The GAG-like protein of the yeast Ty1 retrotransposon contains a nucleic acid chaperone domain analogous to retroviral nucleocapsid proteins.

The reverse transcription process for retroviruses and retrotransposons takes place in a nucleocore structure in the virus or virus-like particle. In retroviruses the major protein of the nucleocore is the nucleocapsid protein (NC protein), which derives from the C-terminal region of GAG. Retroviral NC proteins are formed of either one or two CCHC zinc finger(s) flanked by basic residues and have nucleic acid chaperone and match-maker properties essential for virus replication. Interestingly, the GAG protein of a number of retroelements including Spumaviruses does not possess the hallmarks of retroviral GAGs and in particular lacks a canonical NC protein. In an attempt to search for a nucleic acid chaperone activity in this class of retroelements we used the yeast Ty1 retrotransposon as a model system. Results shows that the C-terminal region of Ty1 GAG contains a nucleic acid chaperone domain capable of promoting the annealing of primer tRNA(i)(Met) to the multipartite primer binding site, Ty1 RNA dimerization and initiation of reverse transcription. Moreover Ty1 RNA dimerization, in a manner similar to Ty3 but unlike retroviral RNAs, appears to be mediated by tRNA(i)(Met). These findings suggest that nucleic acid chaperone proteins probably are general co-factors for reverse transcriptases.

Retroelements form a large family of mobile genetic elements, including retroviruses, retrotransposons (or LTR 1 retrotransposon for long terminal repeats containing retrotransposon) and retroposons (or non-LTR retrotransposon). Despite their tremendous diversity and dispersity among living organisms, their replicative cycle shares several similarities. They all encode a reverse transcriptase (RT) that converts the genomic RNA into a double-stranded DNA subsequently integrated into the cellular genome by integrase or endonuclease (1). Retroelements efficiently spread within eukaryotic organisms and have a major evolutive impact on the genomes that they inhabit. As repeated sequences they represent a high potential for homologous recombination of the host genome and as elements with transcriptional activity, their integration triggers both insertional mutagenesis and deregulation of gene expression (2,3). Furthermore by means of cellular RNAs recruitment and reverse transcription, retroelements have been involved in pseudogene formation and exon shuffling in plants and mammals (4 -6), long term immunity against nonretroviral RNA viruses in mammals (7), and intron loss in yeast (8,9).
The reverse transcription process for retroviruses and probably most LTR retrotransposons occurs in a nucleocore also called nucleocapsid. This nucleoprotein structure is composed of the RNA genome and the primer tRNA coated with nucleocapsid (NC) protein molecules and of the RT and integrase enzymes. The NC protein has one or two highly conserved CX 2 CX 4 HX 4 C zinc finger(s) flanked by basic residues with key functions in particle formation and reverse transcription (10). The NC protein exerts major roles in the replication of retroviruses that can be attributed to its nucleic acid chaperone and match-maker activities (10,11). The NC protein annealing and strand transfer activities are involved in genomic RNA dimerization (12,13), initiation of reverse transcription (12,14,15), and both in minus DNA (16 -20) and plus DNA (21,22) strand transfers required for the generation of the LTRs.
The yeast retrotransposon Ty1 is a LTR-retrotransposon with two overlapping open reading frames TYA1 and TYB1. TYA1 codes for a protein that can assemble into virus-like particles (VLPs) and where reverse transcription occurs while TYB1 codes for the protease, integrase, and RT enzymes similar to those of retroviruses (23)(24)(25)(26)(27). Interestingly, the TYA1 protein appears to be functionally homologous to the retroviral GAG polyproteins, although it does not have the canonical hallmarks. In fact TYA1 protein has no zinc finger motif, no major homology region, and is not processed into matrix, capsid, and NC by the protease (28). Several LTR retrotransposons such as Gypsy, ZAM, 17.6, 297, tom, 412 (Drosophila melanogaster), TED (Autographa californica), or Ta1 (Arabidopsis thaliana), and the members of spumavirus family share these properties in common with Ty1 (29 -32).
In view of the ubiquitous nature of the nucleic acid chaperone and match-maker activities of the NC protein of retroviruses and of a number of LTR retrotransposon, such as Ty3 (33), or retroposon, such as I-Factor (34), we have studied Ty1 as a model system to look for a nucleic acid chaperone in this class of noncanonical GAG proteins. In the present report we show that a domain derived from the TYA1 protein exhibits a nucleic acid chaperone activity promoting primer tRNA i Met annealing to the multipartite primer binding site (PBS) and dimerization of Ty1 RNA in vitro.
