INTRODUCTION

Top Abstract Introduction Materials & Methods Results Discussion References

Adeno-associated virus (AAV) is a human parvovirus that, for its replication, usually requires that its host cell is coinfected with a helper virus, usually an adenovirus or herpesvirus. Infection with AAV in the absence of a helper virus can result in integration of the viral genome into a specific site on chromosome 19. The provirus then enters a latent state until it is reactivated by superinfection of the host cell with a helper virus or by genotoxic stress (for a review, see reference 5). AAV appears to be nonpathogenic in humans; in fact, seropositivity to AAV has been shown epidemiologically to be a protective factor against the development of carcinoma of the uterine cervix (14, 36). Recent data suggest that the cervical neoplasia-associated human papillomaviruses may act as helper viruses for AAV in vivo (53).

The AAV type 2 (AAV2) genome comprises a single-stranded DNA molecule of approximately 4.7 kb which can be divided into early (Rep) and late (Cap) regions (49a). The genome is bounded by inverted terminal repeats (ITRs) which provide the viral origins of replication. The Rep region encodes four related proteins through usage of two promoters and differential splicing. The p5 promoter drives the expression of Rep78 and Rep68, which differ at their C termini due to differential splicing. An internal promoter, p19, directs the synthesis of Rep52 and Rep40, which correspond to the C-terminal regions of Rep78 and Rep68, respectively. The Rep68/78 proteins are essential for the replication of AAV and possess ATP-dependent DNA helicase, site-specific endonuclease, and viral origin of replication binding activities (2, 23, 38, 58). Rep68/78 proteins also bind to the chromosome 19 integration sites and mediate the integration of AAV into the chromosome (27, 35, 57). The Rep40/52 proteins do not bind DNA directly but appear to be required for the accumulation and encapsidation of single-stranded genomes.

Rep proteins also have roles in regulating AAV gene expression. In addition to binding the viral ITRs, a Rep68/78-specific binding site occurs in the p5 promoter (37). In the absence of a helper virus infection, Rep synthesis is tightly repressed; however, in the presence of a helper, both Rep and Cap gene expression is induced. This switch appears to be integrated through the activities of the Rep proteins (32). In the absence of a helper virus, Rep68/78 binding to the p5 promoter site may inhibit transcription by physical occlusion of promoter elements (30). Rep can also inhibit transcription from the p19 promoter in the absence of direct binding to promoter-proximal elements, in a manner that requires an intact Rep nucleotide binding site (22, 29). This suggests that Rep suppression of the p19 promoter is probably mediated through protein-protein interactions. During productive infection, Rep acts as a powerful transactivator of the p19 and p40 promoters, and displacement of Rep68/78 from the p5 promoter binding site may act to derepress this promoter (30, 32, 45). Rep40/52 proteins may act to derepress the p5 promoter by entering into complexes with the larger Rep proteins, reducing their DNA binding activities (45). Rep78, -68, and -52 proteins have the capacity to down regulate gene expression from a variety of heterologous promoters, most of which do not contain sequence homologies with the p5 and ITR Rep binding sites (18, 19, 22, 31, 42, 48). It appears that DNA binding of Rep to specific sites is unlikely to account for transcriptional repression in such cases, and this again suggests that Rep proteins can mediate transcriptional silencer effects via protein-protein interactions.

AAV has also been shown to inhibit oncogenic transformation in vitro, and in many cases this property has been mapped to the Rep proteins. Cotransfection of Rep78-expressing plasmids inhibited transformation by adenovirus E1a+ras, simian virus 40, bovine papillomavirus, and human papillomavirus (17, 19, 25). AAV DNA, when integrated into HeLa cells, resulted in decreased growth, increased anchorage dependence, and enhanced sensitivity to growth factor withdrawal, tumor necrosis factor alpha, or genotoxic agents (51, 52). There is evidence for AAV-induced cell cycle blocks at both the G₁ and G₂ phases of the cell cycle (3, 62). Cells arrested by AAV contain hypophosphorylated Rb and express elevated levels of p21/waf-1/cip-1 protein, which is induced in a p53-independent manner. It is not yet clear whether the up regulation of p21 is a function of the Rep proteins (16). The molecular mechanisms and cellular targets of oncosuppression by AAV are not well understood.

The helper virus functions which are required for AAV replication have been identified for adenovirus and herpes simplex virus (HSV) type 2. The adenovirus E1a, E1b, E2a, E4, and VA genes are sufficient for helper virus function (60). For HSV type 1 the ICP4 transactivator, DNA polymerase, ICP8, origin binding protein, and helicase-primase complex have been implicated (40, 56). During its replication cycle, AAV can inhibit production of its helper virus (4, 8, 9). The mechanisms underlying interference with helper virus replication have not been determined. One report indicated that posttranscriptional down regulation of adenovirus E1b may be involved (43).

