Laboratory of Biochemical Pharmacology, National Institute of Diabetes and Digestive and Kidney Diseases, National Institutes of Health, Bethesda, Maryland 20892

Three rat L1 element integration (target) sites chosen at random can adopt non-B DNA structures in vitro at normal bacterial superhelical densities. These target sites contain, respectively, short, mixed (AT),, tracts that we show can form one or more cruciforms, short (GT),, tracts, or po1ypurine:polypyrimidine regions. These sites share no sequence homology, and a non-B DNA structure appears to be the only feature common to them all. When the right end of the LlRn3 element which forms a complex series of non-B DNA structures including two triplexes, and its target site which undergoes cruciform extrusion, are present on the same supercoiled molecule, they compete for available supercoil energy. The amount of non-B DNA formed at each site varies with pH, the concentration of cations, and the size of the topological domain. The impli- cation of our findings for recombination of L1 elements and for the effect of these elements on contiguous DNA sequences is discussed.

L1’ DNA elements (LINES, long interspersed repeated DNA elements) are present in high copy numbers in all mammalian species examined to date (see Refs. 1-9; and Refs. 10 and 11 for recent reviews). These elements are thought to be mobile based on the frequent occurrence of polymorphisms induced by L1 elements (12-19). The basis of this mobility is not yet known, although some have suggested that L1 elements are retrotransposons (8, 20-23).

Although sites into which L1 elements have inserted do not share any sequence homology (e.g. Refs. 5, 12, 13,17, and 24), it is possible that these insertion or target sites share common properties that may be reflected in their secondary structure. In the first part of this paper we report the results of our examination of three such target sites which we show can all adopt non-B DNA conformations in negatively supercoiled molecules.

The G-rich po1ypurine:polypyrimidine sequence at the right end of L1 elements also assumes a non-B DNA conformation in negatively supercoiled molecules (25). Transposition of an L1 element could thus result in two potential non-B DNA

* The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked “advertisement” in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.

The nucleotide sequence($ reported in thispaper has been submitted to the GenBankTM/EMBL Data Bank with accession number(s) X15742.

$ T o whom correspondence should be addressed. Tel.: 301-496- 2189.

The abbreviations used are: L1, long interspersed (LINE) element: EGTA, [ethylenebis (oxyethylenenitrilo) tetraacetic acid; bp, base pair(?,); kb, kilobase(s).

forming regions sharing the same topological domain. In theory this may alter the properties of both regions since the two sites would now compete for available supercoil energy. In the second part of this paper we describe how these two regions of non-B DNA forming potential interact in their competition for this energy.

The possible role of the non-B DNA structure of L1 element target sites in recombination with L1 DNA and the conse- quences of two non-B DNA structures sharing the same topological domain are discussed.

MATERIALS AND METHODS

Plasmids-The plasmids pTS41, containing the Moloney leukemia virus integration (Mlvi-2) site, and pR15, containing the rat immunoglobulin heavy chain (Igh) locus, were obtained from P. Tsichlis (Fox Chase Cancer Center, Philadelphia, PA). The plasmids pSL3AA, pSL3AAB, and pSL3AABA which contain part of the LlRn3 element and the right side of the LlRn3 DNA target site and pL3TS and pL3TSA, which contain the LlRn3 target site alone, were constructed using a derivative of pBR327, pSS, (constructed by I. Nur, in this laboratory) as a vector. UC15 has been described elsewhere (25).

DNA Preparation-Plasmid DNA was prepared from stationary phase Escherichia coli cells by alkaline lysis. Cruciform-free DNA was prepared by 2 cycles of ethidium bromide-cesium chloride centrifu- gation (26).

Mapping of Bromoacetaldehyde Reactive Sites-Bromoacetalde- hyde was prepared using the method described by Secrist et al. (27) for the preparation of chloroacetaldehyde. Bromoacetaldehyde treatment of supercoiled and linear molecules was carried out as described previously (28). For mapping of the bromoacetaldehyde-reactive sites, bromoacetaldehyde-treated molecules were digested with a suitable restriction endonuclease, radiolabeled if necessary, and treated with S1 nuclease as described in a previous paper (25). The products were resolved by electrophoresis in 0.8% agarose gels in Tris acetate buffer or 5% polyacrylamide gels and Tris borate buffer (29). Autoradi- ographs were scanned using a Bio-Rad model 620 Video Densitome- ter. For mapping at the nucleotide level, bromoacetaldehyde-treated pL3TSA molecules were digested with HindIII, the 5’-end labeled using T4 polynucleotide kinase and [Y-~*P]ATP, and the 3’ end labeled with Klenow polymerase and [ ( U - ~ ~ P I ~ C T P . After purification, the end-labeled fragments were depurinated by treatment with sodium acetate at pH 5.0 and then cleaved at the apurinic sites by @- elimination with piperidine as described elsewhere (25). The products were resolved on 5% polyacrylamide sequencing gels (29:1, acrylamide:bis-acrylamide).

Two-dimensional Gel Electrophoresis-Plasmid topoisomer series were made in the following way. Aliquots of supercoiled DNA were incubated overnight with topoisomerase I in the presence of 0-15 WM ethidium bromide (30). The aliquots were then extracted with phenol and the aqueous phases pooled. Residual phenol was removed by repeated ether extractions and the DNA precipitated with ethanol. The topoisomers were then resolved by electrophoresis at 37 “C for 20 h at 40 V on 2% agarose gels in a buffer containing either 50 mM sodium acetate (pH 5.0) and 1 mM EGTA, with or without 2 mM M$+ or 40 mM Tris acetate (pH 7.5), 50 mM sodium acetate, and 1 mM EDTA, with or without 2 mM M e . The gels were then equili- brated in Tris acetate buffer (pH 7.5) containing 4-5 wg/ml chloro-

20736

Ll Elements and Their Target Sites 20737

quine. rotated 90". and subjected to electrophoresis in the second dimension for 16 h at 40 V in the same buffer. The chloroquine was removed by soaking the gels in water. The gels were then stained with ethidium hromide or blotted to nitrocellulose and hyhridized to a homologous prohe.

