Evidence for the Packaging of Multiple Copies of Tf1 mRNA into Particles and the trans Priming of Reverse Transcription

Abstract

Long terminal repeat (LTR)-containing retrotransposons and retroviruses are close relatives that possess similar mechanisms of reverse transcription. The particles of retroviruses package two copies of viral mRNA that both function as templates for the reverse transcription of the element. We studied the LTR-retrotransposon Tf1 of Schizosaccharomyces pombe to test whether multiple copies of transposon mRNA participate in the production of cDNA. Using the unique self-priming property of Tf1, we obtained evidence that multiple copies of Tf1 mRNA were packaged into virus-like particles. By coexpressing two distinct versions of Tf1, we found that the bulk of reverse transcription that was initiated on one mRNA template was subsequently transferred to others. In addition, the first 11 nucleotides of one mRNA were able to prime, in trans, the reverse transcription of another mRNA.

Much of what is known about the reverse transcription of long terminal repeat (LTR)-containing elements results from the study of retroviruses. Retroviruses are unique among viruses in their ability to incorporate two identical molecules of genomic RNA into a single viral particle (17). Through reverse transcription, both RNAs contribute to the production of the full-length, double-stranded cDNA (4, 5, 15). Reverse transcription in retroviruses begins with the priming of DNA synthesis by specific tRNA species that anneal to primer binding sites (PBS) located just downstream of the 5′ LTR. The virus-encoded reverse transcriptase extends the tRNA primer up to the 5′ end of the mRNA to generate the first intermediate of reverse transcription, called the minus-strand strong-stop DNA (−ssDNA). Several subsequent steps of strand transfer and extension as well as the plus-strand priming are required to complete cDNA synthesis (16). Once completed, the cDNA is inserted into the host genome via the action of integrase.

LTR-retrotransposons are useful models for the study of the retrovirus life cycle. LTR-retrotransposons express Gag, protease, reverse transcriptase, and integrase. These are the same proteins that retroviruses require for their replication. In addition, the mechanisms of reverse transcription of LTR-retrotransposons are the same as those used by retroviruses. Like retroviruses, most LTR-retrotransposons use cellular tRNAs to prime the synthesis of −ssDNA and the polypurine tract to prime plus-strand synthesis (1, 3, 6, 7). Though LTR-retrotransposons have many similarities to retroviruses, little is known about the number of mRNAs that are packaged within an individual virus-like particle (VLP) or the number of mRNAs that are required for the reverse transcription of a full-length cDNA. Here, we take advantage of the genetics of yeast to address these questions.

Tf1 is a highly active, LTR-containing retrotransposon isolated from the fission yeast Schizosaccharomyces pombe (10, 12). The expression of Tf1 in S. pombe results in the assembly of VLPs that contain Tf1 mRNA, Tf1 proteins, and cDNA (11). The question of the number of mRNAs that contribute to the reverse transcription of Tf1 is particularly important because of the unusual mechanism Tf1 uses to initiate reverse transcription. Tf1 employs a novel mechanism of self-priming in which the first 11 bases of the mRNA anneal to the PBS. Following a cleavage event between bases 11 and 12, the 11-nucleotide product serves as the primer of minus-strand reverse transcription (8, 9).

Previous results have demonstrated that single-nucleotide substitutions in either the primer or the PBS severely reduce formation of −ssDNA and Tf1 transposition (8). When second mutations are introduced in the same molecule of Tf1 mRNA to complement the first substitution, the transposition activities are restored. This consequence leads to the question of whether multiple molecules of Tf1 mRNA can be packaged into the same VLP and whether the complementarity results in formation of the priming duplex between two molecules of Tf1 mRNA. To examine the number of mRNAs in a VLP, we developed a system of measuring transposition in S. pombe that expressed two versions of Tf1 mRNA.

