|Home | About | Journals | Submit | Contact Us | Français|
The Ty1 retrotransposon of Saccharomyces cerevisiae is comprised of structural and enzymatic proteins that are functionally similar to those of retroviruses. Despite overall sequence divergence, certain motifs are highly conserved. We have examined the Ty1 integrase (IN) zinc binding domain by mutating the definitive histidine and cysteine residues and thirteen residues in the intervening (X32) sequence between IN-H22 and IN-C55. Mutation of the zinc-coordinating histidine or cysteine residues reduced transposition by more than 4,000-fold and led to IN and reverse transcriptase (RT) instability as well as inefficient proteolytic processing. Alanine substitution of the hydrophobic residues I28, L32, I37 and V45 in the X32 region reduced transposition 85- to 688-fold. Three of these residues, L32, I37, and V45, are highly conserved among retroviruses, although their effects on integration or viral infectivity have not been characterized. In contrast to the HHCC mutants, all the X32 mutants exhibited stable IN and RT, and protein processing and cDNA production were unaffected. However, glutathione S-transferase pulldowns and intragenic complementation analysis of selected transposition-defective X32 mutants revealed decreased IN-IN interactions. Furthermore, virus-like particles with in-L32A and in-V45A mutations did not exhibit substantial levels of concerted integration products in vitro. Our results suggest that the histidine/cysteine residues are important for steps in transposition prior to integration, while the hydrophobic residues function in IN multimerization.
The structural and functional similarities of the Ty retrotransposons of Saccharomyces to mammalian retroviruses suggest similar mechanisms for Ty transposition and retroviral integration. Like retroviruses, Ty elements are bound by long terminal repeats (LTR), contain two overlapping open reading frames analogous to retroviral gag and pol, and undergo proteolytic processing of a Gag/Pol fusion protein. Following reverse transcription of an RNA intermediate within a virus-like particle (VLP), the cDNA is integrated into the host genome by the element-encoded integrase (IN). The two superfamilies Ty1/copia and Ty3/gypsy are distinguished by the order of IN and reverse transcriptase (RT) in the pol region, with IN upstream of RT in Ty1/copia but downstream of RT in Ty3/gypsy (26, 60).
The similarities between retrotransposons and retroviruses extend to the highly conserved sequence homologies in the catalytic active sites of the Pol proteins, protease (PR), RT, and IN (1, 36, 19, 55, 60). A noncatalytic sequence, the zinc binding domain (ZBD), is also conserved in the N-terminal regions of retroviral and retrotransposon INs (36). This domain, characterized by the sequence H-X3-7H-X23-32C-X2C, binds a single zinc ion (10, 37, 42, 67), which stabilizes the N-terminal region of IN and promotes formation of IN tetramers, thereby enhancing the enzyme's catalytic activity (37, 67).
The IN catalytic domain has been well characterized in retroviruses (7, 34) and to a lesser extent in Ty1 (45, 54). The retroviral IN ZBD has also been studied extensively, initially by evaluating the effects of N-terminal deletions (53, 8, 11, 12, 16, 23, 59) or mutations of the histidine or cysteine residues (24, 57, 58) in assays which recapitulate biochemical activities of IN, i.e., 3′ dinucleotide processing and integration of donor molecules representing viral LTR sequences into identical target molecules (7). These results generally indicate that INs with N-terminal deletions or histidine/cysteine mutations are deficient for dinucleotide cleavage and strand transfer but that the site-specific cleavage of a preintegrated donor molecule from its substrate, known as “disintegration” (14), is relatively unaffected.
In vivo studies using virions with IN ZBD histidine and/or cysteine mutations reveal additional defects in viral functions not associated with catalytic activities, such as production of viral progeny (63), postentry viral functions (41, 49), and, in some instances, altered virion morphology (25, 49). These global defects suggest that the ZBDs of retroviral INs play additional roles in the viral life cycle. The structural organization of retroviruses places the IN ZBD adjacent to the C terminus of RT in the Pol fusion protein, whereas in the Ty1/copia superfamily, the positions of IN and RT are reversed, so that the IN ZBD is adjacent to the C terminus of PR rather than RT. Since the functional significance of this difference has not been addressed, the characterization of the Ty1 IN ZBD would both yield information regarding its function in the context of Ty biology and provide a basis for comparison to retroviral studies in view of the reversed pol organization.