RESULTS

L1 Element Target Sites Can Form Non-B DNA Structures in Vitro-Bromoacetaldehyde is a reagent that reacts with the N3 and N4 positions of cytosine, the N1 and N2 of guanine, and the N1 and N6 positions of adenine, positions normally involved in hydrogen bonding in Watson-Crick base pairs (31, 32). Reaction with this reagent is therefore diagnostic of non-B structures involving unpairing of guanine, cytosine, and adenine residues (31). We have used bromoacetaldehyde as a probe for non-B DNA structures in the target sites of L1 elements. Reaction with bromoacetaldehyde was detected in one of two ways: (i) bromoacetaldehyde modifies susceptible bases causing them to become hypersensitive to further modification by various chemicals and subsequent chain cleavage by piperidine; and (ii) supercoiled DNA that has reacted with bromoacetaldehyde remains sensitive to cleavage by S1 nuclease after restriction enzyme digestion, since bromoacetaldehyde adduct formation blocks hydrogen bonding by bromoacetaldehyde reacted bases.

We examined three arbitrarily chosen L1 element target sites: the LlRn3 element target site in the plasmid pLSTS, the Moloney leukemia virus integration 2 (Mlvi-2) locus con- tained in pTS41; and the immunoglobulin heavy chain (Igh) locus in pRI5. The DNA sequence of these sites is shown in Fig. 1.

When supercoiled molecules containing the various target sites were treated with bromoacetaldehyde, digested with a restriction enzyme, and treated with S1 nuclease, two novel bands (Fig. 2, lunes 5, 8, and I I ) were visible after electrophoresis. These novel bands result from S1 nuclease cleavage of

L l R n 3 TARGET SITE

T CCTGCCCCTA AAACCACCTT CCTTCTGGGA AATGAGAAAA TCCACTCATA CAAAATGGAA

TTCCTCAGG ACTGA - (245bp)- TATATTATAT ATATATAATA TATATTATAA TATATATCAT ACTATATATA

Hlvi-2 LOCUS

TCTTCACCAC ATAATCATCA TTGTTTGTCT TCTTTTAAAA AAATGATTTG TCTATTTATA TGTAGTACTC ATCTGTGTGT GTGTGTGTGT GTGTGTGTGT GTGTGTGTGT GTGTGTGTGT GTGTGTACTG TTTGTATGCC TGGTGCCTGC

T

Igh LOCUS T

TTTOTCTTGCI OATCIOCCCCC TGGGTCATGA GAAGACCACT GCCAAAGAAC CTTCTCTTAG CGCACTGAAC CCCTCCCCTG - l367bp) - AAAdAGddGd AddGCCCCCC CCCCGCCCCA GGAGGGCAGG AGACAGTGCA GAGGAGAACA -(147BP)- CTCCCTCACT CCTTTCCTTG - (262bp) ~ GdGddAGCCG GGGAAGAAAA GdGCGGdCCT GAGAGGAddG GGdGCTTTCT

.""" ~.. . . ~ """ ~ ~~~~ ~ ~~ ~ ~

CTCCTTCTCC TGTTTCCTTC TCCTCTTCCC TCTTTACAGG

FIG. 1. DNA sequences of the L l R n 3 e l e m e n t target s i t e (7), the Mlvi-2 locus (23), a n d the rat Igh locus (K. Usdin and A. V. Furano, unpublished data (EMBL accession number X15742)). A, the L1Rn3 target site. The (AT).-rich region is underlined. R, the Mlvi-2 locus. The (GT), sequence is underlined. C, the rat Igh locus. The major po1ypurine:polypyrimidine sequence is underlined. The minor ones are shown in italics. The arrowheads in all panels indicate the sites of L1 element insertion. The sequence of the LlRn3 target site was deduced from the sequence of the DNA flanking the LlRn3 element (7, 24). The sequence of the Mlvi-2 and the Igh target sites was determined by sequencing of the empty (unoccupied) loci (Ref. 23, and K. Usdin and A. V. Furano, unpublished data (EMBL accession number X15742). The point of L1 element insertion in each case was determined by comparison of the sequence of the empty site with that of the corresponding occupied site. In the case of the Mlvi-2 locus 14 bp of target site DNA was duplicated at the point of insertion (24). In the case of the Igh locus, 4 bp (CTAT) of non-L1 DNA was inserted at the site of L1 element integration.

A Hindlll Pvul Hindlll

A

pRl5 Hindlll EcoRlStyI StylHindlll (I@)

PTs41 BglU Hindlll EcoRl Hindlll EcoRl Rglll

Y I A

(Mlvi-2) 1 A

U Ikb

I{ Mw M pL3TS pRIS pTS4 I (kb) (LIRn?) (I@) (Mvi-2)

' 1 2 " 3 4 5 116 7 8 " 9 10 1 1 '

6.56 -I 4.36 -1

2.32 - ~

2.07 - ' 1.35 - 1.08 - 0.87 - 0.60 - O.5h - d

FIG. 2. De tec t ion o f b romoace ta ldehyde r eac t ive s i t e s within three d i f fe ren t L1 element target sites. A, line drawings of the three plasmids used. The solid hars indicate target site DNA and the thin black lines the vector sequences. The nrrofoheads indicate the sites of L1 element insertion. R, aliquots of the plasmids pL.?TS, pRI.5, and pTS41 were digested with HindIII (pL3TS and pR1.5) or RglII (pTS41). These samples were treated with hromoacetaldehyde alongside aliquots of the supercoiled plasmids as described under "Materials and Methods." The supercoiled plasmids were then digested with the appropriate restriction enzyme, and all the samples were then labeled at their 3' ends with Klenow polymerase and dNTPs including [w'"P]dCTP. A portion of each labeled sample was treated with S1 nuclease and subjected to electrophoresis alongside untreated samples on a 0.9% agarose gel. Lane 1, X, HindIII digest; lane 2, rbxli4, HaeIII digest; lanes 3-.5, pL3TS; lanes 6-8, pR1.5; and hnes 9-11, pTS41. Lanes 3, 6, and 9 linear DNA molecules treated with hromoacetaldehyde and SI nuclease. Lanes 4, 7, and IO, supercoiled DNA treated with bromoacetaldehyde, and digested with an appropriate restriction enzyme. Lanes 5 , 8, and 11, supercoiled DNA treated with hromoacetaldehyde, digested with an appropriate restriction enzyme and then treated with S1 nuclease. In the case of the Igh locus (pRI5), a numher of minor bromoacetaldehyde reactive sites are ohserved. These are consistent with the relatively large numher of short po1ypurine:polypyrimidine tracts found in this region.