This assay system is illustrated in Fig. 1, in which two versions of Tf1 mRNA were expressed in S. pombe from separate plasmids. One plasmid contained URA3 as a selectable gene and a version of Tf1 marked with a neo gene (Tf1-neo). The other plasmid contained LEU2 as the selectable marker and a copy of Tf1 without the neo gene (Tf1-unmarked). These two versions were the only source of Tf1 expression, as the yeast strains did not possess genomic copies of Tf1 (12). The expression of both Tf1-neo and Tf1-unmarked was under the regulation of a strong inducible promoter (nmt1). After induction of transcription, the copies of Tf1 mRNA were translated into proteins which coassemble along with the mRNA into VLPs (11). Assuming that the particles of Tf1 copackaged two molecules of mRNA, as is the case for retroviruses, the following predictions were made. In 50% of the particles, one mRNA of Tf1-unmarked would be copackaged with one mRNA of Tf1-neo. Twenty-five percent of the particles would contain two copies of mRNA from Tf1-neo, and the remaining 25% would contain two copies of mRNA from Tf1-unmarked.

FIG. 1.

The assay system for detecting copackaging of Tf1 mRNA in S. pombe. Two versions of Tf1 were expressed from separate plasmids in S. pombe. Tf1-neo was expressed from the plasmid containing URA3 (left). Tf1-unmarked was expressed from the plasmid containing LEU2 (right). If Tf1 VLPs package two copies of mRNA, 50% of the particles would contain one copy of each type of Tf1 mRNA. Due to the presence of a point mutation in the PBS of Tf1-neo and the lack of a neo gene in Tf1-unmarked, G418 resistance could only result from the copackaging of at least one transcript from each version of Tf1 and their subsequent reverse transcription.

To test whether multiple copies of Tf1 mRNA were packaged into VLPs, nucleotide substitutions were included in the PBS of Tf1-neo. Because these mutations blocked the initiation of reverse transcription from the mRNA of Tf1-neo, transposition could only be detected if reverse transcription was first initiated on the mRNA of Tf1-unmarked and subsequently transferred to the mRNA of Tf1-neo. Hence, detection of transposition would be an indication of copackaging of mRNAs.

Transposition activity was detected as G418 resistance in cells that have gained newly transposed copies of Tf1-neo. However, G418 resistance due to transposition can only be detected after the plasmid copy of neo has been evicted (8, 10). The plasmid versions of Tf1-neo were selected against by replica printing cells to Edinburgh minimal medium (EMM) containing 5-fluoroorotic acid. Cells were then replica printed to a medium containing G418, and growth on this medium indicated that Tf1-neo transposed into the genome.

The strains used in this study are listed in Table 1, and the nucleotide substitutions and their relative positions on the PBS duplex of Tf1 are shown in Fig. 2A. High levels of transposition activity were detected in strains serving as positive controls. These strains contained a wild-type copy of Tf1-neo coexpressed with either a wild-type copy of Tf1-unmarked or the vector backbone (Fig. 2B and C, WT/WT and WT/vector). Strains that expressed Tf1-neo with the G178C or C176G mutations in the PBS exhibited low levels of G418 resistance when coexpressed with either the vector backbone or a copy of Tf1-unmarked which possessed the same point mutation (Fig. 2B, G178C/vector and G178C/G178C, and Fig. 2C, C176G/vector and C176G/C176G). However, in the presence of a wild-type copy of Tf1-unmarked, the transposition activities in strains with defective copies of Tf1-neo were restored (Fig. 2B and C, G178C/WT and C176G/WT, respectively). These results indicate that multiple copies of Tf1 mRNA are packaged into a VLP and contribute to the reverse transcription of individual copies of Tf1 cDNA.

TABLE 1.