The various activities of AAV Rep are likely mediated in part through contacts between the Rep proteins and host cell proteins. It has been shown previously that Rep can interact with itself to form higher-order structures, probably hexamers (20, 49). Recently it has been shown that Rep can bind to the transcription factor Sp1, and this may in part account for its ability to regulate transcription of AAV and heterologous promoters (21, 45). Rep has also been found to associate with the high-mobility group 1 protein, an interaction which enhances the DNA binding, helicase, and endonuclease activities of Rep (11).

In this paper we describe the use of the yeast two-hybrid system to identify cellular proteins that interact with AAV2 Rep78. We found that Rep78 but not Rep68 interacts with PRKX, a recently described homolog of the cyclic AMP (cAMP)-dependent protein kinase A (PKA) catalytic subunit (PKA_c). Subsequently, we found that Rep78 also interacts with PKA_c itself. Interaction with Rep78 was found to inhibit the kinase activities of both PRKX and PKA_c. Rep78, but not Rep68, was able to block the induction of CREB-dependent transcription in HeLa cells. These observations suggest that AAV may utilize Rep to subvert cAMP-dependent regulatory pathways in infected cells. The inhibition of these pathways may influence the balance between AAV and helper virus replication.

	MATERIALS AND METHODS

Top Abstract Introduction Materials & Methods Results Discussion References

Construction of plasmid vectors. A maltose binding protein (MBP) fusion construct containing AAV2 Rep68 (pMALcRep68) and a clone containing the entire genome of AAV2 (pAV2) were gifts from Ken Berns. pGBT9-Rep68 was constructed by cloning the Rep68 open reading frame (ORF) from pMALcRep68 into pGBT9 (Clontech) by using the EcoRI and XhoI sites present in both vectors. pGBT9-Rep78 was generated by subcloning an AatII-XhoI fragment comprising nucleotides (nt) 1868 to 2233 of AAV2 (numbering is as for the sequence under GenBank accession no. J01901) from pAV2 into AatII-XhoI-digested pGBT9-Rep68. pGBT9-Rep78-Cterm was generated by eliminating an EcoRI fragment (nt 321 to 1767) from pGBT9-Rep78.

Yeast two-hybrid system library screening. Two-hybrid screening was done in Saccharomyces cerevisiae by using the Clontech Matchmaker system. Briefly, strain HF7c was transformed with pGBT9-Rep78, and the resulting clone was retransformed with a previously amplified HeLa cell cDNA library in the vector pGAD-GH (Clontech). Transformant yeast cells were selected on Trp-Leu-His dropout medium containing 20 mM 3-aminotriazole (3-AT). Approximately 400 colonies were obtained from a screening of approximately 10⁶ independent clones. All 400 colonies were assayed for -galactosidase activity by using a filter assay. Plasmid DNA from positive clones was rescued and used to transform Escherichia coli XL-1 Blue (Stratagene). Clones from the pGAD-GH library were selected, and the plasmids were retransformed into yeast strain SFY526 along with pGBT9-Rep78 or pGBT9. Resultant clones were assayed for -galactosidase activity by using a quantitative solution-based ONPG (o-nitrophenyl--D-galactopyranoside) assay as described by the manufacturer (Clontech). Plasmids from clones which were positive in this assay were analyzed further by DNA sequencing.

pMal-Rep pull-down experiments. MBP (New England Biolabs) and GST (Pharmacia) fusion proteins were expressed in E. coli BLR (Novagen), and proteins were purified according to the manufacturers' instructions. However, for the production of GST-C-terminal protein, the incubation temperature after IPTG (isopropyl--D-thiogalactopyranoside) induction was 30°C. Fusion proteins were immobilized on amylose- or glutathione-Sepharose beads as appropriate, and the protein concentrations were estimated by the Bradford protein assay (Pierce) and Coomassie blue staining of sodium dodecyl sulfate (SDS)-polyacrylamide gels.

PKA assays. One hundred units of purified PKA_c (Promega) was incubated with 0.5 to 1 µg of MBP or GST fusion proteins, immobilized on beads as described above, in a final volume of 50 µl in PBS at 4°C for 10 min, followed by extensive rinsing with PBS. Immobilized PKA activity was then determined by using a kemptide phosphorylation assay (PepTag; Promega) as follows. The beads were resuspended in 25 µl of kinase assay buffer containing kemptide substrate, followed by incubation for 30 min at room temperature. The reaction was terminated by boiling, and the products were examined by 0.8% agarose gel electrophoresis. For experiments on Rep inhibition of PKA activity, the amount of PKA_c required to obtain approximately 50% phosphorylation of substrate peptide in a 10-min reaction was first determined empirically. The assay was then performed with this amount of PKA_c held constant, while increasing amounts of amylose- or glutathione-eluted MBP, MBP-Rep, or GST and GST-C-terminal proteins were added to the reaction mixture. Phosphorylation of the kemptide substrate was then determined as described above.

