|Home | About | Journals | Submit | Contact Us | Français|
Position-specific integration of the retroviruslike element Ty3 near the transcription initiation sites of tRNA genes requires transcription factors IIIB and IIIC (TFIIIB and TFIIIC). Using a genetic screen, we isolated a mutant with a truncated 95-kDa subunit of TFIIIC (TFIIIC95) that reduced the apparent retrotransposition of Ty3 into a plasmid-borne target site between two divergently transcribed tRNA genes. Although TFIIIC95 is conserved and essential, no defect in growth or transcription of tRNAs was detected in the mutant. Steps of the Ty3 life cycle, such as protein expression, proteolytic processing, viruslike particle formation, and reverse transcription, were not affected by the mutation. However, Ty3 integration into a divergent tDNA target occurred exclusively in one orientation in the mutant strain. Investigation of this orientation bias showed that TFIIIC95 and Ty3 integrase interacted in two-hybrid and glutathione S-transferase pulldown assays and that interaction with the mutant TFIIIC95 protein was attenuated. The orientation bias observed here suggests that even for wild-type Ty3, the protein complexes associated with the long terminal repeats are not equivalent in vivo.
Genomic sequence analysis has shown that retroelements account for a significant proportion of the genomes of plants, animals, and microbes. Among various host organisms, the budding yeast Saccharomyces cerevisiae offers one of the most genetically tractable model systems for studying these elements. Currently, five retrotransposons, Ty1 through Ty5, have been identified in S. cerevisiae. Of these retroelements, Ty1, Ty2, Ty4, and Ty5 belong to the copialike family whereas Ty3 belongs to the gypsylike family. Although Ty3 is limited to an intracellular life cycle, It has many similarities to retroviruses, including a number of steps in its life cycle (34). Retrotransposons and retroviruses rely on proteins encoded in the element or virus and in host cells. Full-length Ty3 DNA is approximately 5.4 kb in length (10). It encodes Gag3p and Gag3-Pol3p polyproteins, which are processed into mature proteins in the context of the Ty3 viruslike particle by the Ty3 protease. Gag3p is processed into major structural proteins, capsid (CA) and nucleocapsid. Gag3p-Pol3p is processed into catalytic proteins PR, reverse transcriptase and integrase (IN) (17). The life cycle requires host factors such as transcription and translation machinery and unknown factors involved in such processes as assembly, uncoating, nuclear import, and target site recognition.
Several host factors affect the efficiency or position of retroelement integration. For retroviral integration, in vitro experiments showed enhancement of integration by Ini1 (22), HMG1 (1), and HMG I(Y) (14) and reduction of autointegration by barrier to autointegration factor (29). For yeast retrotransposons, in vivo experiments have identified chromatin-associated proteins that enhance the efficiency of integration into promoter regions of RNA polymerase II (pol II)-transcribed genes or heterochromatic DNA or affect the general efficiency of integration. For example, Ubc2 and CAF-I affect the integration preference of Ty1 (20). Ty5 targets heterochromatic DNA through contacts mediated by Sir proteins (42). Most genomic Ty1 and Ty2 elements are found within 750 bp of the 5′ end of tRNA genes (24), and this targeting presumably involves host proteins.
Ty3 is distinguished from other retroelements by its extreme integration specificity. It integrates within a few nucleotides of the transcription initiation sites of pol III genes on plasmids (9) and of chromosomal tRNA genes (24). There is no sequence similarity among pol III transcription initiation sites, suggesting that the structure of the transcription initiation complex, rather than a consensus DNA sequence, is responsible for the specificity of integration (9). The tRNA and U6 gene transcription preinitiation complexes are composed of transcription factors IIIC and IIIB (TFIIIC and TFIIIB). TFIIIC binds to promoter elements, box B and box A, and recruits the initiation factor TFIIIB, which binds upstream of the initiation site. In the case of the tRNA and U6 genes, TFIIIC is required for transcription in vivo, but in defined in vitro systems TFIIIB can mediate TATA box-dependent transcription in the absence of TFIIIC (39). Similar to what is observed for transcription, only TFIIIB is required for integration upstream of the U6 gene in vitro (40). However, in vitro, chromatographic fractions containing TFIIIB and TFIIIC are required for position-specific integration of Ty3 upstream of a tRNA gene (27). As is the case for transcription in vivo, a point mutation in box B that abolishes TFIIIC binding abrogates the activity of a tRNA or U6 in Ty3-targeting assays (9). Although the in vitro studies suggest that direct contacts must occur between TFIIIB and the preintegration complex (PIC), they did not address whether the in vivo role of TFIIIC in transposition is indirect, that is via loading TFIIIB, or direct, through interactions with the PIC.
In this study, we identified a mutation that caused truncation of the 95-kDa subunit of TFIIIC (TFIIIC95) and severely reduced the recovery of Ty3 integrants in an in vivo assay. The mutant strain had no detectable defect in growth or transcription of tRNAs. Although intermediates of the Ty3 life cycle were not altered, Ty3 elements integrated between a pair of divergent tRNA genes in the mutant strain showed orientation bias. Furthermore, interaction observed between Ty3 IN and TFIIIC95 was attenuated for the mutant protein, suggesting that TFIIIC95 participates directly in docking the PIC. These results provide the first genetic evidence directly linking the targeting of a Ty element to a specific component of the pol III transcription initiation complex. The orientation bias observed here offers insights into the asymmetry of a retroelement PIC.