DNA encoding Ty1 5Ј-RNA under the T7 promoter was generated by PCR using Ty1-H3 clone (GenBank TM accession number M18706), a 5Ј-oligonucleotide containing an EcoRI site, T7 promoter, and 15 nucleotides of the 5Ј-end of Ty1 RNA (5ЈTy1) and a 3Ј-oligonucleotide complementary to positions 563-578 of Ty1 RNA and containing a PstI site (3ЈTy1). The amplified DNA was cloned into pSP64 (Promega) to obtain pTy1-CG73. Template DNA was linearized at position 587 (HindIII) and 5ЈTy1 RNA was generated in vitro using T7 RNA polymerase (Riboprobe, Promega). Mutations in the PBS (see Fig. 6) were introduced using a method based on two successive rounds of PCR (35) with oligonucleotides 5ЈTy1, 3ЈTy1-oligo2-PBS, 5ЈTy1-M2-PBS, 3ЈTy1, and the pTy1-CG73 as template. The amplified DNA was cloned into pSP64 to obtain pTy1-Van5, and RNA synthesis was performed as for Ty1 5Ј-RNA.
DNA encoding tRNA i Met derived from plasmid pHG300 which contains the tRNA i Met sequence under the T7 promoter. pHG300 (36) was cut by HindIII, and BamHI and the insert was subcloned into pSP65 (Promega) to obtain pCG90. Template DNA was linearized at position 73 (BstNI), and tRNA i Met was generated in vitro using T7 RNA polymerase (RiboMax, Promega). Mutations in the tRNA i Met (see Fig. 6) were introduced by PCR using plasmid pCG90 as template, a 5Ј-oligonucleotide containing an EcoRI site, the T7 promoter, the mutations in the 5Ј-acceptor stem, and the 11 first nucleotides of the tRNA i Met (5ЈtRNAimet-KC10Ј) and a 3Ј-oligonucleotide containing a BamHI site, the mutations in the 3Ј-acceptor stem, and the last 11 nucleotides of the tRNA i Met (3ЈtRNAimet-KC10Ј). The amplified DNA was cloned into pSP64 to obtain pVan6 and RNA synthesis was performed as for tRNA i -Met wild-type.
All RNAs were purified by spin-column chromatography (Pharmacia S-300 HR) and dissolved at 1 mg/ml in sterile water. For biotinylated RNA synthesis the final concentration of rNTPs was 1 mM except for rUTP (0.8 mM), and 0.2 mM bio-16-UTP (Sigma) was added. For [ 32 P]UMP-labeled RNA synthesis, the final concentration of rNTPs was 1 mM except for rUTP (0.1 mM) and 10 Ci [␣-32 P]UTP (Amersham Pharmacia Biotech) was added. tRNA i Met was synthesized in vitro using T7 RNA polymerase, purified by 8% PAGE in 7 M urea, recovered, and dissolved at 0.1 mg/ml in sterile water after ethanol precipitation. It was heated at 90°C and cooled down progressively at room temperature in the presence of 1 mM MgCl 2 for proper folding. All mutations were verified by DNA sequencing.
DNA encoding Gag protein and its derivatives were obtained by PCR amplification using Ty1-H3 as template, a 5Ј-oligonucleotide containing a BamHI site (5ЈTy1A-E2 or 5ЈTy1A-D284) and a 3Ј-oligonucleotide containing a BglII site (3ЈTy1A-A283 or 3ЈTy1A-H401). The amplified DNA was cloned into pQE-16 (Qiagen) and then subcloned into the BamHI and HindIII sites of pET-21a (Novagen), which allows expression by coupled in vitro transcription-translation as described by the manufacturer (TNT system, Promega).
