The Ability to Form Intrastrand Tetraplexes is an Evolutionarily Conserved Feature of the 3’ End of Ll Retrotransposons

R. Howell’ and K. Usdin Section on Genomic Structure and Function, Laboratory of Molecular and Cellular Biology, National Institute of Diabetes, and Digestive, and Kidney Diseases, National Institutes of Health, Bethesda, Maryland

Mammalian genomes contain many thousands of members of a family of retrotransposons known as L 1 (or LINE- 1) elements. These elements lack long terminal repeats (LTRs), and are thought to use a retroposition mechanism that differs from that of retroviruses and other LTR-containing retroelements. In order to define those regions of the Ll element that may be important for Ll retroposition, we examined the 3’ untranslated regions (UTRs) of Ll elements from a diverse group of mammals. We show that while the 3’ UTRs of Ll elements from different species share little if any sequence homology, they all contain a G-rich polypurine tract of variable length and sequence which can form one or more intrastrand tetraplexes. This conservation over the 100 Myr since the mammalian radiation suggests that either the G-rich motif itself or a conserved structure such as the tetraplex that can be formed by this motif is a significant structural feature of Ll elements and may play a role in their propagation.

Introduction

Ll elements are mobile genetic elements that are found in all species of placental mammals and in many marsupials (Burton et al. 1986). It is believed that an ancestral version of these elements was present in the last common ancestor of these animals that lived during the late Cretaceous or early Tertiary period about lOO- 150 MYA. After the mammalian radiation, Ll elements continued to generate additional copies of themselves such that they now comprise about lo%-30% of the genetic material of mammals (Usdin et al. 1995). Inser- tion of these copies into new sites in the genome (trans- position) is associated with de novo mutations that in humans have resulted in a number of diseases including hemophilia (Kazazian et al. 1988), colon cancer (Miki et al. 1992), and Duchenne muscular dystrophy (Narita et al. 1993; Holmes et al. 1994). There is evidence to suggest that these elements have had a significant effect on mammalian genomes (Fitch et al. 1990, 1991), and transposition of these elements is also associated with speciation events in murine rodents (Furano et al. 1994; Furano and Usdin 1995; Usdin et al. 1995). In spite of this, relatively little is known about how these elements generate additional copies of themselves and move about the genome.

Ll elements contain two open reading frames (ORFs), ORF 1 and ORF 2. ORF 2 has homology to reverse transcriptase (RT) (Scott et al. 1987; Xiong and Eickbush 1988; Dombroski et al. 1991), and the human ORF 2 has been shown to produce a functional RT ac- tivity in yeast (Mathias et al. 1991). These elements therefore presumably transpose via an RNA intermedi- ate. Unlike retroviruses, Ll elements lack long terminal repeats (LTRs), and since LTRs are vital to the success of the retroviral retroposition strategy, it is believed that

’ Present address: Biotech Research Laboratories, Rockville, Maryland.

Key words: Ll retrotransposons, mammals, tetraplex, evolution.

Address for correspondence and reprints: K. Usdin, Building 8, Room 202, National Institutes of Health, 8 Center Dr MSC 0830, Be- thesda, Maryland 20892-0830. E-mail: ku@helix.nih.gov.

Mol. Biol. Evol. 14(2):144-155. 1997 0 1997 by the Society for Molecular Biology and Evolution. ISSN: 0737-4038

144

Ll elements use a different pathway for their propaga- tion.

In an effort to define sequence elements that might be important for Ll retroposition, we examined the 3’ UTR of Ll elements from a wide variety of mammals. We show that despite a general lack of sequence con- servation in this region, a G-rich polypurine tract (PPT) is a conserved feature of Ll elements from species as diverse as primates, artiodactyls, rodents, lagomorphs, and macroscelids. We demonstrate that the PPTs from these elements, which are significantly longer than the PPTs of retroviruses and LTR-retrotransposons, form in- trastrand tetraplexes. Our observations are discussed in light of current models for retroposition of non-LTR ret- rotransposons.

Materials and Methods Clones

A 382-bp Nco I/BamHI fragment containing the 3’ UTR of the human L1.2 element from pLl.2a (a gift of H. H. Kazazian) was cloned into EcoRI/&zmHI digested pBCKS+ (Stratagene, La Jolla, Calif.). A 109-bp frag- ment containing the 3’ end of the mouse Ll element L 1 Md-A2 was amplified by PCR from a full-length A clone of the mouse element (a gift from C. Hutchison), using primers MdA2.1 (5’-TGCAGAATTCCTTTATA- TGCCCCCAGTACA-3’) and MdA2.2 (5’-ACGT- GAATTCACATTTCCAATGCTATACCA-3’) as fol- lows: A solution of 1 X lo5 p.f.u. of the A clone in TE buffer was boiled for 5 min. The solution was adjusted to 10 mM Tris-HCl (pH 8.0), 25 mM MgCl*, and 200 PM dNTPs. Fifty pmol of each primer and 2.5 U of Tuq polymerase (Gibco BRL, Gaithersburg, Md.) was then added and the reaction mix subjected to 30 cycles of heating and cooling (94°C for 30 s, 55°C for 30 s, and 72°C for 30 s) in a Perkin-Elmer DNA thermocycler. The reaction mix was extracted with an equal volume of phenol : chloroform : isoamyl alcohol (25:24: l), and ethanol precipitated. The PCR product was digested overnight with 10 U EcoRI at 37°C and purified on a 5% PAGE gel. The 109-bp fragment was excised from the gel and eluted overnight at 55°C in 100 mM Tris-

Conserved Structures in Ll Retrotransposons 145

HCl (pH 8.0), 5 n&I EDTA, and 500 mM NaCl. The eluate was passed through a 0.45pm Millipore syringe filter, extracted with an equal volume of phenol, and ethanol precipitated. The gel-purified fragment was then cloned into the EcoRI site of pBCKS+, and the se- quence of the PPT was verified by sequencing using standard protocols. Plasmids were grown in Escherichia coli MBM7070, isolated by alkaline lysis, and purified by CsCl gradient centrifugation according to standard procedures.

