DNA-Binding Activity of Adeno-Associated Virus Rep Is Required for Inverted Terminal Repeat-Dependent Complex Formation with Herpes Simplex Virus ICP8

Abstract

Herpes simplex virus (HSV) helper functions for (AAV) replication comprise HSV ICP8 and helicase-primase UL5/UL52/UL8. Here we show that N-terminal amino acids of AAV Rep78 that contact the Rep-binding site within the AAV inverted terminal repeat (ITR) are required for ternary-complex formation with infected-cell protein 8 (ICP8) on AAV single-strand DNA (ssDNA) in vitro and for colocalization in nuclear replication domains in vivo. Our data suggest that HSV-dependent AAV replication is initiated by Rep contacting the AAV ITR and by cooperative binding of ICP8 on AAV ssDNA.

TEXT

Asubset of six out of seven herpes simplex virus (HSV)-encoded replication functions was shown to provide helper activity for productive replication of adeno-associated virus (AAV). Four of these, the single-strand-DNA (ssDNA)-binding protein infected-cell protein 8 (ICP8) (UL29) and the heterotrimeric helicase-primase complex UL5/UL8/UL52, constitute the minimal four-protein complex of helper functions for AAV DNA replication (32). Further analysis showed that the helicase UL5 and the primase UL52 are primarily needed as structural components of replicative structures, able to recruit AAV Rep and the AAV genome for the initiation of AAV DNA replication (10, 27).

AAV type 2 contains a 4.7-kb linear single-stranded DNA genome flanked by 145-bp inverted terminal repeats (ITR), which comprise the origins of replication where AAV Rep78 and its C-terminal variant Rep68 bind to the Rep-binding site (RBS). By means of its ATP-dependent helicase activity, Rep unwinds the ITR, and its endonuclease activity leads to nicking of the adjacent terminal resolution site (3, 5, 14, 28). The role of Rep78/68 as ori-binding proteins and initiators of AAV DNA replication was analyzed previously using HSV as a helper virus. In coinfection experiments, Rep78 was found to colocalize to HSV ICP8 in nuclear replication compartments in a manner dependent on ITR-flanked single-stranded AAV genomes. Furthermore, direct AAV ssDNA-dependent interaction of purified ICP8 and Rep78 was shown in vitro (10). ICP8 displays high-affinity, cooperative binding to ssDNA (1, 7, 20). Rep78/68 also displays some ssDNA-binding activity (16, 19, 36) but preferentially binds to the double-stranded RBS within the hairpin-shaped AAV ITR (12). We have shown before that ssDNA devoid of AAV ITRs displayed severely reduced ternary-complex formation with wild-type Rep78/68 and ICP8 (10).

In this study, we aimed to identify the Rep domain(s) (Fig. 1A and B) responsible for the interaction with ICP8 on the AAV genome. Ternary-complex formation between Rep78, HSV ICP8, and AAV DNA was analyzed by in vitro pulldown assays with purified glutathione S-transferase (GST)-tagged ICP8 and in vitro-translated ³⁵S-labeled Rep proteins as described before (10). Plasmid-excised, full-length AAV wild-type genomes were gel purified and either used directly as linear dsDNA templates or heat denatured to adopt ssDNA conformation. Subsequent snap-cooling on ice ensured reassociation of the hairpin-shaped ITRs flanking the single-stranded AAV genome (9). Rep78 and ICP8 interact to a low degree in the absence of DNA (Fig. 1C; Fig. 1D, lane 3) without enhancement upon addition of double-stranded full-length AAV DNA (Fig. 1D, lane 5). In contrast, single-stranded AAV genomes significantly enhanced the interaction between Rep78 and ICP8 up to 10-fold (Fig. 1C; Fig. 1D, lane 7).

Fig 1.