The majority of analyses of the N-terminal domains of retroviral INs have targeted the highly conserved histidine/cysteine residues (7), rather than residues within the intervening region between the C-terminal histidine and the N-terminal cysteine, although structural studies of the N termini of human immunodeficiency virus type 1 (HIV-1) and HIV-2 INs indicate that some of the hydrophobic residues in this region are found at the dimer interfaces (13, 21, 22). Mutagenesis studies of the H-C intervening sequences have not determined functional defects arising from altering these hydrophobic residues. Effects have been observed, however, in mutants in which basic amino acids were changed to alanines. In Moloney murine leukemia virus (Mo-MuLV) IN, alanine substitution of closely positioned lysines, K66/K68, caused impaired complementation of the N-terminal domain with the remainder of IN in vitro as well as defective viral replication in vivo (65). Mutation of the basic residue K34 in HIV-1 IN also resulted in lower levels of virion production (39). In contrast, in vitro analysis of an alanine replacement of S39 in HIV-2 IN showed wild-type (WT) levels of 3′ dinucleotide cleavage, strand exchange, and disintegration (57). Two additional works indicate that specific residues between the N-terminal and C-terminal histidines of HIV-1 IN, K14 and Y15, play a role in viral DNA binding and conformational interconversion, respectively (50, 66).
The lack of retrotransposon data and the limited number of observations from the retrovirus literature led us to a comprehensive mutagenesis study of the N-terminal region of Ty1 IN. Here we present a genetic and biochemical analysis of the Ty1 IN histidine/cysteine zinc binding residues (ZBRs) and the intervening sequence between IN-H22 and IN-C55 (X32). This work defines the critical residues for transposition, indicates separate functions for the HHCC and X32 residues, and highlights the functional differences of this region in retrotransposons and retroviruses.
Yeast strain yGS38 (54) was used for quantitative transposition and complementation assays. An isogenic derivative, yGS37, was used for VLP purifications. Strain GRF167 (5) was used for pulse-chase immunoprecipitations, and RDKY1293 (32) was used for glutathione S-transferase (GST) pulldowns.
pGTy1-H3his3-AI, which contains the Ty1 element fused to a GAL1 promoter and marked with the his3-AI transposition indicator gene (15), was used for transposition assays, complementation assays, and biochemical analyses. Site-directed mutations were introduced into Ty1 IN by overlap PCR (29). As a control, WT IN was reconstructed by overlap PCR (WT-overlap). Clones were checked for the presence of the correct mutation by sequence analysis. For GST pulldown experiments, PCR fragments containing either mutant or WT IN were cloned into the XbaI-HindIII restriction sites of the GST expression vector pEG(KT) (44) (a gift of R. Deschenes). Candidate clones were checked by restriction enzyme analysis and transformed into RDKY1293. The pulldown partner for experiments with in-H22A and in-C58A was WT IN. Pulldown partners for GST-in-I28A, GST-in-L32A, and GST-in-V45A were either WT IN or IN with the same mutations as the GST-IN. The pulldown partner INs contained a C-terminal c-myc epitope introduced as part of the downstream PCR primer. The PCR fragment was cloned into the IN expression vector pBDG677 using the XhoI-HindIII restriction sites as previously reported (47), and then the plasmid was converted to a TRP1-based plasmid by microhomologous recombination (40) and transformed into RDKY1293 carrying a pEG(KT)/IN GST fusion plasmid with either WT or mutant IN. Expression of the constructs was evaluated by immunoblotting before pulldown experiments were performed.
The amino acid alignment of the IN N termini of representative retroviruses and functional members of the Ty retrotransposon family was performed using the CLC Combined Workbench 3 program (CLC Bio LLC, Cambridge, MA).
Ty1 transposition was quantitated as the frequency of histidine prototrophy using the pGTy1-H3his3-AI transposition indicator (15). Five early-stationary-phase cultures of each strain grown in synthetic complete medium lacking uracil (SC−ura) with 5% raffinose were diluted to an optical density at 600 nm of 0.06 into SC−ura with 2% galactose (SC−ura+2% galactose) and grown for 5 days at 20°C. The cells were plated on SC−ura+2% glucose plates to determine the titer and on SC−ura−histidine+2% glucose plates to determine the number of histidine prototrophs. Qualitative assays were performed by spreading single colonies to 1-cm by 1-cm patches on SC−ura+glucose plates, followed by incubation at 30°C for 2 to 3 days, and then replica plating to SC−ura+galactose. After incubation at 20°C for 24 h, the galactose-induced cells were replica plated to SC−ura−his+glucose and incubated at 30°C for 2 to 3 days to detect histidine prototrophs. A similar qualitative assay was used to determine if high concentrations of Zn2+ altered the transposition efficiency of the histidine or cysteine mutants. Filter disks soaked in either saturated ZnCl2 or 0.3 M ZnSO4 were placed on a lawn of cells on SC−ura+galactose plates, and the plates were incubated at 20°C for 2 days. The plates were then replica plated and incubated as described above.
