|Home | About | Journals | Submit | Contact Us | Français|
The long terminal repeat (LTR) retrotransposon Tf1 of Schizosaccharomyces pombe integrates specifically into the promoters of pol II-transcribed genes. Its integrase (IN) contains a C-terminal chromodomain related to the chromodomains that bind to the N-terminal tail of histone H3. Although we have been unable to detect an interaction between histone tails and the chromodomain of Tf1 IN, it is possible that the chromodomain plays a role in directing IN to its target sites. To test this idea, we generated transposons with single amino acid substitutions in highly conserved residues of the chromodomain and created a chromodomain-deleted mutant. The mutations, V1290A, Y1292A, W1305A, and CHDΔ, substantially reduced transposition activity in vivo. Blotting assays showed that there was little or no reduction in the levels of IN or cDNA. By measuring the homologous recombination between cDNA and the plasmid copy of Tf1, we found that two of the mutations did not reduce the import of cDNA into the nucleus, while another caused a 33% reduction. Chromatin immunoprecipitation assays revealed that CHDΔ caused an approximately threefold reduction in the binding of IN to the downstream LTR of the cDNA. These data indicate that the chromodomain contributed directly to integration. We therefore tested whether the chromodomain contributed to selecting insertion sites. Results of a target plasmid assay showed that the deletion of the chromodomain resulted in a drastic reduction in the preference for pol II promoters. Collectively, these data indicate that the chromodomain promotes binding of cDNA and plays a key role in efficient targeting.
Long terminal repeat (LTR) retrotransposons are close relatives of retroviruses that propagate by inserting their DNA into the genome of the host. Both retroviruses and LTR retrotransposons encode an integrase (IN) protein that mediates this process. A wealth of information is available about the function of IN from in vitro studies conducted with recombinant proteins (9); however, much less is known about how each domain of IN functions in vivo.
IN proteins of retroviruses and LTR retrotransposons contain three distinct protease-resistant domains. The N-terminal domain has the conserved amino acid motif HX3-7HX23-32CX2C, which binds Zn. The central domain is the catalytic core and contains the conserved DDE motif. The C-terminal domain is the least well conserved and has nonspecific DNA binding activity in vitro (9). Detailed analysis of the C-terminal domains identified two conserved modules present individually or together in a wide variety of INs (33). The GPY/F domain is found in the INs of gamma retroviruses and a broad set of LTR retrotransposons of the Metaviridae family (formerly called the Ty3/Gypsy family) (23, 33). The other conserved module is a chromodomain (CHD) and is found at the C-terminal ends of INs from the Metavirus genus of LTR retrotransposons (26, 33). CHDs are present in a variety of eukaryotic proteins and act as interaction modules for methylation marks such as histone H3 methylated at lysine 9 (11). The presence of CHDs in retrotransposon INs suggests they play a role in the selection of integration sites (33). This possibility is supported by the recent finding that the CHD of the Maggy retrotransposon interacts with histone H3 methylated at lysine 9 (14).
Tf1 of Schizosaccharomyces pombe is an LTR retrotransposon that belongs to the Metavirus genus. Its IN contains both a GPY/F domain and a C-terminal CHD (20, 33). Tf1 exhibits a marked preference for integrating within a window of 100 to 400 nucleotides upstream of open reading frames (ORFs) (3, 6, 42). Recent work found that the insertion of Tf1 upstream of ORFs is due to its recognition of pol II promoters (28). One model for how Tf1 integrates into promoters is that the CHD interacts with chromatin features specific to pol II promoters. However, the CHD of Tf1 is not as conserved as other CHDs, and as a recombinant protein, it fails to interact with any known form of histone protein (H. Ebina and H. L. Levin, unpublished data). This study was undertaken to test the CHD of Tf1 for functions in transposition. In the context of the intact transposon, alanine substitutions were made in conserved residues of the CHD, and a version of Tf1 with an IN that lacked the CHD was generated. The mutations caused substantial reductions in transposition frequencies with little or no decreases in Tf1 protein or cDNA levels. In addition, a genetic assay indicated that for two of the three mutations tested, the cDNA accumulated in the nucleus at wild-type levels. These results indicate that the CHD specifically contributed to integration efficiency. Chromatin immunoprecipitation (ChIP) assays revealed that the contribution of the CHD to integration is in part due to its ability to promote the binding of IN to cDNA. Importantly, integration assays with target plasmids showed that deletion of the CHD disrupted the targeting of integration to pol II promoters.
The yeast strains and the plasmids used in this study are listed in Table S1 in the supplemental material. The oligonucleotide sequences are listed in Table S2 in the supplemental material. The plasmids with mutations in the CHD were constructed by fusion PCR with two overlapping oligonucleotides with the mutant sequence and two PCR oligonucleotides that contained unique restriction sites located on either end of the final PCR product. The CHD point mutations were generated on a 1,183-bp fragment that was cloned into pHL414-2 and pHL449-1 at unique BsrGI and NarI sites.