Histone H1 kinase assay. Unlabelled PRKX was immobilized on S beads as described above. Ten microliters of S beads was incubated at room temperature in kinase assay buffer (20 mM Tris-HCl [pH 7.4], 10 mM MgCl₂) containing 20 nCi of [-³²P]ATP, with or without 0.5 µg of histone H1, in a final volume of 30 µl. For autophosphorylation-inhibition assays, 0.3 or 0.6 µg of MBP or Rep-MBP fusion proteins was added prior to a 45-min incubation.

Cell culture and transfection. HeLa cells were plated at 5 × 10⁵/well on six-well plates 24 h prior to transfection. Cells were transfected with 30 µl of DOTAP (Boehringer Mannheim) per well with 1 µg of pFR-Luc, 1 µg of pFA-CREB, and 1 µg of pcDNA-PRKX, pCIS-PRKX, pFC-PKA, or pCI. Samples were additionally transfected with 2 µg of pKex68, pKex78, or pCI. Cells were harvested 48 h posttransfection and then lysed and assayed for protein content and luciferase activity. Results were expressed as normalized luciferase activity relative to that of the pFR-Luc/pCI/pCI control. The data presented were taken from a minimum of three independent experiments, each conducted in duplicate.

	RESULTS

Top Abstract Introduction Materials & Methods Results Discussion References

Identification of PRKX as a Rep78-interacting protein by using the yeast two-hybrid system. The Rep78 gene from AAV2 was cloned into the yeast shuttle vector pGBT to generate pGBT-Rep78, in which the Rep78 gene was fused to the Gal4 DNA binding domain. Transformation of pGBT-Rep78 into S. cerevisiae HF7c either alone or in combination with the Gal4 activation domain vector (pGAD424) resulted in growth on His dropout medium, indicating that Rep78 has an intrinsic transcriptional transactivation activity in this system. However supplementation of the medium with 20 mM 3-AT, an inhibitor of histidine biosynthesis, effectively suppressed the transactivation by Rep78 (data not shown), and this concentration of 3-AT was used subsequently for library screening. The vector pGBT-Rep78 was used as bait to screen a HeLa cell-derived cDNA library cloned in the vector pGAD-GH. Screening was initially done for growth on His dropout medium; this was followed by a -galactosidase filter assay. Eleven clones that were positive in both assays were isolated. In order to conduct a quantitative assessment of the putative interactions, yeast strain SFY526, which contains the lacZ gene driven by a Gal4-responsive promoter, was transformed with pGBT-Rep78 or pGBT. Both of these clones were subsequently retransformed individually with the 11 pGAD clones isolated from the library. The resulting clones were assayed for -galactosidase production by using a solution-based ONPG assay. For a positive control, a clone containing the pocket protein p130 fused to the DNA binding domain (pGBTp130) and the human papillomavirus E7 protein fused to the activation domain (pGAD-HPV16-E7) was used. As shown in Fig. 1, three of the clones (pGAD6, -12, and -162) in combination with pGBT-Rep78 produced -galactosidase at levels approximately 10-fold higher than that of the negative control (pGBT-Rep78 plus pGAD424) and comparable to that of the positive control (pGBTp130 plus pGAD-HPV16-E7). Activity of these clones was dependent on the presence of Rep78 sequences fused to the DNA binding domain, since negligible -galactosidase activity was observed when pGAD6, -12, or -162 was assayed in combination with the empty vector pGBT9 (not shown). We concluded that clones pGAD6, -12, and -162 encoded proteins which interacted specifically with Rep78.

View larger version (17K): [in this window] [in a new window]   FIG. 1.   Yeast two-hybrid assay for positively interacting clones. SFY526 yeast cells were transformed with pGBT-Rep78 and then retransformed individually with activation domain vectors pGAD6 to -319, which had previously scored positive in a LacZ filter assay. -Galactosidase production was then measured by using a quantitative, solution-based ONPG assay. Control transformations were pGBT-Rep78 plus pGAD424 (GAD vector) and pGBTp130 plus pGAD-HPV16-E7 (p130/GAD-E7).