Yeast strains used

Yeast straina	Plasmid description
	Tf1-neo and URA3	Tf1-unmarked and LEU2	Tf1 allele (Tf1-neo/Tf1-unmarked)
YHL6632c	pHL891-19d	pHL1451-2c	WTb/WT
YHL6751-2,4	pHL1704-2	pHL1451-2	G178C in PBS/WT
YHL6752-1,2	pHL1704-2	pHL868-1	G178C in PBS/G178C in PBS
YHL6753-2,3	pHL1704-2	pHL869-1	G178C in PBS/C5G in primer
YHL6754-3,4	pHL1704-2	pHL870-1	G178C in PBS/G7C in primer
YHL6755-2,4	pHL1705-1	pHL1451-2	C5G in primer/WT
YHL6756-2,4	pHL1705-1	pHL869-1	C5G in primer/C5G in primer
YHL6757-2,3	pHL1705-1	pHL868-1	C5G in primer/G178C in PBS
YHL6758-3,4	pHL1705-1	pHL867-1	C5G in primer/C176G in PBS
YHL6868-2,4	pHL1706-2	pHL1451-2	G7C in primer/WT
YHL6869-3,4	pHL1706-2	pHL870-1	G7C in primer/G7C in primer
YHL6870-1,3	pHL1706-2	pHL867-1	G7C in primer/C176G in PBS
YHL6871-1,3	pHL1706-2	pHL868-1	G7C in primer/G178C in PBS
YHL6872-2,3	pHL1703-5	pHL1451-2	C176G in PBS/WT
YHL6873-1,2	pHL1703-5	pHL867-1	C176G in PBS/C176G in PBS
YHL6874-1,3	pHL1703-5	pHL870-1	C176G in PBS/G7C in primer
YHL6875-1,3	pHL1703-5	pHL869-1	C176G in PBS/C5G in primer
YHL7375	pHL415-2	pSP1e	PR-fs/empty vector
YHL7439	pHL1995-1	pHL1451-2	C176G + C5G/WT
YHL7440	pHL1995-2	pHL1451-2	C176G + C5G/WT
YHL7441	pHL1996-1	pHL1451-2	G178C + G7C/WT
YHL7442	pHL1996-2	pHL1451-2	G178C + G7C/WT
YHL7462	pHL1703-5	pSP1	C176G/empty vector
YHL7463	pHL1704-2	pSP1	G178C/empty vector
YHL7464	pHL891-19	pSP1	WT/empty vector

^aAll plasmids were transformed into YHL912 (h⁻ ura4-294 leu1-32). 

^bWT, wild type. 

^cReference 14 

^dReference 13 

^eReference 2 

FIG. 2.

Results of transposition assays indicated that multiple copies of Tf1 mRNA copackaged and contributed to the reverse transcription of full-length Tf1 cDNA. (A) The nucleotide sequences and substitutions in the PBS duplex. The arrows indicate the positions of the single-base substitutions. (B and C) Results of transposition assays of strains that contained defective copies of Tf1-neo coexpressed with various versions of Tf1-unmarked. The mutations of Tf1-neo are G178C (B) and C176G (C). Patches in rows 1 and 2 represent two independent transformants of the same strains. Identical patches of a representative transformant are shown for rows 3 through 6. In both panels, row 1 shows a strain (YHL7464) containing a wild-type sequence of Tf1-neo coexpressed with a vector lacking Tf1. Rows 2 to 4 are strains YHL7463, YHL6752, and YHL6751 (B) and YHL7462, YHL6873, and YHL6872 (C). Strains in row 5 are YHL7441 (B) and YHL7439 (C). Shown in row 6 as a negative control is a strain (YHL7375) that expressed Tf1-neo with a frameshift in protease and coexpressed a vector lacking Tf1. The WT/WT control (YHL6632) is shown in the last row. The patches in each panel were from the same plate. All plates were treated identically. (D) The cis-mutant/WT configuration was generated to test the possible role of trans priming in reverse transcription. A second mutation was introduced into Tf1-neo, disabling the potential for trans priming between the two versions of Tf1 mRNA.