the plasmid molecule at a site (or sites) that has reacted with bromoacetaldehyde, since no such bands result from S1 cleavage of similar substrates that have not been treated with bromoacetaldehyde (data not shown) or in samples that were not treated with S1 nuclease (lanes 4, 7, and IO). This reactivity is supercoil-dependent as indicated by the lack of bromoacetaldehyde reactivity, and consequent S1 nuclease sen- sitivity, of linear molecules (lanes 3, 6, and 9).

The bromoacetaldehyde reactive region within each of these target sites was mapped to within about 20 bp on polyacrylamide gels by reference to a number of restriction sites. The results were confirmed by the construction of various dele- tions (data not shown). The bromoacetaldehyde reactivity was mapped to a region containing a (GT)26 tract in the Mlvi- 2 locus, an area containing a 55-bp poly-purine: polypyrimidine tract in the Igh locus and a region rich in

20738 Ll Elements and Their Target Sites

short mixed (AT), tracts in the L1Rn3 element target site (underlined in Fig. 1). The LlRn3 element target site was mapped at the nucleotide level by modifying the bromoacetaldehyde-reacted bases with sodium acetate (pH 5.0), and then cleaving the phosphodiester backbone of the DNA molecule at those modified bases by treatment with piperidine ( 2 5 ) . The results obtained are shown in Fig. 3, and are summarized in Fig. 4. These results were confirmed by primer extension on bromoacetaldehyde-treated templates as described previously (%), and match the S1 nuclease cleavage pattern for bromoacetaldehyde-treated molecules (data not shown).

STRAND: TOP BOTTDM

BAA : - - + + - + + " Mp2+ : + - + -

1 2 3 4 5 6 7

7T

T T T A T A T T A T A T A T A T A A T A

A T

T A T T A T

A A

A T

T

FIG. 3. Fine mapping of the bromoacetaldehyde reactive sites in the LlRn3 target site. pL3TSl was treated with bromoacetaldehyde at pH 5.0 with or without 2 mM Mg', digested with HindIII, treated with either Klenow polymerase and [a-:"P]dCTP to label the 3' ends, or T4 polynucleotide kinase and [y-"PIATP fol- lowed by filling in with Klenow polymerase and nonradioactive dNTPs, to label the 5' ends. The DNA was then modified by reaction with 0.1 M sodium acetate (pH 5.0), cleaved with piperidine as described previously, and subjected to electrophoresis alongside a Maxam and Gilbert sequencing ladder on a 5% polyacrylamide sequencing gel. In this reaction Gs that have not been modified by bromoacetaldehyde are also cleaved. Lanes 2-4, top strand; lanes 5- 7 , bottom strand. Lane 1, T + C cleavage; lane 2, sodium acetate- piperidine treated DNA (no bromoacetaldehyde); lanes 3 and 4 , bromoacetaldehyde-treated DNA (pH 5.0 with and without M e , respectively); lanes 5-7, same as lanes 2-4 but for the bottom strand.

A

FIG. 4. Schematic representation of the bromoacetaldehyde reactivity of the LIRn3 target site. The bromoacetaldehyde- reactive bases in pL3TSl at pH 5.0 in the presence of M P are indicated by uertical lines on the sequences in A, and the bromoacetaldehyde-reactive bases detected at pH 5.0 in the absence of Mg2' are shown in B. The thicker lines indicate more strongly reactive bases which most likely represent the loops of the several cruciform structures tha t are formed in this region. The more weakly reactive bases might represent bases in the loop of less stable alternate cruciforms, various stable intermediates in the cruciform extrusion process, or both.

The reactive bases in the LlRn3 target site fall predominantly within the mixed (AT), tract referred to above. In the presence of Mg', a number of discrete sites within this region are bromoacetaldehyde-reactive. The pattern of reactivity is most consistent with either the formation of a number of different cruciform structures in which the most reactive bases fall within the loops of the various stem and loop structures formed on each strand, or with the existence of a number of stable intermediates in the cruciform extrusion pathway. In the absence of M$+, the reaction with bromoacetaldehyde is stronger and involves more bases. The formation of these structures is probably pH-independent since the same reaction pattern is observed at pH 5.0 (Fig. 3) and pH 7.5 (data not shown).

The Non-B DNA Structures Formed by a n Ll Element and Its Target Site Can Interact-We previously showed that the G-rich po1ypurine:polypyrimidine sequence at the right end of the rat L1 element adopts several non-B conformations including two triplexes (25) as shown by reactivity with bromoacetaldehyde and other chemical reagents. Therefore L1 element insertion into a target site could bring two potential non-B DNA forming sequences into the same topological domain, where they could perhaps interact by competing for available supercoil energy.

In order to investigate this possibility, we studied the bromoacetaldehyde reactivity of pSLSAAB, a plasmid that contains both the LlRn3 po1ypurine:polypyrimidine sequence and the LlRn3 DNA target site. Fig. 5 shows that when pSL3AAB is treated with bromoacetaldehyde, digested with EcoRI and then treated with S1 nuclease, the ethidium bromide-stained bands labeled Lla, Llb, TSa, TSb, and I are produced. Bands labeled TS are produced by cleavage at the bromoacetaldehyde-reactive region in the target site, and bands labeled L1 correspond to the expected fragments that would be produced by cleavage at the bromoacetaldehyde- reactive region in the L1 element. The I band corresponds to the fragment that would be generated by cleavage a t both bromoacetaldehyde-reactive sites on the same plasmid.