Isolation and Characterization of Ll Elephant&us edwardii

Elements from

A library of randomly sheared genomic DNA from the elephant shrew Elephant&us edwardii in Lambda Zap II was a gift of G. Wistow. An Ll probe was syn- thesized by PCR from 100 ng of total genomic DNA from Rattus nowegicus using two primers, LlDl (5’- GCCAATCCATGAGCATGGGAGATCT-3’) and LlD2 (5’-ACTAAAATCAGGGACTAGACAAGGC-3’), that are homologous to regions of the second open reading frame of the Ll element conserved in A4us domesticus and Rattus nowegicus. The PCR mix contained 100 PM of dGTP dATP and TTP and 50 PCi of [(x-32P]-dCTP (DuPont/NEN, Boston, Mass.; 3,000 Ci/mmol). This probe was purified by G-50 chromatography, denatured by heating to 37°C in the presence of 0.1 M NaOH for 5 min, neutralized by addition of 0.1 M Tris-HCl (pH 7.5) and 0.1 M HCl, and hybridized to plaque lifts from the E. edwardii library at 55°C in a solution containing 2 X SSC, 1 X Denhardt ‘s solution, 1% SDS, and 100 kg/ml sheared salmon sperm DNA. Filters were washed three times for 5 min each at room temperature in 2 X SSC before autoradiography. Eight positive clones were isolated, and phagemid was rescued from the E. coli XLl-Blue cells containing these clones by infection with ExAssist Helper phage and infection of E. coli SOLR cells according to standard procedures. Sequenc- ing was carried out using a Sequenasem (Tabor and Richardson 1987) kit (Amersham, Arlington Heights, Ill.) according to the manufacturer’s instructions. The sequences of the individual clones were compared to one another and to sequences in the GenBank database.

Sequence Analysis

Analysis of the sequence of the 3’ UTRs from dif- ferent species was carried out using the GCG software package (GCG, Madison, Wis.) and Blast (NCBI, Be- thesda, Md.).

RNA Synthesis

RNA was synthesized by in vitro transcription us- ing the pBCKS +-derived plasmids described above. Five micrograms of plasmid template was linearized with Pvu II and incubated in 80 mM Hepes-KOH (pH 7.5), 40 mM DTT, 1 r&I spermidine, 6 mM MgC12, 750 FM rNTPs, 0.5 U inorganic pyrophosphatase (Sigma, St. Louis, MO.), 3 U Prime RNase Inhibitor (5’ 3’, Inc., Boulder, Colo.), and 75 U of T3 or T7 RNA polymerase (Stratagene, La Jolla, Calif.) for 2 h at 37°C. Five units of RNase-free DNase (Gibco BRL, Gaithersburg, Md.)

was added to the reaction and the mixture was incubated for 1 h at 37°C. Reactions were extracted with phenol : chloroform : isoamyl alcohol, and precipitated with eth- anol. Transcript size and yield were monitored by elec- trophoresis of a fraction of this material in a 1% agarose gel containing formaldehyde.

Tetraplex-Formation Assays

Primers for these reactions were end-labeled with [y - 32P]ATP (DuPont/NEN, Boston, Mass.; 3,000- 6,000 Ci/mmol) using T4 Polynucleotide Kinase (Epi- centre Technologies, Inc., Madison, Wis.), and a K+-free kinase buffer (50 mM Tris-HCl, pH 8.0, 10 r&I MgCl,). Each primer extension reaction contained 2 nM of tem- plate, 0.16 nM of primer, 10 PM each of dATP dCTP dTTP and 7-deaza-dGTP 50 mM Tris-HCl (pH 9.3), 2.5 mM MgC12, 5 units of Tuq DNA polymerase (Gibco BRL, Gaithersburg, Md.) and one of the following di- deoxynucleotides at the concentration indicated in pa- rentheses: ddATP (0.3 n&I), ddGTP (0.017 mM), ddCTP (0.2 n&I), ddTTP (0.6 mM), in a total volume of 6 p,l. Cations were added to this mixture where in- dicated to a final concentration of 40 mM. Reactions were subjected to 30 cycles of heating and cooling (30 s at 95”C, 30 s at 55”C, and 30 s at 72°C). One half volume (3 p,l) of stop buffer (95% [v/v] formamide, 10 mM EDTA [pH 9.51, 10 mM NaOH, 0.1% xylene cy- anol, 0.1% bromophenol blue) was then added and re- actions were heated at 90°C for 5 min and loaded on a 6% polyacrylamide sequencing gel.

RNA analysis was done using -l/100 of the tran- script generated as described above. The transcript was mixed with 2.5 pmol of 32P-labeled primer and heated for 3 min at 94°C. KC1 was then added where indicated to a final concentration of 40 mM. Samples were then cooled to 37”C, and primer extensions were performed at this temperature for 20 min using 10 U of AMV re- verse transcriptase (Sigma, St. Louis, MO.) in a solution of 1 X AMV-RT buffer (Sigma, St. Louis, MO.) con- taining 200 PM dNTPs. Three microliters of stop buffer was added to each sample and 3 ~1 of this mixture was electrophoresed on a 6% sequencing gel.

Dimethylsulphate (DMS) Modification

The oligonucleotide used in these studies (HUMGl) contained the first 30 bases of the human PPT flanked by 10 bases of non-L1 sequence at the 5’ end and 25 bases at the 3’ end (5’-GTACGA- ATTCTGGGGACTGTGGTGGGGTCGGGGGA- GGGGGGATCAACGTAACACTTT-3’ where the PPT is shown in bold font). End-labeled oligonucleotides were diluted in 50 mM Tris-HCl, pH 9.3,2.5 mM MgC12 or TE to a concentration of approximately 13 nM, over- laid with a drop of mineral oil and denatured for 5 min at 94°C. Immediately following denaturation, KC1 was added to some of the samples to a final concentration of 40 n-M. The samples were heated for 30 s at 94”C, 30 s at 55”C, and 30 s at 72°C on a Perkin-Elmer ther- mocycler. The aqueous phase was transferred to 1.5-ml screw-capped tubes. DMS modification and piperidine cleavage of labeled oligonucleotides was performed as