The strains and plasmids used in this work are listed in Tables Tables11 and and2.2. Media and standard techniques for yeast were as previously described (35). The haploid yeast strain yMA1322 used to generate mutants was derived from YPH500 (36). First, YPH500 was transformed with the LYS2 gene excised from pDP6 (32), resulting in the lysine prototrophic strain yMA1241. An ochre allele of the LYS2 gene was generated by PCR amplification of the wild-type LYS2 gene present on the pDP6 plasmid template, with primers 489 and 485 (sequences of the oligonucleotides used in this study are shown in Table Table3).3). This amplification converts Tyr31 to an ochre codon. The PCR product was transformed into yMA1241, and transformants were plated onto α-aminoadipate medium to select for lysine auxotrophs. Multiple lysine auxotrophs were transformed with the pTIT plasmid (L. Yieh, unpublished work), which carries the SUP2bo suppressor tDNA, and SUP2bo-dependent lysine prototrophs were selected. One of the strains carrying a suppressible lys2o allele was designated yMA1322. A LEU2 gene excised with BamHI and NarI from plasmid YEp351 was transformed into yMA1322 to obtain leucine-prototrophic strain yMA1342, the reference strain which is referred to as wild type in this study.
Strain yMA1343 (25-41A mutant) was identified in the genetic screen as having a reduced frequency of transposition-dependent activation of the suppressor tRNA, sup2bo. The pMA1922 (pTFC1::mTn) plasmid was generated by the plasmid gap rescue method (16) from yMA1343. Briefly, pTFC1 (yCpCS7) DNA (11) was cleaved with AatII and HpaI to generate a gap within the TFC1 coding region flanking the mTn3 insertion site. This gapped plasmid was transformed into yMA1343, and Ura+ transformants carrying plasmids repaired by gene conversion from the TFC1 locus were selected. Plasmids (pTFC1::mTn) from Ura+ colonies were rescued by transforming DNA from those strains into Escherichia coli HB101. The ApaI-SacII fragment containing part of the TFC1 gene with an mTn3 insertion was excised from pTFC1::mTn plasmid and transformed into the yMA1322 strain, and Leu+ transformants were selected to obtain yMA1344. DNA isolated from yMA1344 was analyzed by Southern blotting using a TFC1-specific probe to confirm the TFC1 disruption in this strain.
To minimize homologous recombination, which generates background in the helper-donor transposition assay, the RAD52 gene was deleted using a knockout construct. The pBJC302 plasmid (12) cleaved with PvuII was transformed into the wild-type (yMA1322) or tfc1 strain, and transformants were selected on synthetic complete medium lacking uracil (SC-Ura). Subsequent loss of the URA3 gene from these cells was selected on 5-fluoroorotic acid medium. Gene disruption was confirmed by Southern and PCR analysis.
Yeast two-hybrid plasmids for Ty3 IN were constructed by cloning a BamHI fragment with Ty3 IN sequence into pMA424 (Gal4 BD fusion) and pGAD2F (Gal4 AD fusion) vectors. Constructs for Ty3 IN domains amino-terminally fused to Gal4 BD were generated by single-stranded mutagenesis (28) from pAS IN by using oligonucleotides 471 (pAS IN-A), 472 (pAS IN-C), 473 (pAS IN-AB), and 470 (pAS IN-BC) and from pAS IN-BC by using oligonucleotide 473 (pAS IN-B). Plasmids pAS95, pACT95, and pACT55 were described previously (30). pAS95ΔC was constructed by cloning the Ncol fragment of TFC1 from pAS95 into the NcoI site of pAS1-CYH2. The same NcoI fragment was excised from pAS95 and the backbone was religated to yield pAS95(C).
To generate glutathione S-transferase (GST) fusion constructs, the pGEX2T vector (Pharmacia) was linearized with BamHI and blunt ends were created by filling in with Klenow polymerase. The SmaI-SalI and NcoI-SalI fragments of TFC1 from pAS95 were treated with Klenow polymerase to generate blunt ends and cloned into the pGEX2T vector described above to create pGSTτ95 and pGSTτ95ΔC, respectively. The BamHI fragment containing the IN-A domain from pAS IN-A was treated with Klenow and cloned into the same vector to create pGST IN-A.
To generate templates for in vitro transcription and translation reactions, fragments of IN were amplified by PCR from pJK788 (J. Kirchner, unpublished work) using primers 762 and 763 (A), 762 and 764 (N), and 762 and 765 (AB). Each PCR product was cloned into the pCRII-Topo vector (Invitrogen). Plasmids containing IN fragments downstream of the T7 promoter were identified by restriction and sequence analyses.