Lysates from galactose-induced cultures were purified by sucrose gradient and differential centrifugation according to established procedures (18, 28). For in vitro integration assays, VLPs were further purified by density gradient centrifugation through a 2-ml step gradient consisting of 5% increments between 5% and 50% of Opti-Prep density gradient medium (Sigma-Aldrich, St. Louis, MO) in VLP buffer A (15 mM KCl, 10 mM HEPES-KOH [pH 7.8], 5 mM EDTA, 10% sucrose). The resuspended pellet from the linear sucrose gradient was layered onto the gradient and centrifuged at 4°C for 3 h at 201,078 × gavg in a Beckman TLA-100 tabletop ultracentrifuge. The gradient was fractionated into 2-drop fractions and assayed for exogenous RT activity (27). Three (or four) fractions displaying the highest levels of RT activity were pooled and stored at −80°C.
Purified VLPs were analyzed by immunoblotting using polyclonal antisera B2 for IN, B8 for RT, and anti-VLP for Gag as previously described (28). Recombinant His6-IN, GST-IN, and IN-c-myc were detected by B2 antiserum. Horseradish peroxidase (HRP)-conjugated anti-GST was purchased from GE Healthcare (Piscataway, NJ), and mouse monoclonal antibody to c-myc was purchased from Roche Diagnostics (Indianapolis, IN). Secondary ECL HRP-linked anti-rabbit and HRP-linked anti-mouse antibodies were purchased from GE Healthcare (Piscataway, NJ). Cross-reactivity was detected by chemiluminescence.
Four independent colonies of yeast strain yGS38 (54) containing pGTy1-H3his3-AI (WT) or with pGTy1-H3his3-AI carrying the IN mutation in-H22A, in-I28A, in-K33A, in-Y39A, in-V45A, or in-C58A or pr-1682 (a point mutation in the PR active site) (54) were grown in SC−ura+5% raffinose at 30°C for 3 days. From this culture, an amount equal to 0.1 optical density unit (600 nm) was transferred to 5 ml SC−ura+2% galactose and grown at 20°C for 5 days. DNA was extracted by the method of Hoffman and Winston (30), and 6 μg of each DNA was digested overnight with DraI and SphI (New England Biolabs, Beverly, MA). The SphI site is in the vector just downstream of the Ty1 LTR, and digestion with this enzyme was necessary to prevent an overwhelming signal from the expression plasmid. DNAs were separated on a 0.9% Tris-borate-EDTA gel at 80 V for 10 h, and Southern hybridizations were performed using a [α-32P]CTP-labeled DNA probe derived from nucleotides −50 to −190 upstream of the HIS3 gene. The resulting cDNA fragment is expected to comprise 650 bp. Following imaging on a Typhoon Trio phosphorimager (GE Healthcare, Piscataway, NJ), the blots were stripped and hybridized with a probe derived from the single-copy yeast gene CPR7. This hybridization served as a loading and normalization control. Quantitation was performed using ImageQuant TL software (GE Healthcare, Piscataway, NJ).
Galactose-induced cells coexpressing GST WT or mutant IN fusions and WT or mutant IN-c-myc were lysed by glass bead vortexing in lysis buffer (20 mM Tris-HCl [pH 7.5], 0.6 M NaCl, 10 mM MgCl2, 1 mM EDTA, 0.12% Triton X-100). The high salt concentration in the lysis buffer served to enhance the solubility of the c-myc IN. Following lysis, additional lysis buffer without NaCl was added slowly while mixing to reduce the NaCl concentration to 0.4 M to prevent the high salt concentration from interfering with binding of the protein to GST beads. A 400-μg aliquot of protein from the cleared lysates was incubated in 1 ml lysis buffer (0.4 M NaCl) and 50 μl glutathione-Sepharose 4B beads (GE Healthcare, Piscataway, NJ) with gentle mixing at 4°C for 2 h. The beads were washed four times with 1.0 ml lysis buffer with 0.4 M NaCl to remove unbound proteins and then resuspended in 40 μl sample loading buffer, boiled, centrifuged, and loaded onto a 10% Tris-glycine Novex gel (Invitrogen, La Jolla, CA). Immunoblotting was performed as previously described. Lanes exhibiting similar signals for GST-IN input and GST-IN pulldown were selected for quantitation by Quantity One software (Bio-Rad, Hercules, CA).
