Efficient Expression of the Adeno-Associated Virus Type 5 P41 Capsid Gene Promoter in 293 Cells Does Not Require Rep
http://www.100md.com
《病菌学杂志》
Department of Molecular Microbiology and Immunology, Life Sciences Center, University of Missouri—Columbia, Columbia, Missouri 65211
ABSTRACT
Efficient expression of the adeno-associated virus type 5 (AAV5) P41 capsid gene promoter required adenovirus E1A and/or E1B; however, in contrast to what was observed for expression of the AAV2 capsid gene promoter (P40), neither adenovirus infection nor the large Rep protein was required. Although both the AAV2 and the AAV5 large Rep proteins efficiently bound the (GAGY)3 Rep-binding element, the AAV5 large Rep protein transactivated transcription of the inducible AAV2 P40 promoter much less well than AAV2 large Rep. Differences in their activation potentials were mapped to the amino-terminal region of the proteins, and the poorly transactivating AAV5 Rep protein could competitively inhibit AAV2 Rep transactivation.
INTRODUCTION
Appropriate levels of capsid protein gene expression must be achieved and maintained throughout parvovirus infection. Although parvoviruses that have a single promoter (e.g., B19 and Aleution mink disease virus) utilize posttranscriptional mechanisms to specifically control capsid protein levels (12, 19), those parvoviruses with internal capsid gene promoters (e.g., minute virus of mice and adeno-associated virus [AAV]) control capsid protein accumulation by regulated transcriptional activation of these promoters (6, 14). Transcriptional transactivation, mediated by the large nonstructural proteins of these viruses (NS1 and Rep, respectively), assures the appropriate timing and expression of these promoters and hence the levels of their encoded RNA and protein products.
Transactivation of P40 by AAV type 2 (AAV2) Rep requires binding to the transcription template at either the AAV2 inverted terminal repeat (ITR) or the P5 promoter and has been proposed to work, at least in part, by stabilizing a loop-like structure that localizes the P5 promoter, and presumably P5-associated transcription factors, to the P40 promoter (11). Efficient activation by Rep also requires coinfection by adenovirus (Ad), even in 293 cells which constitutively express the Ad5 E1A, a well-characterized transcription activator, and E1B proteins.
AAV5, another serotype of the human AAVs, shares only 64% overall nucleotide identity with AAV2. The AAV2 and AAV5 Rep proteins share 67% identity. Both AAV2 and AAV5 Rep proteins bind the Rep-binding element (RBE) in both AAV2 and AAV5 ITRs (2). However, the AAV2 and AAV5 Rep proteins cannot substitute for one another with respect to Rep-mediated DNA replication; because of sequence differences within their terminal resolution sites (TRS), each protein can fully process only its native origin (2). Additionally, in contrast to AAV2, the AAV5 large Rep gene promoter (P7) contains only a poor-consensus RBE, suggesting that that the activation of the AAV5 capsid gene promoter (P41) may be governed differently from its AAV2 counterpart.
Here, we show that in contrast to what was observed for other AAV serotypes and other parvoviruses with internal capsid gene promoters, the constitutive activity in 293 cells of the AAV5 P41 promoter in RepCap constructs is high and is not significantly activated further by its large Rep protein or Ad5 infection.
MATERIALS AND METHODS
Cells and virus. HeLa and 293 cells were propagated as previously described (8). Transfections, using Lipofectamine and the Plus reagent (Invitrogen, Carlsbad, CA), were performed as previously described (14), and when Ad5 was coinfected, this was done 5 h after transfection at a multiplicity of infection of 5.
Plasmid constructs. (i) RepCap and RepStopCap plasmids. AAV2 RepCap (containing AAV2 nucleotides [nt] 145 to 4492) and AAV5 RepCap (containing AAV5 nt 185 to 4448) have been described previously (13, 14). AAV1 RepCap (containing AAV1 nt 181 to 4566), AAV3 RepCap (containing AAV3 nt 181 to 4560), and AAV6 RepCap (containing AAV6 nt 181 to 4560) were constructed by inserting AAV1- and AAV3-specific sequences amplified directly from these viruses (obtained from the American Type Culture Collection), and AAV6-specific sequences amplified from the pAAV6 infectious clone (15), into the polylinker of pBluescript SK(+) (Stratagene, La Jolla, Calif.), between the EcoRI and XbaI sites. The AAV4 RepCap plasmid was a gift from J. A. Chiorini (3). The AAV2 RepStopCap and AAV5 RepStopCap constructs have premature-termination codons introduced at nt 489 and 480, respectively, in the amino termini of their respective Rep proteins. To create the AAV1, AAV3, AAV4, and AAV6 RepStopCap plasmids, frameshift mutations were introduced in the parent RepCap plasmids within the large Rep-coding region at nt 743, 523, 781, and 729, respectively, leading to the termination of these proteins shortly downstream.
(ii) P40 and P41 minimal transcription unit test plasmids. P40VP contains AAV2 nt 1700 to 4492. P41Cap contains AAV5 nt 1637 to 4448. Sequences from AAV2 P5 (nt 146 to 320) (14), AAV5 P7 (nt 168 to 358) (13), the AAV2 ITR (nt 1 to 145) (14), or the AAV5 ITR (nt 1 to 183) (13) were inserted into P40VP at nt 1700 (123 nt upstream of the P40 TATA box) or into P41VP at nt 1637 (243 nt upstream of the P41 TATA box), to create P5P40VP, P7P40VP, V2ITRP40VP, and V5ITRP40VP or P5P41VP, P7P41VP, V2ITRP41VP, and V5ITRP41VP, respectively, as shown in the diagram in Fig. 2.