146 Howell and Usdin

A I I I I I I I I

0 1 2 3 4 5 6 7

5’ UTR ORF 1 3’ UTR

A-rich region

3’ end of ORF 2 3’ UTR A-rich

3’ end of a R. nowegicus Ll element region

FIG. 1.-A, The basic organization of a mammalian Ll element. Ll elements contain a 5’ untranslated region of variable size that in some mammals has been shown to have promoter activity (Swergold 1990; Minakami et al. 1992), two open reading frames (ORF 1, which encodes an RNA-binding protein [Hohjoh and Singer 19961, and ORF 2, which encodes a reverse transcriptase [Hattori et al. 1986; Fanning and Singer 1987; Scott et al. 1987; Dombroski et al. 1991; Mathias et al. 1991]), and a 3’ untranslated region (3’ UTR) that includes an A-rich region. B, Comparison of Ll elements from H. sapiens and R. norvegicus. Ll elements of different species differ mostly in the length and sequence of their 5’ and 3’ untranslated regions with the open reading frames showing much higher levels of sequence conservation than do the untranslated regions. The 3’ ends of representative Ll elements from human (GB:HUMTNL22) and rat (GB:RATLIN3A) were compared using the Compare subroutine from the GCG package. The output of this program was plotted using the Dotplot subroutine. The 3’ end of each ORF 2 is shown in gray, and the 3’ UTR is shown in white on the element shown alongside each axis. The arrows indicate the 3’ end of ORF 2. The 3’ ends differ in length by -478 bp.

follows: The oligonucleotide suspensions were diluted with 180 p,l of DMS reaction buffer. Half to one mi- croliter of DMS was added to tubes with and without KCl, respectively, and incubated at 18°C for 1 min. The reactions were stopped by precipitation with 1 ml of n-butanol. Pellets were resuspended in 100 ~1 of 1 M piperidine, incubated for 30 min at 9O”C, and then bu- tanol precipitated. Samples were resuspended in 6-10 ~1 of distilled HZ0 to which a l/2 volume of Sequenase Stop buffer (Amersham, Arlington Heights, Ill.) had been added, heated to 90°C for 3 min and electropho- resed at 2,500 V on a 20% polyacrylamide gel contain- ing 7 M urea. The gel was then wrapped in plastic film and exposed to X-ray film at -70°C. The autoradiogram was scanned into a Power Macintosh 7500 using an Agfa Arcus II flatbed scanner, and the resultant image was analyzed using NIH Image (developed by Wayne Rasband at the National Institutes of Health and avail- able on the Internet at http://rsb.info.nih.gov/nih-image/).

Results The 3’ End of Ll Elements is Not Conserved Between Species

The region between the termination codon for the Ll second open reading frame (ORF 2) and the A-rich region located at the 3’ end of Ll elements is referred to as the 3’ untranslated region (3’ UTR) (see fig. 1A). While the sequence of the ORF 2 of Ll elements is highly conserved, their 3’ UTRs are much more highly diverged as evidenced by the wide variation in size (ta- ble 1) and the fact that sequence homology between any two elements from different species drops off dramati- cally 3’ of the ORF 2 termination codon (fig. 1B).

All Ll Elements in GenBank with Intact 3’ Ends Contain a G-rich Polypurine Tract

Examination of the 3’ end of Ll elements found in GenBank (Release 95.0) indicated that the presence of

Conserved Structures in Ll Retrotransposons 147

a G-rich polypurine tract (PPT) close to the A-rich re- gion is a conserved feature of Ll elements from all spe- cies for which the 3’ end is known (i.e., cows and goats (artiodactyls), rabbits (lagomorphs), primates and ro- dents) (table 1). The PPT ranges in size from about 34 bases in goats to 62 in rabbits. The distance between the end of ORF 2 and the start of the PPT varies from -42 bp in humans to 1,260 bp in rabbits, but the dis- tance between the 3’ end of the PPT and the start of the A-rich tail varies over a smaller range: -22 bp in arti- odactyls to 129 bp in humans.

Characterization of Ll Elements from the Macroscelid Elephantulus edwardii

While the evolutionary relationships of placental mammals are not fully understood, it is believed that primates, rodents, lagomorphs, and even artiodactyls form a relatively closely related group (Graur, Duret, and Gouy 1996). To determine whether Ll elements from more distantly related animals also contain a PPT, we isolated Ll-containing clones from a library of ge- nomic DNA from the elephant shrew Elephant&us ed- wardii. E. edwardii is a member of the macroscelidae, a group of animals that are believed to be one of the earliest offshoots of the eutherian lineage (de Jong, Leu- nissen, and Wistow 1993). About 5% of clones in a library of randomly sheared DNA from this animal were positive with an ORF 2 probe generated by PCR from genomic DNA from the rat Rattus norvegicus. This is consistent with an Ll copy number of about 18 000 assuming a genome size of 2.9 X lo9 bp and an average insert size of 8 kb.

We sequenced eight clones that hybridized with the ORF 2 probe. One of these clones contains 256 bases from the central portion of ORF 2. This region of ORF 2 showed 71% sequence identity to bases 3891 to 4145 of GB:RATLIN3A, an Ll element from R. norvegicus, and a similar amount of sequence identity to regions of the human (GB:HUMTNL22), and the rabbit (GB: LCLlOC5R) Ll elements. The human Ll element by comparison shows about 72% sequence identity to the rat element over this region, and the rabbit element shows about 75% sequence similarity to both the human and the rat Ll elements. The remaining elephant shrew clones contain smaller amounts of ORF 2 sequence. The elephant shrew ORF 2 sequences are 99% identical to one another, consistent with the extent of the intraspe- ties sequence identity seen in the ORF 2 of other ani- mals.