Following transformation into S. pombe, various versions of Tf1 were assayed for transposition and recombination as described previously (1). Briefly, all the mutants of Tf1 elements were first grown as patches on agar plates containing EMM (13) supplemented with 2 g/liter dropout mix (an equal-weight mix of all amino acids plus 2.5 times more adenine than the amino acids; no uracil is present) and a final concentration of 10 μM thiamine to repress the transcription of Tf1 driven by the nmt1 promoter. After 2 days of incubation at 32°C, patches of cells were replica printed to similar EMM agar plates that lacked thiamine to induce the transcription of Tf1. After growing for 4 days at 32°C, induced patches were replica printed to EMM agar containing 1 g/liter 5-fluoroorotic acid (5-FOA), the dropout mix, and 50 μg/ml of uracil to eliminate the plasmid with Tf1-neo. Transposition was measured by a final replica print to YES (yeast extract plus dropout mix) plates supplemented with 1 g/liter 5-FOA and 0.5 g/liter G418. These plates were examined after 36 h of growth at 32°C to observe integration of Tf1-neo.
Quantitative measurements of transposition frequencies were performed as described for a different transposon (12). In brief, strains were grown on EMM agar plates (with the dropout recipe described above) that lacked thiamine. These cells were then resuspended at an optical density at 600 nm (OD600) of 5.0 and used to make a series of 10-fold dilutions starting with 108 cells/ml and finishing with 104 cells/ml. One hundred microliters of the cells from the three lowest dilutions was then spread onto FOA plates, and 0.1 ml of cells from the three highest dilutions was spread onto the YES plates containing FOA and G418. The transposition frequency, i.e., the percentage of the FOAr colonies that were also G418r, was then calculated.
Homologous recombination between the cDNA of Tf1 and the Tf1-neoAI sequence in the expression plasmid (pHL449-1) was detected by printing induced patches from the EMM lacking thiamine directly to plates with YES-G418 (1). Quantitative measurements of homologous recombination were performed using an approach similar to that for the quantitative transposition assay. Strains were grown on EMM agar plates (with the dropout recipe described above) that lacked thiamine. These cells were then resuspended at an OD600 of 5.0 and used to make a series of 10-fold dilutions starting with 108 cells/ml and finishing with 104 cells/ml. One hundred microliters of the cells from the three lowest dilutions were spread onto YES plates, and 0.1 ml of cells from the three highest dilutions were spread onto the YES plates containing G418. The homologous recombination frequency, i.e., the percentage of the YES colonies that were also G418r, was then calculated.
Proteins were extracted from cultures grown under induction conditions according to previously established protocols (1). Equal amounts of total protein were run on sodium dodecyl sulfate (SDS)-10% polyacrylamide gels and transferred to Immobilon-P membranes (Millipore, Bedford, MA). The ECL Plus detection system (Amersham Biosciences) was used in conjunction with polyclonal antisera at a 1:10,000 dilution for the detection of IN (production bleed of rabbit 657), Gag (production bleed of rabbit 660) (31), and reverse transcriptase (RT) (production bleed of rabbit R-5514, a generous gift of Amnon Hizi).
Fifty milliliters each of YHL9510-3, YHL9511-2, YHL9512-1, YHL9514-1, YHL9515-3, and YHL9516-2 was harvested at an OD600 of 1.0 and extracted with a bead beater (Mini Beadbeater; Biospec Products) in 200 μl of extraction buffer (15 mM KCl, 10 mM HEPES-KOH [pH 7.8], 5 mM EDTA, 5 mM dithiothreitol, 2 mM phenylmethylsulfonyl fluoride, 0.7 μg/ml pepstatin, 0.5 μg/ml leupeptin, 1 μg/ml aprotinin, and one protease inhibitor cocktail tablet/50 ml [Roche Complete protease inhibitor cocktail tablet, EDTA free]) and 100 μl of glass beads. Six hundred micrograms of extracted proteins was immunoprecipitated with 30 μl of protein A-agarose beads (Invitrogen) prewashed and coupled with anti-IN antibody (polyclonal anti-IN, exsanguination bleed HL5851; Covance). After incubation for 25 min at 4°C, the beads were washed twice with 800 μl of extraction buffer. The protein bound to the beads was eluted in 35 μl of 2× sample buffer with a 5-min incubation at 95°C. Samples were run on an SDS-10% polyacrylamide gel and transferred to Immobilon-P membranes. Immunoblotting was performed using anti-IN antibody (production bleed of rabbit 657) as the primary antibody at a 1:5,000 dilution and horseradish peroxidase-conjugated light-chain-specific monoclonal mouse anti-rabbit immunoglobulin G (Jackson ImmunoResearch Laboratories, Inc.) as the secondary antibody at a 1:10,000 dilution. Bands were detected using chemiluminescence (ECL Plus; Amersham Biosciences).
cDNA preparations were performed as described previously (1). In brief, cells were grown to stationary phase in the absence of thiamine, and the total DNA extracted from 109 cells (100 OD600 units) was resuspended in 100 μl of Tris-EDTA (TE). Two micrograms of each DNA sample (about 5 μl) was digested with BstXI, and the products were separated on 1.0% Tris-borate-EDTA agarose gels. The DNA fragments were transferred to Genescreen Plus (Perkin-Elmer Life Sciences Inc.) overnight by capillary transfer, and the filters were hybridized with a 1.0-kb neo probe derived from a BamHI digest of pHL765 (pBS2 containing a BamHI fragment of neo).