(iii) pHelper and Rep-expressing plasmids. pHelper plasmid was from Stratagene (TX). Rep expression constructs were driven by the human immunodeficiency virus (HIV) promoter within the HIV long terminal repeat promoter and have been described previously (13, 14).
(iv) AAV2/AAV5 chimeric Rep plasmids and the P5P40 luciferase reporter plasmid. Chimeric Rep constructs were based on the HIV-driven, AAV2, and AAV5 Rep-expressing parental constructs previously described (13, 14). The numbers in the name of each construct designate exactly where the chimeric borders were made. All fusions were made in frame and sequenced to ensure that they were as predicted. The V2 211 (V5) series was based on the V2 211 parental construct. To generate the P5P40 luciferase construct, the AAV2 P5 promoter (nt 146 to 320 from AAV2) was cloned upstream of the luciferase gene driven by AAV2 P40 (nt 1700 to 1883), to provide an RBE, but was cloned in an inverted orientation relative to the direction of P40 transcription to exclude potential P5-generated luciferase activity.
(v) Chimeric Rep-binding domain plasmids and the P5P40 minimal-transcription-unit test plasmid. The parental RepBD.TZ.AD and RepBD.TZ plasmids were gifts from Matt Weitzman (Salk Institute, San Diego, Calif.) and contain a modified leucine zipper dimerization domain from GCN4 (TZ), and in the case of RepBD.TZ.AD, the activation domain from herpesvirus VP16 (1). RepBD.TZ.AD- and RepBD.TZ-derived constructs containing either AAV5 or chimeric Rep sequences (V5 101-121, V5 121-161, and V5 161-181) were constructed by replacing the amino-terminal 244-amino-acid (aa) Rep-binding domain from the original AAV2-binding domain construct with either AAV5 or chimeric Rep sequences. The P5P40 minimal-transcription-unit test construct is the same as that described above.
RNase protection assays. Total RNA was isolated 36 to 41 h posttransfection as previously described (9, 16). RNase protection assays were performed as previously described (9, 16), using homologous antisense probes which spanned nt 1781 to 1923 (for AAV1), 1767 to 1906 (for AAV2), 1764 to 1903 (for AAV3), 1824 to 1957 (for AAV4), and 1766 to 1908 (for AAV6) of the individual P40 promoters and nt 1846 to 1985 of the P41 promoter of AAV5.
Luciferase assays. Luciferase assays were performed according to reference 6. Briefly, 293 cells grown in 12-well plates were transfected with the P5P40 luciferase reporter construct together with pHelper and the Rep expression constructs described in the text at a 1:4:0.25 ratio. In addition, 0.05 μg per well of cytomegalovirus (CMV)-driven -galactosidase reporter was cotransfected as an internal control. The total amount of DNA transfected into each well was kept at 0.6 μg. Thirty-six hours after transfection, cells were lysed with lysis buffer containing 1% Triton X-100. Equal amounts of the sample were used to test luciferase activity and -galactosidase activity according to the manufacturer's instructions (Applied Biosystems [Tropix], Bedford, MA). Each experiment represents the average values for duplicate samples from three individual experiments (error bars are shown in the relevant figures), each normalized to -galactosidase activity.
Rep inhibition assays. The activity of the P5P40 luciferase construct in response to an HIV-driven AAV2 Rep plasmid was first assayed (data not shown), and the lowest amount of Rep-expressing plasmid giving the highest level of activity was determined. Luciferase activity was then monitored in assays in which the total DNA concentration was kept constant, following addition of decreasing amounts of the HIV-driven AAV2 Rep-expressing plasmid, using either empty vector or the HIV-driven AAV5 Rep-expressing plasmid to make up the balance. Each experiment represents the average values for duplicate samples from three individual experiments (error bars are shown in the relevant figures), each normalized -galactosidase activity.
RESULTS
The constitutive activity in 293 cells of AAV5 P41 in RepCap constructs is high and is not significantly activated further by its large Rep protein or Ad5 infection. Maximal levels of activation of AAV2 P40, in the presence of Ad, occur from full-length, ITR-containing clones. While the ITR can supply the RBE required for Rep activation of AAV2 P40, it is likely that these maximal levels of activation are partially attributable to amplification of templates due to replication, because, as mentioned above, in the presence of adenovirus, the AAV2 P5 RBE can fully support P40 activation in nonreplicating AAV2 RepCap constructs (14). Thus, for our analyses, we chose to utilize ITR-deleted RepCap constructs to examine the effect of Ad and Rep on the activation of the capsid gene promoters of various AAV serotypes, thereby eliminating expression differences due to variation in viral replication under these different conditions. Although the AAV5 P7 promoter does not contain a consensus RBE, the AAV5 RepCap construct could be included in this analysis because the level of RNA generated from the AAV5 capsid gene promoter (P41), in an ITR-lacking AAV5 RepCap clone, was surprisingly high (13).