Rep68 does not interact with PRKX. Rep68 is a spliced variant of Rep78 in which amino acids 530 to 621 are absent and are replaced with the sequence LARGHSL (Fig. 2a). Rep78 thus contains a unique C-terminal domain comprising six CXXH or CXXC motifs, which are typical of zinc finger domains found in a number of transcription factors. In order to investigate whether Rep68 could interact with PRKX, the Rep68 gene was first cloned into pGBT9 to produce pGBT-Rep68. This construct was transformed in combination with pGAD-PRKX, followed by determination of -galactosidase production (Fig. 2b). The combination of Rep78 and PRKX produced high levels of -galactosidase activity, as expected, while activity of the negative control Rep78 plus pGAD424 was low but detectable, consistent with the previously observed transactivation activity of Rep78. On the other hand, Rep68 showed no detectable -galactosidase activity in combination with either pGAD-PRKX or the pGAD424 control vector. We concluded that Rep68 does not interact with PRKX in the yeast two-hybrid system. Moreover, this showed that Rep78, but not Rep68, displays intrinsic transactivation activity in this system. The positive control comprising pGBT-p107 and pGAD-HPV16-E7 showed rather low activity compared to the Rep78/PRKX and to the p130/HPV-16-E7 (Fig. 1) combinations. This may be due to poor expression or low tolerance for p107 in this system.

View larger version (15K): [in this window] [in a new window]   FIG. 2.   Yeast two-hybrid assay to localize the interaction domain between Rep78 and PRKX. (a) pGBT-Rep78 contains the full-length Rep78 ORF for amino acids 1 to 621. In pGBT-Rep68 the intron from amino acid 529 is spliced out and the sequence LARGHSL occurs at the splice acceptor site (light grey). An EcoRI fragment extending from the pGBT cloning site to AAV nt 1767 was deleted to produce pGBT-Cterm, which encodes Rep amino acids 483 to 621. (b) Strain SFY56 yeast cells were transformed with pGBT and pGAD vectors in the combinations indicated on the x axis. -Galactosidase activity was then measured by ONPG assay. Error bars indicate standard deviations.

The C-terminal region of Rep78 interacts with PRKX. Since the interaction between Rep and PRKX was restricted to Rep78, we examined whether the Rep78 C-terminal region alone was sufficient for interaction with PRKX. A pGBT vector encoding only the C-terminal 139 amino acids from Rep78 was produced (Fig. 2a), and the resulting clone, pGBT-Cterm, was assayed for -galactosidase activity as described above. In combination with pGAD-PRKX, the Rep78 C-terminal clone showed -galactosidase activity approaching that of the full-length Rep78 clone, indicating that the Rep78 C-terminal region interacted with PRKX (Fig. 2b). pGBT-Cterm in combination with the empty binding domain vector pGAD-424 also showed clearly detectable activity. This confirmed that the C-terminal region possesses intrinsic transcriptional transactivation abilities and suggests an involvement of the zinc binding domain in this phenomenon. The levels of -galactosidase activity observed with the C-terminal domain constructs were consistently lower than those obtained with the full-length Rep78 clone (Fig. 2b and data not shown). While it is clear from these experiments that the C-terminal region of Rep78 is essential for the interaction with PRKX, regions outside the C-terminal region may contribute to the strength of the interaction.

Interaction between a pMal-Rep78 fusion protein and PRKX in vitro. In order to determine whether Rep78 and PRKX interact directly in vitro, pull-down experiments were conducted. The Rep78- and Rep68-coding sequences were expressed as MBP fusions in E. coli, and the proteins were purified by immobilization on amylose beads. The PRKX fragment from pGAD-PRKX was cloned into the T7 promoter-driven vector pCITE4a, and the PRKX protein was expressed and metabolically labelled in RRL. The Rep68 and luciferase proteins were also metabolically labelled in RRL for use as controls (Fig. 3b). The labelled proteins were then used in pull-down reactions in combination with equalized amounts of the immobilized MBP fusions described above, and the amounts of labelled protein associated with the beads were determined by SDS-PAGE. As shown in Fig. 3a, MBP-Rep78 did not associate with labelled luciferase; however, a strong interaction was observed upon incubation with labelled PRKX, as evidenced by the appearance of a cluster of bands of approximately 43 to 46 kDa. Labelled PRKX did not interact with MBP alone. MBP-Rep68 was not able to associate with labelled PRKX, consistent with the previous observation with the yeast two-hybrid system. In order to ensure that the MBP-Rep68 was competent for protein-protein interaction, we incubated the MBP-Rep68 with RRL-labelled Rep68 generated from a T7-driven Rep68 expression vector (p68-myc/his). MBP-Rep68 was found to self-associate by using this method, confirming and extending a previous observation made for Rep78 (22, 49).