The relative amount of each band and hence the number of molecules containing a non-B structure a t one or other site (or both) depends on conditions such as pH and the presence or absence of M$+ and/or Na'. At pH 5.0 with Mp", the non-B DNA structure in the LlRn3 element is readily formed


A

M W p H 5 0 pH75

B kb

~ 1 2 . 7 4 5 6 7 8

23.1

9.4

6.6

4.4

2.3 2. I

C

FORMATION OF A NON-B DNA STRUCTURE

at GHP region I at target site

(Mg2+) PH5.0 +++ p H 5 0 +++

(no Mg2+)

pH75 + pH73 -H"l-

(no Mg2+ or Na+)

+ +++ +++ +++

FIG. 5. The effect of pH and Mg" on the bromoacetaldehyde reactivity of the right end of the LlRn3 element and its target site. A, pSL3AAB was treated with bromoacetaldehyde at the indicated pH values, with (+) and without (-) 2 mM MgCI, (and Na' in the case of lane 4), digested with EcoRI, treated with S1 nuclease, and subjected to electrophoresis on a 0.8% agarose gel in Tris acetate buffer a t 100 V for 4 h as described under "Materials and Methods." The labels Lla and Llb indicate the two bands that result from bromoacetaldehyde reaction in the L1Rn3 element. The labels TSa and TSb indicate those bands produced after reaction a t the target site. The letter I indicates the band produced after bromoacetaldehyde reaction a t both sites. The letter L corresponds to the linear plasmid. The same symbols and labels are used in B. The portion of the photograph shown within curued brackets is an overexposure to allow visualization of band I. B, the possible S1 cleavage products for bromoacetaldehyde-treated pSL3AAB. The open bar indicates L1 DNA, the solid bar, target site DNA, and the thin black line, the vector sequences. The open arrowhead indicates the position of the LlRn3 element site of non-B DNA forming potential, the filled arrowhead indicates the position of the region of non-B DNA forming potential in its target site. C , table indicating the relative amounts of non-B DNA structure formed a t each site. The acronym GHP is used to indicate the L1 element G-rich homopurine (po1ypurine:polypyrimidine) sequence. The number ofplus signs indicates approximately how much of one non-B DNA structure is formed relative to the other, under the same reaction conditions.

(lane I). The relatively small amount of TSa and the approximately equivalent amounts of bands Llb and TSb indicates that in addition to molecules reactive only at the L1 element site (bands Lla and Llb), many of the molecules reactive at the target site are also reactive at the L1 site. At this pH, in the absence of M$+ (Fig. 5, lane 2), all molecules which have reacted at the L1 DNA site, also contain a non-B structure in the target site as indicated by the disappearance of band L l b (lane 2). This experiment also shows that significantly more molecules are reactive with bromoacetaldehyde in the absence of M$+ than in its presence (cf. the relative amounts of the intact plasmid and the S1 fragments in lanes 1 and 2). At pH 7.5 in the presence of M$+ and Na' (lane 3), the alternate structures in the target site predominates (the bands labeled TS are stronger than the L1 bands), but in the absence of both ions, almost all the molecules are reactive a t both sites simultaneously (lane 4).

The size of the fragments in band I varies with pH and the presence or absence of M e and/or Na'. At pH 7.5, in the absence of Mg2' and Na+, these fragments are about 200 bp shorter than at pH 5.0 in the presence of M e . This difference is probably due to a combination of two factors. (i) Fine mapping of the L1 non-B DNA forming region indicates that a t pH 5.0 the major reactivity is in the region of the polypurine:polypyrimidine sequence involved in triplex formation,

whereas a t pH 7.5 sequences 3' of this region are more strongly reactive (results not shown). (ii) Fine mapping of the bromoacetaldehyde-reactive bases in the target site indicate that the non-B DNA structure formed in this region is more extensive in the absence of M$+ than in its presence (Fig. 3).

The above results indicate that two non-B structures can be formed on the same negatively supercoiled molecule. The data also show the extent to which the formation of each structure can be modulated by experimental conditions. To determine whether these results were due to an interaction between the two non-B DNA structures, we carried out two sets of experiments: in one we compared the behavior of the L1 po1ypurine:polypyrimidine tract in the presence or absence of the target site non-B DNA forming region and in the second we compared the effect of reducing the size of the topological domain (and thus the amount of available negative supercoil energy) on the formation of the L1 and the target site structures.

To make quantitative comparisons, we used end-labeled restriction fragments and determined the relative amount of each S1 digestion fragment by densitometric scan of autora- diographs of the polyacrylamide gels. In these experiments the fragment corresponding to band I in Fig. 5, would not be detected since it would not contain a labeled restriction enzyme site.


Fig. 6 shows that the presence of the target site has very little effect on the amount of non-B DNA formed by the L1 element under conditions that favor triplex formation by the pol.ypurine:polypyrimidine tract (the presence of M$+; pH 5.0). The amount of reaction is reduced by only 15% (cf. lanes 1 and 2). However, under conditions that do not favor triplexes (the absence of Mg"; pH 7.5), the presence of the target site reduces the amount of bromoacetaldehyde reactivity in the L1 sequence by 62% (cf. lanes 7 and 8). This result shows that there is direct competition between the two non- B DNA forming regions for the available supercoil energy.

There is little bromoacetaldehyde reactivity in the L1 element at pH 7.5 in the presence of Mg'" whether or not the target site was present (Fig. 6, lanes 5 and 6). These results are very different to those obtained in the absence of M C a t this pH (lanes 7 and 8). We presume that other deformable sites in the remainder of the molecule are favored under these conditions and compete with the L1 sequence for supercoil energy. The large number of minor bromoacetaldehyde-reactive bands seen in molecules treated in the presence of M$+ support this idea (cf. lane 5 or 6 with lane 7 or 8, Fig. 6).