As shown in Fig. 1A, expression in 293 cells of both the upstream P5 and P19 promoters and the P40 promoters within Rep-expressing RepCap constructs of various AAV serotypes was high, both in the absence and in the presence of Ad5. However, the P40 promoters of AAV1, AAV2, AAV3, AAV4, and AAV6 were all activated by Ad5 significantly more than was the P41 promoter of AAV5 (Fig. 1A, compare lanes 1 through 8, 11, and 12 to lanes 9 and 10), both as measured directly (Fig. 1C) and as a percentage of total RNA (Fig. 1D, compare white bars to black bars). Identical amounts of RNA were used for each protection. Because no suitable internal transfection control whose expression was independent of Ad5 was identified, direct activation levels (Fig. 1C) were calculated as the averages for at least three separate complete experiments that included all test plasmids, and various plasmid preparations were tested. Southern analysis was initially used to detect and standardize for cell-associated plasmid DNA after transfection as previously reported (11). However, for our experiments, this was not proportional to DNA that had entered the nucleus (data not shown) and so was also deemed an unsuitable control for these purposes. As previously reported (14), in 293 cells the activity of the P40 promoters in RepCap constructs remained relatively low compared to the activities of the P5 and P19 promoters, even in the presence of Ad5, and was significantly lower than that seen during AAV2 infection.
That the expression of the AAV5 P41 promoter was high in these assays was surprising because the AAV5 RepCap plasmid is not predicted to contain a Rep-binding site. This suggested that either AAV5 Rep activation of P41 did not require binding to the transcription template or P41 activity was independent of Rep under these conditions. The latter seemed to have been the case, because in contrast to what was observed for the other AAV serotypes, the activity of the P41 promoter also remained high in similar constructs in which expression of the Rep protein was abolished by the introduction of a translation termination signal at nt 480 in the region of the gene encoding the amino terminus of its product (RepStopCap plasmids), both in the presence and in the absence of Ad5 (Fig. 1B, compare lanes 13 through 20 and 23 through 24 to lanes 21 and 22; also Fig. 1D). The addition of homologous Rep proteins in trans to AAV1, AAV2, AAV3, AAV4, and AAV6 RepStopCap constructs enhanced their P40 promoters to various extents; however, AAV5 P41 was not further activated (data not shown). These results suggested that basal activity of P41 in 293 cells was constitutively high and not further enhanced by either Ad5 infection or Rep.
Expression of AAV5 P41 was also independent of the AAV5 Rep protein in HeLa cells, and in these cells, the difference between AAV5 P41 and AAV2 P40 was even more marked. In HeLa cells, the basal levels of both the upstream promoters and the capsid gene promoters of both AAV2 and AAV5 RepCap constructs were considerably less than those seen in 293 cells (Fig. 1E, lanes 1 and 5, respectively), and the levels of both of these promoters were thus dramatically enhanced by adenovirus (Fig. 1E, lanes 2 and 6, respectively). Full levels of activation for the AAV2 promoters in HeLa cells were dependent upon the Rep protein (Fig. 1E, compare lanes 1 and 2 to lanes 3 and 4); however, as in 293 cells, the activity of the AAV5 constructs was independent of Rep expression (Fig. 1E, compare lanes 5 and 6 to lanes 7 and 8). Furthermore, these results suggested that the higher activity of AAV5 P41 in 293 cells shown in Fig. 1B was likely due to the resident Ad E1A and/or E1B gene products. Thus, while expression of the AAV2 P40 promoter requires Ad5 infection plus Rep in 293 cells, expression of AAV5 P41 can be sustained by E1A and E1B alone.
Compared at optimal alignment, there are significant differences between the two promoters that may account for the higher basal activity of P41 (Fig. 1F). Interestingly, the Sp1-50 binding site, which is important for AAV2 Rep activation of P40 (11), is significantly altered in P41. A comprehensive comparison of the two promoters is in progress.
Although the P41 promoter did not require Rep for activation, because the AAV5 RepCap constructs are predicted to lack an AAV5 Rep-binding site, the intrinsic transcription-transactivating potential of the AAV5 Rep protein needed to be tested in another way.
In contrast to the AAV2 Rep protein, the AAV5 Rep protein is a poor transactivator of the inducible AAV2 P40 promoter. To examine the activation potential of AAV5 Rep further, we chose to examine the transactivation of the AAV2 P40 promoter resident within a series of minimal constructs to which various naturally occurring potential Rep-binding sites were attached. As expected, the AAV5 P7 promoter region, which lacks a consensus binding site for either the AAV2 or the AAV5 Rep proteins, conferred to P40 only a modest amount of activation by either the AAV2 or the AAV5 Rep proteins (Fig. 2A, lanes 7 to 9) compared to the background parent P40VP (Fig. 2A, lanes 1 to 3). When a Rep-binding element was supplied by the AAV2 P5 promoter (P5P40VP), activation by AAV2 Rep increased approximately sevenfold over background levels (Fig. 2A, compare lanes 4 and 5); however, activation by AAV5 Rep remained approximately twofold over background levels (Fig. 2A, compare lanes 4 and 6). When the AAV2 ITR was used to supply the RBE (V2ITRP40VP), transactivation by AAV2 Rep, but not by AAV5 Rep, was enhanced further yet (Fig. 2A, compare lane 10 to lanes 11 and 12, respectively). The AAV5 Rep protein exhibited somewhat higher activation of AAV2 P40 when the AAV5 ITR was linked in cis (V5ITRP40VP), approximately three- to fourfold over background levels, similar to levels achieved by AAV2 Rep with this construct (Fig. 2A, compare lane 13 to lanes 14 and 15). The AAV5 and AAV2 Rep proteins accumulated to similarly high amounts following transfection of 293 cells as detected by Western blot analysis (data not shown), and the AAV5 Rep protein was also shown to bind these elements as well as did the AAV2 Rep protein, as measured by the ability to target the VP16 activator to a test template (see below).