View larger version (42K): [in this window] [in a new window]   FIG. 3.   Rep78 interaction with PRKX in vitro. (a) MBP fusion proteins containing Rep78, Rep68, or Mal sequences alone were immobilized on amylose beads and then incubated with 35S-labelled PRKX, Rep68, or luciferase (Luc) protein synthesized in RRL. After washing, labeled protein which was associated with the MBP fusion protein was visualized by 12% PAGE and autoradiography. The lane markers indicate the MBP fusion/RRL protein combination. Lane M, molecular mass markers (in kilodaltons). (b) Input RRL proteins corresponding to 10% of the amount added to the pull-down reaction. (c) Unlabelled, S-tagged PRKX was synthesized in RRL and immobilized on S beads, followed by incubation with 35S-labelled, RRL-derived Rep78 or luciferase. Bound Rep78 was visualized by PAGE and autoradiography. (d) A GST fusion protein containing the C-terminal 139 amino acids of Rep78 was used in pull-down experiments with 35S-labelled, RRL-derived PRKX or luciferase. Input lanes contain 10% of the amounts added to the pull-down reactions.

PRKX demonstrates autophosphorylation and histone H1 kinase activity. Although PRKX shows high degrees of homology to the kinase domain of mammalian PKA_c subunits, kinase activity of PRKX has not been demonstrated previously. We sought to demonstrate a kinase activity for PRKX, initially with histone H1 as a substrate. Unlabelled, S-tagged PRKX was synthesized in RRL and immobilized on S beads. After washing, the protein was incubated in kinase buffer containing [-³²P]ATP with or without histone H1. Surprisingly, when PRKX alone was incubated with kinase buffer without histone H1, a ³²P-labelled band was visualized at approximately 46 kDa, corresponding to the size of PRKX itself (Fig. 4a, lane 2). Silver staining showed that the only detectable protein present was PRKX (not shown). These data suggested that PRKX was capable of autophosphorylation. In order to ensure that this activity was associated with PRKX and not a contaminant carried over from the RRL, S beads were incubated with RRL containing luciferase and then were washed and used in the kinase assay. No phosphorylated proteins were detected in this control (Fig. 4a, lane 1). Additional control experiments using a full-length, untagged PRKX clone confirmed that the N-terminal S tag carried on PRKX was not the substrate for autophosphorylation (not shown).

View larger version (40K): [in this window] [in a new window]   FIG. 4.   Kinase activity of PRKX protein in vitro. (a) S-tagged PRKX was synthesized in RRL and purified on S beads (S-bds) prior to addition to a 30-min kinase assay conducted in the presence (lane 3) or absence (lane 2) of histone H1 as a substrate. As a positive control, purified PKAc was used in the kinase assay with (lane 5) or without (lane 4) histone H1. As a negative control, S beads were incubated with luciferase (LUC)-programmed RRL and then added to the kinase assay (lane 1). 32P-labelled phosphorylated proteins were visualized by PAGE and autoradiography. Positions of phosphorylated PRKX and histone H1 are indicated. Lane M, molecular mass markers (in kilodaltons). (b) Same as in panel a except that incubation of the kinase reaction mixture was extended to 60 min. In this experiment histone H1 was added to the negative control reaction (lane 1). (c) S-bead-purified PRKX was incubated with the fluorescence-labelled PKA substrate kemptide (LRRASLG). Phosphorylated kemptide was separated from unphosphorylated substrate by agarose gel electrophoresis and visualized on a UV transilluminator. As a control, S beads preincubated with luciferase-programmed RRL were used in the kinase reaction. (d) S-bead-purified PRKX was autophosphorylated in a [-32P]kinase reaction in the presence of 300 ng (lanes 3, 5, and 7) or 600 ng (lanes 4, 6, and 8) of added MBP or MBP-Rep fusion protein as indicated. Lane 1, PRKX but no MBP; lane 2, S beads from luciferase-programmed RRL and no MBP.

PRKX can activate CREB-dependent transcription. The close homology between PRKX and PKA_c, combined with the observation that PRKX could phosphorylate the PKA kemptide substrate, suggested that PRKX might be capable of modulating PKA-responsive transcriptional regulatory pathways through activation of CREB. We investigated whether PRKX could active CREB-dependent transcription by utilizing a transient-cotransfection model. In this system, a CREB activation domain/Gal4 DNA binding domain fusion protein is used to activate a luciferase reporter controlled by the Gal1 upstream activating sequence. Thus, when CREB is phosphorylated by activation of the PKA pathway, luciferase expression is strongly induced.