RNA Affinity Precipitation-Streptavidin magnetic beads were coated with biotinylated Ty1 5Ј-RNA 30 min at room temperature in a total volume of 300 l in a buffer containing 20 mM Tris⅐Cl, pH 7.0, 1 M NaCl, 0.1 mM MgCl 2 , 10 M ZnCl 2 . After a wash, beads were incubated 30 min at room temperature in a total volume of 300 l of buffer containing 20 mM Tris⅐Cl, pH 7.0, 30 mM NaCl, 0.1 mM MgCl 2 , 10 M ZnCl 2 and 1 mg/ml bovine serum albumin. 5 l of translation product were incubated with 100 l of beads in a final volume of 300 l in a buffer containing 20 mM Tris⅐Cl, pH 7.0, 100 mM NaCl, 0.1 mM MgCl 2 , 10 M ZnCl 2 , 1.5 g/ml salmon sperm DNA, and 1 mg/ml bovine serum albumin. Repeated washes were performed by increasing the concentration of NaCl from 0.03 to 1 M NaCl to minimize nonspecific interactions with proteins. Beads were resuspended in sample buffer, heated 5 min at 90°C, and loaded on a 15% SDS-PAGE gel. The gel was fixed, dried, and autoradiographed.
For sedimentation assay, the complexes were pelleted at 10,000 ϫ g for 10 min at 4°C. Pellet and supernatant were separated, and radioactivity in each fraction was counted.
Reverse Transcription-Reactions were performed as described previously (40). After 10 min at 30°C to allow nucleic acid annealing in 10 l (see above), the reaction volume was increased to 25 l by adding 1 pmol of MLV-RT (Life Technologies, Inc.), dNTPs at 0.25 mM each, 60 mM NaCl, and 2.5 mM MgCl 2 . Incubation was for 1 h at 37°C, after which the reaction was stopped and processed as for the analysis of tRNA annealing (see above). After phenol-chloroform extraction nucleic acids were ethanol-precipitated, recovered by centrifugation, dissolved in 15 l of formamide, denatured at 95°C for 2 min, and 2-6 l were analyzed by 8% PAGE in 7 M urea and 50 mM Tris borate, 1 mM EDTA, pH 8.3. 5Ј-32 P-Labeled ⌽X174 DNA HinfI markers (Promega) were used for size determination (not shown). In the case of self-primed reverse transcription, 32 P-tRNA was not used but dNTPs were at 0.25 mM each except for dCTP at 0.01 mM with 2 Ci of [ 32 P]dCTP (Amersham Pharmacia Biotech) per reaction.

RESULTS
The C-terminal Domain of p54 TYA1 Directs RNA Binding-Reverse transcription of Ty1 RNA occurs in VLPs in which Ty1 RNA is specifically packaged (23,41). The major protein of Ty1 VLPs is the product of TYA1, which is expressed as a 58-kDa precursor and processed by cleavage at position 401 to remove a small acidic C-terminal peptide to give rise to the p54 TYA1 protein (28). As already shown the p58 TYA1 and p54 TYA1 can bind RNA and DNA (24). Computer analysis indicates that the C-terminal domain of the p54 TYA1 protein possesses three stretches of basic amino acids, which could mediate nucleic acid binding (cf. Fig. 1A). To examine this possibility, we set up an assay with biotinylated Ty1 5Ј-RNA coupled to streptavidinmagnetic beads. p54 TYA1 and different deletion mutants (see These results confirm that the p54 TYA1 protein can interact with RNA as described previously (24) and suggest that the C-terminal part of this protein is sufficient for RNA binding.
Binding of the C-terminal Domain of p54 TYA1 to RNA Generates High Molecular Weight Nucleoprotein Complexes-To further investigate the properties of the C-terminal domain of the p54 TYA1 protein, we synthesized by Fmoc/o-pentafluorophenyl ester chemistry a 103-amino acids peptide encompassing amino acids Asn 299 to His 401 and named TYA1-D. The RNA binding ability of this TYA1-D peptide was examined by bandshift assays. 32 P-Labeled Ty1 5Ј-RNA was incubated with the TYA1-D peptide at increasing protein to nt molar ratios, and complexes were analyzed by PAGE in nondenaturing conditions. As a negative control we used a GST-cEngrailed2 fusion protein that belongs to the homeoprotein family known to bind to nucleic acids (42). At a protein to nt ratio of 1:20 to 1:10 virtually all the RNA molecules were present in high molecular weight nucleoprotein complexes (see Fig. 2A, top of gel). Since no intermediary complexes could be detected, the binding of the TYA1-D peptide to Ty1 RNA probably takes place in a cooperative manner. The TYA1-D peptide was found to bind to other RNAs and DNAs in a similar manner (cf. Fig. 3 and data not shown), suggesting that it has a strong affinity for nucleic acids. We have also investigated the ability of the TYA1-D⅐RNA complexes to be pelleted by centrifugation. As shown in Fig. 2B all the radiolabeled RNA was found in the pellet at a 1:12 protein:nt molar ratio. Taken together, these results indicate that the C-terminal domain of the Ty1 p54 protein can interact with nucleic acids generating high molecular weight complexes.