The inserts of two of the clones (Ee3 and Ee6) extend 3’ of ORF 2. Ee3 contains a long A-rich region 225 bases from the ORF 2 termination codon. Homol- ogy between Ee3 and Ee6 extends to the end of the insert in Ee6. The last base of the insert in this clone corresponds to the base 33 bases 5’ of the A-rich region in Ee3 (fig. 2). This suggests that the entire sequence 3’ of the stop codon for ORF 2 in Ee6 is part of the shrew Ll element. We probed the E. edwardii library with an oligonucleotide from immediately 5 ’ of the A-rich re- gion in Ee3 and found that about 5% of clones were positive. This is consistent with the number we would

148 Howell and Usdin

Ee3 1

Ee6 1

Ee3 51

Ee6 51

Ee3 101

Ee6 101

Ee3 150

Ee6 151

Ee3 200

Ee6 201

Ee3 243

Ee6 251

Ee3 291

Ee6 301

Ee3 329

Ee6 351

Ee3 341

Ee6 401

Ee3 369

Ee6 451

Ee3 409

Ee6

GGCCGACATCGGGATTGGGGTAAGACTAAGAAAAACTGGATGAAATATGT III I IIIIIII I IIIIIIIIIII IIIIlIIII IIIIIIIIIIII GGCTGGCATCGGGGTGGGGGTAAGACTGAGAAAAACTAGATGMiATATGT

TTACGGCATTGAAGTCTTTGGGGGAGAGAATATTTTA.......TATTTT IIIIIIIIIIIIII II IIIIIIIIIIIIII IIII III II TTACGGCATTGAAGCCTATGGGGGAGAGAATAGTTTAAAAGTATTATATT

AGGGCATTATGTTTGTCCTA.TAGTACGGCTCGAGGGTT.CAGAGTGGAT IIIIIIIIIIII II II IIIII IIIIIII IIIIIII II

TAGGCATTATGTTTATCGTAGGAGTACTCAATGAGGGTTGAGGGTTGCAGAGTGTAT

TCTCTTTTA.........GGGGTTGGGGGATGT...GGGGGGGGATTCGC I I I I I IIIIIIIIIII II IIIIlIII I I TTATATTTAAGGTTGGGGTGGGTTGGGGGAGGTGCGGGGGGGGGGATTGT

GTACCAATGAGG......................................

I I I I I I I I I I I I

. . . . . . . . . . . . . . . . . . . . . . GGGCGGGGAGAGGGGTGCAACAATAAAT IIII IIIIII IIIIIIIIIIII III

GAATGGAATCCAMATCTCTGTAGATGTTCATTGGAGTGATA

I I I I I I GAATGG............................................

AATAAAATAAAAATAAAAAAATA

. . . . . . . . . . . . . . . . . . . . . . .

100 3’ end of ORF 2

100

150 J stop codon

199

200

242

250

290

300

328

350

340 PPT

400

368

450

408

A-rich region

FIG. 2.-Sequence alignment of the 3’ end of two E. edwardii Ll elements. The 3’ end of the ORF 2’s of Ee3 and Ee6 are shown in the grey box. The polypurine tracts (PPTs) are shown in bold type with the additional bases in the PPT of Ee6 shown in the white box. The A-rich region in Ee3 is underlined.

expect for the 3’ UTR since the copy number should be similar to that of the ORF 2 sequence. A probe homol- ogous to a sequence 3’ of the A-rich region gave no hybridizing plaques in 2,000 clones screened (data not shown). Given that Ll elements from other animals all terminate in an A-rich region and that the sequence 3’ of the A-rich region is apparently not present at high copy number, we believe that the A-rich region in Ee3 represents the 3’ end of the elephant shrew Ll element. As can be seen from the sequence shown in bold type in figure 2, E. edwardii Ll elements also contain a PPT at their 3’ ends, with the 3’ end of the PPT being -51 bases 5’ of the A-rich tail. The PPT of Ee3 is 57 bp long, while the PPT of Ee6 contains an additional 59 bases (underlined in fig. 2). The additional bases include a string of 18 pyrimidines followed by a 3%bp purine- rich tract. As expected, the 3’ UTR of the elephant shrew Ll is not homologous to the 3’ UTR of Ll ele- ments from other mammals.

The PPT of Mammalian Ll Elements Can Form Intrastrand Tetraplexes

While all mammalian Ll elements have a PPT, the PPTs are not homologous. The retention of a G-rich mo- tif in the absence of sequence conservation suggests that some other property of the PPTs, perhaps the ability to form a particular secondary structure, may be con- served. G-rich sequences can form a variety of different structures including hairpins, triplexes, and tetraplexes. Certain stable secondary structures block DNA synthe- sis in vitro (Baran, Lapidot, and Manor 1991; Dayn, Samadashwily, and Mirkin 1992; Weitzmann, Wood- ford, and Usdin 1996), and the characteristics of the blocks can be diagnostic of the underlying secondary structure.

For example, we have previously shown that in- trastrand tetraplexes produce blocks to DNA synthesis with certain unique diagnostic features (Woodford,

Conserved Structures in Ll Retrotransposons 149

FIG. 3.-Hydrogen-bonding scheme for the G4 tetrads (Zimmer- man, Cohen, and Davies 1975) . Tetraplexes are held together by hy- drogen bonds between four different G-rich motifs. These motifs can be on two or four different DNA strands in the case of intermolecular tetraplexes or on a single DNA strand in the case of intramolecular tetraplexes. G4 tetrads form the basic unit of these structures. The hy- drogen-bonded residues in the tetrad are numbered. Tetrad formation involves Hoogsteen base-pairing between the N’ and O6 residues, and the N* and N7 residues on adjacent guanines.

triplexes -which are formed by an interaction between

Howell, and Usdin 1994; Usdin and Woodford 1995; Howell et al. 1996; Weitzmann, Woodford, and Usdin

two complementary strands, blocks due to tetraplexes

1996). Intrastrand tetraplexes involve the interaction of four G-rich regions on a single nucleic acid strand. Four G’s, one from each of the four G-rich regions, interact

are seen only when the G-rich strand is used as the tem-

to form a planar tetrad held together by Hoogsteen base interactions (fig. 3). A series of successive tetrads make up the stem of the tetraplex which is bounded by loops formed by the bases between the G-rich regions. Unlike

1990). As expected, other known noncanonical DNA structures such as hairpins, triplexes, and Z-DNA show no such K+-specific effect. We have shown using a va- riety of synthetic and naturally occuring tetraplex-form- ing sequences that K+-dependent DNA synthesis arrest makes a reliable, sensitive and specific assay for intra- strand tetraplexes (Woodford, Howell, and Usdin 1994; Usdin and Woodford 1995; Howell et al. 1996; Weitz- mann, Woodford, and Usdin 1996).