Yeast cultures (100 ml) were inoculated at an OD600 of 0.05 and were grown to an OD600 of 1.0 in EMM (dropout mix minus uracil) without thiamine and cross-linked with 1% formaldehyde (Mallinckrodt) at room temperature for 30 min. The reaction was quenched with glycine at a final concentration of 300 mM for 5 min. Cells were then washed twice with chilled 1× phosphate-buffered saline. Cell pellets were resuspended in 400 μl of lysis buffer (50 mM HEPES-KOH [pH 7.8], 140 mM NaCl, 1 mM EDTA, 1% Triton X-100, 1% deoxycholate, 1 mM phenylmethylsulfonyl fluoride, 0.7 μg/ml pepstatin, 0.5 μg/ml leupeptin, 1 μg/ml aprotinin, and one protease inhibitor cocktail tablet/50 ml [Roche Complete protease inhibitor cocktail tablet, EDTA free]). After the addition of acid-washed glass beads (Sigma, G-8772) the cells were lysed by three cycles of 1 min of bead beating (Mini Beadbeater; Biospec Products) with 1 min of resting in between. The glass beads were washed with 450 μl of lysis buffer, which increased the lysate volume to 850 μl. Lysates were sonicated eight times for 30 s at 60% amplitude (Branson Sonifier 450) and centrifuged at 13,000 × g for 30 min at 4°C. Fifty microliters of clarified lysate was used as the whole-cell extract (WCE) DNA as well as for each IP reaction.
Fifty microliters of Dynabeads coated with sheep anti-rabbit IgG (Dynal Biotech) was incubated with 50 μl of antibody (polyclonal anti-IN, exsanguination bleed HL5851 [Covance], raised against full-length IN lacking the His6 tag that was purified as recombinant protein from Escherichia coli) at 4°C overnight and further incubated with 50 μl of clarified lysate at 4°C for 5 h. Beads were washed twice with lysis buffer, followed by two washes with lysis buffer containing 0.5 M NaCl, followed by two washes with wash buffer (10 mM Tris-HCl [pH 8.0], 0.25 M LiCl, 0.5% NP-40, 0.5% deoxycholate, 1 mM EDTA) and finally one wash with TE (10 mM Tris-HCl [pH 8.0], 1 mM EDTA). All washes were done at 4°C for 5 min each. Beads were eluted with 100 μl of 50 mM Tris-HCl (pH 8.0), 10 mM EDTA (pH 8.0), and 1% SDS at 65°C for 15 min. The eluate was transferred to a fresh tube and pooled with a final bead wash of 150 μl TE with 0.67% SDS at 65°C for 10 min. For input DNA, 200 μl of TE with 0.67% SDS was added to 50 μl of lysate. To reverse the cross-links, all samples were incubated at 65°C overnight and then combined with 140 μl TE and 100 μg of proteinase K and incubated at 55°C for 4 h. Samples were extracted twice with phenol-chloroform-isoamyl alcohol and once with chloroform. DNA was ethanol precipitated with 20 μg glycogen (Sigma), washed with 70% ethanol, resuspended in 20 μl of TE with 1 μg of RNase A, and incubated at 37°C for 1 h. PCR analyses (initial denaturation of 94°C for 5 min; 27 cycles of 94°C for 30 s, 55°C for 45 s, and 72°C for 1 min; and a final step of 72°C for 7 min) were done in 25-μl reaction volumes with 2 μl of template DNA (undiluted for IP and 1:1,000 dilution of WCE) and 3 μCi of [α-33P]dATP (specific activity, 3,000 Ci/mmol). Primers were designed to amplify products of 521 bp for cDNA and 250 bp for fbp1. The cDNA primer HL1576 bridges the artificial intron (AI) in neo (pHL449-1 contains Tf1-neoAI) and can be extended only if annealed to neo that lacks the intron. This configuration allows the Tf1 cDNA to be amplified without signal from the Tf1-neoAI in the plasmid. PCR products were separated on 6% Tris-borate-EDTA polyacrylamide gels, and band intensities were quantified using the Phosphoimage Analyzer (Storm 840; Amersham Biosciences) and the ImageQuant (5.2) software. For multiplex PCR, fold enrichment values for each strain were calculated as follows: [cDNA(IP)/fbp1(IP)]/[cDNA(WCE)/fbp1(WCE)]. To test whether the mutations in the CHD reduced the precipitation of IN by the anti-IN antibody, IP assays were conducted with extracts made from the mutants.