Fig. 7 shows the effect of altering the size of the topological

A kb I l l l l l l l ,

0 1 2 3 4 5 6 7 8

pSL3AAB Ava I

L L II L l b I I 1 sa I1 TS b I UL!JM

Ava I

pSL3AABADX Ava 1 V Ava I

I Lla 11 Llb' I

B

M pH 5.0 pH 7.5

I + - II + - I

1 2 3 4 5 6 7 8

Llh Llh" - - - - TSh

. Llh'

TSa ,

Lla

- TSa

- Lla

FIG. 6. The effect of the presence of the target site on the bromoacetaldehyde reactivity of plasmids containing the LlRn3 bromoacetaldehyde-reactive site. A shows the two plasmids used. The symbols and labels correspond to those described in the legend to Fig. 5 except that the bromoacetaldehyde-dependent S1 nuclease products of pSL3AABADX are labeled Lla and Llb', respectively. R shows the bromoacetaldehyde reactivity of the two plasmids. The bromoacetaldehyde reactions were carried out at the indicated pH values with (+) and without (-) 2 mM MgCI,, the plasmids were digested with AuaI, end labeled with Klenow polymerase and [n-"'PIdCTP, treated with S1 nuclease, and subjected to electrophoresis on a 1% agarose gel as described under "Materials and Methods." Lanes 1, 3, 5 , and 7, pSL3AAB; lanes 2, 4, 6. and 8, pSL3AARADX.

A kh

pSL3AA Ava 1

I A v a I

I L13 II I.lh I

pSL3AAB Ava I I V T L Ava I

I Lla II L l h I I TSa 11 I TSh

I -L!J pSL3AABA

Lla", AvuAva ' U- -1 TSh I U ! A U

B pSL3AA pSL3AAR pSL3AARA ' pH 5.0 pH 7 .5 ' ' pH 5 . 0 pH 7.5 I ' pH 5.0 pH 7.5 I

1 2 3 4 5 6 7 8 ') IO I I 1.7 r + - 1 1 + _ I I + - 1 1 + - I r+ - - I + - 1

L - m-m TSa - Lla -

Lla - 'Sh -

rsa -

FIG. 7. The effect of pH and MgCl, on the bromoacetaldehyde reactivity of various sized plasmids that contain both the right end of the LlRn3 element and its target site. A shows line drawings of the three plasmids used. The position of the restriction site a t which these plasmids were cut and labeled is indicated. The symbols and labels used are the same as described in the legend to Fig. 5. B shows the results obtained after these plasmids were treated with bromoacetaldehyde at the indicated pH values, in the presence (+) or absence (-) of 2 mM MgCl,, digested with AuaI, end- labeled with Klenow polymerase and [tr-:"P]dCTP, treated with S1 nuclease, and subjected to electrophoresis on a 1% agarose gel.

domain and hence the number of negative supercoiled turns (7) on the amount of each non-B DNA structure formed. To do this we used plasmids of different lengths which at normal bacterial superhelical density ( 5 = -0.06), would have the following values of T: pSL3AA (12 kb), ? = 69; pSL3AAB (7.5 kb), T = 43; and pSL3AABA (2.9 kb), T = 17. In pSL3AA and pSL3AAB the proportion of each non-B DNA structure formed is similar and the results are consistent with those shown in Fig. 5 (Fig. 7, lanes 1-8). The fragment L la in molecules reacted at pH 7.5 is slightly longer than the Lla fragment produced from molecules reacted at pH 5.0 in the presence of Mg2' for the reason discussed earlier in reference to the size of Band I (Fig. 5). Both L la bands are seen in molecules a t pH 5.0 reacted with bromoacetaldehyde in the absence of Mg", a condition which significantly reduces the amount of triplex in the po1ypurine:polypyrimidine region (e.g., lane 6, and Ref. 25).

In the case of pSL3AABA which is only 2.9 kb in size, only the target site non-B structure is observed (Fig. 7, lanes 9- 12). We surmised that the lack of detectable bromoacetalde-


hyde reactivity in the polypurine:pol~ypyrimidine region of pSL3AAB.l was due to a limitation in the amount of available supercoil energy in a plasmid of this size. This implies that the target site is more readily deformed than the L1 site and that once deformation takes place insufficient supercoil energy remains for non-B DNA formation at the L1 site. In order to test this idea we examined the energetics of the non- B DNA transitions using two-dimensional gel electrophoresis of topoisomers of plasmids containing either the target site alone (pL3TSA), or the LlRn3 po1ypurine:polypyrimidine sequence (UC15). The plasmid, UC15 contains almost the entire length of the L1 po1ypurine:polypyrimidine sequence from pSL3AA cloned into pUC19, and the non-B structure of this plasmid does not differ from that of the full length po1ypurine:polypyrimidine sequence (25). It was necessary to use Uc15 because we were unable to propagate small molecules containing the full length po1ypurine:polypyrimidine region in the absence of the target site sequence in E. coli.