View larger version (11K): [in this window] [in a new window]   FIG. 5.   Transfection assay for CREB-dependent transcriptional activation. HeLa cells were cotransfected with Gal4 DNA binding domain/CREB fusion protein vector pFA-CREB and pFR-Luc, a luciferase reporter construct containing a minimal promoter and Gal1 upstream activating sequence. Samples were further cotransfected with the PRKX expression vector pcDNA-PRKX and with the pKex78 or pKex68 vector which express exclusively Rep78 and Rep68, respectively. Forty-eight hours later, cells were lysed and luciferase activity was determined. Results are presented relative to the levels found in cells transfected with pFR-Luc/pFA-CREB only. Means of at least three independent experiments, each conducted in duplicate, are shown, with standard deviations represented as error bars.

Rep78, but not Rep68, inhibits activation of CREB-dependent transcription by PRKX. The effects of Rep proteins on CREB-dependent transcriptional activation were studied by using the cotransfection system described above. When a clone expressing exclusively Rep78 (pKex78) (22) was cotransfected into HeLa cells along with pFA-CREB, pFR-Luc, and pcDNA-PRKX, reduced activation of luciferase activity was observed (Fig. 5, third bar). This suppression was not observed when pKex68, which expresses exclusively Rep68, was used (fourth bar). This suggests that Rep78 interacts with and inhibits PRKX in cells; however, we cannot exclude the possibility that effects of Rep78 on transcription may contribute to the observed inhibition of CREB-dependent activation (see Discussion).

Rep78 interacts directly with PKA_c and inhibits its catalytic activity. Since PRKX is a close homolog of PKA_c, we investigated whether PKA_c might also interact with Rep78, notwithstanding the fact that PKA_c was not isolated from the yeast two-hybrid library screen. Purified PKA_c (100 U) was mixed with purified MBP-Rep78, MBP-Rep68, or MBP immobilized on amylose beads. After extensive washing, protein associated with the beads was assayed for kinase activity by using the PKA-specific kemptide target. As shown in Fig. 6a, immobilized kinase activity was associated with MBP-Rep78 but not with MBP-Rep68 or MBP alone. In the absence of preincubation with PKA_c, MBP-Rep78 did not demonstrate any kinase activity (not shown). We concluded that Rep78 interacts with PKA_c in vitro and that, since both the Rep78 and the PKA_c were present as purified proteins, the interaction most probably involves a direct physical contact between the two proteins.

View larger version (38K): [in this window] [in a new window]   FIG. 6.   In vitro interaction between PKAc and Rep78. (a) Purified PKAc (100 U) was incubated with 0.5 to 1.0 µg of MBP (lane 2), MBP-Rep68 (lane 3), or MBP-Rep78 (lane 4) fusion protein immobilized on amylose beads. After washing, the beads were added to a kemptide phosphorylation assay. Phosphorylated kemptide was visualized by agarose gel electrophoresis and UV transillumination. In lane 1, PKAc was added directly to the kemptide assay. (b) Same as in panel a except that GST (lane 1) or GST-C-terminal protein (lane 2) was used to capture the PKAc. (c) One unit of purified PKAc was added to a 20-min kemptide assay, along with 0.3, 0.6, or 0.9 µg of MBP (lanes 1 to 3), MBP-Rep68 (lanes 4 to 6), or MBP-Rep78 (lanes 7 to 9) fusion protein. (d) One unit of PKAc was added to an 8-min kemptide assay with 30, 60, or 90 ng of GST (lanes 1 to 3) or GST-C-terminal protein (lanes 4 to 6).

Rep78, but not Rep68, inhibits activation of CREB-dependent transcription by PKA_c. We used the CREB-dependent transcription system to investigate whether Rep78 proteins could block downstream functions in the PKA pathway. HeLa cells were cotransfected with pFA-CREB, pFR-LUC, and pFC-PKA, in addition to either pKex78, pKex68, or an equal amount of pCI vector. As shown in Fig. 7, reporter gene activity was greatly induced by the cotransfection of pFC-PKA. Addition of pKex78 to the assay resulted in severe repression of the CREB-dependent promoter activation. Little or no repression was observed when pKex68 was used. These data indicated that Rep78 can block PKA induction of CREB-dependent transcription in cells. While direct effects of Rep78 on transcription may contribute, these observations strongly suggest that Rep78 can block CREB activation by binding to and inhibiting the kinase activity of PKA_c.

View larger version (12K): [in this window] [in a new window]   FIG. 7.   Transfection assay for Rep-mediated suppression of CREB-dependent transcriptional activation. HeLa cells were transfected with pFR-Luc and pFA-CREB. Some samples were cotransfected with the PKAc expression vector pFC-PKA and pKex68 or pKex78. Results are presented relative to the levels found in cells transfected with pFR-Luc/pFA-CREB only. Means of at least three independent experiments, each conducted in duplicate, are shown, with standard deviations represented as error bars.