The targeting plasmid assay was conducted as previously described (28). Target plasmids were introduced into S. pombe by transformation with lithium acetate followed by selection on EMM dropout medium lacking leucine. These strains were then retransformed with the Tf1 expression plasmid by selecting on EMM lacking leucine and uracil. Strains were arranged on EMM−Leu−Ura+B1 plates in 2-cm2 patches. Following 24 h of incubation at 32°C, these plates were replica printed to EMM−Leu−Ura−B1 to induce transposition. After 4 days of incubation at 32°C, the strains were printed to EMM−leu+FOA+B1 to select against the Tf1 donor plasmid. Following 1 day of incubation at 32°C, the plates were again printed to EMM−leu+FOA+G418. After 2 days of growth at 32°C, the patches were washed in 0.5 ml of distilled water, resuspended in 0.5 ml citrate phosphate sorbitol buffer with Zymolyase 100T (50 mM phosphate, 50 mM citrate, 1.2 M sorbitol, 2 mg/ml Zymolyase 100T [Seikagaku Corporation]), and incubated at 37°C for 1 hour. After the incubation, the cells were resuspended in 100 μl TE. Twelve microliters of 10% SDS was added, and the cells were incubated at 65°C for 5 min; 33 μl of 5 M potassium acetate was then added, and the samples were mixed well and then placed on ice for 30 min. The cells were then centrifuged at 4°C for 10 min at 20,800 × g. The supernatant was phenol extracted and ethanol precipitated. This material was then electroporated into bacteria to obtain target plasmids. Colonies with resistance to kanamycin contained target plasmids with insertions.
The C-terminal end of Tf1 IN has a 72-amino-acid CHD. To test whether this domain plays a critical role in Tf1 integration, we generated alanine substitutions in three of the most conserved residues of the CHD (Fig. (Fig.1).1). V1290, Y1292, and W1305 were chosen because they correspond to residues in heterochromatin protein 1 that make key hydrophobic contacts between beta sheets two and three, which form the backbone that stabilizes the domain (22, 36). In addition, mutations in other CHD proteins in the amino acids that correspond to Y1292 and W1305 have been found to disrupt function (35, 36). We also introduced a frameshift mutation at the beginning of the CHD to produce a version of Tf1 with an IN that lacked the CHD. We observed the transposition activity of the mutant elements with an assay that has been described previously (1, 29).
The transposition assay is based on strains of S. pombe that express Tf1-neo from an inducible nmt1 promoter present on a multicopy expression plasmid. Integration of Tf1-neo causes cells to become resistant to G418. To observe transposition, patches of cells were induced for Tf1-neo expression, grown on medium that selects against the expression plasmid (5-FOA), and transferred to medium containing G418 (Fig. (Fig.2A).2A). Frameshift mutations in Tf1-neo that block the expression of IN (INfs) or of RT and IN (PRfs) greatly reduce transposition (Fig. (Fig.2A).2A). When cells were grown under these conditions, the three alanine substitutions in the CHD and the CHD truncation (CHDΔ) significantly reduced transposition for each of four independent transformants of the expression plasmid (Fig. (Fig.2A,2A, T1 through T4). The residual growth of the CHD mutants on the G418 plates was equivalent to or less than that of Tf1 with the INfs, indicating that the bulk of transposition was eliminated. The remaining growth was due to homologous recombination between cDNA and the endogenous copies of Tf elements (E. Sweeney and H. L. Levin, unpublished).
A quantitative transposition assay was conducted to measure the activities of the mutant transposons (Fig. (Fig.2B).2B). Two independent transformants of each mutant were tested. The results indicated that CHDΔ caused a 14-fold reduction in transposition, whereas the V1290A and Y1292A substitutions caused 9- and 5-fold drops in transposition, respectively. The most severe defect was detected for the strain expressing the W1305A version of Tf1 IN. This strain showed a 100-fold drop in transposition compared to the wild-type level of transposition. The INfs version of Tf1 showed a 28-fold drop in transposition.
One possible explanation for the reduction in transposition of the IN mutants is that the mutations caused defects in the expression or stability of IN. To test this possibility, immunoblot analysis was performed on total cell proteins extracted from strains expressing wild-type Tf1-neo, INfs, PRfs, and the four CHD mutants (Fig. (Fig.3A).3A). Two independent transformants of each CHD mutation were tested. The Y1292A, W1305A, and CHDΔ mutants produced wild-type levels of IN. The strain with V1290A expressed 83% as much IN as wild-type IN as determined by densitometry. As expected, frameshift mutations at the beginning of IN (INfs) or in PR (PRfs) blocked expression of IN (Fig. (Fig.3A).3A). The expression of Gag was not changed by any of the mutations except PRfs, which produces a Gag fused to an N-terminal fragment of PR.