For complementation tests, pGTy1-H3his3-AI mutants in-H22A, in-I28A, and in-V45A and pGTy1-H3his3-AI-WT were converted from URA3-based plasmids to TRP1 plasmids by microhomologous recombination (40). Quantitative transposition assays were performed using the his3-AI transposition indicator as described above. The transposition efficiency was defined as the number of His+ Trp+ Ura+ cells divided by the number of Trp+ Ura+ cells present after galactose induction. The complementation efficiency was calculated by dividing the transposition efficiency obtained with a given strain in which one or both plasmids carried an IN mutation by the transposition efficiency obtained with the control strain in which both the URA3 and TRP1 plasmids carried WT IN as previously described (48).
The method for an in vitro integration assay for recombinant IN has been previously published (46). Briefly, this assay monitors the IN-mediated insertion of a linear donor molecule into a supercoiled plasmid target and the products visualized by Southern blotting. The instability of the ZBR INs precluded the use of mutant recombinant IN or native VLP-associated IN, but we were able to alter the assay to accommodate IN synthetically stabilized by an N-terminal GST fusion (see the Methods section in the supplemental material). For X32 mutants, a 10-μl aliquot of purified VLPs with similar endogenous RT activity (27) was used as the source of IN. The Southern blot was visualized by phosphorimaging, and products were quantitated using ImageQuant TL software (GE Healthcare, Piscataway, NJ).
The sequence alignment of the N-terminal regions of INs from four functional Ty elements and six representative retroviruses (Fig. (Fig.1)1) illustrates the conserved nature of the histidine and cysteine residues as well as certain residues of the intervening sequence (Ty1-X32). The similarity of amino acids I28, S31, L32, I37, T38, Y39, D44, V45, and S48 in retrotransposons and/or retroviruses suggested that these residues are important for IN functions and, consequently, interesting candidates for mutagenesis. As four of the eight similar residues are hydrophobic, we included in our analysis additional, less well conserved, hydrophobic residues, I51 in X32 and I16, L59, and I60 adjacent to the histidine/cysteine residues. We also analyzed N24, E42, and K33, as these residues do not exhibit similarity, although K33 is the nearest basic residue to Mo-MuLV K68, which has been shown to be important for complementation of strand transfer and 3′ dinucleotide cleavage in vitro and virus production in vivo (65).
We quantitated the transposition frequencies of pGTy1-H3his3-AI elements having alanine replacements at the ZBR IN-H17, IN-H22, IN-C55, and IN-C58 as well as 13 residues in X32 and the 3 hydrophobic residues adjacent to IN-H17 and IN-C58. As a control for position specificity, we also substituted alanines for two cysteines outside of the ZBD, IN-C176 and IN-C200. Each of the ZBR mutations caused a profound deficiency in Ty1 transposition ranging from 4,666- to 12,727-fold below WT levels (Table (Table1,1, experiment 1), while the transposition efficiency of a reconstructed WT element (WT-overlap) was identical to that of the original WT element. Alanine substitutions for IN-C176 and IN-C200 resulted in a modest decrease in transposition, 21-fold and 2-fold, respectively, indicating that the transposition defects with in-C55A and in-C58A are position specific rather than residue specific. For comparison, we included the previously characterized IN mutant, in-2600, in which an in-frame linker insertion encoding four additional amino acids disrupts the critical 35-amino-acid interval between the catalytic IN D154 and E190 (45). The 200-fold decrease in transposition with in-2600 is comparable to that in previous reports (48, 54) and illustrates the importance of the ZBRs in transposition, as a single-residue replacement of any of the ZBRs resulted in an even greater transposition defect than disruption of the catalytic domain.
The transposition defect exhibited by the HHCC residues was not due to our choice of alanine as a replacement residue, as an H-to-C interchange and replacement of in-H22A and in-C58A with glutamic acid, arginine, or serine residues failed to rescue the transposition defect in qualitative patch assays (see Fig. S1 in the supplemental material). We also attempted to rescue the transposition defect by placing filter disks soaked in either saturated ZnCl2 or 0.3 M ZnSO4 on a lawn of IN-ZBR mutant cells during pGTy1 induction prior to replica plating to SC−ura−his+glucose plates. Unlike the restoration of growth of the Gal4p ZBD mutant, P25L (33), no increase in histidine prototrophs appeared around any of the disks compared to control disks saturated with water (data not shown), indicating that the availability of additional Zn2+ ions could not reverse the transposition defect exhibited by IN-ZBR mutants.
