PerspectiveMolecular Biology

Ring Around the Retroelement

See allHide authors and affiliations

Science  09 Jan 2004:
Vol. 303, Issue 5655, pp. 182-184
DOI: 10.1126/science.1093514

Retrotransposons containing long terminal repeats (LTRs) are mobile DNA elements that replicate via RNA intermediates. In their structure and mobility they resemble retroviruses, except that they are unable to move from cell to cell. Among the best characterized of the retrotransposons is the Ty1 retroelement of the budding yeast Saccharomyces cerevisiae (1). LTR-retrotransposons co-opt the host cell machinery to transcribe their genomic DNA elements into RNA—the RNA of retrotransposons serves both as genomic RNA and mRNA—and to reverse transcribe the RNA into cDNA. The cDNA then becomes integrated at a new chromosomal site (transposition) in the nucleus (see the figure, A). It is well established that the transposition of the Ty1 retrotransposon is helped by a host-encoded enzyme that degrades the intron lariats (RNA loops) of pre-mRNAs after splicing. But how can an enzyme that degrades intron lariats promote the transposition of a retrotransposon that does not contain introns? The report by Cheng and Menees (2) on page 240 of this issue suggests an intriguing and unexpected solution.

LTR-retrotransposons resemble retroviruses in many ways. For example, they are transcribed from their genomic DNA by polymerase II, have mRNAs with caps and polyadenylated tails, and have their mRNAs translated on cytoplasmic ribosomes. In addition, retrotransposon RNAs are reverse transcribed into double-stranded cDNAs by a retrotransposon-encoded reverse transcriptase (RT), and two copies of these RNAs are packaged into virus-like particles (VLPs) in the cytoplasm. The replication cycle is completed by movement of the cDNAs to the nucleus and their insertion into new chromosomal sites by a retrotransposon-encoded integrase.

The LTRs of these retrotransposons contain transcription initiation and termination signals and three sequence segments called U3, R, and U5 (see the figure, A, line 1). Transcribed LTR sequences are crucial for the synthesis of the first (minus) strand of the retrotransposons's cDNA. The primer-binding site adjacent to the 3' end of the 5' LTR is complementary to the 3' end of a transfer RNA that primes synthesis of “strong-stop” cDNA, a DNA strand complementary to the R and U5 RNA sequences. (Strong-stop DNA is the very first product of reverse transcription, and the product “stops” because the RT has reached the 5' end of the RNA.) The template for that cDNA is mostly degraded by the RT-associated ribonuclease H (RNase H) and, with the assistance of nucleocapsid protein, the cDNA then transfers or jumps to the downstream LTR to prime the synthesis of the remaining minus-strand cDNA. Synthesis of the second (plus) cDNA strand also involves strand transfer gymnastics with different sequences.

Reverse transcription of the Ty1 RNA element of yeast.

(A) Depicted is the conversion of single-stranded Ty1 RNA into a double-stranded cDNA copy. Shown in the Ty1 DNA are the U3, R, and U5 sequences of the 5' and 3' LTRs, the primer binding site, and sites for transcription initiation and polyadenylation. The initial transcript lacks 5'-U3 and 3'-U5, and the replication cycle replaces those missing sequences before the completed cDNA is integrated into a new genomic site. A color code is used to denote the LTR elements in all panels. Each cDNA strand is initiated at an internal position, as shown, and the rest of the cDNA is made after a strand transfer event. The new minus strand is shaded in gray; the new plus strand is shaded in green. (B) Model for RNA branching during the replication cycle of Ty1. The Cheng and Menees study (2) suggests a way in which intramolecular branching of Ty1 transcripts may promote initiation of cDNA synthesis and promote strand transfer followed by debranching, which promotes elongation of the initial strong-stop DNA (steps B1 to B6). (C) A model for cDNA synthesis in yeast cells lacking debranching activity. Asterisks indicate the site of the branched nucleotide where high-frequency errors are found in several retroelements, including Ty1.


A screen for host genome mutations that lower the frequency of transposition of the S. cerevisiae Ty1 retroelement yielded a recessive mutation that inhibited transposition by about 85% (3). The wild-type gene involved, DBR1, was cloned by complementation of the Ty1 transposition defect. This gene encodes the RNA debranching enzyme (Dbr1p), which hydrolyzes 2'-5' bonds in the intron lariats formed by the spliceosome (3). Yeast cells lacking DBR1 exhibit normal growth and mRNA splicing, although they do accumulate excised intron lariat RNAs.

During the dozen years since the DBR1 gene was cloned, various hypotheses have been postulated for its role in the transposition of retrotransposons. There was a lingering concern that inhibition of transposition in dbr1-deficient yeast cells could be an indirect consequence of the high levels of lariat RNAs in their nuclei. Because dbr1 mutant cells transcribe Ty1 RNA, make Ty1-encoded proteins, and assemble VLPs that contain Ty1 RNA but show an apparent kinetic defect in the accumulation of full-length Ty1 cDNA, Dbr1p may somehow promote cDNA synthesis (5, 6).