Rep78 domains involved in AAV ssDNA-dependent interaction with HSV ICP8. (A) Individual domains of Rep78 are shown as rectangles. The N terminus unique to Rep78 and Rep68 (amino acids 1 to 225) is gray, the central region common to all four Rep proteins is white, and the C terminus specific to Rep78 and Rep52 derived from unspliced mRNAs is black. Domains involved in functional activities are represented by solid black lines. Approximate amino acid positions are shown as boundary marks as compiled from the literature. The respective references are given in the text. (B) Plasmid constructs for Rep78 and mutants thereof. The mutated amino acids and the amino acid positions are indicated. (C) AAV DNA-dependent interaction of AAV Rep78 with HSV ICP8 was analyzed by in vitro GST pulldown assays. In vitro-translated ³⁵S-labeled Rep78 and its mutants were incubated with GST-ICP8. Assays were performed either in the absence of DNA, in the presence of double-stranded wild-type AAV-2 DNA, or in the presence of heat-denatured, single-stranded wild-type AAV-2 DNA. Bound Rep proteins were analyzed by autoradiography in sodium dodecyl sulfate-polyacrylamide gel electrophoresis (SDS-PAGE) gels. The results of phosphorimager quantifications in three to five independent experiments are given as means ± standard deviations. Significances were calculated for ssDNA versus dsDNA by Student's t test. Significance levels are indicated as follows: ***, P < 0.001; **, P < 0.01; *, P < 0.05. (D) Representative autoradiogram and results of five independent in vitro pulldown experiments either with GST alone or with wild-type Rep78, performed as described for panel C. The lane labeled “Rep78input” was loaded with 50% of the amount of Rep included in the binding assay. (E) Results of an experiment performed as described for panel C with AAV plasmid-excised dsITR or hairpin-structured ssITR (nt 1 to 181).

In order to delineate the domain(s) involved in ternary-complex formation, a series of Rep mutants (Fig. 1B) were assessed. As shown in Fig. 1C, ssDNA-dependent complex formation was maintained with the single-amino-acid exchange mutant RepK340H. RepK340H is impaired in ATP-binding precluding DNA helicase activity in vitro (3, 4, 35) and AAV DNA replication in vivo (2, 17). The ability of RepK340H to bind to the hairpin-structured AAV ITR (4, 23, 24, 33) and its site-specific endonuclease activity are retained (5). Likewise, Rep78 mutants with deletions of the C terminus encompassing the nuclear localization signal (M1/481) retained the ability to interact with ICP8 and AAV ssDNA in vitro (Fig. 1C). In contrast, N-terminal-deletion mutants of Rep, shown to be replication negative in vivo (13, 17), were entirely deficient for ternary-complex formation in vitro (Fig. 1C, Rep52 and Met172). The N terminus of Rep78/68 comprises the DNA-binding domain that mediates site-specific binding to the RBS within the AAV ITR and also the domain for endonuclease activity (Fig. 1A). In addition, endonuclease requires binding to the RBS to exert its activity (5). To differentiate between the two activities, the exclusively endonuclease-deficient mutant RepY156F (5) was generated, which proved to retain ternary-complex formation (Fig. 1C). Rep mutants with exclusive ITR-binding defects were designed as follows. Based on crystal structure analysis of the N terminus of AAV-5 Rep78 bound to the RBS of the AAV-5 ITR (12), the critical amino acids of AAV-2 Rep78 that contact the ITR were aligned to positions R107, R136, and R138 (11, 12). These amino acid positions were mutated individually or in combination, as shown in Fig. 1B. In pulldown assays, all R-to-A exchanges at position 107 (R107A, R107A/K136A, and R107A/R138A) led to a complete loss and the mutations K136A and R138A to a severe reduction in ternary-complex formation (Fig. 1C). To further narrow down the presumed site of interaction, the experiment was repeated with the isolated AAV ITR (nucleotides [nt] 1 to 181) in a double-stranded conformation, or a hairpin-structured, partially single-stranded conformation (Fig. 1E). Similar to the entire AAV-2 genome, Rep78wt, RepK340H, and RepY156F retained the capacity for ssDNA-dependent complex formation with ICP8, whereas the RBS binding-deficient mutant RepR107A/R138A lost this activity. Together, these data show that the ability of Rep78 to directly contact the AAV ITR is required for ternary-complex formation with ICP8.