Shuttle mutagenesis was performed as previously described (6). Briefly, DNA was prepared from library pools separately mutagenized with mini-Tn3::lacZ/LEU2 (kindly provided by M. Snyder, Yale University), cleaved with NotI, and transformed into strain yMA1322 carrying pTM45 and pPK689 (pCH2bo19V) by the lithium acetate procedure (21). Genomic loci disrupted by mTn3 insertion in each mutant of interest were amplified by vectorette PCR (http://genome-www.stanford.edu/group/botlab/protocols/vectorette.html) and identified by sequence analysis. Briefly, total yeast DNA prepared from the mutant strain was cleaved with RsaI and ligated with preannealed anchor bubble primers (primer 637 and 638). One-tenth of the reaction mixture was used as the input for amplification of the genomic disruption using primers 639 and 642. PCR mixtures were prepared as previously described (31). Following an initial incubation at 95°C for 2.5 min, 35 cycles of PCR consisting of 20 s at 92°C, 30 s at 67°C, and 2.5 min at 72°C were performed. PCR products were separated by electrophoresis, and each product was excised. DNA was extracted by using the QiaexII kit (Qiagen) and sequenced by using primer 642 and a Thermosequenase kit (Amersham). Each sequence generated was compared to complete S. cerevisiae genomic DNA using the Blastn search of the Saccharomyces Genome Database (http://genome-www.stanford.edu /Saccharomyces/).
A target-specific genetic assay (25) was slightly modified to screen mutants for the Ty3 transposition phenotype. Leu+ mutant transformants or wild-type strain yMA1342, carrying pTM45 and pPK689, were patched onto SC-His-Trp-Leu (synthetic dropout medium with glucose). After incubation for 24 h at 30°C, each plate was replica plated to SC(Gal)-His-Trp-Leu (galactose as the carbon source) for induction of Ty3 expression. After 48 h of growth at 30°C on this medium, yeast cells expressing Ty3 were replica plated to minimal medium (with glucose and supplemented with uracil) for detection of retrotransposition events. The latter plates were incubated at 30°C for 5 days, and the number of papillae within each patch was compared. The cells on SC-His-Trp-Leu plates after 1 day at 30°C were replica plated to minimal plates with glucose and uracil as the negative control.
Whole-cell extracts (WCEs) were prepared from 10-ml cultures as described previously (31). Then 20-μg portions of WCEs were fractionated by sodium dodecyl sulfate-polyacrylamide gel electrophoresis (SDS-PAGE), transferred to nitrocellulose membranes (Hybond ECL; Amersham), and incubated with rabbit polyclonal antibodies to CA and IN. Secondary antibodies to rabbit immunoglobulin G (IgG) were detected by the ECL system.
RNA-free total yeast DNA (1 μg) was digested with EcoRI, separated by electrophoresis, transferred to a nylon membrane (Duralon UV; Stratagene), and immobilized by UV cross-linking in a Stratalinker 1800 (Stratagene). Hybridization was performed with 32P-labeled internal BglII fragment of Ty3.
Yeast cells grown for 6 h with 2% raffinose or 2% galactose were harvested. Total RNA was extracted, separated in an 8% polyacrylamide–8.3 M urea gel by electrophoresis, transferred to a GeneScreen Plus membrane (Stratagene), and probed with 32P-labeled oligonucleotide specific for mature tRNA. Primer extension analysis was performed as previously described (26).
Yeast transcription extracts were prepared from 12 liters of stationary-phase cultures as previously described (23). Briefly, cell pellets were washed, resuspended, and lysed with glass beads in a bead-beater chamber. Centrifugation of lysates for 1 h at 100,000 × g yielded S100 supernatant. S100 extract was fractionated by (NH4)2SO4 precipitation. The pellet precipitated by 35 to 70% ammonium sulfate was resuspended and loaded on to a BioRex70 ion-exchange column. Elution with 500 mM NaCl from this column yielded BR500 extracts.
In vitro transcriptions were performed with 30 μg of BR500 as previously described (23). Each reaction mixture contained 110 mM NaCl, 8 mM MgCl2, 20 mM HEPES (pH 7.8), 250 μM each ATP, CTP, and UTP, 15 μM GTP, 10 μCi of [α-32P]GTP (16.7 Ci/mmol), 200 ng of template tRNA gene containing plasmid, and 800 ng of pIBI20 as nonspecific DNA, in a total volume of 40 μl. The reaction mixtures were incubated for 30 min at 30°C, and the reactions were terminated by the addition of 160 μl of stop solution (27 mM EDTA, 0.27% SDS, 0.33 μg of salmon sperm DNA per μl, 1.33 M LiCl). The transcription products were separated by electrophoresis on an 8% polyacrylamide gel and visualized by autoradiography.
Yeast two-hybrid filter assays were performed as previously described (37). Two-hybrid constructs were transformed into the SF526 strain, and the filter assay for β-galactosidase activity was conducted on at least three different transformants.
GST pulldown assays were performed as previously described (37). Expression of fusion proteins was induced by isopropyl-β-d-thiogalactopyranoside (IPTG) at a final concentration of 500 μM for 3 h at 37°C in E. coli HB101, and fusion proteins were purified by batch binding to glutathione-Sepharose beads. Ty3 IN or its domains were labeled with [35S]methionine in a coupled transcription-translation system (Promega) in vitro. Labeled protein was incubated with GST fusion proteins or GST alone bound to the beads equilibriated with 0.3% bovine serum albumin fraction V (Sigma) in buffer C (20 mM HEPES [pH 7.8], 100 mM NaCl, 20% glycerol, 1 mM EDTA, 1 mM dithiothreitol). After incubation for 30 min at room temperature with gentle agitation, the beads were washed three times with buffer C. The proteins retained on the beads were eluted by boiling in sample buffer and were separated by SDS-PAGE (10% polyacrylamide). Each gel was fixed in 40% methanol and 10% acetic acid, soaked in Amplify (Amersham), dried, and exposed to film.