	DISCUSSION

Top Abstract Introduction Materials & Methods Results Discussion References

In this study we have identified the catalytic subunit of the cAMP-dependent PKA and its X-linked homolog PRKX as cellular targets for AAV Rep78. The interaction was demonstrated in the yeast two-hybrid system, in vitro, and in mammalian cells with a CREB-dependent transcription system. In the case of PKA_c, a direct interaction between the purified proteins was demonstrated. We consider it likely, therefore, that this represents a physiological interaction which occurs during AAV infection or replication.

PRKX was identified originally by positional cloning as a candidate for the chondrodysplasia punctata gene (26). It was shown that the mRNA was expressed in a wide range of adult and fetal tissues and that the gene is associated with a site of chromosomal instability. Little characterization of the protein encoded by PRKX has been conducted. Although PRKX is a close homolog of PKA_c, it is not a PKA isoform. The homology between isoforms is typically greater than 90% identity in the catalytic domain, whereas the identity with PRKX in this region is only 49 to 55%. The closest known homolog of PRKX is Drosophila DC2 (24). The major Drosophila homolog of PKA_c is DC0, and DC2 is unable to complement DC0 mutants, suggesting that their functions differ (39). By analogy, these observations suggest that PRKX and PKA might have related but separable functions.

We show here that PRKX exhibits kinase activity towards histone H1 and a peptide containing the PKA consensus recognition sequence. We further showed that a PRKX expression vector can induce CREB-dependent transcriptional activation, presumably by in vivo phosphorylation of CREB. In these respects, PRKX appears to function similarly to PKA_c. However, some of the evidence suggests that PKA and PRKX may not be functionally analogous. First, PRKX was found to undergo autophosphorylation in vitro, whereas autophosphorylation of PKA_c does not occur. Histone H1 kinase activity of PRKX was detected only after prolonged incubation of the kinase assay mixture. The observation that autophosphorylation occurred preferentially in the presence of excess histone H1 suggests that the autophosphorylation event might be involved in the activation of PRKX activity towards heterologous substrates. Moreover, although PRKX could activate CREB-dependent transcription, the degree of activation was far less than that achieved with a PKA_c expression vector. This effect may be partly due to differences in PKA and PRKX expression levels, because the natural start codon of PRKX is in a weak Kozak consensus context (26). Replacement of the start codon with one close to the consensus increased the activation of CREB-dependent transcription to approximately 20-fold (data not shown), which is more than the 8-fold induction observed with the natural start codon (Fig. 5) but still less than the 80-fold induction typically observed with the PKA_c expression vector (Fig. 7). Taken together, these observations suggest that the substrate preferences and regulatory mechanisms of PRKX and PKA may differ and that a rigorous comparison of PRKX and PKA activities is warranted.

PKA is a relatively ubiquitous enzyme, so the question arises as to why PKA did not emerge from the yeast two-hybrid system screen. Similarly, we did not detect interactions involving the HMG-1 and Sp1 proteins, which have been shown, by other methods, to interact with AAV Rep (11, 21, 45). Screening of the yeast two-hybrid library was difficult because of an endogenous transactivation activity of Rep78 and because yeast clones containing Rep78 grew very poorly. Therefore, the number of potential interactions screened was suboptimal, and there may be other Rep-interacting proteins which have not yet been identified in the screen, which is continuing.

In yeast, both the transactivation activity and growth inhibition phenotypes were localized to the C-terminal region of Rep78 (Fig. 2 and data not shown). The interaction with PRKX and PKA was clearly mapped to this C-terminal domain, which distinguishes Rep78 from Rep68. This domain contains motifs which may be involved in the formation of three zinc fingers and contains other sequence elements which have been implicated in the transactivation functions of adenovirus E1a (22, 55). Few studies have been able to address differences between Rep78 and Rep68 functions, partly because vectors expressing exclusively individual Rep proteins have only recently become available. For Rep68, a slightly reduced capacity to down regulate heterologous promoters has been noted, although the major sequences required for promoter down regulation are located in the N-terminal region (22). Rep68-MBP fusion protein produced in E. coli has been shown to be sufficient to sponsor AAV replication in vitro (54). Recently it has been shown that Rep78 proteins can self-associate (20, 49). As shown in the present paper, self-association can also be observed for Rep68 (Fig. 3). Therefore, a biological function which is specific only to Rep78 remains to be clarified. The Rep78 C-terminal domain which interacts with PRKX and PKA_c is shared by the p19-derived Rep52 protein. This suggests that Rep52 also may interact with these cellular targets; however, this interaction has yet to be demonstrated experimentally.