Since the mutations in the CHD did not reduce expression of IN significantly, we explored the possibility that they inhibited transposition by reducing cDNA levels. DNA was extracted from S. pombe cells expressing the wild-type and mutant versions of Tf1-neo. These DNAs were then digested with BstXI and blotted. The resulting membrane was hybridized with a neo-specific probe. Wild-type Tf1-neo produced a cDNA band of 2.1 kb, derived from the right end of the transposon (Fig. (Fig.3B).3B). The 9.5-kb band represented the neo-containing fragment from the expression plasmid and served as an internal control for levels of material loaded in each lane. While the INfs version of Tf1-neo generated wild-type levels of cDNA, the version that expressed no RT (PRfs) lacked both the 2.1-kb band and the single LTR circle intermediate that migrated as a 5.5-kb species.
By densitometry, we quantified the amount of cDNA produced by the CHD mutants. The signal from the 2.1-kb cDNA band was normalized to the signal from the 9.5-kb plasmid band. The CHDΔ, V1290A, and Y1292A mutants produced 87%, 91%, and 83.4% of the cDNA that the wild type generated, indicating that these mutations did not significantly reduce reverse transcription (Fig. (Fig.3B).3B). A somewhat greater reduction in cDNA production was detected for the W1305A mutant, which produced 66.7% of the 2.1-kb cDNA that was observed for wild-type.
RT and IN can interact and function together during reverse transcription and integration. Human immunodeficiency virus (HIV) type 1 IN and RT have been shown to interact (45), and in the case of avian leukosis sarcoma virus, RT consists of alpha and beta chains, where beta is a fusion of RT and IN (8, 15, 39). The RT of Ty3 has a similar alpha-beta configuration (18, 25). In addition, the interaction of Ty1 IN and RT has been shown to be necessary to activate RT (44). Likewise, the RT and IN of Tf1 have been shown to interact with each other in two-hybrid experiments (43). We therefore examined the possibility that the mutations in the CHD decreased the stability of RT in a way that reduced IN function. Immunoblots probed with anti-RT antibody revealed that the levels of RT were not changed except in the V1290A mutant, which showed a small reduction equivalent to the reduction in IN observed for this mutant (Fig. (Fig.3C).3C). As expected, PRfs blocked expression of RT, while INfs had no effect on RT production.
Other than the W1305A substitution, which showed a 33% decrease, mutations in the CHD did not significantly reduce cDNA levels. Despite these normal levels of cDNA, it is possible that the mutations in the CHD prevented the entry of the cDNA into the nucleus. To detect the levels of reverse transcripts that accumulated in the nucleus, we used a previously described modification of the transposition assay that detects homologous recombination between Tf1 cDNA and the Tf1 sequence in the expression plasmid (1).
For the recombination assay, we used a version of the Tf1 (Tf1-neoAI) that contained an AI that disrupted the reading frame of neo. The neo gene was in the opposite orientation from Tf1 and was therefore transcribed as an antisense mRNA. The intron was in the sense strand and could not be spliced out of the neo mRNA to provide G418r. However, the intron could be spliced out of the Tf1 transcript so that reverse transcription of Tf1 produced active versions of neo. Unlike the transposition assay, the recombination assay did not include the removal of the plasmid expressing Tf1 before cells were replica printed to medium containing G418. The cells became resistant to G418 only after the cDNA recombined with the expression plasmid or was inserted by IN. The frequencies of homologous recombination were significant, such that the resistance to G418 was not dependent on IN expression (Fig. (Fig.4,4, INfs). Strains expressing the PRfs mutation lacked RT and consequently showed no growth on G418 due to the absence of cDNA. The strain expressing Tf1 that had mutations in the nuclear localization signal of GAG showed significantly reduced growth on G418, as the cDNA in this case was unable to enter the nucleus (24). Strains expressing the CHD mutants that produced near-wild-type levels of cDNA (CHDΔ, V1290A, and Y1292A) showed growth on G418 that was comparable to that of the wild type (Fig. (Fig.4).4). This observation indicated that mutations in the CHD did not inhibit the import of cDNA into the nucleus.
We conducted a quantitative recombination assay to measure the levels of reverse transcripts that accumulated in the nucleus (Table (Table1).1). In a method similar to the quantitative assay for transposition, patches of cells induced for expression of Tf1-neoAI were diluted in series and plated onto YES medium with and without G418 to determine the fraction of cells that acquired an active neo. Two independent transformants of each mutant were tested. The level of G418 resistance produced by INfs was 0.9% and represented the amount of homologous recombination that occurred in the absence of integration. The V1290A and Y1292A mutants produced equal or slightly more resistance to G418, indicating that they accumulated wild-type levels of cDNA in the nucleus. The transposon with CHDΔ produced 73% as many G418r colonies as INfs, indicating that this mutation caused a small reduction in the amount of cDNA accumulated in the nucleus.