In a separate experiment, we measured the transposition efficiencies of X32 mutants (Table (Table1,1, experiment 2). Alanine replacement of the hydrophobic residues I28, L32, I37, and V45 in the X32 sequence resulted in a decrease in transposition in the range of 85- to 688-fold, while mutation of the nonhydrophobic residues N24, K33, Y39, and E42 showed little effect. In addition, we observed no transposition defect in mutants in-S31A, in-T38A, in-D44A, and in-S48A (data not shown). The in-I51A mutant displayed a WT level of transposition. Of the ZBR-adjacent residues, only in-I16A displayed a transposition defect. Taken together, these results suggest that the hydrophobic residues including and N-terminal to IN-V45 are pivotal for one or more IN functions. Our subsequent work centered on identifying the steps in transposition that were affected by these N-terminal mutations.
The transposition process requires the assembly of Ty Gag and Pol proteins into VLPs (27); therefore, we examined VLPs by immunoblotting for the presence of IN, RT, and Gag (Fig. (Fig.2).2). For comparison, we included WT VLPs and the transposition-competent mutants in-N24A, in-K33A, in-E42A, and in-I51A. In addition, we analyzed Ty1 proteins from VLPs containing the cysteine mutations in-C176A and in-C200A. IN, if present, was below the level of detection in the ZBR mutants, in-H17A, in-H22A, in-C55A, or in-C58A (Fig. (Fig.2A),2A), but all the X32 mutants, whether transposition competent or transposition defective, displayed WT levels of IN, as did the control cysteine mutant in-C200A. However, we frequently observed a slight decrease in the level of IN in the in-C176A mutant. Although RT was detected in the ZBR mutants, (Fig. (Fig.2B),2B), the level was drastically reduced compared to WT. The RT in the X32 and the in-C176A and in-C200A mutants appeared to be unaffected. A processing defect of the Gag of the ZBR mutants was also observed (Fig. (Fig.2C).2C). This result illustrates a critical difference between the ZBR mutations and the X32 mutations in protein stability and/or PR function.
Immunoblot analysis of trichloroacetic acid-precipitated whole-cell extracts from the ZBR mutants was performed to determine if the absence of IN and RT could be attributed to premature VLP dissociation or the inability of IN and RT to remain associated with the VLP. These immunoblots were identical to those shown in Fig. Fig.2,2, indicating that the absence of IN and RT from VLPs was not a VLP-specific observation (data not shown).
To further address whether the absence of IN in ZBD mutant VLPs was due to IN instability, inefficient proteolytic processing, or a combination of both defects, we performed a pulse-chase immunoprecipitation of in-H22A and a glucose chase of GAL-induced recombinant in-H22A and in-C58A. These results (see Fig. S2 and S3 in the supplemental material) indicated that the mutant VLP proteins were not processed correctly by the Ty1 PR and that mutant INs were stable for less than 90 min when expressed independently. Thus, the absence of Ty1 IN and RT resulted from both protein instability and proteolytic processing defects.
A critical step in Ty1 transposition is the RT-mediated production of the element cDNA. The level of cDNA can be assayed by Southern blotting using a probe specific to a marked region of the element (Fig. (Fig.3A)3A) and normalizing the signal to that of a chromosomal gene. The ZBR mutants in-H22A and in-C58A exhibited an eight to ninefold reduction in the amount of cDNA compared to the WT (Fig. (Fig.3B).3B). In contrast, the hydrophobic residue mutants in-I28A, in-L32A, in-I37A, and in-V45A exhibited little or no cDNA deficiency compared either to the WT or to the transposition-competent X32 mutants in-K33A and in-Y39A. Consequently, the transposition defect observed with these mutants could not be attributed to a lack of cDNA available for integration.
Structural studies indicating that hydrophobic residues play a role in IN-IN dimerization in HIV-1 (13) and HIV-2 (21, 22) led us to evaluate IN-IN interactions by GST pulldown experiments with the transposition-defective X32 mutants, in-I28A, in-L32A, in-I37A, and in-V45A (Fig. (Fig.4).4). Two of these residues, L32A and V45A, represent residues that are highly conserved among retrotransposons and retroviruses; I37A is moderately conserved; and the IN-I28 residue is conserved among retrotransposons but not retroviruses. We also included two transposition-competent mutants, in-K33A and in-Y39A. The mutants in-I28A and in-I37A exhibited modest interaction with WT IN (Fig. (Fig.4,4, middle row) but not with IN carrying the identical mutation (Fig. (Fig.4,4, bottom row). The in-V45A mutant did not pull down WT IN efficiently, and no c-myc-in-V45A pulldown was detected. The in-L32A mutant, which was the most highly conserved but the least compromised of the four hydrophobic residues for transposition, demonstrated a WT level of interaction when coexpressed with either WT IN or its cognate IN mutant. The mutants in-K33A and in-Y39A, which demonstrated only slight reductions in transposition were not compromised for interaction with coexpressed IN carrying the identical mutations (Fig. (Fig.4,4, top row). These results suggest that the transposition defects of the in-I28A, in-I37A, and in-V45A mutants are related to the inability of the mutant INs to form oligomeric complexes, while the in-L32A mutant may either have a less severe defect or play a less critical role in transposition.