The purine-rich strand of the Ll PPTs from hu- mans, mice, and elephant shrews all form a series of K+-dependent blocks to DNA synthesis in vitro (fig. 4). The locations of these blocks are shown in figure 5. No such polymerase stops are seen in the absence of mono- valent cation, or in the presence of other monovalent cations such as Na+ (fig. 4). The complementary strand shows no such stops (data not shown), and formation of these blocks is independent of template concentration over a wide range, consistent with the formation of in- trastrand tetraplexes.

In the case of the human PPT, at least two different tetraplexes are formed as evidenced by the two distinct K+-dependent stop sites at the 30fh and 3Sh bases from the 5’ end of the PPT (GjO and T35, respectively, in fig. 5). We know from previous work that the polymerase stops at the base immediately preceding the tetraplex (Weitzmann, Woodford, and Usdin 1996). Therefore, in the tetraplex causing the stop at G30 the 3’-most base in the structure is Gz9. The 3’-most base in the tetraplex responsible for the stop at T35 corresponds to the last base in the PPT (G&. The mouse PPT produces one major cluster of blocks to DNA synthesis at the 3’ end of the PPT as well as a number of other chain termi-

plate, and, form even in the absence of the complemen- tary strand. These blocks are independent of template concentration, as one might expect of an intrastrand structure, and, unlike a triplex formed between the tem- plate and the nascent strand, the blocks to DNA synthe- sis occur at the 3’ end rather than in the middle of the of the G-rich sequence (Weitzmann, Woodford, and Us- din 1996). Incorporation of 7-deazaguanine into the tem- plate abolishes both the tetraplex and the blocks to DNA synthesis, and under these reaction conditions, guanines on the template show a pattern of alternating dimethyl- sulphate (DMS) protection and reactivity immediately 5’ of the blocks to DNA synthesis that is diagnostic of tetraplexes (Woodford, Howell, and Usdin 1994; Usdin and Woodford 1995; Howell et al. 1996; Weitzmann, Woodford, and Usdin 1996). Most striking, however, is the absolute dependence of these blocks to DNA syn- thesis on the presence of K+. This K+ dependence does not reflect a nonspecific requirement for ion screening

shown), and both -a discrete and a diffuse set of stops in the case of Ee6 (figs. 4 and 5). The additional set of

nation products (figs. 4 and 5). The elephant shrew Ll

stops in Ee6 is due to the additional bases in the PPT

elements produced one major region of K+-dependent

of this element that divide it into two parts, each capable of forming one or more tetraplexes.

blocks to DNA synthesis in the case of Ee3 (data not

Characterization of One of the Tetraplexes Formed by the Human PPT

Because the human PPT forms at least two differ- ent tetraplexes that would complicate structural analysis, we carried out dimethylsulphate (DMS) modification of an oligonucleotide containing bases l-30 of the PPT (see fig. 5 and ikterials and Methods). This sequence is sufficient to produce the structure responsible for the GsO stop, but not the one underlying the stop at T35. DMS reacts with the N7 of G’s. Hydrogen bonding in G,-tetrads involves the N7 residue of all participating G’s (see fig. 3), and these G’s are thus protected from modification by DMS. DMS reactive and nonreactive

since other monovalent cations do not have the same bases can be - distinguished because piperidine only effect. Rather, it reflects the unique ability of K+ to sta- cleaves those G’s that have reacted with DMS. The bilize tetraplexes. K+ has this effect because its ionic DMS/piperidine cleavage products for the human Ll radius is such that it can fit inside the tetraplex stem and PPT are shown on the left side of fig. 6. form a coordination complex with the O6 atoms in ad- Our previous work showed that tetraplex-forming jacent tetrads (Pinnavaia et al. 1978; Sen and Gilbert sequences form hairpins in the absence of monovalent

Conserved Structures in Ll Retrotransposons 151

A base 1 of

PPT stop site

Template 5 , strand

-3 ’ 4 ----

direction of DNA synthesis

B 30 T35

H. sapiens 0 t 5 ’ -GGGGACTGTGGTGGGGTC GGGGGAGGWXGAGGGTAGCATT- 3 ’

M. domesticus 5 I- GGGGGAGTGGGTGGGC #

VACTTTTG- 3 ’

E. edwardii 5 ’ GGGGTGGGTTGGGGGAGGTGC WV*

GOGGOGOGGGATTGTGTACCAATGAGGGAGTT

TTTCTTTTCGCTTTTCAAGGAGAGGAGCTCAAGGAGAOOAGCAAA BGGGGTGAGGTGGC

GTGGGCAGGGAGAAGGGTGCAAC~ 3 ’

FIG. 5.-Location of K+-dependent blocks to DNA polymerase in the human, mouse, and elephant shrew 3’ UTRs. A, Schematic repre- sentation of the templates used in the experiments shown in figure 3. The PPT is shown as a white box with the sequence 3’ of the PPT shown as a black line. Our numbering convention makes the 5’-most base in the PPT base 1. This base is indicated by the grey arrow. The direction of DNA synthesis is shown as the dashed arrow. A block to DNA synthesis (stop site) is shown by the black arrow. B, Positions of the K+- dependent DNA polymerase stop sites in the human, mouse, and elephant shrew 3’ UTRs. The PPT is shown in bold type, The positions of the discrete blocks to DNA synthesis shown in figure 3 are marked by arrows above the sequence. Broad K+-dependent regions of premature chain termination are indicated by the shaded boxes. The two major polymerase stop sites in the human Ll PPT at G,, and T,, are shown above the human Ll sequence. The sequence underlined in the human PPT was incorporated into the oligonucleotide used to determine the structure responsible for the stop at G,,-(see figs. 6 and 7).

active as the same bases in the absence of K+, and the most protected bases are lo%-25% as reactive as the corresponding bases in the K+-free reaction.