These results suggested that the AAV5 Rep protein was a significantly less potent activator of the inducible P40 promoter than was AAV2, even when targeted to the transcription template. As expected, the AAV5 P41 promoter was only poorly inducible by either AAV2 or AAV5 Rep, regardless of the nature of the RBE linked in cis, primarily because of its higher basal activity (Fig. 2B).
A difference in activation potential between AAV2 and AAV5 Rep can be mapped to the N-terminal domains of the molecules. Although AAV2 and AAV5 Rep proteins share significant sequence homologies and can both bind to a (GAGY)3 consensus binding site with similar affinities in vitro (2), they activated the inducible AAV2 P40 promoter to significantly different levels when targeted to the transcription template by either the AAV2 P5 or the ITR RBE. To determine regions of the two Rep proteins that might distinguish their transactivation properties, we made a series of chimeras between the two (diagramed in Fig. 3A) and tested them for their ability to activate the inducible AAV2 P40 promoter. Constructs expressing luciferase from a P40 promoter with an RBE-containing AAV2 P5 element linked upstream in an orientation inverted relative to the direction of transcription of P40 (Fig. 3B, top) were cotransfected in 293 cells with the pHelper plasmid (Invitrogen), which supplies Ad5 E2A, E4orf6, and virus-associated (VA) RNA. All of the wild-type and chimeric Rep proteins assayed in the following experiments accumulated to similar levels as assayed by Western blotting (data not shown).
Consistent with results presented in the previous sections, the full-length AAV2 Rep protein was a strong activator in this system (Fig. 3B, lane V2). The full-length AAV5 Rep protein, however, which bound the AAV2 P5 RBE to an extent similar to that for AAV2 Rep in the heterologous binding assay described below, showed dramatically reduced activation potential (Fig. 3B, lane V5). Analysis of chimeras in which carboxyl-terminal residues of two proteins were exchanged demonstrated that the amino-terminal regions of these Rep proteins governed their transactivation properties in these assays. Both V2 362 and V2 211, which contained the AAV2 Rep amino-terminal domain, retained transactivation activities similar to that of AAV2 Rep, while V5 362 and V5 211, which contained the amino-terminal region of AAV5 Rep, were as inactive in this assay as they were for the wild-type AAV5 Rep protein (Fig. 3B). Based on these results, chimeras within the first 211 amino acids were generated within the AAV5 Rep background (within the V2 211 parent) to attempt to identify regions which were responsible for differences between the activation properties of these proteins. These chimeras exhibited variability in their ability to transactivate the test template. V2 211-generated Rep proteins, in which either aa 181 to 211 or aa 161 to 181 were replaced with residues of AAV5 Rep, retained approximately half of the transactivation activity of the parent protein [Fig. 3B, lanes V2 211 (V5 181-211) and V2 211 (V5 161-181), respectively]. Replacement of V2 211 amino acids 101 to 121 with those of AAV5 reduced the activation activity of the molecule further [Fig. 3B, lanes V2 211 (V5 101-121)], while the replacement of amino acids 1 to 57, 57 to 101, or 121 to 161 reduced activity to the very low levels seen for the AAV5 wild-type Rep molecule [Fig. 3B, lanes V2 211 (V5 1-57), V2 211 (V5 57-101), and V2 211 (V5 121-161), respectively]. These results suggested that a region that lies within the previously identified DNA-binding region of AAV2 Rep governs its activation properties. While the full 211 amino termini of these Rep proteins are important, these results generally suggest that the amino-terminal 100 amino acids are most critical for activity.
Although the chimeras were made from two parents, each competent for binding to an RBE, three amino-terminal chimeras were further analyzed to reveal any potential differences in their ability to bind to the P5 RBE, which might account for differences in their activation activities (Fig. 4). This was done using a heterologous reporter assay, in which DNA binding was measured indirectly via targeting of the VP16 activation domain. The amino-terminal 244-aa Rep-binding domains from the wild-type or chimeric Rep proteins assayed above, fused with a modified leucine zipper dimerization domain from GCN4 (TZ) (1) and the VP16 activation domain, were assayed for levels of VP16-dependent activation of a P5P40 test plasmid (diagramed in Fig. 4A, top). As expected from previously published reports (2), the AAV2 and AAV5 amino-terminal binding domains (V2 BD.TZ.AD and V5 BD.TZ.AD, respectively) exhibited similar levels of binding to the P5 RBE (as measured by VP16 activation of P40 [Fig. 4A, compare lanes 3 and 4 to lanes 5 and 6]). Three of the chimeras tested for activation as described above, V5 161-181 BD.TZ.AD, which exhibited the highest level of activation, V5 101-121 BD.TZ.AD, which exhibited an intermediate level of activation, and V5 121-161 BD.TZ.AD, which exhibited little if any activation, all exhibited similar levels of binding to the P5 RBE as measured in this assay (Fig. 4A, compare lanes 7 and 8, 9 and 10, and 11 and 12, respectively). Thus, the chimeras bound to the AAV2 P5 RBE as well as the two parents, and gross differences in binding could not explain differences in their activation properties. Activity over vehicle transfection alone (SK) was not detected following transfection of a Gal4VP16 chimera (Gal-VP) (Fig. 4A, compare lanes 1 and 2) or when the VP16 activation domain was not present on the Rep-binding domain/GCN4 leucine zipper domain fusions (Fig. 4A, compare lane 2 to lanes 3, 5, 7, 9, and 11). Activation of the P5 promoter by VP16 showed a similar profile (Fig. 4A, arrowhead). All Rep fusion proteins were expressed at similar levels in these assays (data not shown). These results suggested that the transactivation potentials of the two Rep proteins could be separated from their binding activities, at least in these simple binding assays.