Fig. 8 shows the results of these experiments. Electropho- resis in the first dimension (top to bottom) was carried out under acidic or neutral conditions that approximated those used for the structural analysis (see “Materials and Meth- ods”), and the second dimension (left to right) was performed in the presence of chloroquine. A spur in the curve of topoisomers (indicated with the arrowhead in Fig. 8) is diagnostic of a non-B DNA transition, and the number of superhelical turns required for the non-B DNA transition can be determined by counting the bands that separate the spur from the band of molecules with T of 0. The mobility shift caused by the non-B DNA transition indicates the number of superhelical turns that are unwound (for a more complete discussion

Targcr Site plasmid 1-1 ZTzht

LI plasmid ( 1 n o

+v(F:- .Us:* +Mg?*

FIG. 8. Two-dimensional electrophoretograms of plasmids pL3TSA (a, b, d, and e) and UC15 (c and t), at pH 5.0 (a-c) and pH 7.5 (d-0. with (a, c, d, and t) and without ( b and e) Mg2+. Electrophoresis was carried out as described under “Materials and Methods”; the first dimension was from top to bottom and the second dimension was from left to right. The second dimensions of the gels shown in a, b, d, e and f were carried out in the presence of 5 pg/ml chloroquine, and the second dimension of the gel in c was carried out in the presence of 4 pg/ml chloroquine. The photographs in a, b, and d-fare of ethidium bromide-stained gels. c is a photograph of a contact print made from an autoradiogram of a gel that was transferred to nitrocellulose and probed with a homologous probe. The “0” refers to molecules with T = 0. In this set of experiments the UC15 molecules appear to have a 0 higher than that of the other plasmids used. We believe that this is not a plasmid-specific phenom- enon but is related instead to the relative amount of ethidium bromide used in the generation of the topoisomer series. We have no evidence to suggest that the 5 of UC15 differs from the other plasmids in uiuo (data not shown).

on the interpretation of two-dimensional gels see Refs. 33- 35).

In this way we were able to determine that at pH 5.0 in the presence of M$+, the target site sequence undergoes a broad transition to a non-B DNA conformation starting at a superhelical density of -0.040 (?. = 10 in the case of pL3TSA) and results in the loss, once fully extruded, of about four superhelical turns (Fig. 8a) . This indicates that about 40 bp are involved in the non-B DNA structure(s) under these conditions (consistent with the data shown in Figs. 3 and 4). The broadness of the transition also suggests that the complex pattern of bromoacetaldehyde reactivity observed (Figs. 3 and 4) may be due, a t least in part, to the existence of stable intermediate(s) in the extrusion pathway. At pH 5.0 in the absence of Mg?’ (Fig. 86) the transition of the target site sequence, which is also broad, occurs at a higher superhelical density, but involves greater unwinding (six rather than four turns, i.e. 60 bp rather than 40 bp; Fig. 8b).

At pH 5.0 in the presence of Mg?+, the L1 polypurine:polypyrimidine non-B transition occurs a t a higher superhelical density than the non-B DNA transition in the target site (3 = -0.052 or ? = 14 for UC15) (Fig. 8c) and removes only two to three superhelical turns. The lower superhelical density required for extrusion of the target site non-B DNA structure means that the target site structure would form more readily than the L1 polypurine: polypyrimidine triplexes under these conditions. In larger plasmids of native bacterial superhelical density that contain both sites removal of four turns on extrusion of the cruciform(s) would leave more than sufficient superhelical density to allow triplex formation, cf. pSL3AAB (Fig. 7, lane 5 ) , but in smaller plasmids such as pSL3AABA, this would result in the reduction of the average plasmid superhelical density from around -0.060 (7 = 17) to -0.047 (T 13), which is less than that required for non-B DNA formation by the L1 element (Fig. 7, lane 9). This explains why a non-B DNA transition is not detected for this plasmid even under favorable conditions using DNA of native bacterial superhelicity.

At pH 7.5, a non-B DNA transition is still observed in two- dimensional gels for pL3TSA, a t values of 3 lower than the average bacterial superhelical density ( a = -0.060) (Fig. 8, d and e ) . While the amount of unwinding associated with the transition is the same as at pH 5.0, the point in the topoisomer series a t which the transition occurs is displaced two topoisomers up the series, i.e. the transition occurs at a slightly higher superhelical density at this pH (3 = -0.046 in the presence of M e ) . This is somewhat surprising since proton- ated bases are presumably not involved in the cruciform(s), and perhaps reflects a general destabilization of the DNA duplex associated with protonation. Curiously, the non-B DNA transition also appears sharper a t pH 7.5, than at pH 5.0 (cf. Fig. 8, a, b, d, and e ) .

At pH 7.5, Uc15 shows a small non-B DNA transition at 3 = -0.066 (Fig. Sf). Since at this pH in the presence of M F , the bromoacetaldehyde-reactive sites of UC15 molecules a t normal bacterial superhelical density (3 = -0.060) are located mainly in the vector and not in the L1 po1ypurine:polypyrimidine region (25), we attribute this transition to a site in the vector. Whatever abnormal structure exists in the L1 po1ypurine:polypyrimidine region at pH 7.5 in the presence of Mg2‘ (e.g. Fig. 6, lane 6 (25)), it is either too unstable to detect by two-dimensional gel electrophoresis or it involves little or no unwinding of the helix.

DISCUSSION

We have previously shown that the po1ypurine:polypyrimidine sequence at the right end of the rat L1 element


adopts several non-B DNA structures as assessed by reaction with bromoacetaldehyde, a reagent to which Watson-Crick duplex is unresponsive (25). In this paper we report that portions of three L1 element target (insertion) sites are also bromoacetaldehyde-reactive, i.e. they can also adopt non-B DNA conformations. These target sites are (i) a site in the Mlvi-2 locus (23) (ii) a site in the immunoglobulin heavy chain locus (23, 25), and (iii) the target site for the LlRn3 element (7).

The Mlvi-2 locus contains a (GT)26 sequence situated within 37 bp of the site of L1 element insertion. Our data indicate that a non-B DNA conformation is readily adopted by this region in vitro. Various alternating purine and pyrimidine sequences can adopt Z DNA conformations under certain circumstances (36-39). While this work was in progress the bromoacetaldehyde-reactive nature of B-Z junctions (40, 41) and of 2-DNA stretches themselves (42) was reported. The non-R DNA conformation adopted by the (GT)26 sequence is probably of the Z DNA-type.

The major non-B DNA forming site in the Igh locus maps to the po1ypurine:polypyrimidine rich sequence located 921 bp downstream of the L1 element site of insertion. While this is relatively far from the insertion point, the intervening region contains many shorter po1ypurine:polypyrimidine tracts (e.g. those situated 19, 389, and 676 bp from the point of L1 DNA insertion) that react to a lesser extent with bromoacetaldehyde (see minor bands Fig. 2, lune 3 ) , and that may contribute to the distortion of the helix. The non-B DNA structure of this region might involve the formation of a triplex or some other non-B structure (25).