Since multiple copies of Tf1 mRNA are present during the reverse transcription of individual cDNAs, the above results introduced the possibility that the priming of Tf1 reverse transcription could occur in a trans configuration. That is, the first 11 nucleotides of one mRNA could potentially anneal to the PBS of another copy of Tf1 mRNA. To evaluate this possibility, strains were generated that contained a copy of Tf1-neo with a mutation in the PBS and an additional mutation in the primer that did not complement the PBS. A G7C substitution was introduced in the version of Tf1-neo containing G178C, and a C5G substitution was introduced in the version of Tf1-neo containing C176G. These strains also contained a plasmid with the wild-type copy of Tf1-unmarked, and this configuration was called cis mutant. The introduction of a second substitution in the transcript of Tf1-neo disabled the potential for any trans-priming interactions between transcripts from separate plasmids encoding Tf1 (Fig. 2D). These strains (G178C+G7C/WT and C176G+C5G/WT) exhibited transposition frequencies equivalent to those obtained by the G178C/WT and C176G/WT strains in which Tf1-neo contained the substitution only in the PBS (Fig. 2B and C). This result indicated that cis priming of the WT transcript of Tf1-unmarked followed by strand transfer to the neo-containing transcript was the predominant mode of reverse transcription that was responsible for the majority of the transpositions in the cis-mutant-WT strains.

The conclusion that trans priming was not the predominant mode of initiating reverse transcription was based on the assumption that no more than two mRNAs serve as template for the synthesis of an individual cDNA. However, if more than two copies of mRNA were packaged, there exists the possibility that reverse transcription could have initiated from trans priming between two WT transcripts of Tf1-unmarked. In addition, it is possible that cis priming was the preferred mode but that in the absence of intramolecular priming, reverse transcription could be repaired by priming in trans. In light of these possibilities, we tested whether strains were capable of priming in trans when both versions of Tf1 were defective for cis priming.

Pairs of Tf1 were coexpressed in strains that possessed mutations in the primer of one mRNA and mutations in the PBS of the other mRNA. Each of these elements individually was defective for priming reverse transcription (8). If trans priming can occur, then the combination of defective elements should rescue the transposition defects. Strains with four combinations of Tf1-neo and Tf1-unmarked elements were tested for their potential to trans prime. These strains are termed trans-wild-type (T-WT), trans-identical (T-ID), trans-complementary (T-C), and trans-noncomplementary (T-NC). Their potential to trans prime is depicted in Fig. 3.

FIG. 3.

The models for trans priming in strains with four combinations of Tf1-neo and Tf1-unmarked. The four combinations are trans-wild-type (T-WT), trans-identical (T-ID), trans-complementary (T-C), and trans-noncomplementary (T-NC). A single trans-priming duplex has the potential to form in T-WT and T-NC configurations through the interactions between the two wild-type sequences. No trans-priming potential exists in the T-ID configuration. In the T-C configuration, two potential trans-priming duplexes could form through the complementarity of the two wild-type and two mutated duplexes.

In the T-WT configuration, a defective Tf1-neo with a mutation in either the primer or the PBS is coexpressed with a WT Tf1-unmarked. In addition to cis priming of the WT Tf1-unmarked, one configuration of trans priming was possible. The example of T-WT drawn in Fig. 3 has the potential to form a trans duplex between the primer of the Tf1-unmarked and the PBS of the Tf1-neo. The T-ID strains contained the identical mutation in both the Tf1-neo and the Tf1-unmarked. Neither cis priming nor trans priming could occur in the T-ID strains. The T-ID strains therefore served as the baseline controls. In the T-C configuration, not only could trans priming occur through the wild-type sequences of Tf1-neo and Tf1-unmarked but it also could result from the combination of the PBS and the primers with mutations that reestablish complementarity (Fig. 3). If priming occurs in trans, this combination of defective elements would show higher levels of transposition than strains with two identical versions of Tf1 with the substitutions. However, this increase in transposition may be limited because only particles that packaged both versions of the Tf1 mRNA could trans prime (Fig. 1). In the T-NC strains, trans priming could only occur between the wild-type sequences of Tf1-neo and Tf1-unmarked. In addition, no cis priming could take place.