To recover plasmids with Ty3 insertions, pEH2b19V target plasmid together with pTM45 (helper) and pDLC348 (donor) plasmids were transformed into rad52Δ versions of the wild-type (yMA1356) and tfc1 (yMA1357) strains. The transposition assay was performed as previously described (9). Transposition was induced by growing transformants for 4 to 6 days at 30°C on SC-Trp-Ura-His, containing galactose as the carbon source. These cells were patched to yeast extract-peptone-dextrose (YPD) containing 700 μg of G418 per liter to allow loss of Ty3 plasmids and to enrich for plasmids with Ty3 insertions. G418-resistant cells that had lost the URA3-marked donor plasmid were selected on medium containing 5-fluoroorotic acid, and colonies that retained the target plasmid were identified on SC-His medium containing G418. DNA was isolated from these cells, and plasmids containing Ty3-N insertions were recovered by transformation of E. coli and selecting for kanamycin and ampicillin resistance. Plasmid DNA was isolated from a single E. coli transformant per galactose-induced colony to ensure that the Ty3-N insertions were independent. The position and the orientation of Ty3-N insertions were determined by sequence analysis.
Insertional mutagenesis (6) coupled with a genetic assay (25) was used to identify host factors that affect retrotransposition of Ty3 (Fig. (Fig.1).1). In the assay, Ty3 transcription is induced by growth on galactose and Ty3 integration is detected using a plasmid-borne tDNA target. The target tRNAVal gene is positioned so that it interferes with expression of a neighboring, divergent ochre suppressor tRNATyr gene, sup2bo. In addition, the latter is inactivated by a tract of pyrimidines on the nontranscribed strand in the transcription initiation region. Ty3 position-specific integration into this target both alleviates the interference between the divergent genes and changes the sequence composition upstream of the suppressor, thereby activating its expression. Haploid yeast mutants were generated as described in Materials and Methods. Leu+ mutants and wild-type strain yMA1342, carrying pTM45 with a galactose-inducible Ty3 element and pPK689 with the divergent tDNA target, were patched onto SC-His-Trp-Leu plates and replica plated to SC(Gal)-His-Trp-Leu for induction of Ty3 expression. These patches were replica plated to minimal medium supplemented with uracil. Cells that had undergone transposition and activated the suppressor expression grew in the absence of adenine and lysine and were identified as papillae within each patch. The relative frequency of retrotransposition was determined by comparison between the number of papillae arising from mutant and wild-type patches (Fig. (Fig.1).1). A total of 27,000 mutants containing mTn3 insertions generated from various pools of the insertion library were screened for the Ty3 phenotype. Details of the screen with a complete list of the genes isolated will be described in a separate paper.
One mutant, 25-41A, exhibited an 11-fold decrease in transposition (data not shown) by a quantitative version of the divergent tRNA gene target assay. To identify the host mutation responsible for the Ty3 phenotype, vectorette PCR was used as described in Materials and Methods to amplify part of the mTn3 insertion and flanking genomic DNA. The sequence of the resulting PCR product was determined and compared to the Saccharomyces Genome Database. This search identified TFC1, which encodes an essential 95-kDa subunit of TFIIIC (TFIIIC95). This subunit has a helix-turn-helix motif and an acidic C-terminal domain (38). TFIIIC95 binds to tDNA (15) and is centrally located within TFIIIC over the box A promoter element (5). Sequence analysis indicated that mTn3 insertion at nucleotide position 1571 of TFC1 coding region would lead to expression of a truncated protein with the N-terminal 526 amino acids instead of the 649-residue full-length protein (Fig. (Fig.2A).2A).
To test whether the mTn3 insertion into the TFC1 gene was solely responsible for reduced Ty3 retrotransposition phenotype, the TFC1::mTn allele was recovered from the 25-41A mutant strain by the plasmid gap repair method (16) using a gapped plasmid-borne TFC1 gene. The mutant allele was then used to disrupt TFC1 in the wild-type strain by homologous recombination. Southern analysis with a TFC1-specific probe confirmed that this gene was disrupted in the reengineered strain as in the original mutant (data not shown). The resulting strain showed the same severely reduced transposition phenotype as the 25-41A mutant (Fig. (Fig.2B).2B). Next, the effect of the wild-type TFC1 gene on the transposition frequency in the mutant strain was tested. The wild-type and mutant strains were transformed with either a control plasmid, pRS316, or a low-copy plasmid carrying TFC1, pTFC1, and retrotransposition was tested for multiple transformants. The mutant transformants with pRS316 showed severely reduced retrotransposition, but those with pTFC1 exhibited wild-type level of retrotransposition (Fig. (Fig.2C).2C). These results showed that the reduced-transposition phenotype was caused by truncation of TFC1 gene and that the mutant allele was recessive since the mutant phenotype was rescued by a plasmid-borne wild-type gene.