The pattern of DMS reactivity in which four regions of more protected G’s alternate with regions of relatively strongly reactive G’s is consistent with tetra- plex formation. This pattern of DMS reactivity is inde- pendent of oligonucleotide concentration over a wide range, indicative of the formation of an intrastrand struc- ture. This is consistent with our observation that no in- termolecular association of the oligonucleotide occurs under our reaction conditions as assessed by nondena- turing gel electrophoresis (data not shown). The most parsimonious interpretation of the pattern of DMS mod- ification is that the most protected guanines participate in guanine tetrads in the stem of the tetraplex, with the more reactive G’s being located in the loops. The partial protection of these bases in the presence of K+ can be explained by inter- and intraloop interactions and/or conformational polymorphism (Usdin and Woodford 1995). A model for the structure of this tetraplex is shown in figure 7. This structure is unusual in a number of respects since in order to account for the pattern of DMS protection we have had to invoke an AGGG-tet- rad, and a number of G-G base pairs in which one gua- nine is the obligatory N7 donor (e.g., Gig, which we suggest is base-paired with G&. We would expect that the DMS protection pattern of the full-length PPT would be more complex since two different structures are pres- ent. However, since these structures are both tetraplexes,

we would predict that significant DMS protection of most of the guanines would still be evident.

RNA from the PPT Also Forms a Strong Secondary Structure

In order to test for the formation of secondary structure in the PPT of the Ll transcript, we synthesized an RNA template from the human element L1.2, and from the mouse element LlMd-A2. Figure 8 shows that RNA transcripts from both the human and mouse Ll elements form strong blocks to DNA synthesis by an RNA-dependent DNA polymerase (reverse transcrip- tase, RT). However, this block to RT is not K+-depen- dent, since it is seen even in the absence of monovalent cation. Since these blocks to DNA synthesis occur in the same positions as the blocks seen on DNA tem- plates, it is possible that the underlying molecular basis is similar. We have previously shown that large DNA tetraplexes are stable in the absence of K+ (Weitzmann, Woodford, and Usdin 1996). RNA hairpins are more stable than DNA hairpins (Antao and Tinoco 1992), and the RNA version of the Ll DNA tetraplex may be suf- ficiently stable so as to be independent of K+. Since the human Ll PPT is one of the shortest Ll PPTs known, formation of a stable secondary structure by this PPT makes it very likely that similar structures formed by the longer PPTs in Ll transcripts from other animals will also be stable.

Discussion

The 3 ’ UTRs of previously characterized Ll ele- ments from primates, rodents, lagomorphs and artiodac-

Conserved Structures in Ll Retrotransposons 153

ence of an additional 59 bp of sequence in the PPT of Ee6. The additional bases included a short pyrimidine tract followed by a long string of purines. This is the first report of such length variants in Ll PPTs.

The 3’ UTRs of these two Ll elements are 85% identical (92% if the PPT is excluded). This compares to a sequence identity of 98% for two rat Ll elements (GB:RATLIN3A and GBRNMLVI2R) (Usdin et al. 1995). Assuming a neutral substitution rate of 1% per Myr, this level of sequence divergence indicates that the two elephant shrew Ll elements last shared a common ancestor about 4 MYA. The ORF 2 region of Ee3 and Ee6 are 99% identical. The difference in the extent of divergence of ORF 2 and the 3’ UTR suggests that like Ll elements in other animals, the 3’ UTR is changing more rapidly than ORF 2. This presumably reflects a lack of selection at the sequence level for the 3’ UTR. The extent of ORF 2 sequence homology and the fact that both reading frames are still open suggests that these ORFs evolved under selective pressure for much of of the 4 Myr since the elements diverged. Our data thus indicate that Ee3 and Ee6 represent members of two different Ll lineages/subfamilies that have been evolving independently for some time. The presence of a PPT in Ll elements from this representative of an ancient mammalian lineage supports the hypothesis that a PPT is an evolutionarily conserved feature of Ll el- ements.

While this work was in progress, a partial sequence of an Ll element from Didelphis virginiana, a North American marsupial, was published (Dorner and Paabo 1995). However, the sequence of this element does not extend as far 3’ as an A-rich region although there is a 36-base-long PPT 294 bases downstream of the ORF 2 termination codon. If this PPT is indeed located at the 3’ end of the full-length Ll element, it would provide additional support for the argument that a PPT is a func- tionally important feature of Ll elements.

While the sequence of the PPTs varies among dif- ferent genera, we have found that the PPT from animals as diverse as elephant shrews, rabbits, mice and humans form a set of related DNA secondary structures known as intrastrand tetraplexes. We also demonstrate directly that RNA transcripts of both the human and mouse PPTs are capable of forming intrastrand folded structures as well, and it is likely that RNA versions of the PPTs from other species would do likewise.

functionally interchangeable, and it has therefore been suggested that some conserved secondary structure in these 3’ UTRs might be important for recogition of the transcript by the R2 reverse transcriptase (Luan and Eickbush 1995). By analogy with R2 elements, it may be that the Ll PPT provides the structure necessary for recognition of the Ll transcript by the Ll RT. Once this recognition has been accomplished, any impediment to reverse transcriptase could be removed by one of a va- riety of proteins that are able to remove secondary struc- tures from nucleic acids (Waga, Mizuno, and Yoshida 1990). The frequency of successful tetraplex removal need not be 100% and may represent one level at which propagation of this element is controlled.

The RNAs of retroviruses and LTR-retrotranspo- sons form parallel dimers (Coffin 1979). The role of retroelement dimerization is still unclear, but there is evidence to suggest that this process is important for reverse transcription of the retroelement by facilitating template strand switching (Jones, Allan, and Temin 1993). It has also been suggested that dimerization is important for the generation of retroelement genetic di- versity, and in packaging (Fu and Rein 1993). Dimeri- zation occurs via RNA-RNA interactions since dimers are stable even after removal of all proteins (Duesberg 1968) and can form spontaneously in the absence of proteins if the concentrations of particular cations are high enough (Marquet et al. 1991; Paoletti et al. 1993). Sequences that form intrastrand (unimolecular) tetra- plexes can often also form bimolecular tetraplexes (Sen and Gilbert 1990), so it is possible that two Ll tran- scripts may associate in this way to form dimers. Other ways the G-rich element may affect Ll retroposition in- clude increasing the efficiency of pre-mRNA processing in a manner analogous to the effect of the G-rich ele- ment upstream of the SV40 late polyadenylation signal (Bagga et al. 1995), or increasing the stability of the Ll transcript, thus increasing the probability that successful retroposition will occur.

Sequence Availability

The DNA sequences reported in this paper have been deposited in the GenBank database under acces- sion numbers U62039 (Ee3) and U62040 (Ee6).