The data above indicate that the mutations in the CHD greatly reduce transposition but have little or no effect on the amounts of IN, cDNA, and cDNA accumulated in the nucleus. These results suggest that the mutations may inhibit a late step in integration, such as the ability of IN to bind its substrate. To investigate whether the mutations in the CHD reduced the binding of IN to the Tf1 cDNA, we used ChIP. Initially, we conducted an IP assay to determine whether the anti-IN antibody used for ChIP was equally efficient in precipitating the mutant versions of Tf1 IN. An immunoblot of the IP showed that the INs produced by the Tf1-V1290A and Tf1-Y1292A mutants showed weak interactions with the ChIP anti-IN antibody. However, a very strong interaction was detected for the CHDΔ mutant (Fig. (Fig.5A).5A). Thus, we choose CHDΔ to analyze the binding of IN to cDNA with ChIP.
For this assay we used Tf1-neoAI. To detect IN bound specifically to cDNA, we designed PCR primers that amplified only the neo in the cDNA copies of Tf1 and not the plasmid-encoded Tf1-neoAI. The oligonucleotide HL1576 bridged the AI and primed only when the intron was spliced out, and thus it could not amplify the plasmid copy of neoAI (Fig. (Fig.5B).5B). The primer pair HL1576 and HL1040 detected a 521-nucleotide cDNA that included the downstream LTR. This primer set was used in PCR on DNA prepared either from immunoprecipitated chromatin fractions or from whole-cell crude extracts used as an input control. To account for differences in loading or the variance in amplification in separate PCRs, we used multiplex PCR and normalized the enrichment of the cDNA sequences relative to a PCR product of fbp1. The amplification of the whole-cell input from cells expressing wild-type Tf1-neoAI produced similar amounts of signals from the cDNA and the fbp1 gene (Fig. (Fig.5C).5C). The results from three independent experiments were averaged for each strain (Fig. (Fig.5D).5D). The ChIP of wild-type Tf1-neoAI using anti-IN antibody produced a 6.6-fold enrichment of the cDNA. We tested whether the cDNA signal was indeed specific for cDNA copies of neo by analyzing a strain with a D779N substitution in RT that disrupts reverse transcription (30). No cDNA was detected in this strain in either the WCE or the IP, demonstrating that these primers were specific for cDNA. Strains expressing the PRfs and the INfs were also analyzed as controls. Since the PRfs does not express RT, no cDNA signals were detected. The INfs produces normal levels of cDNA but does not express IN. A greatly reduced cDNA product in the IP demonstrated that cDNA detected in the IP of the wild type specifically interacts with IN. To test whether the cDNA bound to IN was unintegrated, we performed ChIP with cells that expressed the R786H mutation in the RNase H domain of RT. This mutation does not affect the production of cDNA, but because processing of the plus-strand primer of reverse transcription is defective, the cDNA cannot be integrated into the host genome (2). Detection of wild-type levels of cDNA in the ChIP relative to the fbp1 signal for this strain (Fig. 5C and D) demonstrated that the cDNA signal was indeed derived from IN bound to unintegrated cDNA. The strain that expressed the CHDΔ mutation exhibited a threefold decrease in the binding of IN to the cDNA compared to the strain expressing the wild type (Fig. (Fig.5D).5D). Thus, it appears that the CHD played an important role in mediating binding of IN to the cDNA.
The CHD of Maggy interacts with histone H3 methylated at lysine 9, and this directs it to regions of heterochromatin (14). Although the CHD of Tf1 is less conserved than that of Maggy, we tested whether it played a role in directing integration to promoters. To determine whether there were defects in integration into target sites, we used a plasmid-based targeting assay that measures the integration activity of specific sequences (28). This assay is based on a strain of S. pombe that contains both a plasmid that expresses Tf1-neo and a plasmid with the target sequences. The neo gene in Tf1 causes G418 resistance upon successful integration. Target plasmids extracted from individual G418-resistant patches were introduced into bacteria and selected for resistance to ampicillin. Because the neo gene encodes resistance to kanamycin, the target plasmids with Tf1-neo insertions resulted in colonies that were resistant to both ampicillin and kanamycin. These plasmids were then sequenced to determine the positions of Tf1 insertions. Insertions mediated by IN produce signature duplications of base pairs at the target sites. Only insertions that resulted in duplications of 5 base pairs, the type typical of Tf1, were included in the analyses described below.
The strong preference of Tf1 for integrating in an intergenic region containing a pair of divergent promoters for the ade6 and bub1 genes has been established (28). The target plasmid contained the ORF of ade6, its upstream intergenic region, and a portion of the adjacent bub1 ORF (Fig. (Fig.6A).6A). When the bub1-ade6 target plasmid is introduced into a strain of S. pombe that expresses wild-type Tf1-neo, 95% (41 of 45) of the insertions in the plasmid occur in the divergent pair of promoters within a 160-bp window (Fig. (Fig.6A6A).