Although TFC1 is essential (38), the mutant with a truncated allele is viable. To test if this mutation causes a growth defect and thus might indirectly affect the transposition frequency, serial dilutions of wild-type and mutant cultures in mid-log phase were spotted onto YPD medium and incubated at 30°C for 2 days or at 37°C for 3 days. The mutant strain grew as well as the wild-type strain at both temperatures (Fig. (Fig.3A).3A). Growth curves of liquid cultures also showed no significant difference in growth rate between the two strains (data not shown). These results indicated that even at 37°C, the mutant strain has no growth-limiting defect in pol III transcription.
To test more specifically whether pol III transcripts were limiting in the mutant strain, cells carrying a plasmid with the SUP2b tRNA gene (pDLC356) (9) were grown with raffinose or galactose as the carbon source and total RNA was prepared. Northern blot analysis (Fig. (Fig.3B)3B) showed no significant difference in mature tRNAMet levels between samples from the mutant and the wild type. To determine the accuracy of pol III transcription initiation in the mutant, reverse primer extension analysis was performed on the same RNA samples with an oligonucleotide primer specific for the SUP2b pre-tRNA intron (Fig. (Fig.3C).3C). Comparison of results from mutant and wild-type samples showed identical sizes and amounts of the primer extension products, indicating that neither the transcription initiation site nor the amount of pre-tRNA was dramatically affected in the mutant strain. In addition, no dramatic difference in transcription activity was detected by in vitro transcription experiments using pol III transcription extracts prepared from wild-type and mutant strains (data not shown). To determine if Ty3 insertions into the divergent target could activate the expression of the suppressor tRNA gene, a target plasmid containing a Ty3 insertion was transformed into the tfc1 strain. Transformants grew readily on media requiring expression of the sup2bo gene. These results showed that the decrease in Ty3 retrotransposition was not due to a dramatic decrease in pol III transcription activity.
To find how truncation of TFC1 affects Ty3 transposition, the amounts of Ty3 CA, IN, and DNA intermediates in wild-type and mutant cells were compared. For these assays, wild-type and mutant strains carrying pTM45 and either pRS316 or pTFC1 were grown to early log-phase (absorbance at 600 nm [A600], 0.2 to 0.4) in synthetic medium with raffinose as the carbon source. Expression of Ty3 was induced by addition of galactose to a final concentration of 2%. After 6 h, the cells were harvested and proteins or DNA was isolated. Immunoblot analysis with anti-Ty3 CA and anti-Ty3 IN antibodies showed no significant difference in the amounts or processing of CA or IN protein between the wild-type or the mutant extracts with and without pTFC1 (Fig. (Fig.4A,4A, lanes 2 to 5). Results of immunoblot analysis of VLPs prepared from these strains were consistent with these results (data not shown).
After particle assembly and protein maturation, Ty3 DNA is reverse transcribed from the genomic RNA template. Total DNA was extracted from wild-type and mutant cultures, and Southern blot analysis was performed using a radiolabeled Ty3-specific probe. Quantitative analysis of the blot using PhosphorImager and ImageQuant software (Molecular Dynamics) showed that similar amounts of full-length cDNA were present in both wild-type and mutant strains irrespective of the presence of pTFC1 (Fig. (Fig.44B).
The previous finding that TFIIIC is essential for in vitro integration of Ty3 at a tRNA gene (27), coupled with the results described above, suggested that either the efficiency or the position of Ty3 integration was defective in the tfc1 mutant. To address the possibility that Ty3 integration was altered in the mutant strain so that integration events failed to occur in positions that activated sup2bo, Ty3 insertions in vivo were selected independent of the sup2bo expression. To facilitate the recovery of insertions into target plasmids, the ochre anticodon of sup2bo was changed to the wild-type anticodon (sup2b) and the target plasmid was converted into a high-copy plasmid. Wild-type and tfc1 mutant strains were transformed with pTM45; the modified target, pEH2b19V; and pDLC348, carrying a galactose-inducible, Neo-marked Ty3 (Ty3-N). Ty3-N insertions into the target plasmid were selected in yeast and plasmid DNA was recovered in bacteria as described in Materials and Methods. The Ty3 insertion sites in 15 plasmids recovered from the wild-type strain and 14 recovered from the tfc1 strain were determined by sequencing. All the Ty3-N insertions from the wild-type strain and 13 of 14 insertions from mutant strains were within the 5-bp window expected to activate sup2bo expression (Fig. (Fig.5A).5A). A total of 12 of the 14 insertions recovered from the mutant and 10 of the 15 insertions from the wild type occurred at position −19. These data confirm the previous finding that tDNAVal is the target for the vast majority of insertions, although the spacing of the two genes results in Ty3 insertions occurring much closer to the 5′ end of sup2b. In addition, the characteristic 5-bp duplication of flanking sequence was observed for all 10 insertions for which the sequence at both Ty3-plasmid junctions was determined. Thus, the tfc1 strain showed no significant alteration in integration specificity.
Inspection of Ty3-N insertions in tfc1 as a set showed that they were distinct from insertions in the wild-type strain in their orientation. Whereas 9 of 15 insertions in the wild-type strain occurred so that the transcriptional orientation of Ty3-N was opposing that of the tRNAVal target gene, all 14 insertions in the tfc1 mutant were in that orientation (Fig. (Fig.5A).5A). A chi-square test of the mutant indicates that this distribution is not random (P < 0.005). Lack of recovery of Ty3-N in the same transcriptional orientation as the tDNAVal was not due to inability to select for Ty3-N in that orientation in the mutant background, since plasmids with Ty3-N insertions recovered from wild-type cells and transformed into mutant cells readily conferred resistance to G418 (data not shown). These data indicated that the orientation of Ty3 integration into the target plasmid was biased in the mutant strain.