Since both the AAV2 and the AAV5 Rep proteins bound to the AAV2 P5 RBE yet show significant differences in activation activity, we tested whether the AAV5 Rep protein could compete for, and thus inhibit, AAV2 Rep activation of a P5 RBE-linked reporter template. As seen in Fig. 4B, the addition of excess AAV5 Rep-expressing plasmid, but not the plasmid vehicle, inhibited activation of these reporter templates in response to a broad range of AAV2 Rep-expressing plasmids.
DISCUSSION
Parvoviruses generate capsid-encoding mRNAs either from an internal promoter (6, 14) or by RNA processing of a large pre-mRNA generated from a single promoter (12, 19). In previously characterized examples, expression from internal capsid gene promoters is low in all cell types examined prior to activation either by a viral protein alone (the large nonstructural viral protein for the Parvovirus genera and goose parvovirus) or by the viral Rep protein plus helper virus (for the AAV group of viruses). In this work, we show that within RepCap constructs lacking ITR sequences, expression in 293 cells of the AAV5 P41 capsid gene promoter is high and is not further activated by either its large Rep protein or Ad5 infection. In addition, the AAV5 Rep protein is itself a poor activator of the inducible AAV2 P40 promoter. The sequences immediately upstream of the P41 and P40 promoters are quite different, and this may play a role in the different activities of these promoters in minimal constructs.
Does this imply that, unlike for other parvoviruses, expression of the AAV5 capsid proteins is unregulated during infection It may be that even though P41 basal activity is high within minimal expression constructs in 293 cells, capsid expression from the viral genome may be regulated during infection in response to Rep and the presence of helper virus. This is currently being examined. High basal activity of P41 requires Ad E1A and/or E1B, suggesting that helper virus gene products, and perhaps Rep, play a role in P41 expression during infection.
Our results also suggest that efficient transcriptional activation by AAV2 Rep is likely dependent on the nature of its interaction with the RBE-binding site in vivo. Surprisingly, AAV2 Rep activation of the V5ITRP40VP construct was less than for those containing an authentic AAV2 RBE. Although AAV2 Rep binds both of these sites with similar affinities in in vitro gel shift assays (2), AAV2 Rep binding to the AAV5 ITR is not sufficient to fully activate the AAV2 P40 promoter. Similarly, a synthetic minimal RBE site alone is not sufficient to allow Rep-mediated transactivation of the AAV2 P40 promoter; flanking sequences including the TRS are required for optimal transactivation (data not shown) (14). In addition, AAV2 and AAV5 Rep proteins cannot substitute for one another during DNA replication, because the TRS sequences of the AAV2 and AAV5 ITRs are functionally different (2). Thus, the natures of AAV2 Rep and AAV5 Rep binding to their respective ITRs in vivo are different, and this may account for their differences in activation potential. Alternatively, these binding sites may themselves differ in their ability to bind additional cellular factors required for the activation process.
Our examination of chimeras between the AAV2 and AAV5 Rep proteins has revealed a separation between the RBE-binding and transactivation capabilities of Rep. Amino-terminal chimeras which transactivated strongly, intermediately, or weakly showed indistinguishable binding to the AAV2 P5 RBE in heterologous binding assays. Thus, while the amino termini of both AAV2 and AAV5 Rep are interchangeable in regard to binding of the RBE in our assay, the amino terminus of AAV2 Rep contains regions required for transactivation that are lacking in AAV5 Rep.
An independent activation domain of Rep has not yet been identified; however, the DNA-binding domain of Rep has been localized to the amino-terminal region (7, 10, 17, 18), and the structure of the AAV5 Rep DNA-binding domain (aa 1 to 197) has been determined (4). This region of AAV5 Rep protein displays a ferredoxin-like fold, which is topologically homologous to numerous other DNA-binding proteins, most notably simian virus 40 T antigen and the DNA-binding domain of the replication initiation protein E1 of bovine papillomavirus. Interestingly, the core elements responsible for this high homology (4, 1, 3, 2, B, and C) (4) reside in the region of AAV5 Rep that, when transferred to the analogous region of AAV2 Rep, most significantly impaired its ability to activate in our assays (1 to 57: 1; 57 to 101: B, 2, and 3; 101 to 121: C; 121 to 161: 4).