The TYA1-D Peptide Exhibits Nucleic Acid Annealing Activity-Because the previous properties of TYA1-D peptide were reminiscent of nucleic acid chaperone proteins, we have tested its effect on the annealing of two complementary single-strand DNA oligonucleotides. For this purpose we used a 32 P-labeled Tar(Ϫ) and a Tar(ϩ) oligonucleotides derived from the HIV-1 5Ј-sequence as described previously (39). The annealing of Tar(Ϫ) to Tar(ϩ) is most probably prevented at 30°C due to extensive secondary structures (see Fig. 3C, lane 1, and Fig.  3A, lanes 1-7, with Tar(Ϫ) structures migrating as two different forms in the gel). Addition of the TYA1-D peptide promoted a rapid annealing of the two complementary oligonucleotides with most of the duplexes formed at 10 min and at a peptide to nt molar ratio of 1:5 (cf. Fig. 3A, lane 15, Fig. 3, B and C, lane 14). To further characterize the annealing activity of the TYA1-D peptide, we compared it to nucleocapsid protein NCp9 of the yeast retrotransposon Ty3. Interestingly the Ty3 NCp9 and Ty1 TYA1-D were found to behave similarly regarding TAR oligonucleotide annealing (see Fig. 3C, lanes 4 -8 for NCp9 and lanes 10 -14 for TYA1-D, and Fig. 3D). In conclusion the TYA1-D peptide has strong nucleic acid annealing activity.
The TYA1-D Peptide Promotes the Annealing of Primer tRNA i Met onto Ty1 RNA-The next step was to study the TYA1-D annealing activity in a context relevant to Ty1 replication. For this purpose, we have designed an in vitro model system comprising the 5Ј-part of Ty1 genomic RNA (5ЈTy1 RNA) with its multipartite primer binding site (43,44), 32 Plabeled primer tRNA i Met (45) and TYA1-D. As shown in Fig. 4, TYA1-D allowed the annealing of primer tRNA i Met to Ty1 RNA at a peptide:nt ratio of 1:10 to 1:5 (cf. Fig. 4, A and B). Furthermore the TYA1-D peptide promoted the dimerization of Ty1 RNA and tRNA at a similar protein to nt ratios (Fig. 4A,  lanes 9 -10, and Fig. 4B, lanes 4 and 5). Interestingly RNA dimerization did not occur in the absence of primer tRNA (Fig.  4A, lanes 1-5). These results indicate that the TYA1-D peptide promotes the annealing of primer tRNA onto genomic RNA and Ty1 RNA dimerization.
To compare the chaperoning activity of the TYA1-D peptide with NC proteins from different retroelements, we tested the ability of NCp10 of Moloney murine leukemia virus, NCp7 of the HIV-1, and NCp9 from the yeast Ty3 retrotransposon or of its deletion mutant without the zinc finger (NCp9 dd) to direct  lanes 1-3). Complete products are indicated by a star. The translation products were incubated with magnetic beads coupled to streptavidin preincubated without (D, lanes 1-3) or with biotinylated 5ЈTy1 RNA (D, lanes 4 -6).
primer tRNA i Met annealing and Ty1 RNA:tRNA dimerization. Clearly NCp9 was as efficient as TYA1-D in promoting tRNA annealing and Ty1 RNA:tRNA dimerization (cf. Fig. 5, compare  lanes 1-4 with lanes 13-16). The NCp9 dd mutant was less efficient, since higher protein:nt ratios were required to obtain a high level of tRNA annealing and dimerization (cf. Fig. 5,  compare lanes 1-4 with lanes 17-20). On the other hand, NCp7 and NCp10 were almost completely inactive in these processes FIG. 2. Formation of sedimentable, high molecular weight nucleoprotein complexes between the Ty1 RNA and the TYA1-D peptide. Complexes were formed by incubating 32 P-labeled Ty1 5Ј-RNA and increasing doses of the TYA1-D peptide (lanes 7-14) or GST-engrailed fusion protein (lanes 1-6) at 30°C. A, the complexes were resolved by PAGE in nondenaturing conditions, and the gel was autoradiographed (A) or submitted to centrifugation and radioactivity was counted (B). The protein to nt molar ratios used in A are indicated at the top of the lanes.  5-8 and lanes 9 -12). This comparative study indicates that all NC proteins are not equivalent in the Ty1 model system, since only NCp9 from the Ty3 retrotransposon was as active as the TYA1-D peptide in vitro despite the absence of sequence homologies.