The 11-fold reduction in the recovery of Ty3 insertions into the divergent target in the mutant strain was significantly greater than the 40 to 50% reduction predicted for loss of insertions in the same orientation as the target tDNA. We first tested whether Ty3 insertions in the two orientations affect sup2bo expression differently and thus contribute to the bias in detection of Ty3 integration. BR500 extracts from the wild-type strain were used to transcribe the tRNA genes on target plasmids with Ty3 integrated at different positions and in each orientation (Fig. (Fig.5B).5B). Plasmid templates carrying either SUP2b (Fig. (Fig.5B,5B, lane 2) or tDNAVal (lane 3) alone were included to identify the various RNA species generated. As expected, SUP2bo pre-tRNA was transcribed inefficiently from the divergent tRNA gene target plasmid compared to SUP2b alone (lane 4 compared to lane 2). Target plasmids with Ty3 integrated, at positions −19, −18, −17, and −7 relative to the 5′ end of the mature tRNA, in the same transcriptional orientation as the tDNAVal yielded slightly more unprocessed and processed RNA species than did the target plasmid alone (compare lanes 6, 7, 8, 10, and 4). Surprisingly, target plasmids with Ty3 at positions −19 and −15 and in the opposite orientation generated lower levels of pre-tRNAs and processed species than did the the target alone (compare lanes 5, 9, and 4). Thus, only Ty3 insertions in the same orientation as tDNAVal activated transcription. Similar results were obtained using extracts from the mutant strain (data not shown). If Ty3 insertion in vivo affected sup2bo expression similarly to what was observed in vitro, insertions in one orientation might not have been detected in the mutant or wild-type strains by the suppressor activation assay. To test this hypothesis, cells with independent Ty3 insertions into pPK689 in the wild-type strain were selected as described for Fig. Fig.1.1. Of 30 insertions, 27 were found to be in the same orientation as the target tDNAVal by Southern blot analysis (data not shown). This suggested that in the wild-type background, where insertions of Ty3-N in both orientations were observed, Ty3 insertion activated sup2bo only when inserted in the same transcriptional orientation as tDNAVal. Thus, although insertions probably occurred in the tfc1 strain for Ty3, as they did for Ty3-N, they would not have been in the orientation that activated sup2bo.
The fact that pol III transcription did not appear to be altered in the tfc1 mutant suggested that the orientation bias was not due to an indirect effect on the number of initiation complexes. Rather, it suggested that specific contacts between the Ty3 PIC and the preinitiation complex might be lacking in the mutant. The possibility that TFIIIC95 and Ty3 IN might interact directly was therefore investigated. As shown in Fig. Fig.6B,6B, TFIIIC95 fused to the DNA-binding domain of Gal4 (Gal4 BD; amino acids 1 to 147), when tested with Ty3 IN fused to the activation domain of Gal4 (Gal4 AD; amino acids 768 to 881) gave a robust signal in a two-hybrid assay. This signal was abrogated when the C-terminal region was deleted from TFIIIC95 (TFIIIC95ΔC), suggesting that this region is important for interaction with Ty3 IN. In addition, the C-terminal domain of TFIIIC95 (TFIIIC95C) gave a weak but significant signal when tested against IN. Full-length IN or IN-AB (amino acids 1 to 304) fused to the Gal4 BD did not show interaction with TFIIIC95 fused to Gal4 AD. However, equivalent interactions are not always observed in the two-hybrid assay in each of the two possible expression contexts (4). Testing of individual domains showed that the IN-A domain (amino acids 1 to 61), but not the other domains, interacted with TFIIIC95 (Fig. (Fig.6C).6C).
To further investigate the IN-TFIIIC95 interaction, GST fusions with full-length and truncated TFIIIC95 were expressed in E. coli and purified on glutathione-Sepharose beads. These GST-TFCIII95 fusion proteins were tested for interaction with IN labeled with [35S]methionine produced in a coupled transcription-translation system (Promega) (Fig. (Fig.6D).6D). In accordance with two-hybrid results, full-length IN interacted strongly with GST-TFIIIC95 and truncated TFIIIC retained significantly less 35S-IN (compare the two right lanes). Moreover, in a GST pulldown assay, 35S-labeled TFIIIC95 showed weak interaction with GST–IN-A (data not shown). Because the IN-A contains few methionine residues, it does not incorporate [35S]methionine at a significant level. Therefore, IN-N (amino acids 1 to 150) was tested for interaction with GST-TFIIIC95. This experiment showed that the IN-N domain was sufficient to interact with TFIIIC95, but no differential interaction was observed between IN-N and GST-TFIIIC95 or GST-TFIIIC95ΔC (Fig. (Fig.6E).6E). These data suggested that the N-terminal domain of IN was minimally required to interact with TFIIIC95 but that a larger fragment, possibly full-length IN, was required to distinguish between the full-length TFIIIC95 and TFIIIC95ΔC. The fact that the C terminus of TFIIIC95 was important for optimal protein-protein interaction with IN argued that both TFIIIC95 and IN are involved in mediating Ty3 integration orientation.