For the current experiment, the same conditions established by Leem et al. (28) were used to test Tf1-neo with the mutations in CHD. For the V1290A mutant, insertions at the 160-bp window constituted 83% (24 out of 29 events) of the plasmid events (Fig. (Fig.6B).6B). In strains that expressed Y1292A, 29 out of 39 events (74%) of the plasmid insertions occurred in the 160-bp window (Fig. (Fig.6C).6C). Compared to the 95% level of promoter insertions produced by wild-type IN, these frequencies for the alanine mutations produced modest reductions in targeting that were not statistically significant.
Assaying integration into promoters by the Tf1 with CHDΔ was difficult because the efficiency of integration into the target plasmid was significantly reduced. When isolating insertions in the bub1-ade6 target plasmid, only 10 events could be isolated. However, out of the 10 events, 1 (10%) was in the 160-bp window (Fig. (Fig.6D).6D). This decrease of insertions into the target window is significant, with a P value of 0.021 (Fisher's exact test).
To test whether the CHD functioned in directing integration at other promoters, we assayed two additional target plasmids. Previous experiments with a target plasmid demonstrated that Tf1-neo integrates with a strong preference for an enhancer element (UAS1) in the promoter of fbp1 (28). In particular, 83% (19 of 23) of the insertion events are within a 90-bp window upstream of the fbp1 ORF, with positions 4972 and 4982 being the two dominant insertion positions (Fig. (Fig.7A).7A). Tf1-CHDΔ exhibited a dramatic lack of preference for the UAS1 element, and only two out of nine events (22%) were found to occur in the promoter region of fbp1 (Fig. (Fig.7B).7B). Importantly, none of these events were at the dominant position 4972 or 4982. This lack of integration into positions 4972 and 4982 was significant, with a P value of 0.042 (Fisher's exact test). Previous experiments also showed that the promoter region of the gene SPCC4F11.03c is a strong target of integration (Fig. (Fig.7C).7C). Just two out of seven events (28%) generated by Tf1-CHDΔ targeted the promoter region of SPCC4F11.03c (Fig. (Fig.7D).7D). Together, these data indicate that Tf1 IN requires the CHD for target site preference.
Three alanine substitutions in the CHD of Tf1 IN and one CHD-deleted mutant caused significant reductions in transposition frequency. The results of the immunoblotting showed that, apart from a small reduction caused by V1290A, the mutations did not alter the levels of the IN and RT expressed by Tf1. DNA blots indicated that, except for the 33% reduction caused by W1305A, the levels of cDNA produced were also not significantly changed. By measuring homologous recombination, we determined that the mutations in Tf1 had little or no effect on the accumulation of cDNA in the nucleus. Nevertheless, all four of the mutants exhibited substantial reductions in transposition. Thus, these drastic reductions in transposition likely resulted from defects that occurred specifically at the integration step of the transposition cycle.
The mutation CHDΔ reduced the amounts of cDNA that bound IN. Although this result indicates that the CHD contributed to the binding of cDNA, experiments that tested each domain of Tf1 IN for DNA binding activity revealed that as a recombinant fragment, the CHD itself does not bind DNA (10). Instead, strong DNA binding activity in the C-terminal domain was mapped to the GPY/F domain, a region of IN just upstream of the CHD. Together, these findings suggest that Tf1 cDNA may have bound the GPY/F domain and that the deletion of the CHD reduced this binding by generating a conformational change. Testing the GPY/F domain in vivo for a role in binding cDNA will be difficult since single amino acid substitutions in the conserved residues caused IN to be degraded (10). Alternatively, it is also possible that the interaction detected between IN and the cDNA is indirect and that another factor is responsible for binding the cDNA.
The ChIP assays revealed that CHDΔ caused a 2.7-fold reduction in binding of IN to cDNA. Although this defect in cDNA binding was substantial, this does not appear to be sufficient to account for the 14.3-fold reduction in transposition caused by the mutation. Assuming a direct interaction between IN and cDNA, one possibility is that the mutations may have inhibited binding of IN to both ends of the cDNA. Since our ChIP assay measured only binding of IN to the downstream LTR, the cumulative defect in cDNA binding could potentially be the square of what was detected. Thus, the defect in binding cDNA for CHDΔ could be a large as 7.3-fold. Currently, we are unable to test this possibility directly because the ChIP assay as designed with the AI cannot be used to test binding of IN to the upstream LTR. Primers that amplify the upstream LTR of the cDNA would also detect copies of Tf1 in the expression plasmid. It is also possible that the 2.7-fold reduction in binding of CHD-deleted IN to cDNA was an underestimate of the true defect. Since the anti-IN antibody used for ChIP was more efficient in precipitating CHDΔ than wild-type IN (Fig. (Fig.5A),5A), the defect in the interaction between IN and cDNA is likely to be greater than 2.7-fold.
It is possible that the mutations in the CHD reduced the catalytic activity of IN. Such mutations may allow interaction with target DNA but have an indirect effect on the catalytic site. This kind of defect is consistent with the mutations that did not significantly lower levels of IN, cDNA, or the amounts of cDNA available for homologous recombination in the nucleus. However, one result that argues against a defect in catalytic activity is that as a recombinant protein, the CHDΔ form of IN had four to seven times greater strand transfer activity than wild-type IN (19).