Mutations in Primer tRNA i Met or in the Multipartite PBS Impaired Ty1 RNA:tRNA Dimerization in Vitro Induced by the TYA1-D Peptide-As previously suggested for Ty3, in vitro RNA:tRNA dimerization is probably mediated by the 5Ј-tRNA acceptor stem (from G2 to C13), which is palindromic (see Fig.  6C, wt panel). To examine if this hypothesis also holds true for Ty1, mutations in primer tRNA i Met and compensatory mutations in the PBS (cf. Fig. 6A for Ty1 RNA structure and Fig. 6B for a description of mutations) have been performed. It should be noted that the tRNA i Met mutations were introduced in the 5Јand 3Ј-strands of the acceptor stem to maintain the stem structure. We selected these mutations made in vivo, since the PBS or tRNA i Met mutations alone abolished Ty1 transposition, while using both mutations partially restored transposition (pKC10 and pKC66 constructed by Chapman et al. (45)). The only differences between the in vitro and in vivo mutations are at positions A1G/U72C of tRNA i Met instead of A1U/U72A to allow in vitro transcription by T7 RNA polymerase (46).
As shown in Fig. 6D, mutations in tRNA i Met or the multipartite PBS drastically decreased tRNA annealing to Ty1 RNA (compare lanes 4-6 or lanes 7-9 with lanes 1-3) and completely inhibited dimerization induced by the TYA1-D peptide. When compensatory mutations in tRNA and Ty1 RNA were used (lanes 10 -12), tRNA annealing was restored but not RNA dimerization. Interestingly the mutant tRNA i Met could not form homodimers unlike wild-type tRNA i Met (see Fig. 6D, lanes 11   and 12). These results indicate that Ty1 RNA dimerization probably occurs via a tRNA:tRNA hybridization process. Thus each strand of the tRNA i Met acceptor stem appears to achieve distinct functions, since the 3Ј-strand allows priming of reverse transcription once hybridized to the PBS while the 5Ј-strand promotes RNA:tRNA dimerization (cf. Fig. 6, B and  C).
The TYA1-D Peptide Can Direct Specific Initiation of Reverse Each mutation in the 3Јacceptor stem of tRNA i Met has been compensated by a mutation in the 5Ј-acceptor stem to maintain the tRNA structure. Both wild-type and mutant tRNAs have been modified at position 1 by introducing a G to allow efficient in vitro transcription by T7 RNA polymerase. C, effect of the mutations introduced in the 3Ј-acceptor stem of tRNA i Met on the putative palindromic tRNA:tRNA interaction. D, mutations in the 5Јend of tRNA i Met acceptor stem alters Ty1 tRNA:RNA dimerization but not tRNA annealing to the PBS. Nucleoprotein complexes were formed using either wt (lanes 1-6) or mutant (lanes 7-12) Ty1 5Ј-RNA in the presence of wild-type (lanes 1-3 and 7-9) or mutant (lanes 4 -6 and 10 -12) 32 P-labeled primer tRNA i Met with increasing doses of the TYA1-D peptide as described under "Experimental Procedures." Nucleoprotein complexes were processed as described previously. The gel has been autoradiographed to visualize 32 P-labeled tRNA. M and D are monomeric and dimeric Ty1 5Ј-RNA or tRNA i Met , respectively. The protein to nt molar ratios are indicated at the top of the gel.