For both transcription and transposition in vivo, the box B element, which is bound by TFIIIC, is required by TATA-containing genes and TATA-less genes. However, similar to pol III, in vitro Ty3 requires TFIIIC and TFIIIB for interaction at TATA-less tRNA genes but only TFIIIB at a TATA-containing gene (40). This would seem to suggest that the role of TFIIIC in transcription and transposition is indirect, that is, to load TFIIIB. However, subunits of pol III do interact directly with TFIIIC (13, 19), raising the possibility that TFIIIC contributes directly to recruitment of pol III and potentially Ty3 as well. In this work, we describe the recovery from a large-scale screen of a novel, viable mutant that has a truncated subunit of TFIIIC. In the mutant background, Ty3 transposition is position specific but occurs in only one orientation in the context of a synthetic divergent tRNA gene target. Investigation of the basis of this effect indicated interactions between the nonconserved, N-terminal domain of Ty3 IN and the acidic C-terminal domain of TFIIIC95. While these data did not demonstrate that a direct contact between the Ty3 PIC and TFIIIC is required for specificity, they did provide the first evidence that both TFIIIC and IN could interact directly during Ty3 integration. In addition, they show the potential of the two Ty3 ends for dramatically different integration activity.
Studies of the yeast and human TFIIIC transcription factors have shown that aspects of structure and function are conserved. Both complexes are comprised of multiple subunits: yeast TFIIIC (yTFIIIC) contains six subunits, and human TFIIIC (hTFIIIC) has at least nine polypeptides (39). The human homologue of yTFIIIC95 is a 63-kDa protein (hTFIIIC63) (19) containing helix-loop-helix and C-terminal acidic domains. Yeast TFIIIC95 contacts the internal promoter box A element (5) and interacts with the 55-kDa subunit of TFIIIC (30). Interestingly, hTFIII63 shows physical interactions with hTFIIIC90 and hTFIIIB90 (Brf), as well as with a subunit of pol III (19). In yeast, a point mutation in TFC1 affects TFIIIB complex formation (S. Jourdain et al., unpublished work). This observation also suggests interaction of TFIIIC95 with TFIIIB. Although the C-terminal acidic region is conserved between the human and yeast proteins, its function is not known. Presumably it is not required for DNA binding or pol III interaction under normal growth conditions, since neither the deletion mutant isolated in this screen nor a strain expressing TFIIIC95 with β-galactosidase fused at the C-terminal end showed obvious growth defects (11). This domain is also not essential for interaction of TFIIIC95 with TFIIIC55 (S. Jourdain, unpublished work). However, the genetic and biochemical experiments reported here suggest that the TFIIIC95 C-terminal region could be a protein-protein interaction domain.
Two-hybrid and GST pulldown assays indicated that the Ty3 IN amino-terminal domain is likely to mediate at least some of the contact between the PIC and TFIIIC. Retroelement IN proteins can be divided into amino-terminal, core, and carboxyl-terminal domains. The N-terminal domain contains a Zn2+-binding HHCC motif but is otherwise not well conserved (2). This domain is required for multimerization of IN (41) and strand transfer but not for disintegration, the reverse of strand transfer, in vitro (7). The Ty3 IN amino-terminal sequence is almost 100 amino acids longer than that of human immunodeficiency virus IN and also contains a zinc-binding motif (17). The present study showed that the first 61 amino acids of Ty3 IN can interact weakly with TFIIIC95 and that the interaction is enhanced by the presence of the HHCC domain (data not shown and Fig. Fig.6E).6E). The amino-terminal domain also contains several patches of charged residues. Ty3 IN mutants with substitutions of alanine for basic amino acids at positions 53 and 54 and with substitutions of alanine for basic amino acids from positions 62 and 63 fall to transpose but are only slightly reduced for replicated cDNA (33), suggesting a defect at a late step in the life cycle. It will be of interest to determine whether these mutants display orientation bias in integration.