The new study by Cheng and Menees presents intriguing evidence for an unexpected solution to the puzzle of how the Dbr1p lariat debranching enzyme influences Ty1 transposition (2). First, the authors examined the 5' end of Ty1 RNA in VLPs from wild-type and mutant yeast cells. They found that Ty1 RNAs from VLPs of wild-type cells have both capped and uncapped 5' ends, but the 5' ends of VLP RNAs from dbr1 mutant cells could not be amplified, even after in vitro decapping treatment. Next, they showed that the 5' ends of VLP RNAs from the dbr1 mutant strain could be amplified if the RNA was first treated with an extract containing debranching activity. Although no intron is known to be spliced from Ty1 transcripts, the Cheng and Menees findings suggest that the Ty1 RNA in the VLPs contains a 2'-5' branch similar to that formed during pre-mRNA splicing. Further characterizations of the VLP RNA suggest a structure in which the 5' end of the first nucleotide of the Ty1 transcript (the first nucleotide of the R sequence of the upstream LTR) is joined to the 2'OH of the last nucleotide of the U3 sequence in the downstream LTR, although direct physical evidence for this structure is still needed. The proposed structure of the branch is difficult to prove using polymerase chain reaction and related methods because its juxtaposition of U3 and R sequences is virtually indistinguishable from the native structure in the 3' LTR of linear Ty1 RNA.

Like all mRNAs, Ty1 transcripts have a 3'-guanosine diphosphate (GDP) cap added after transcription to the α phosphate of the initial nucleotide of the transcript. Although it remains possible that some other transcript of Ty1 could be the precursor for this branching reaction, the simplest notion is that the lariats are formed by an intramolecular reaction of the capped RNA in which the 2'OH of the last nucleotide in the U3 sequence of the 3' LTR attacks the bond between the α and β phosphates of the capped end, releasing GDP and forming a lariat joined by a 2'-5' phosphodiester bond (see the figure, B). It is intriguing that some deoxyribozymes catalyze branching at that bond of the 5' nucleotide triphosphate in model RNA molecules (7).

The nature of the machinery that would catalyze such a reaction is unknown, but both the structure of the substrate and the sequences involved do not implicate the spliceosome. This behavior of Ty1 RNA is new because it does not occur at a 3'-5' phosphodiester bond and corresponds to only the first half of a splicing reaction: The released cap is not spliced to a downstream element of the transcript. Only a fraction of Ty1 transcripts are found in VLPs, with the majority serving as mRNAs in the cytoplasm and some in the nucleus. The authors suggest that branching may divert some transcripts toward packaging in VLPs instead of serving as mRNAs (2). They propose that these branches may facilitate minus-strand transfer, a key step in first-strand cDNA synthesis by LTR retrotransposons. That transfer is promoted by the RNase H and nucleocapsid proteins, but its exact mechanism—in particular, how the target sequences are found—is not known. The authors note that the branch in Ty1 RNA places the 5' LTR and the strong-stop cDNA right at the 3' LTR (see the figure, B) and suggest that this proximity promotes intramolecular transfer of the initial cDNA to the 3' LTR. Because the branch should impede elongation of the transferred cDNA (see the figure, B, line 5), synthesis of full-length cDNA would appear to be enhanced by hydrolysis of the branch by Dbr1p after strand transfer (see the figure, B, line 6). Note that the presence of cDNA near the branch would make such a Dbr1p substrate quite distinct from a standard intron RNA lariat.

Orchestrating the branching and debranching reactions in Ty1 RNAs seems necessary to propel the transposition pathway forward, but we do not yet understand how that occurs. It is not known in which intracellular compartment the branching reaction takes place, but it is likely that the debranching reaction occurs in cytoplasmic VLPs because cDNA synthesis takes place there and branched RNAs accumulate in VLPs in dbr1 mutant cells.

Transposition of retrotransposons is not eliminated in the dbr1 mutant cells, but is reduced by about 85%. The authors suggest that the Ty1 RT may occasionally yield full-length cDNA by reading through the 2'-5'branch (perhaps without strand transfer) (see the figure, C). Several groups have shown that various RTs can read-through a 2'-5' branch but only inefficiently [e.g., (4)]. When read-through of a branch site occurs, the RT often incorporates a nucleotide that differs from the Watson-Crick partner of the branched template nucleotide (see the figure, C, asterisks). Remarkably, LTR retrotransposons make high-frequency “errors” that correspond precisely to the last nucleotide of the U3 sequence of the 3' LTR (that is, the position of the branch nucleotide in Ty1 RNA) (8, 9). The finding that an internal deletion of Ty1 inhibits transposition by ∼90% and that the level of transposition is not further inhibited in dbr1 mutants (5) suggests that internal Ty1 sequences may be required for branching. Also, a long-range base-pairing interaction in Ty1 RNA has been reported that controls cDNA synthesis from Ty1 transcripts (10). That interaction, and likely others, may control the stepwise formation of a folded RNA structure that promotes this branching reaction and chaperones the convoluted reverse transcription process. The existence of branched RNA could also help to explain why the accumulation of strong-stop DNA is so easily observed in disrupted retroviral and Ty1 particles but is not readily observed in vivo, where branching and debranching could facilitate strong-stop transfer.

Cheng and Menees have reported intriguing features of Ty1 transcripts in VLPs that provide a new model for the previously undefined role for a 2'-5' phosphodiesterase in the Ty1 transposition mechanism. Experimental extensions of their research will provide further details of this unexpected role for branching in the replication cycle of this LTR retrotransposon. If branching proves to be general for other LTR retrotransposons—even for retroviruses—then the Cheng and Menees work will have provided an important new insight into the process of reverse transcription.


  1. 1.
  2. 2.
  3. 3.
  4. 4.
  5. 5.
  6. 6.
  7. 7.
  8. 8.
  9. 9.
  10. 10.
View Abstract

Navigate This Article