To analyze whether in vitro ternary-complex formation is reflected in vivo by the ability of AAV-Rep and HSV-ICP8 to colocalize in nuclear replication domains, their distribution was analyzed by confocal microscopy. We had previously demonstrated colocalization of Rep and ICP8 upon coinfection of wild-type AAV and HSV (10) and upon cotransfection of plasmids coding for wild-type AAV-2 and for the minimal set of HSV helper proteins, consisting of ICP8 and the helicase-primase complex UL5/UL8/UL52 (27). Full-length AAV-2 plasmids were generated that expressed Flag-tagged versions of Rep78wt, RepK340H, RepY156F, or N-terminal amino acid exchange mutants. In the presence of helper virus, the plasmids mediated comparable Rep expression but, with the exception of Rep78wt, had lost DNA replication properties (Fig. 2A and B). Subcellular localization of Rep and ICP8 was quantified by confocal microscopy 40 h after cotransfection of expression plasmids for Rep and the four HSV helper functions as described before (27). When transfected alone, Rep and all mutants thereof displayed a homogenous nuclear distribution pattern (data not shown). In the presence of HSV replication proteins, Rep78wt followed the punctate distribution pattern of HSV replication foci and colocalized to ICP8 (Fig. 2C and D), as described before (27). In contrast, the DNA binding-deficient mutants RepR107A/K136A and RepR107A/R138A hardly ever colocalized to ICP8 above threshold levels (Fig. 2C and D). The helicase-deficient mutant RepK340H, despite its ability to form the ternary complex in vitro, never colocalized to ICP8, whereas the helicase-proficient mutant RepY156F occasionally colocalized to ICP8 in HSV replication foci (Fig. 2C and D).

Fig 2.

In vivo analysis of AAV DNA-dependent nuclear colocalization of Rep78 mutants and ICP8. (A) Analysis of AAV DNA replication by Rep78 and mutants thereof. HeLa cells were transfected with AAV-2 constructs expressing Rep78 or mutants thereof, as described in the text (data not shown for RepY156F). Sixteen hours later, cells were infected with helper virus (Ad-2) and were harvested 24 h postinfection. Low-molecular-weight DNA was prepared by Hirt extraction and digested with DpnI. Equal amounts of DNA were separated on agarose gels, and Southern blots were hybridized with a ³²P-labeled DNA probe spanning the AAV cap. Replicated AAV DNA was visualized by autoradiography. RF1 and RF2, monomeric and dimeric replicative forms. Similar results were obtained with helper HSV (data not shown). (B) Rep expression was analyzed by Western blot analysis of cell extracts processed in parallel to the Southern blot in panel A. Rep expression was detected by the anti-Rep monoclonal antibody 303.9. (C) AAV DNA-dependent nuclear colocalization of Rep78 and ICP8. BHK cells grown on coverslips were cotransfected with AAV plasmids expressing Flag-tagged Rep78 wild-type protein or mutants, together with plasmids for the minimal set of HSV helper genes, the primase/helicase complex UL5/UL8/UL52 and the ssDNA binding protein ICP8 (UL29). Cells were fixed and permeabilized by formaldehyde-Triton treatment 40 h later and stained with antibodies against HSV ICP8 and the Flag tag of AAV Rep, followed by fluorophore-labeled secondary antibodies, as described previously (27). Cells were analyzed by confocal microscopy with a Zeiss LSM 510 microscope. The images represent cross sections of 0.8 μm. Anti-ICP8 reactivity is displayed in green, while anti-Flag (Rep) reactivity yields red. Merged foci (yellow) indicate colocalization. (D) Quantification of AAV DNA-dependent nuclear colocalization of Rep78 and ICP8 displayed in panel C. To determine the extent of colocalization, the correlation coefficient (R) of green (ICP8) and red (Rep) fluorescence above a fixed background value was determined with the help of the Zeiss LSM 510 software for a total of 15 cells for each individual construct, and values are means ± standard errors of the means (*, P < 0.05; **, P < 0.01; ***, P < 0.001). (E) Quantification of colocalization of Rep78 and ICP8 in U2OS cells transduced with a monomeric rAAV subtype 2 vector at an MOI of 1 × 10⁴ genomic particles/cell as a source of AAV ssDNA. Cells had been transfected with plasmids expressing Rep and the minimal HSV helper genes. Colocalization was quantified in six individual cells per construct, as described for panel D. (F) The initial steps of Rep-dependent initiation of AAV DNA replication at the AAV ITR are displayed in consecutive order: DNA binding of Rep to the RBS followed by Rep-dependent helicase activity, leading to ssDNA strands, followed by the Rep-dependent endonuclease step. The degree to which Rep78wt and the mutants support these activities are indicated by arrows.