Furthermore, the structure of AAV5 Rep aa 1 to 197 complexed with the AAV5 ITR RBE has also been determined (5). Seven amino acids that make contact with the RBE are conserved between the two proteins (5), yet they fall into the same region (aa 121 to 161) which our assays identify as functionally different between the two with regard to transactivation. However, differences between the two proteins in this region—specifically, that AAV2 Rep has a single Arg at residue 138 rather than the Lys137 and Lys138 residues of AAV5 Rep and that for AAV5, the downstream Gly139 and Gly140 are followed by Ala141 rather than the three sequential glycines found for AAV2 Rep—have led to the prediction that the interaction of AAV5 with its RBE may be qualitatively different from the interaction of the other AAV Rep proteins with their RBEs (5). Thus, while the conserved similarities between the AAV2 and AAV5 Rep-binding domains may help explain why the two proteins behave comparably in our simple binding assays, it will be important to determine whether differences within the region required for transactivation which have been predicted to result in subtly different DNA-binding properties account for differences in the transactivating activities of the two Rep proteins.
ACKNOWLEDGMENTS
We thank Lisa Burger and Fang Cheng for excellent technical assistance and Greg Tullis for critical reading of the manuscript.
This work was supported by PHS grants RO1 AI46458, RO1 AI56310, and RO1 AI21302 from NIAID to D.J.P.
REFERENCES
Cathomen, T., D. Collete, and M. D. Weitzman. 2000. A chimeric protein containing the N terminus of the adeno-associated virus Rep protein recognizes its target site in an in vivo assay. J. Virol. 74:2372-2382.
Chiorini, J. A., S. Afione, and R. M. Kotin. 1999. Adeno-associated virus (AAV) type 5 Rep protein cleaves a unique terminal resolution site compared with other AAV serotypes. J. Virol. 73:4293-4298.
Chiorini, J. A., L. Yang, Y. Liu, B. Safer, and R. M. Kotin. 1997. Cloning of adeno-associated virus type 4 (AAV4) and generation of recombinant AAV4 particles. J. Virol. 71:6823-6833.
Hickman, A. B., D. R. Ronning, R. M. Kotin, and F. Dyda. 2002. Structural unity among viral origin binding proteins: crystal structure of the nuclease domain of adeno-associated virus Rep. Mol. Cell 10:327-337.
Hickman, A. B., D. R. Ronning, Z. N. Perez, R. M. Kotin, and F. Dyda. 2004. The nuclease domain of adeno-associated virus rep coordinates replication initiation using two distinct DNA recognition interfaces. Mol. Cell 13:403-414.
Lorson, C., and D. J. Pintel. 1997. Characterization of the minute virus of mice P38 core promoter elements. J. Virol. 71:6568-6575.
McCarty, D. M., T. H. Ni, and N. Muzyczka. 1992. Analysis of mutations in adeno-associated virus Rep protein in vivo and in vitro. J. Virol. 66:4050-4057.
Mouw, M. B., and D. J. Pintel. 2000. Adeno-associated virus RNAs appear in a temporal order and their splicing is stimulated during coinfection with adenovirus. J. Virol. 74:9878-9888.
Naeger, L. K., R. V. Schoborg, Q. Zhao, G. E. Tullis, and D. J. Pintel. 1992. Nonsense mutations inhibit splicing of MVM RNA in cis when they interrupt the reading frame of either exon of the final spliced product. Genes Dev. 6:1107-1119.
Owens, R. A., M. D. Weitzman, S. R. Kyostio, and B. J. Carter. 1993. Identification of a DNA-binding domain in the amino terminus of adeno-associated virus Rep proteins. J. Virol. 67:997-1005.
Pereira, D. J., and N. Muzyczka. 1997. The adeno-associated virus type 2 p40 promoter requires a proximal Sp1 interaction and a p19 CArG-like element to facilitate Rep transactivation. J. Virol. 71:4300-4309.
Qiu, J., F. Cheng, L. R. Burger, and D. Pintel. 2006. The transcription profile of Aleutian mink disease virus in CRFK cells is generated by alternative processing of pre-mRNAs produced from a single promoter. J. Virol. 80:654-662.
Qiu, J., R. Nayak, G. E. Tullis, and D. J. Pintel. 2002. Characterization of the transcription profile of adeno-associated virus type 5 reveals a number of unique features compared to previously characterized adeno-associated viruses. J. Virol. 76:12435-12447.
Qiu, J., and D. J. Pintel. 2002. The adeno-associated virus type 2 Rep protein regulates RNA processing via interaction with the transcription template. Mol. Cell. Biol. 22:3639-3652.
Rutledge, E. A., C. L. Halbert, and D. W. Russell. 1998. Infectious clones and vectors derived from adeno-associated virus (AAV) serotypes other than AAV type 2. J. Virol. 72:309-319.
Schoborg, R. V., and D. J. Pintel. 1991. Accumulation of MVM gene products is differentially regulated by transcription initiation, RNA processing and protein stability. Virology 181:22-34.
Weitzman, M. D., S. R. Kyostio, B. J. Carter, and R. A. Owens. 1996. Interaction of wild-type and mutant adeno-associated virus (AAV) Rep proteins on AAV hairpin DNA. J. Virol. 70:2440-2448.
Yang, Q., and J. P. Trempe. 1993. Analysis of the terminal repeat binding abilities of mutant adeno-associated virus replication proteins. J. Virol. 67:4442-4447.