The results of transposition assays are shown for sets corresponding to the substitutions G178C, C176G, C5G, and G7C in Tf1-neo (Fig. 4). In all four panels, the T-WT strains showed high levels of transposition activity compared to WT/WT strains. This is in contrast to the T-ID strains that resulted in drastically reduced transposition frequencies (Fig. 4). The T-C strains possessed the potential for trans-priming interactions, and in three of the four cases tested, they showed a distinct increase in transposition activity over that of the T-ID strains (Fig. 4). The exception was the set corresponding to the C5G substitution in Tf1-neo (Fig. 4C). The T-NC strains possessed versions of Tf1-neo and Tf1-unmarked with substitutions that did not complement each other (Fig. 3). In Fig. 4A and B, there was a reduction in transposition frequencies compared to those of the T-C strains. However, in Fig. 4C and D, no reduction was observed, suggesting that in these cases the wild-type sequence in the priming duplex was more active for trans-priming interactions than mutant duplexes with restored complementarity.

FIG. 4.

Results of transposition assays indicating that the reverse transcription of Tf1 mRNA can be primed in a trans configuration. Each panel represents one of four nucleotide substitutions in Tf1-neo. From top to bottom, independent transformants are shown for T-WT, T-ID, T-C, and T-NC strains. Strain numbers from top to bottom are as follows: (A) YHL6751, YHL6752, YHL6753, YHL6754; (B) YHL6872, YHL6873, YHL6874, YHL6875; (C) YHL6755, YHL6756, YHL6757, YHL6758; (D) YHL6868, YHL6869, YHL6870, YHL6871. Note that a duplicate patch of YHL6757 is present. The patches in each panel were from the same plate. All plates were treated identically.

If the high levels of G418 resistance produced by the T-C and T-NC strains were due to trans priming and the subsequent insertion of the cDNA, there should be a corresponding increase in Tf1 cDNA that accumulated in these strains. DNA blot analysis was used to measure cDNA levels in strains that contained the C5G and C176G in Tf1-neo. Total DNA was extracted from these strains and digested with ApaI. After separation by agarose gel electrophoresis, the DNA was transferred to a GeneScreen membrane and probed with a fragment of DNA specific for Gag (Fig. 5). Using this probe, three sizes of bands corresponding to the Tf1 cDNA were detected. The 4.2-kb band represents the accumulated amount of linear cDNA produced by Tf1-neo and Tf1-unmarked. The 4.5- and 5.5-kb bands are likely to be single LTR circles generated by Tf1-unmarked and Tf1-neo, respectively (unpublished data). In addition, fragments of DNA from the plasmids, containing LEU2 and URA3, were visualized as 12.6- and 10.5-kb bands, respectively. The intensities of these plasmid bands served as loading controls to compare the levels of Tf1 cDNA production. As expected for strains with high transposition activity, T-WT strains for both sets resulted in high levels of cDNA expression (Fig. 5, WT/C5G and WT/C176G). While Tf1 cDNA bands were not prominent for either set of T-C and T-NC strains, faint bands and smeared signals of cDNA were detected. The cDNA signals in both T-C strains (C5G/G178C and C176G/G7C) were noticeably darker than those produced by T-ID strains. Therefore, the higher levels of cDNA signals correspond to the strains with the potential to trans prime. One discrepancy is that the increased levels of cDNA in the T-C strain of C5G in Tf1-neo rarely showed a corresponding increase in transposition activity compared to that of the T-ID strain. One explanation is that the T-ID strain of C5G showed higher levels of background transposition than the other T-ID strains.

FIG. 5.