In the case of the LlRn3 element target site, the non-B DNA structure was found to be confined predominantly to a 54-bp area rich in short, mixed (AT),, tracts located 256 bp away from the site of the LlRn3 element insertion (Fig. 1). It has been suggested that long alternating (AT), sequences adopt an altered conformation that is prone to torsional deformation in vitro (43). Such sequences are thought to undergo subsequent cruciform extrusion under certain conditions (44-48). Our data indicate that such behavior can also be adopted even by short mixed (AT),, tracts where we have evidence from fine mapping of the bromoacetaldehyde-reactive sites, for the formation of either a number of different cruciforms or a number of stable intermediates in the extrusion process. The nature of the non-B DNA transition re- vealed by two dimensional gels of plasmids containing the target site, i.e. a spur that under certain conditions is broad rather than sharp, would be consistent with the latter interpretation.

Since the non-B DNA structures adopted by the polypurine:polypyrimidine sequence of the L1 element (see “Results” and Ref. 25) and each of the three target sites examined here are supercoil-dependent, we would predict that when these structures are present within the same topological domain, they would compete for the available supercoil energy (49). This competition could affect the transitional behavior of each region even if separated by a considerable distance. TO examine this interaction we analyzed the formation of non-B DNA structures by the L1Rn3 element and its target site when present on the same supercoiled molecule. Using bromoacetaldehyde reactivity as a probe for non-B DNA, we found that although under certain conditions both sites on a single molecule can be quite reactive (Fig. 5), there is a hierarchical relationship in their competition for negative supercoil energy which is determined by the experimental conditions. For example, at pH 5.0, in the presence of Mg2‘ and sufficient negative supercoil energy, the non-B DNA

structure(s) in the L1 element predominate and the bromoacetaldehyde reactivity of this site is hardly altered by the presence of the target site (Fig. 6, lunes 1 and 2). These conditions favor triplex formation by the L1 polypurine:polypyrimidine sequence (25). Under conditions unfavor- able to these structures (pH 7.5), the target site significantly reduces the bromoacetaldehyde reactivity in the L1 sequence (Fig. 6, lunes 7 and 8).

When supercoil energy is limiting (e.g. in the 2.9-kb plasmid pSL3AABA), the cruciform structure(s) adopted by the target site predominates and precludes detectable deformation of the L1 po1ypurine:polypyrimidine sequence (Fig. 7). These results indicate that under all conditions, deformation of the target sequence occurs at lower superhelical density than the L1 sequence, and that once formed the residual superhelical energy in molecules as short as 2.9 kb would be below the threshold required for deformation of the L1 sequence even under the most favorable conditions (pH 5.0, with Mg+). These conditions were supported by the results obtained by two-dimensional gel analysis of the non-B DNA transitions of the target site and L1 element (Fig. 8).

In conclusion, bromoacetaldehyde reactivity, and thus an abnormal structure, appears to be a common feature of rat L1 target sites. This non-B DNA character could affect chro- matin structure and the susceptibility of this region to endog- enous nucleases. Thus these sites may be prone to strand scission making them good substrates for illegitimate recombination events such as the insertion of an L1 element (50- 52). One consequence of L1 DNA insertion would be to bring into the same topological domain two potentially deformable sequences that were once isolated. Competition between these two sites for supercoil energy could be modulated in vivo by variations in pH, cations, the presence of cellular components such as spermine, spermidine and histones, and by variations in the size of the topological domain or the superhelical density (e.g. during transcription (53)), that may favor one or other structure. This competition could affect both the behavior of the L1 po1ypurine:polypyrimidine sequence and that of those L1 element target sites whose biological properties are sensitive to variations in supercoil energy, and this effect might operate over considerable distances.

REFERENCES 1. Manuelidis, L. (1980) Nucleic Acids Res. 8, 3247-3258 2. Meunier-Rotival, M., Soriano, P., Cuny, G., Strauss, F., and

Bernardi, G. (1982) Proc. Natl. Acad. Sci. U. S. A. 79, 355-359 3. Grimaldi, G., Skowronski, J., and Singer, M. F. (1984) EMBO J.

4. Hattori, M.. Hidaka, S., and Sakaki, Y. (1985) Nucleic Acids Res. 3,1753-1759

13, 781317827 5. Soares, M. B., Schon, E., and Efstratiadis, A. (1985) J. Mol. Euol.

22,117-133 6. Burton, F. H., Loeb, D. D., Voliva, C. F., Martin, S. L., Edgell,

M. H., and Hutchinson, C. A., 111. (1986) J. Mol. Biol. 187,

7. D’Ambrosio, E., Waitzkin, S. D., Witney, F. R., Salemme, A., and

8. Fanning, T., and Singer, M. (1987) Nucleic Acids Res. 15, 2251-

9. Fanning, T., and Singer, M. F. (1987) Biochirn. Biophys. Acta

291-304

Furano, A. V. (1986) Mol. Cell. Biol. 6,411-424

2260

910,203-212 10. Rogers, J. H. (1985) Znt. Reu. Cytol. 93, 187-279 11. Singer, M. F., and Skowronski, J. (1985) Trends Biochern. Sci.

12. Economou-Pachnis, A., Lohse, M. A., Furano, A. V., and Tsichlis,

13. Lakshmikumaran, M. S., D’Ambrosio, E., Laimins, L. A., Lin, D. T., and Furano, A. V. (1985) Mol. Cell. Bid. 5 , 2197-2203

14. Burton, F. H., Loeb, D. D., Chao, S. F., Hutchison, C. A., 111, and Edgell, M. H. (1985) Nucleic Acids Res. 13, 5071-5084

10,119-122

N. (1985) Proc. N ~ t l . Acad. Sci. U. S. A. 82, 2857-2861


15. Hattori, M., Kuhara, S., Takenaka, O., and Sakaki, Y. (1986) Nature 321, 625-628

16. Loeb, D. D., Padgett, R. W., Hardies, S. C., Shehee, W. R., Comer, M. W., Edgell, M. H., and Hutchison, C. A., 111. (1986) Mol. Cell. Biol. 6, 168-182

17. Katzir, N., Rechavi, G., Cohen, J. B., Unger, T., Simoni, F., Segal, S., Cohen, D., and Givol, D. (1985) Proc. Natl. Acad. Sci. U. S.