Acknowledgments

Current models for non-LTR element retroposition The authors wish to thank M. Neale Weitzmann invoke a nick or staggered break in a chromosome, as- sociation of the element transcript with the 5’ end of the

and Kerry J. Woodford for critical reading of this manu-

nick with the free 3’ end acting as a primer for reverse script, and Herbert Tabor and Anthony V. Furano for

transcription. The 3’ OH- group at the end of the cDNA their advice and many helpful discussions.

strand is then ligated to the free 5’ end of the chromo- LITERATURE CITED somal nick, the transcript removed by an RNaseH activ- ity followed by DNA repair to fill in the gaps (Schwarz- ANTAO, V. I?, and I. TINOCO JR. 1992. Thermodynamic param-

Sommer et al. 1987; Luan et al. 1993). In the case of eters for loop formation in RNA and DNA hairpin tetra-

R2 elements, Ll -like non-LTR containing retrotranspo- loops. Nucleic Acids Res. 20:819-824.

sons in insects, the last 250 bases of the 3’ UTR are BAGGA, I? S., L. I? FORD, E CHEN, and J. WILUSZ. 1995. The

essential for reverse transcription. Like Ll elements, the G-rich auxiliary downstream element has distinct sequence

3’ UTRs from different insects show no sequence sim- and position requirements and mediates efficient 3’ end pre-

ilarity (Luan and Eickbush 1995). Nevertheless, they are mRNA processing through a trans-acting factor. Nucleic Acids Res. 23:1625-1631.

154 Howell and Usdin

BARAN, N., A. LAPIDOT, and H. MANOR. 1991. Formation of DNA triplexes accounts for arrests of DNA synthesis at d(TC)n and d(GA)n tracts. Proc. Natl. Acad. Sci. USA 88: 507-511.

BURTON, I? H., D. D. LOEB, C. E VOLIVA, S. L. MARTIN, M. H. EDGELL, and C. A. HUTCHISON. 1986. Conservation throughout mammalia and extensive protein-encoding ca- pacity of the highly repeated DNA long interspersed se- quence one. J. Mol. Biol. 187:291-304.

COFFIN, J. M. 1979. Structure, replication, and recombination of retrovirus genomes: some unifying hypotheses. J. Gen. Virol. 42: l-26.

DAYN, A., G. M. SAMADASHWILY, and S. M. MIRKIN. 1992. Intramolecular DNA triplexes: unusual sequence require- ments and influence on DNA polymerization. Proc. Natl. Acad. Sci. USA 89: 11406-l 1410.

DE JONG, W. W., J. A. M. LEUNISSEN, and G. J. WISTOW. 1993. Eye lens crystallins and the phylogeny of placental orders: evidence for a macroscelid-paenungulate clade? Pp. 5-12 in E S. SZALAY, M. J. NOVACEK and M. C. MCKENNA, eds. Mammalian phylogeny: placentals. Springer, New York.

DOMBROSKI, B. A., S. L. MATHIAS, E. NANTHAKUMAR, A. E SCOTT, and H. H. J. KA~AZIAN. 1991. Isolation of an active human transposable element. Science 254: 1805-l 808.

DORNER, M., and S. PAABO. 1995. Nucleotide sequence of a marsupial LINE-l element and the evolution of placental mammals. Mol. Biol. Evol. 12:944-948.

DUESBERG, I? H. 1968. Physical properties of Rous Sarcoma Virus RNA. Proc. Natl. Acad. Sci. USA 60:1511-1518.

FANNING, T., and M. SINGER. 1987. The LINE-l DNA sequenc- es in four mammalian orders predict proteins that conserve homologies to retrovirus proteins. Nucleic Acids Res. 15: 225 l-2260.

FITCH, D. H., W. J. BAILEY, D. A. TAGLE, M. GOODMAN, L. SIEU, and J. L. SLIGHTOM. 199 1. Duplication of the gamma- globin gene mediated by Ll long interspersed repetitive el- ements in an early ancestor of simian primates. Proc. Natl. Acad. Sci. USA 88:73967400.

FITCH, D. H., C. MAINONE, M. GOODMAN, and J. L. SLIGHTOM. 1990. Molecular history of gene conversions in the primate fetal gamma-globin genes. Nucleotide sequences from the common gibbon, Hylobates lar. J. Biol. Chem. 265:781- 793.

Fu, W., and A. REIN. 1993. Maturation of dimeric viral RNA of Moloney murine leukemia virus. J. Virol. 67:5443-5449.

FURANO, A. V., B. E. HAYWARD, I? CHEVRET, E CATZEFLIS, and K. USDIN. 1994. Amplification of the ancient murine Lx family of long interspersed repeated DNA occurred dur- ing the murine radiation. J. Mol. Evol. 38:18-27.

FURANO, A. V., and K. USDIN. 1995. DNA “fossils” and phy- logenetic analysis. J. Biol. Chem. 270:1-4.

GRAUR, D., L. DURET, and M. GOUY. 1996. Phylogenetic po- sition of the order Lagomorpha (rabbits, hares and allies). Nature 379:333-335.

HATTORI, M., S. KUHARA, 0. TAKENAKA, and Y. SAKAKI. 1986. Ll family of repetitive DNA sequences in primates may be derived from a sequence encoding a reverse tran- scriptase-related protein. Nature 321:625-628.

HOHJOH, H., and M. E SINGER. 1996. Cytoplasmic ribonucle- oprotein complexes containing human LINE-l protein and RNA. EMBO J. 15630-639.

HOLMES, S. E., B. A. DOMBROSKI, C. KREBS, C. D. BOEHM, and H. H. KA~AZIAN JR. 1994. A new retrotransposable human Ll element from the LRE2 locus on chromosome lq produces a chimaeric insertion. Nat. Genet. 7: 143.

HOWELL, R., K. J. WOODFORD, M. N. WEITZMANN, and K. USDIN. 1996. The chicken B-globin gene promoter forms a

novel “cinched” tetrahelical structure. J. Biol. Chem. 271: 5208-5214.

JONES, J. S., R. W. ALLAN, and H. M. TEMIN. 1993. Alteration of location of dimer linkage sequence in retroviral RNA: little effect on replication or homologous recombination. J. Virol. 67:3151-3158.