The point mutations in the CHD and the CHDΔ IN showed comparable reductions in transposition frequencies (Fig. (Fig.2).2). However, only the CHDΔ mutation caused a profound effect on target recognition. In the absence of the CHD, Tf1 demonstrated a drastically reduced ability to target the pol II promoter regions of fbp1, ade6, and SPCC4F11.03c. The disparity between the CHDΔ and the point mutations in their ability to direct integration to the promoters indicates that the CHD has two separate functions. One function contributed significantly to the frequency of integration. The other function of the CHD, as revealed by CHDΔ, contributed to the selection of target sites. The role of the CHD in positioning integration is likely due to interactions with a chromatin factor. Our data do not address whether the single amino acid mutations in the CHD may have lowered integration frequencies by partially disrupting an interaction. The deletion of the CHD may have fully disrupted the interaction, resulting in both low integration frequencies and an inability to recognize pol II promoters. Alternatively, the interactions responsible for integration efficiency may be distinct from those required for target selection.
Cellular proteins have been shown to play a role in directing the position of integration of retrotransposons. For example, the Sir4 protein recruits Ty5 IN and promotes integration at heterochromatic regions in the telomere and the silent mating loci of the budding yeast Saccharomyces cerevisiae (46). Similarly, Ty3 IN requires transcription factors of RNA Pol III to direct highly selective integration into Pol III transcript start sites (5, 38). For HIV, integration occurs preferentially in highly active transcription units (32, 34, 40), and the cellular protein LEDGF binds the IN of HIV type 1 and mediates the selection of target sites (7, 41).
It has been recently reported that Tf1 integration is directed to pol II promoters by transcription activators such as Atf1p (28). That work demonstrates that Atf1p plays a role in recruiting Tf1 IN to the promoter of fbp1 but not to the divergent promoters of bub1-ade6. However, our study shows that the absence of the CHD not only disrupted the targeted integration at the promoter of fbp1 but also caused similar defects in targeting the promoters of bub1-ade6 and SPCC4F11.03c. From these results, we speculate that the CHD of Tf1 IN may function together with multiple factors of chromatin that are commonly found at promoters where Tf1 integrates. This feature of chromatin could be a variety of transcription factors that together are responsible for many targets of Tf1 integration. Atf1p belongs to the basic leucine zipper family of transcription factors and shares its highly conserved bZIP motif with other transcription factors found in S. pombe, such as Pap1p, Atf31p, Zip1p, and Atf21p (4). One or more of these proteins, together with the CHD, could be required for targeting Tf1 to the bub1-ade6 and SPCC4F11.03c promoters.
The relatively few insertion events that were observed in the bub1-ade6 and fbp1 promoters in the absence of the CHD were due to residual integration activity of the truncated IN. Although CHDΔ no longer preferred the promoter sequences, the insertions were not randomly positioned. Interestingly, the absence of the CHD caused integration to occur at nucleotide 340 of the β-lactamase gene, as demonstrated by 6 insertion events out of 10 in the plasmid with the bub1-ade6 promoter and 5 insertion events out of 9 in the fbp1 plasmid. Thus, in the absence of the CHD, Tf1 developed a new integration preference. Since the β-lactamase gene is from E. coli, this preference is likely a result of fortuitous interactions. One possibility is that IN lacking the CHD may retain low levels of binding to some transcription or chromatin factor that binds to the β-lactamase gene. Alternatively, the DNA sequence of the β-lactamase gene at nucleotide 340 may have a structural perturbation that enhances integration. The integration sites of several transposons have structural features that are recognized by the transposases (17, 21, 27, 37).
The selection of target sequence for integration by retroviruses such as HIV and murine leukemia virus (MLV) has important consequences for both the virus and the host. Retroviral integration can affect host gene expression due to insertion of viral promoters or enhancers. Because of their target site preferences, the use of retroviral vectors in gene therapy is associated with great risk. MLV vector-based treatment of X-linked severe combined immunodeficiency led to the activation of the cellular proto-oncogene LMO-2 and contributed to the development of leukemia (16). These setbacks have generated intense interest in understanding integration site selection. Tf1 exhibits a similar preference for insertion as MLV, as both elements target the promoters of genes. The molecular mechanism of MLV integration is still unknown. In light of the role of the Tf1 CHD in mediating target site selection, further studies of Tf1 integration may lead to improved understanding of how retroviruses target integration.
We thank Amnon Hizi for the gift of integrase antibodies and Jia-Hwei Lin for creating an initial set of mutations in the CHD. We are grateful for the guidance and input provided by Kei-Bang Nam before his untimely death. We thank Susan Vitale for assistance with statistical analysis.
This research was supported by the Intramural Research Program of the NIH from the Eunice Kennedy Shriver National Institute of Child Health and Human Development.
Published ahead of print on 24 December 2008.
†Supplemental material for this article may be found at http://jvi.asm.org/.