Intragenic complementation analysis provides an in vivo estimation of molecular interactions. To further evaluate the interaction defect that we observed with mutants in-I28A and in-V45A in the GST pulldown experiments, we assayed these mutants for interactions with two previously characterized IN mutants, the catalytic mutant in-2600 (45) and the C-terminal nuclear localization signal (NLS) mutant in-K596,597G (35, 48) (Fig. (Fig.5).5). Quantitative transposition experiments revealed a twofold increase in transposition when in-I28A was coexpressed with the in-2600 mutant compared to that of in-I28A coexpressed with identically mutated IN, although this difference was not statistically significant. However, when the in-I28A mutant and the NLS mutant in-K596,597G were coexpressed, transposition increased 10-fold. Similar results were observed with the in-V45A mutant in that transposition increased 2.5-fold when coexpressed with the in-2600 mutant, but unlike the in-I28A mutant, the transposition increase was the same when in-V45A was coexpressed with the NLS mutant. These results support our observations from the GST-IN pulldown experiments, suggesting that the inability of in-I28A and in-V45A mutants to self-interact leads to transposition defects. In addition, the transposition level observed when in-I28A and in-V45A were coexpressed was not significantly different than that of each mutant coexpressed with its cognate mutant partner, suggesting that the functional defects are similar or that spatial constraints prevent interaction between these two X32 mutants. ZBR mutant in-H22A, although not trans-dominant negative, did not interact with either in-2600 or the in-K596,597G NLS mutant (see Table S1 in the supplemental material).
We have previously characterized concerted integration products resulting from the IN-mediated insertion of a linear donor molecule into a supercoiled plasmid target (46). Concerted integration products result from the coupled joining of two donor molecules to the plasmid target (bimolecular concerted integration), leading to the linearization of the donor-target product, or from the insertion of both ends of a single donor into a plasmid target (RFII). Nonconcerted donor-target complexes consisting of only one end of a single donor integrated into the target also represent integration products observed in this assay (Fig. (Fig.6A6A).
As the X32 mutants exhibited stable IN in VLPs, we carried out the concerted integration assay with purified VLPs having the X32 mutations in-L32A, in-K33A, and in-V45A as well as the WT (Fig. (Fig.6).6). The mutants in-L32A and in-V45A, which were defective for transposition, showed significantly reduced (in-L32A) or absent (in-V45A) detectable in vitro integration products. The mutant in-K33A yielded an integration product, although less than observed with WT VLPs. We also performed an in vitro integration assay with synthetically stabilized ZBR mutants in-H22A and in-C58A (see the Methods section in the supplemental material and Fig. S4 in the supplemental material). No integration products were detected with either mutant.
The strict conservation of the N-terminal H-X3-7H-X23-32C-X2-C and the central catalytic DxD35E motifs in retroviral and retroelement INs suggests that these motifs carry out similar functions in the retroelement and retroviral life cycles. However, a comparison between the roles of the N-terminal histidine/cysteine ZBRs in retroviruses and LTR retrotransposons has been limited due to the lack of information regarding the role of the IN ZBR in retrotransposons. Moreover, the intervening Ty1 IN X23-32 sequence contains several residues that are conserved among retroviruses and retrotransposons, suggesting similar and/or important functions, but this region has not been extensively scrutinized by mutagenesis for defects in integration, retroviral infectivity, or virus production.
The transposition defect exhibited by the four X32 mutants in-I28A, in-L32A, in-I37A, and in-V45A, as well as in-I16A, suggested a function related to the hydrophobic nature of these residues, as alanine substitution for nucleophilic, acidic, basic, or aromatic residues resulted in WT levels of transposition. Additional characterization of these transposition-compromised mutants revealed WT levels of IN, RT, and fully processed Gag in VLPs and WT levels of Ty1 cDNA, but there were deficiencies in processes that require the interaction of IN molecules, such as GST pulldown and intragenic complementation, as well as activity in an in vitro integration assay. The in-L32A mutant, representing the most highly conserved residue in this region among retrotransposons and retroviruses, exhibited the least severe transposition defect. Additionally, this mutant appeared to retain the capacity to interact with both WT and other in-L32A molecules but was still profoundly deficient in an in vitro integration assay. With this possible exception, these results suggest that the determinant for Ty1 IN-IN interaction resides, at least in part, in the N-terminal region of IN and is comprised of hydrophobic residues.