Based on previous in vivo results showing that the TFIIIC binding site is required for Ty3 integration at TATA-containing and TATA-less targets (9) and on in vitro results with TATA-containing genes showing that TFIIIB is sufficient for Ty3 integration, a model was proposed that TFIIIC was required for integration as the TFIIIB loading factor but was not directly involved in contacts with the PIC. Results from the present in vivo study have contributed to significant revision and extension of this model for Ty3 position specific integration. This revised model (Fig. (Fig.7)7) has four features. (i) TFIIIB is a major determinant of Ty3 targeting and can target integration in either orientation (Fig. (Fig.7A).7A). This feature is based on the observation that TFIIIB is sufficient to target in vitro integration in both orientations at the TATA-containing SNR6 gene. No integration is observed for TFIIIC alone (40). (ii) The Ty3 PIC is asymmetric. The existence of a Ty3 tDNA target makes it possible to define orientation for Ty3 insertions, and because Ty3 inserts with dramatic orientation bias relative to the tDNAVal target gene, we know that the Ty3 PIC itself cannot be symmetric. A completely symmetric Ty3 PIC could not display insertion bias at any insertion site. Although the synthetic, divergent tDNA target showed the asymmetric behavior of the Ty3 PIC in a particularly dramatic pattern, previous observations are also consistent with asymmetric behavior of the Ty3 PIC at genomic targets. In a study of 91 independent insertions into the genome, examination of seven pairs of independent integrations, each at the same genomic tRNA gene, showed an overall distribution of orientations similar to the distribution of orientations of preexisting genomic insertions, but each individual pair of insertions at a particular target occurred in the same orientation. The probability that both pair members at each site would be in the same orientation is quite low (8). This result would be consistent with differential interaction of at least one end of the cDNA in the PIC with the target complex but also suggests that individual tRNA genes might differ with respect to features that affect orientation, such as strand sequence or TFIIIC occupancy. Figure Figure77 shows how preferential interactions between TFIIIB and U5, coupled with interactions between TFIIIC and U5 but not U3, could lead to insertions in either orientation at a genomic tDNA (Fig. (Fig.7A)7A) or at the divergent target in the wild-type background (Fig. (Fig.7B)7B) but might lead to biased integration at the divergent target in the tfc1 mutant background (Fig. (Fig.7C).7C). (A detailed explanation of the diagram is given in the legend.) Although preferential interactions between the factors and the ends of the Ty3 PIC would be consistent with our observations, the specific interactions shown in Fig. Fig.77 are for illustrative purposes; there are no data that demonstrate particular preferential interactions between U3 and U5 or specific protein domains exposed at these ends and TFIIIB or TFIIIC. (iii) TFIIIC is probably associated with at least some targets during integration. It was shown in the present study that TFIIIC95 interacts with IN. The distribution of Ty3-N insertions for the divergent target is broader in the wild type than in the tfc1 strains, indicating that an interaction between IN and TFIIIC95 could influence integration site selection in vivo. Interestingly, the pattern of integrations observed in this study in the presence and absence of the C-terminal domain of TFIIIC95 is similar to the pattern of in vitro integration at a SNR6 target in the presence and absence, respectively, of TFIIIC (L. Yieh et al., unpublished data). (iv) Contact between Ty3 IN and the TFIIIC95 C-terminal domain, while not essential, either facilitates integration in the same transcriptional orientation as the target tDNA or impedes it in the opposite orientation (Fig. (Fig.7,7, band C). The synthetic target is neither transcribed nor used as a transposition target as efficiently as a wild-type tRNA gene (25). Among other possibilities, this could result from attenuated TFIIIB function caused by steric interference with proper TFIIIB binding upstream of the target tRNAVal gene by TFIIIC or TFIIIB bound to the divergent sup2bo gene. With loss of the C-terminal domain of TFIIIC95, Ty3 insertions in the same transcriptional orientation as tDNAVal in the divergent target did not occur; however, insertions were still observed in this orientation at chromosomal tRNA gene targets (M. Aye et al., unpublished data). These observations argue that some feature of the divergent tDNA target, such as attenuated TFIIIB function, causes increased dependence on TFIIIC (Fig. (Fig.7B)7B) for Ty3 insertions in the same orientation as the target tDNAVal and that this dependence is associated with the TFIIIC95 C-terminal domain. Ty3 IN mutations that disrupt the interaction between IN and TFIIIC95 could be used to directly test this aspect of the model since they would be predicted to bias insertion in a similar way to the TFIIIC95 truncation.
One novel aspect of the Ty3 integration model introduced here is that the ends of the element, in spite of containing perfect inverted repeats for IN recognition, clearly have different integration activities and are likely to reflect this in their interactions with TFIIIB and TFIIIC. The key observation of orientation bias was possible only because a large set of Ty3 integrations can be observed at a specific target, a situation which does not occur for retroviruses. Orientation specificity has also been observed in Tn7, a bacterial transposon with a defined target, (3). While difficult to elucidate in retroviruses, orientation has clear implications for the effect of proviral enhancers and promoters on flanking genes (18) and is a relatively unexplored aspect of the retrovirus-host interaction.
Although Ty3 is similar to retroviruses in organization and proteins encoded, it has a high degree of position specificity. Ty1 to Ty4 are each associated with tRNA genes; however, this is the first report of direct contact between a member of the PIC and a pol III transcription factor. The contribution of this contact for integration may vary for each target, and our ability to use a targeted insertion assay was crucial for distinguishing different effects of the tfc1 mutation at different loci. These studies suggest that retroelements with targeting properties might be also be useful as reporters of protein occupancy at the loci where they insert. In cases where members of a gene family may be differentially associated with protein factors, these elements could offer insights into whether the distribution into different states reflects some kinetic parameter of the population as a whole or the differential behavior of some subset of genes or cells where transposition occurred.
We thank M. Snyder for providing the mTn3::lacZ/LEU2 library and C. Friddle for providing the vectorette PCR protocol. We also thank E. Chen for assistance with the mutant screen, M. H. Nymark-McMahon and L. Yieh for assistance with VLP and BR500 preparations, J. Steffan for assistance with GST pulldown assays, T. Menees for helpful discussions, and A. Sentenac for critical reading of the manuscript.
This work was supported by Public Health Service grant GM33281 to S.B.S. and by the Synthesis and Structure of Biological Macromolecules training grant GM07311-24 (M.A.).