Yoto, Y., J. Qiu, and D. J. Pintel. 2006. Identification and characterization of two internal cleavage and polyadenylation sites of parvovirus B19 RNA. J. Virol. 80:1604-1609.(Chaoyang Ye, Jianming Qiu)
ABSTRACT
Efficient expression of the adeno-associated virus type 5 (AAV5) P41 capsid gene promoter required adenovirus E1A and/or E1B; however, in contrast to what was observed for expression of the AAV2 capsid gene promoter (P40), neither adenovirus infection nor the large Rep protein was required. Although both the AAV2 and the AAV5 large Rep proteins efficiently bound the (GAGY)3 Rep-binding element, the AAV5 large Rep protein transactivated transcription of the inducible AAV2 P40 promoter much less well than AAV2 large Rep. Differences in their activation potentials were mapped to the amino-terminal region of the proteins, and the poorly transactivating AAV5 Rep protein could competitively inhibit AAV2 Rep transactivation.
INTRODUCTION
Appropriate levels of capsid protein gene expression must be achieved and maintained throughout parvovirus infection. Although parvoviruses that have a single promoter (e.g., B19 and Aleution mink disease virus) utilize posttranscriptional mechanisms to specifically control capsid protein levels (12, 19), those parvoviruses with internal capsid gene promoters (e.g., minute virus of mice and adeno-associated virus [AAV]) control capsid protein accumulation by regulated transcriptional activation of these promoters (6, 14). Transcriptional transactivation, mediated by the large nonstructural proteins of these viruses (NS1 and Rep, respectively), assures the appropriate timing and expression of these promoters and hence the levels of their encoded RNA and protein products.
Transactivation of P40 by AAV type 2 (AAV2) Rep requires binding to the transcription template at either the AAV2 inverted terminal repeat (ITR) or the P5 promoter and has been proposed to work, at least in part, by stabilizing a loop-like structure that localizes the P5 promoter, and presumably P5-associated transcription factors, to the P40 promoter (11). Efficient activation by Rep also requires coinfection by adenovirus (Ad), even in 293 cells which constitutively express the Ad5 E1A, a well-characterized transcription activator, and E1B proteins.
AAV5, another serotype of the human AAVs, shares only 64% overall nucleotide identity with AAV2. The AAV2 and AAV5 Rep proteins share 67% identity. Both AAV2 and AAV5 Rep proteins bind the Rep-binding element (RBE) in both AAV2 and AAV5 ITRs (2). However, the AAV2 and AAV5 Rep proteins cannot substitute for one another with respect to Rep-mediated DNA replication; because of sequence differences within their terminal resolution sites (TRS), each protein can fully process only its native origin (2). Additionally, in contrast to AAV2, the AAV5 large Rep gene promoter (P7) contains only a poor-consensus RBE, suggesting that that the activation of the AAV5 capsid gene promoter (P41) may be governed differently from its AAV2 counterpart.
Here, we show that in contrast to what was observed for other AAV serotypes and other parvoviruses with internal capsid gene promoters, the constitutive activity in 293 cells of the AAV5 P41 promoter in RepCap constructs is high and is not significantly activated further by its large Rep protein or Ad5 infection.
MATERIALS AND METHODS
Cells and virus. HeLa and 293 cells were propagated as previously described (8). Transfections, using Lipofectamine and the Plus reagent (Invitrogen, Carlsbad, CA), were performed as previously described (14), and when Ad5 was coinfected, this was done 5 h after transfection at a multiplicity of infection of 5.
Plasmid constructs. (i) RepCap and RepStopCap plasmids. AAV2 RepCap (containing AAV2 nucleotides [nt] 145 to 4492) and AAV5 RepCap (containing AAV5 nt 185 to 4448) have been described previously (13, 14). AAV1 RepCap (containing AAV1 nt 181 to 4566), AAV3 RepCap (containing AAV3 nt 181 to 4560), and AAV6 RepCap (containing AAV6 nt 181 to 4560) were constructed by inserting AAV1- and AAV3-specific sequences amplified directly from these viruses (obtained from the American Type Culture Collection), and AAV6-specific sequences amplified from the pAAV6 infectious clone (15), into the polylinker of pBluescript SK(+) (Stratagene, La Jolla, Calif.), between the EcoRI and XbaI sites. The AAV4 RepCap plasmid was a gift from J. A. Chiorini (3). The AAV2 RepStopCap and AAV5 RepStopCap constructs have premature-termination codons introduced at nt 489 and 480, respectively, in the amino termini of their respective Rep proteins. To create the AAV1, AAV3, AAV4, and AAV6 RepStopCap plasmids, frameshift mutations were introduced in the parent RepCap plasmids within the large Rep-coding region at nt 743, 523, 781, and 729, respectively, leading to the termination of these proteins shortly downstream.
(ii) P40 and P41 minimal transcription unit test plasmids. P40VP contains AAV2 nt 1700 to 4492. P41Cap contains AAV5 nt 1637 to 4448. Sequences from AAV2 P5 (nt 146 to 320) (14), AAV5 P7 (nt 168 to 358) (13), the AAV2 ITR (nt 1 to 145) (14), or the AAV5 ITR (nt 1 to 183) (13) were inserted into P40VP at nt 1700 (123 nt upstream of the P40 TATA box) or into P41VP at nt 1637 (243 nt upstream of the P41 TATA box), to create P5P40VP, P7P40VP, V2ITRP40VP, and V5ITRP40VP or P5P41VP, P7P41VP, V2ITRP41VP, and V5ITRP41VP, respectively, as shown in the diagram in Fig. 2.