In the intragenic complementation analysis, the mutants in-I28A and in-V45A demonstrated nearly normal levels of transposition when coexpressed with WT IN. Although this result seems contradictory to the GST pulldown experiments, especially in the case of in-V45A, it may reflect a difference between the in vivo and in vitro integration reactions. In the in vivo complementation analysis, WT IN should be able to form WT-WT multimers and carry out integration even in the presence of a mutant IN, regardless of the mutation. With IN mutants that are defective in IN-IN interaction, this activity may, in fact, be enhanced as the defective IN molecules fail to occlude interaction sites on WT molecules, unlike the in-2600 and in-K596,597G mutants, which are WT with respect to their N termini. A prediction of this hypothesis is that in-I28A or in-V45A coexpressed with in-2600 should exhibit similar levels of complementation as in-2600 coexpressed with itself, since the IN N-terminal mutant is refractory in the reaction. A similar prediction could also be made for the IN N-terminal mutant coexpressed with in-K596,597G. Comparison of in-V45A plus in-2600 with in-2600 plus in-2600 and of in-V45A plus in-K596,597G with in-K596,597G plus in-K596,597G (Fig. (Fig.5)5) indicates that this prediction holds, not only for this example but also for in-I28A paired with either in-2600 or in-K596,597G.
Although retroviral integration requires homomeric IN complexes (7), the interacting domains vary among retroviruses. For Rous sarcoma virus and avian sarcoma virus INs, the multimerization determinants have been mapped to the core and C-terminal domains (2, 12, 6). For HIV IN mapping, the multimerization domain is not straightforward, as all three domains assume a dimer conformation in solution (17, 20, 21, 38, 52). However, structural studies indicating that specific hydrophobic residues in the N termini of HIV-1 and HIV-2 IN, including four in the X27 sequence, are found at the IN dimer interface support our hypothesis that Ty1 IN-IN interaction relies on specific hydrophobic residues in the N terminus. As our analysis was confined to the region between IN-I16 and IN-I60, residues outside these boundaries may contribute to intermolecular interactions, but our observation that mutants in-I51A, in-L59A, and in-I60A demonstrated WT transposition levels suggested that downstream residues were not involved in interactions, and the alignment of Ty and retroviral INs (Fig. (Fig.1)1) revealed no remarkable similarities N terminal to I16. Interestingly, another hydrophobic residue (position 46 in Fig. Fig.1,1, the residue following Ty1 IN V45) is highly conserved among retroviral INs but not in retrotransposons. Although this residue was not implicated in IN-IN interactions (13, 21, 22), the degree of conservation invites evaluation for other retroviral defects.
In addition to similarity, we also based our choice of candidates for mutation on the inclusion of different classes of amino acids with the expectation that other IN functions might be attributed to this region. However, our results revealed only a minor, if any, reduction in transposition efficiency with these mutants, suggesting that multimerization is the primary, and perhaps the only, function of this sequence.
A mutagenesis study of the charged residues in X32 of Mo-MuLV demonstrated that alanine substitutions at K66/68 and K73 were impaired for in vitro 3′ processing and strand transfer and in vivo infectivity, but these mutants retained their ability to form dimers (65). By comparison, we examined one charged residue, K33, which did not show a significant decrease in transposition when mutated. A comprehensive survey of HIV-2 IN examined the 3′ processing, strand transfer, and disintegration activities of 36 IN mutants including S39, which is adjacent to the ZBR residue C40 (57). All the in vitro activities of this mutant were similar to those of the WT.
Our analysis of the ZBR mutants in-H17A, in-H22A, in-C55A, and in-C58A was limited by the instability and processing defects exhibited by these mutants whether expressed in the context of element or expressed independently as a recombinant protein. Although the instability of the recombinant protein could be partially relieved by the addition of an N-terminal GST moiety, this synthetically stabilized protein was still inactive in GST pulldown and in vitro integration assays.