Transcription-As observed above (see Figs. 4 -6), the TYA1-D peptide supports annealing of replication primer to the Ty1 RNA template within a nucleoprotein complex. This prompted us to examine wether this complex was competent for the initiation of Ty1 RNA reverse transcription. To this end RT and dNTPs were added to preformed complexes composed of Ty1 RNA, 32 P-labeled tRNA i Met and TYA1-D. With or without prior annealing of primer 32 P-tRNA i Met , minus strong-stop cDNA (minus ss-cDNA) could only be detected after overexposure (cf. Fig. 7A, lanes 1 and 2). The addition of TYA1-D peptide resulted in ss-cDNA synthesis (lanes 4 and 5), indicating that the nucleoprotein complexes were functional. In presence of an excess of the TYA1-D peptide (at a peptide:nt ratio of 1:5), ss-cDNA synthesis was inhibited (lane 6) as described previously for retroviral NC proteins (20,47,48). The fact that heat annealing does not allow efficient initiation of minus ss-cDNA as does the TYA1-D peptide suggests that the tRNA-template initiation complex formed in the two conditions may be different regarding recruitment of the RT enzyme and thus the transition from initiation to elongation as shown for HIV-1 (15,47).
To further investigate the role of the TYA1-D peptide in the specificity of reverse transcription initiation, we examined wether it could inhibit nonspecific cDNA synthesis (cf. Fig. 7B). Such initiation events can arise by intramolecular interactions or folding back of the template and have been called selfpriming (20,47). Incubating Ty1 5Ј-RNA template with RT and dNTPs (including [ 32 P]dCTP) resulted in the synthesis of a large number of cDNA products (see Fig. 7B, lane 1). However, addition of increasing amounts of the TYA1-D peptide caused a strong inhibition of self-priming (lanes 2-5) as observed before with retroviral NC proteins (20,47,48).
Thus, the TYA1-D peptide directs specific initiation of re-verse transcription by tRNA annealing to the PBS of Ty1 RNA while inhibiting self-primed cDNA synthesis.

DISCUSSION
Ty1, a yeast LTR retrotransposon distantly related to oncoretroviruses and lentiviruses, encodes a major protein TYA1, considered to be equivalent to retroviral GAG, although it lacks most of the hallmarks of GAG. We have investigated the nucleic acid binding properties of the p54 TYA1 protein and mapped the nucleic acid-binding domain within the C-terminal region of p54 TYA1 (Fig. 1). In an attempt to extensively characterize the properties of this domain, it was synthesized in vitro as a highly pure 103-amino acid peptide, designated TYA1-D. This TYA1-D peptide binds both RNA and DNA in vitro and forms high molecular weight nucleoprotein complexes (Fig. 2).
Also the TYA1-D peptide was found to promote the annealing of complementary DNA oligonucleotides in a manner very similar to NCp9 of the yeast retrotransposon Ty3 and of retroviral NC proteins (Fig. 3), raising the possibility that the TYA1-D peptide has nucleic acid chaperone and match-maker activities similar to previously characterized NC zinc finger proteins (10,33,34). Using an in vitro reconstituted Ty1 replication system, the TYA1-D peptide was able to direct the hybridization of primer tRNA i Met to the multipartite PBS of Ty1 RNA, allowing elongation of tRNA i Met by RT (Figs. 4 and 7). Furthermore due to its nucleic acid chaperone and match-maker activities, the TYA1-D domain is expected to direct the DNA strand transfers during Ty1 reverse transcription as shown for retroviral NC proteins (16 -22), and this is presently under investigation. Both the TYA1-D peptide and Ty3 NCp9 appeared to be efficient in primer annealing to the PBS and in the formation of Ty1 RNA:tRNA i Met dimers, while HIV-1 NCp7 was poorly active and MLV NCp10 inactive. The common use of replication primer tRNA i Met in Ty1 and Ty3 (45,49), the nature of their multipartite PBS, and the extended tRNA i Met -PBS interactions (33,44) might explain the differences seen between Ty1 TYA1-D peptide and Ty3 NCp9, and the retroviral NC proteins. In agreement with this, HIV-1 primer tRNA Lys3 probably also interacts with genomic sequences flanking the PBS (50), but to a lesser extent than in Ty1 and Ty3. The nucleic acid binding and chaperoning activities of the TYA1-D peptide suggest that the C-terminal domain of p54 TYA1 probably is a major factor driving genomic RNA and tRNA i Met packaging required for the formation of functional Ty1 VLPs. In agreement with this notion, binding of tRNA i Met to the multipartite PBS appears to mediate its recruitment into VLPs (51). Biochemical data presented in Fig. 4 are the first direct evidences indicating that Ty1 RNA can exist in a dimer form. According to genetic data (52) intermolecular DNA strand transfers take place in the course of Ty1 replication in agreement with Ty1 RNA dimers. The observation that Ty1 RNA dimerization only occurs in presence of tRNA i Met leads us to propose a model where two genomic RNAs are linked via two tRNA i Met molecules (see Fig.  8). Dimerization of tRNA i Met would be mediated by 5Ј-5Ј interactions involving a 12-nucleotide palindrome (position 2-13, GCGCCGUGGCGC, see Fig. 6C). In support of this dimerization model, mutations in the 5Ј-tRNA palindrome impaired dimerization (Fig. 6D). Interestingly, Chapman et al. (45) found that these tRNA mutations (cf. Fig. 6C) inhibited retrotransposition, probably by interfering with tRNA i Met annealing to the PBS and reverse transcription. However, mutating the PBS to insert compensatory mutations did not restore a wild-type level of retrotransposition, but only 15% of it. In the light of our in vitro data, it is tempting to speculate that this might be due to the low level of Ty1 RNA:tRNA dimerization.