18. Kazazian, H. H., Jr., Wong, C., Youssoufian, H., Scott, A. F., Phillips, D. G., and Antonarakis, S. E. (1988) Nature 332,

19. Morse, B., Rothberg, P. G., South, V. J., Spandorfer, J. M., and

20. Fanning, T. G. (1983) Nucleic Acids Res. 11, 5073-5091 21. Voliva, C. F., Jahn, C. L., Comer, M. B., Hutchison, C. A., 111,

and Edgell, M. H. (1983) Nucleic Acids. Res. 11, 8847-8859 22. Voliva, C. F., Martin, S. L., Hutchison, C. A,, 111, and Edgell, M.

H. (1984) J. Mol. Biol. 178, 795-813 23. Skowronski, J., and Singer, M. F. (1985) Proc. Natl. Acad. Sci.

U. S . A. 82,6050-6054 24. Furano, A. V., Somerville, C. C., Tsichlis, P. N., and D’Ambrosio,

E. (1986) Nucleic Acids Res. 14, 3717-3727 25. Usdin, K., and Furano, A. V. (1989) J. Bid. Chem. 264, 15681-

15687 26. Sinden, R. R., Broyles, S. S., and Pettijohn, D. E. (1983) Proc.

Natl. Acad. Sci. U. S. A. 80, 1797-1801 27. Secrist, J. A,, 111, Barrio, J. R., Leonard, N. J., Villar-Palosi, C.,

and Gilman, A. C. (1977) Science 177, 279-280 28. D’Ambrosio, E., and Furano, A. V. (1987) Nucleic Acids. Res. 15,

3155-3175 29. Maniatis, T., Fritsch, E. F., and Sambrook, J. (1982) Molecular

Cloning: A Laboratory Manual, Cold Spring Harbor Laboratory, Cold Spring Harbor, NY

30. Collier, D. A,, Griffin, J . A., and Wells, R. D. (1988) J . Bid. Chem. 263,7397-7405

31. Kowhi-Shigematsu, T., Gelinas, R., and Weintraub, H. (1983) Proc. Natl. Acad. Sci. U. S. A. 80, 4389-4393

32. Kowhi, Y., and Kowhi-Shigematsu, T. (1988) Proc. Natl. Acad. Sci. U. S. A. 85, 3781-3785

A. 82, 1054-1058

165-166

Astrin, S. M. (1988) Nature 333, 87-90

33.

34.

35.

36.

37.

38.

39.

40. 41.

42.

43.

44.

45. 46. 47.

48.

49.

50.

51.

52.

53.

Peck, L. J., and Wang, J. C. (1983) Proc. Natl. Acad. Sci. U. S. A. 80,6206-6210

Lee, C.-H., Mizusawa, H., and Kakefuda, T. (1981) Proc. Natl. Acad. Sci. U. S. A. 78, 2838-2842

Gellert, M., O’Dea, M. H., and Mizuuchi, K. (1983) Proc. Natl. Acad. Sci. U. S. A. 80, 5545-5549

Wells, R. D., Miglietta, J . J., Klysik, J., Larson, J. E., Stirdivant, S. M., and Zacharias, W. (1982) J. Bid. Chem. 257, 10166- 10171

Nordheim, A,, and Rich, A. (1983) Proc. Natl. Acad. Sci. U. S. A.

Singleton, C. K., Kilpatrick, M. W., and Wells, R. D. (1984) J.

Haniford, D. B., and Pulleyblank, D. E. (1983) Nature 302,632-

Htun, H., and Dahlberg, J. E. (1989) Science 243, 1571-1576 McClellan, J . A., Palecek, E., and Lilley, D. M. J . (1986) Nucleic

Kowhi-Shigematsu, T., Manes, T., and Kowhi, Y. (1987) Proc.

McLean, M. J., Larson, J. E., Wohlrab, F., and Wells, R. D.

Vogt, N., Marrot, L., Rousseau, N., Malfoy, B., and Leng, M.

Lilley, D. M. J. (1985) Nucleic Acids Res. 13, 1443-1465 Sullivan, K. M., and Lilley, D. M. J. (1986) Cell 47,817-827 Sullivan, K. M., and Lilley, D. M. J. (1986) J. Mol. Biol. 193,

Murchie, A. I. H., and Lilley, D. M. J. (1987) Nucleic Acids Res.

Ellison, M. J., Fenton, M. J., Ho, P. S., and Rich, A. (1987)

Orr-Weaver, T. L., Szostak, J. W., and Rothstein, R. J. (1981)

Lin, F.-L., Sperle, K., and Sternberg, N. (1984) Mol. Cell. Biol. 4,

Song, K.-Y., Chekuri, L., Rauth, S., Ehrlich, S., and Kucherlapti,

Wu, H-Y., Shyy, S., Wang, J. C., and Liu, L. F. (1989) Cell 53,

80,1821-1825

Biol. Chem. 259,1963-1967

634

Acids Res. 14,9291-9309

Natl. Acad. Sci. U. S. A. 84, 2223-2227

(1987) Nucleic Acids Res. 15, 6917-6935

(1988) J . Mol. Biol. 201, 773-776

397-404

15,9641-9654

EMBO J. 6,1513-1522

Proc. Natl. Acad. Sci. U. S. A. 78, 6354-6358

1020-1034

R. (1985) Mol. Cell. Biol. 5, 3331-3336

433-440

insertion of l1 elements into sites that can form non-b dna

Documents