The high degree of conservation of an N-terminal histidine/cysteine motif among retroviral INs has fostered numerous studies of the effects of mutation or deletion of these residues. In vitro experiments with purified recombinant mutant INs and linear DNA substrates representing both donor and target molecules have shown that, in general, the steps most affected by these mutations are the initial 3′ dinucleotide cleavage and the subsequent strand transfer. The reverse disintegration reaction is usually unaffected (7). The abundance of in vitro data obtained with purified recombinant retroviral INs underscores the difference in stability of retroviral and Ty1 INs. In addition, our observation that the instability of the downstream RT and defects in protein maturation by the Ty1 PR are associated with N-terminal ZBR mutations suggests that this region plays different roles in transposition than the HHCC domain plays in retroviral infectivity, as these extended effects on retroviral RT stability and protein maturation are not observed in retroviruses. The difference in retroviral and retrotransposon IN ZBR effects may reflect the difference in pol organization, as the IN and RT are reversed relative to each other in retroviral and Ty/copia elements. This notable difference is particularly intriguing in light of the report by Bizub-Bender et al. (4) that experiments with monoclonal antibodies to different epitopes of HIV-1 IN indicate an association between the N and C termini. If a similar intramolecular interaction occurred with Ty1 IN after Gag/Pr cleavage but before PR/IN processing, this interaction could position PR near the IN/RT cleavage site. Alternatively, if the interaction occurred after PR/IN cleavage, the folded IN might recruit PR to the IN/RT cleavage site. In retroviruses, the IN N-C interaction would not place either the IN N terminus or PR near a cleavage site, due to IN being the most distal protein in the retroviral Pol arrangement. Whether an IN N-C terminus interaction occurs is not known, but this possibility illustrates how the different pol arrangement of Ty1 could affect transposition. In fact, we constructed a Ty1 element in which the positions of IN and RT were reversed while preserving all the PR cleavage sites (43). Expression of the resulting “flipped” element revealed no Pol proteins and a partially processed Gag similar to that of the HHCC mutants reported here (S. P. Moore and D. J. Garfinkel, unpublished results). This result supports the hypothesis that the order of the Pol proteins in the Ty1/copia family is functionally significant and that these elements have evolved in a manner that ensures the maintenance of this order.
While our assays do not permit us to directly address the influence of IN on RT stability, a required postproteolytic interaction between Ty1 IN and RT has been reported (51, 62, 61) as well as several observations of retroviral IN-RT interactions (31, 56, 64, 68). Wilhelm and Wilhelm (62) have proposed that Ty1 RT is a component of a heteromeric complex that includes IN and element cDNA in a preintegration complex and that RT remains associated with IN even after nuclear import of the preintegration complex. Their experiments included an IN-RT PR cleavage site mutant in which IN residues 45 to 520 are also deleted, removing both cysteines of the ZBD and the entire central catalytic core domain. This mutant produces an 80-kDa fusion protein which is detectable by immunoblotting, and VLPs containing this fusion possess RT activity. This result is interesting compared with our observation that mutation of either of these cysteines leads to both IN and RT instability, but the deletion of an additional 462 residues beyond the ZBD in their construct precludes an extensive comparison.
Another interesting observation is that in HIV-1 a premature stop codon at the beginning of IN, or at position 107, results in decreased virion production as measured by release of capsid protein into the culture supernatant, but this effect is reversed when PR is inactivated either by mutation or by a PR inhibitor (9). Considering the even closer proximity of the IN ZBD to PR in Ty1 than in retroviruses, it is possible that IN mutations that disrupt the conformation of this region of IN also lead to aberrant PR function prior to PR cleavage of the PR-IN-RT polypeptide. Although it would be possible to examine the effect of PR inactivation on the ZBR mutants, a direct comparison could not be made as Ty1 VLPs are not released into the culture supernatant, and the HIV-1 mutant whole-cell lysates did not exhibit the defective phenotype (9).
Until the structure of Ty1 IN is solved, it is not possible to definitively determine the roles of specific amino acids in IN structure or to propose structure-function relationships. However, our results indicating that I28 and V45 mutants are compromised for GST pulldowns when coexpressed with their cognate mutant IN but less so when coexpressed with WT IN, and our transposition complementation data with mutations in other domains of IN suggest that the hydrophobic residues function in IN-IN interaction, while the histidine/cysteine residues profoundly affect one or more early steps in retrotransposition, resulting in a subsequent failure in the cascade of events required for reverse transcription, proteolytic processing, and integration. Overall, our work illustrates the importance of the N-terminal domain of IN for Ty1 transposition, delineates different roles played by the histidine/cysteine residues compared to the hydrophobic X32 residues, and highlights the structural similarities and differences in retroviral and retrotransposon mechanisms.
We thank Timothy Veenstra and Li-Rong Yu for protein sequencing analysis and Tammy Schroyer for preparation of figures.
This work was sponsored by the Center for Cancer Research of the National Cancer Institute, Department of Health and Human Services.
The contents of this publication do not necessarily reflect the views or policies of the Department of Health and Human Services, and mentions of trade names, commercial products, or organizations does not imply endorsement by the U.S. government.
Published ahead of print on 1 July 2009.
†Supplemental material for this article may be found at http://jvi.asm.org/.