In many aspects Ty1 is much different from other retrotransposons like Ty3 or retroviruses, for example in respect to the As a control annealing of primer tRNA i Met has also been performed by heat (20 min at 70°C, lane 2). dNTPs with [ 32 P]dCTP and MLV-RT were added to the complexes to allow cDNA synthesis (1 h at 30°C). cDNA products were resolved by 8% PAGE in denaturing conditions, and the gel was autoradiographed. The horizontal black arrow shows ss-cDNA of 91 nt covalently linked to tRNA (72 nt). It should be noted that a faint band could be seen lanes 2 and 3 after overexposure. B, inhibition of self-priming. Conditions were the same except that primer tRNA was absent and that dNTPs added to perform reverse transcription contained [ 32 P]dCTP. For both panel the vertical black arrow indicates the direction of electrophoresis. 5Ј-32 P-Labeled ⌽X174 HinfI DNA markers (Promega) were used for size determination (not shown). The protein to nt molar ratios are indicated at the top of the gel. structure and maturation of its GAG precursor, called TYA1. In fact, Ty1 does not contain a retroviral (CX 2 CX 4 HX 4 C) zinc finger motif, the major homology region, and is not processed into mature matrix, capsid, and NC proteins. The sole maturation event occurs by cleavage of p58 resulting in the release of p54. The cleavage product, a very acidic peptide of 40 amino acids, cannot be detected and is probably degraded (28). It is likely that the TYA1-D peptide will exhibit nucleic acid chaperone and match-maker activities in the context of the fulllength p54 TYA1 protein. As already shown for other nucleic acid chaperone proteins like hnRNP C1/C2, the chaperone activity was found in both the full-length protein and in the RNAbinding domain alone (53). Furthermore in HIV-1, the nucleic acid chaperone activity was found in the mature NCp7, in the precursor NCp15 (13) and in the Gag polyprotein (54). Thus p54 TYA1 might be a multidomain protein with several associated functions rather than a polyprotein-like in retroviruses. The mapping of a minimal nucleic acid chaperone domain in TYA1-D is under way. A number of retroelements such as Gypsy, ZAM, 17.6, 297, tom, 412, TED, or Ta1 and the Spumaviruses share these properties in common with Ty1, since they all possess a domain with stretches of basic amino acids with possible nucleic acid chaperone properties (29 -32).
Due to the general property of single-stranded nucleic acids to fold into and to be kinetically trapped in several conformations (55), reverse transcriptases might be especially sensible to the existence of nonfunctional initiation complexes. Based on our datas and previous reports on retroviruses (10), Ty3, another yeast very distant LTR retrotransposon (33,38) and I factor, a long interspersed nuclear element-like element in Drosophila (34), we propose that RT activity, whatever the reverse transcription mechanism, should be associated with a nucleic acid chaperone protein that acts in setting up functional and specific initiation complexes. This would be achieved through two complementary mechanisms: first by annealing the primer to the template and second by inhibiting nonspecific priming events. In agreement with this notion, it has recently been reported (56) that hnRNP A1, a nucleic acid-binding protein known to exhibit a strong chaperoning activity (53), probably interacts with telomeric repeats so as to stimulate telomere elongation by telomerase, distantly related to RTs (57). Thus functional interactions between RTs and a nucleic acid chaperone protein might be a general feature of genome maintenance in retroelements and their cellular hosts.