To test in vivo colocalization on authentic AAV ssDNA, U2OS cells were infected with recombinant AAV (rAAV) vectors at a multiplicity of infection (MOI) of 1 × 10⁴ genomic particles/cell after cotransfection with plasmids for Rep and the four HSV helper genes. The data displayed in Fig. 2E confirmed the drop in colocalization to ICP8 of the N-terminal mutant RepR107A/R136A compared to that of Rep78wt (P < 0.01). In contrast, RepY156F showed a high and significant degree of colocalization to ICP8 (P < 0.05) just slightly below that of Rep78wt, whereas RepK340H never colocalized with either plasmid- or virus-derived AAV template DNA (Fig. 2D and E). Although both RepK340H and RepY156F can bind to the AAV ITR, only RepY156F unwinds it to expose ssDNA (5), but it cannot proceed with DNA replication due to its endonuclease defect (Fig. 2F). Obviously, the unwound AAV ITR exposes sufficient ssDNA for ICP8 binding. RepK340H, which is entirely defective in ITR unwinding only in vitro, binds to ssDNA templates (Fig. 1) but is unable to generate ssDNA for in vivo ternary-complex formation.

In summary, colocalization of Rep and ICP8 in nuclear replication domains depends on the ability of Rep to bind and unwind the AAV ITR, generating ssDNA regions. The in vitro data show that neither helicase nor endonuclease activities of Rep are needed as such for ternary-complex formation as long as AAV ssDNA is present. In addition, the reduced in vivo colocalization of ICP8 and certain Rep78 mutants likely reflects their inability to support AAV DNA replication to generate and amplify sufficient AAV ssDNA templates (Fig. 2F).

HSV ICP8 is characterized by highly cooperative and DNA sequence-independent ssDNA-binding (21) with an apparent binding constant (K_α) for monomeric ICP8 on ssDNA in the range of 1 × 10⁷ M⁻¹ (1, 7, 20). Electron microscopy confirmed the highly cooperative but unspecific nature of ICP8 binding to ssDNA (18, 25) whereas Rep molecules on the AAV ssDNA genome bind exclusively to the hairpin-shaped ITR that contains the RBS (14, 15, 34, 37). For Rep78/68 DNA-binding affinities to the isolated RBS of up to 8 × 10¹⁰ M⁻¹ were calculated (3, 8, 22), while Rep68 also binds to unrelated ssDNA (16), though with a lower binding affinity, around 2 × 10⁸ M⁻¹ (19, 36). Obviously, the affinity of Rep for the AAV ITR is roughly 2 orders of magnitude higher than its affinity for ssDNA. These data, taken together with the above findings, suggest that Rep78/68 serves as the driving force for ternary-complex formation by binding to the RBS within the ITR. Due to the high degree of cooperativity, ICP8 quickly covers available ssDNA regions. We cannot say at present whether the two processes are coupled or take place independently. Rep domains interacting with heterologous proteins have been exclusively mapped to the C terminus (6, 8, 26, 29–31), and the ICP8 domain for interactions with heterologous proteins was described as separate from that engaged in cooperative ssDNA binding (21). In an alternative scenario, formation of the ternary complex could therefore be initiated by protein-protein interaction of ICP8 and Rep with subsequent binding to the AAV genome. Our data extend previous evidence that Rep78 serves as origin-binding protein on the AAV ITR (27). In analogy to the HSV ori-binding protein (UL9), AAV Rep78 binds and unwinds the AAV ITR and recruits ICP8, the HSV helicase-primase complex, and additional replication factors to initiate AAV DNA replication.