The cDNA levels produced by Tf1-neo and Tf1-unmarked indicate that reverse transcription was able to prime in a trans configuration. Results from two sets of strains, containing C5G or C176G in the Tf1-neo, are shown. Total DNA was extracted from strains that were induced for Tf1 transposition. DNA was digested with ApaI and probed with a Gag-specific sequence (an EcoRI fragment of 748 bases). The 4.2-kb band represents the accumulated amount of Tf1 cDNA produced by Tf1-neo and Tf1-unmarked. The 4.5- and 5.5-kb bands are likely the single LTR circles resulting from reverse transcription of Tf1-unmarked and Tf1-neo, respectively. The 12.6- and 10.5-kb bands resulted from fragments of DNA from the LEU2 and URA3 plasmids, respectively.

The same filter that was exposed to X-ray film was also scanned for radioactivity with a phosphorimager (Storm 860; Molecular Dynamics). We measured the total level of cDNA, including intermediates, by scanning the length of each lane except for the plasmid bands. This signal was normalized to the sum of the bands produced by the two plasmids. The quantitative results are shown in Table 2. The cDNA produced by the strains with the potential to prime reverse transcription in trans, C5G/G178C and C176G/G7C, produced 1.9 and 2.3 times more cDNA than their corresponding T-ID strains, respectively.

TABLE 2.

Quantification of cDNA on DNA blot

Tf1-neo/ Tf1-unmarked	Total amt of plasmid	Total amt of cDNA	% cDNA in plasmid
WT/WT	86,840.0	360,400.0	100
PR-fs/vector	22,420.0	26,100.0	28.1
WT/C5G	51,660.0	183,300.0	85.5
C5G/C5G	58,320.0	75,310.0	31.1
C5G/G178C	55,540.0	139,300.0	60.4
C5G/C176G	68,500.0	87,890.0	30.9
C176G/WT	71,310.0	272,400.0	92.0
C176G/C176G	62,510.0	80,240.0	30.9
C176G/G7C	62,520.0	182,800.0	70.5
C176G/C5G	66,000.0	173,300.0	63.3

The fact that the strains with the potential to trans prime produced increased levels of cDNA as well as transposition activity provides strong evidence that one copy of Tf1 mRNA can prime in trans the reverse transcription of another Tf1 mRNA. The trans priming of Tf1 reverse transcription is also supported by the unusual observation that half of the minus-strand strong-stop cDNA is full length and half is missing 11 nucleotides at the 3′ end (8). The full-length species could not have been the result of self-priming in cis because the nucleotide primer would be missing from the 5′ end of the mRNA, whereas trans-primed cDNA would be full length.

That the rescues in transposition produced by trans priming in the T-C strains were modest could be attributed to two factors. The possibility that retrotransposon particles are similar to retroviruses in that they package two copies of mRNA suggests that only half of the particles would contain the combination of the two types of mutant RNAs necessary for trans priming to occur (Fig. 1). In addition, our previous finding that half of the minus-strand strong-stop cDNA is full length and half is missing 11 nucleotides at the 3′ end suggested that trans priming accounted for only half of the initiation events. Therefore, the highest level of transposition detected in the T-C strains should be no more than ¼ of the level of wild-type Tf1.

trans priming is an unusual mechanism of initiating reverse transcription that may be unique to elements that are self-priming (13). The finding that strains blocked for trans priming (Fig. 2B and C, row 5) showed little reduction in transposition indicates that cis priming was likely an important mode of initiation of reverse transcription. This also suggests that trans priming may serve as a rescue mechanism for elements that have lost the ability to prime reverse transcription in cis. Nevertheless, the high levels of transposition activity produced by the T-WT and cis-mutant-WT strains was direct evidence that a substantial amount of the cDNA synthesis was initiated on one mRNA template and subsequently transferred to others. The reverse transcription of retroviruses has been reported to exhibit high degrees of template switching from one mRNA to another during the synthesis of cDNA (4, 5, 15). Our results provide the evidence that LTR-retrotransposons are similar to retroviruses in that they too use multiple copies of mRNA during reverse transcription.