KAZAZIAN, H. J., C. WONG, H. YOUSSOUFIAN, A. E SCOTT, D. G. PHILLIPS, and S. E. ANTONARAKIS. 1988. Haemophilia A resulting from de novo insertion of Ll sequences repre- sents a novel mechanism for mutation in man. Nature 332: 164-166.

LUAN, D. D., and T. H. EICKBUSH. 1995. RNA template re- quirements for target DNA-primed reverse transcription by the R2 retrotransposable element. Mol. Cell. Biol. 15:3882- 3891.

LUAN, D. D., M. H. KORMAN, J. L. JAKUBCZAK, and T. H. EICKBUSH. 1993. Reverse transcription of R2Bm RNA is primed by a nick at the chromosomal target site: a mecha- nism for non-LTR retrotransposition. Cell 72:595-605.

MARQUET, R., I? BAUDIN, C. GABUS, J. L. DARLIX, M. Mou- GEL, C. EHRESMANN, and B. EHRESMANN. 199 1. Dimeri- zation of human immunodeficiency virus (type 1) RNA: stimulation by cations and possible mechanism. Nucleic Acids Res. 19:2349-2357.

MATHIAS, S. L., A. E SCOTT, H. H. J. KA~AZIAN, J. D. BOEKE, and A. GABRIEL. 1991. Reverse transcriptase encoded by a human transposable element. Science 254: 1808-l 8 10.

MIKI, Y., I. NISHISHO, A. HORII, Y. MIYOSHI, J. UTSUNOMIYA, K. W. KINZLER, B. VOGELSTEIN, and Y. NAKAMURA. 1992. Disruption of the APC gene by a retrotransposal insertion of Ll sequence in a colon cancer. Cancer Res. 52:643-645.

MINAKAMI, R., K. KUROSE, K. ETOH, Y. FURUHATA, M. HAT- TORI, and Y. SAKAKI. 1992. Identification of an internal cis- element essential for the human Ll transcription and a nu- clear factor(s) binding to the element. Nucleic Acids Res. 20:3139-3145.

NARITA, N., H. NISHIO, Y. KITOH, Y. ISHIKAWA, Y. ISHIKAWA, R. MINAMI, H. NAKAMURA, and M. MATSUO. 1993. Inser- tion of a 5’ truncated Ll element into the 3’ end of exon 44 of the dystrophin gene resulted in slopping of the exon during splicing in a case of Duchenne muscular dystrophy. J. Clin. Invest. 91: 1862-1867.

PAOLETTI, J., M. MOUGEL, N. TOUNEKTI, I? M. GIRARD, C. EHRESMANN, and B. EHRESMANN. 1993. Spontaneous di- merization of retroviral MoMuLV RNA. Biochimie 75:68 l- 686.

PINNAVAIA, T. J., C. L. MARSHALL, C. M. METTLER, C. L. FISK, H. T. MILES, and E. D. BECKER. 1978. Alkali metal ion specificity in the solution ordering of a nucleotide, 5’-gua- nosine monophosphate. J. Am. Chem. Sot. 100:3625-3627.

SAMADASHWILY, G. M., A. DAYN, and S. M. MIRKIN. 1993. Suicidal nucleotide sequences for DNA polymerization. EMBO J. 12:4975-4983.

SCHWARZ-SOMMER, Z., L. LECLERCQ, E. GOBEL, and H. SAE- DLER. 1987. Cin4, an insert altering the structure of the Al gene in Zea mays, exhibits properties of non-viral retrotransposons. EMBO J. 146:3873-3880.

SCOTT, A. E, B. J. SCHMECKPEPER, M. ABDELRAZIK, C. T. COMEY, B. O’HARA, J. l? ROSSITER, T. COOLEY, l? HEATH, K. D. SMITH, and L. MARGOLET. 1987. Origin of the human Ll elements: proposed progenitor genes deduced from a consensus DNA sequence. Genomics 1: 113-125.

SEN, D., and W. GILBERT. 1990. A sodium-potassium switch in the formation of four-stranded G6DNA. Nature 344: 410-414.

Conserved Structures in Ll Retrotransposons 155

SWERGOLD, G. D. 1990. Identification, characterization, and cell specificity of a human LINE-l promoter. Mol. Cell. Biol. lo:67 18-6729.

TABOR, S., and C. C. RICHARDSON. 1987. DNA sequence anal- ysis with a modified bacteriophage T7 DNA polymerase. Proc. Natl. Acad. Sci. USA 84:4767-4771.

USDIN, K., I? CHEVRET, E M. CATZEFLIS, R. VERONA, and A. V. FURANO. 1995. Ll (LINE-l) retrotransposable elements provide a “fossil” record of the phylogenetic history of Murid rodents. Mol. Biol. Evol. 12:73-82.

USDIN, K., and K. J. WOODFORD. 1995. CGG repeats associ- ated with DNA instability and chromosome fragility form DNA structures that block DNA synthesis in vitro. Nucleic Acids Res. 23:4202-4209.

WAGA, S., S. MIZUNO, and M. YOSHIDA. 1990. Chromosomal protein HMGl removes the transcriptional block caused by the cruciform in supercoiled DNA. J. Biol. Chem. 265: 19424-19428.

WEITZMANN, M. N., K. J. WOODFORD, and K. USDIN. 1996. The development and use of a DNA polymerase arrest assay for the evaluation of parameters affecting intrastrand tetra- plex formation. J. Biol. Chem. 271:20958-20964.

WOODFORD, K. J., R. M. HOWELL, and K. USDIN. 1994. A novel K+-dependent DNA synthesis arrest site in a com- monly occurring sequence motif in eukaryotes. J. Biol. Chem. 269:27029-27035.

XIONG, Y., and T H. EICKBUSH. 1988. Similarity of reverse tran- scriptase-like sequences of viruses, transposable elements, and mitochondrial introns. Mol. Biol. Evol. 5:675-690.

ZIMMERMAN, S. B., G. H. COHEN, and D. R. DAVIES. 1975. X-ray fiber diffraction and model-building study of polyguanylic acid and polyinosinic acid. J. Mol. Biol. 92:181-192.

THOMAS EICKENJSH, reviewing editor

Accepted October 28, 1996