PKA activity could be detected by a pull-down assay with MBP-Rep78 (Fig. 6a); however, Rep78 was also found to inhibit kinase activity (Fig. 6c and d). In the pull-down assays, PKA_c was present in excess, and the duration of the kinase assay was longer than that in the inhibition assays. The kinase activity detected in the pull-down assays probably results from free PKA_c which dissociated from the complexes during the kinase assay, suggesting that inhibition may be reversible. In the in vitro assays, we sometimes observed that Rep68 could inhibit phosphorylation, although to a much lesser extent. This suggests that, although Rep68 clearly cannot bind PRKX and PKA, domains outside the amino acid 529 to 621 intro region might contribute to inhibition of kinase activity. These domains might have a weak effect in in vitro assays where high concentrations of purified Rep68 proteins are present and the requirement for tight binding is reduced. A detailed analysis of the structural requirements of both PKA and Rep proteins for binding and kinase inhibition is in progress.

In cotransfection assays, Rep78 clearly inhibited CREB-dependent transcriptional activation; however, the mechanism by which this is achieved is not certain. The in vitro data suggest that inhibition of PRKX and PKA kinase activities makes the major contribution, since the system is highly dependent on CREB phosphorylation. However one must also consider that Rep has suppressive activities on heterologous promoters (18, 19, 31, 42). The CREB, PRKX, PKA, and Rep expression constructs used in this system all utilized the CMV immediate-early promoter-enhancer, which is relatively insensitive to Rep suppression (1). The suppression of heterologous promoters by Rep has been shown to be mediated, at lest in some cases, through Sp1 sites (21). The Gal4-CREB-responsive promoter in the reporter construct pFR-Luc is very simple, comprising the Gal upstream activating sequence and the TATA box from adenovirus E1b. The E1b promoter Sp1 site is not present in the pFR-Luc construct; therefore, direct effects of Rep on transcription from this promoter should be minimal. This view is supported by the observations that Rep68 was unable to suppress CREB-dependent transcriptional activation from this promoter (Fig. 5 and 7). Accordingly, we favor the interpretation that Rep78 inhibits the kinase activity of PRKX and PKA in cells, thereby reducing CREB phosphorylation and consequently suppressing reporter gene activation.

What might be the functional consequences of the interaction between Rep78 and PKA or PRKX? One possibility is that the Rep78 is a substrate for PRKX and/or PKA. AAV Rep proteins are known to be phosphorylated, and their phosphorylation state changes during the replication cycle (10). However, we have not been able to demonstrate phosphorylation of MBP-Rep fusion proteins by either PRKX or PKA (data not shown).

A number of viruses have subsumed PKA/CREB pathways to regulate their gene expression; examples include human T-cell leukemia virus type 1, human immunodeficiency virus type 1, adenovirus, HSV, and Epstein-Barr virus (12, 15, 28, 33, 41, 47, 50, 59, 63). The cAMP response pathway components are frequently used to control switches from early to late or latent to lytic replication modes. For example, adenovirus E1a protein can transactivate genes containing CREs via an interaction with CBP/p300 (44). In addition, the E1a, E3, and E4 promoters of adenovirus contain CREs (34). The HSV latency-associated transcript and UL9 gene promoters are cAMP responsive, and this response has been associated with induction of lytic replication (33). The HSV ICP4 protein has been demonstrated to provide a crucial helper function for both HSV and AAV replication (40, 56). ICP4 is a substrate for PKA, and its phosphorylation promotes HSV lytic replication (59). These observations suggest that one function of AAV Rep78 might be to inhibit the utilization of cAMP pathways by helper viruses and thereby inhibit productive replication of the helper.

Few viruses are known to inhibit cAMP pathways. The hepatitis C virus NS3 protein has been shown to bind to PKA_c, inhibiting its catalytic activity and nuclear translocation (6, 7). The poliovirus 3Cpro protease has been shown to cleave CREB (61). The roles of these interactions in the respective virus life cycles are not yet understood. Both AAV2 and the autonomous parvovirus minute virus of mice (MVMp) have CRE elements in the promoters controlling the major nonstructural proteins (13, 46). The function of the AAV2 p5 promoter CRE has not been extensively studied; however, cAMP-dependent stimulation of the CRE of minute virus of mice has a negative regulatory effect (46). This supports a model wherein Rep78 suppression of PRKX/PKA may be involved in an autoregulatory loop controlling Rep gene expression. Accordingly, we suggest that the function of the Rep-PRKX/PKA interaction may be to influence the balance between AAV and helper virus replication.