(iii) pHelper and Rep-expressing plasmids. pHelper plasmid was from Stratagene (TX). Rep expression constructs were driven by the human immunodeficiency virus (HIV) promoter within the HIV long terminal repeat promoter and have been described previously (13, 14).
(iv) AAV2/AAV5 chimeric Rep plasmids and the P5P40 luciferase reporter plasmid. Chimeric Rep constructs were based on the HIV-driven, AAV2, and AAV5 Rep-expressing parental constructs previously described (13, 14). The numbers in the name of each construct designate exactly where the chimeric borders were made. All fusions were made in frame and sequenced to ensure that they were as predicted. The V2 211 (V5) series was based on the V2 211 parental construct. To generate the P5P40 luciferase construct, the AAV2 P5 promoter (nt 146 to 320 from AAV2) was cloned upstream of the luciferase gene driven by AAV2 P40 (nt 1700 to 1883), to provide an RBE, but was cloned in an inverted orientation relative to the direction of P40 transcription to exclude potential P5-generated luciferase activity.
(v) Chimeric Rep-binding domain plasmids and the P5P40 minimal-transcription-unit test plasmid. The parental RepBD.TZ.AD and RepBD.TZ plasmids were gifts from Matt Weitzman (Salk Institute, San Diego, Calif.) and contain a modified leucine zipper dimerization domain from GCN4 (TZ), and in the case of RepBD.TZ.AD, the activation domain from herpesvirus VP16 (1). RepBD.TZ.AD- and RepBD.TZ-derived constructs containing either AAV5 or chimeric Rep sequences (V5 101-121, V5 121-161, and V5 161-181) were constructed by replacing the amino-terminal 244-amino-acid (aa) Rep-binding domain from the original AAV2-binding domain construct with either AAV5 or chimeric Rep sequences. The P5P40 minimal-transcription-unit test construct is the same as that described above.
RNase protection assays. Total RNA was isolated 36 to 41 h posttransfection as previously described (9, 16). RNase protection assays were performed as previously described (9, 16), using homologous antisense probes which spanned nt 1781 to 1923 (for AAV1), 1767 to 1906 (for AAV2), 1764 to 1903 (for AAV3), 1824 to 1957 (for AAV4), and 1766 to 1908 (for AAV6) of the individual P40 promoters and nt 1846 to 1985 of the P41 promoter of AAV5.
Luciferase assays. Luciferase assays were performed according to reference 6. Briefly, 293 cells grown in 12-well plates were transfected with the P5P40 luciferase reporter construct together with pHelper and the Rep expression constructs described in the text at a 1:4:0.25 ratio. In addition, 0.05 μg per well of cytomegalovirus (CMV)-driven -galactosidase reporter was cotransfected as an internal control. The total amount of DNA transfected into each well was kept at 0.6 μg. Thirty-six hours after transfection, cells were lysed with lysis buffer containing 1% Triton X-100. Equal amounts of the sample were used to test luciferase activity and -galactosidase activity according to the manufacturer's instructions (Applied Biosystems [Tropix], Bedford, MA). Each experiment represents the average values for duplicate samples from three individual experiments (error bars are shown in the relevant figures), each normalized to -galactosidase activity.
Rep inhibition assays. The activity of the P5P40 luciferase construct in response to an HIV-driven AAV2 Rep plasmid was first assayed (data not shown), and the lowest amount of Rep-expressing plasmid giving the highest level of activity was determined. Luciferase activity was then monitored in assays in which the total DNA concentration was kept constant, following addition of decreasing amounts of the HIV-driven AAV2 Rep-expressing plasmid, using either empty vector or the HIV-driven AAV5 Rep-expressing plasmid to make up the balance. Each experiment represents the average values for duplicate samples from three individual experiments (error bars are shown in the relevant figures), each normalized -galactosidase activity.
RESULTS
The constitutive activity in 293 cells of AAV5 P41 in RepCap constructs is high and is not significantly activated further by its large Rep protein or Ad5 infection. Maximal levels of activation of AAV2 P40, in the presence of Ad, occur from full-length, ITR-containing clones. While the ITR can supply the RBE required for Rep activation of AAV2 P40, it is likely that these maximal levels of activation are partially attributable to amplification of templates due to replication, because, as mentioned above, in the presence of adenovirus, the AAV2 P5 RBE can fully support P40 activation in nonreplicating AAV2 RepCap constructs (14). Thus, for our analyses, we chose to utilize ITR-deleted RepCap constructs to examine the effect of Ad and Rep on the activation of the capsid gene promoters of various AAV serotypes, thereby eliminating expression differences due to variation in viral replication under these different conditions. Although the AAV5 P7 promoter does not contain a consensus RBE, the AAV5 RepCap construct could be included in this analysis because the level of RNA generated from the AAV5 capsid gene promoter (P41), in an ITR-lacking AAV5 RepCap clone, was surprisingly high (13).
As shown in Fig. 1A, expression in 293 cells of both the upstream P5 and P19 promoters and the P40 promoters within Rep-expressing RepCap constructs of various AAV serotypes was high, both in the absence and in the presence of Ad5. However, the P40 promoters of AAV1, AAV2, AAV3, AAV4, and AAV6 were all activated by Ad5 significantly more than was the P41 promoter of AAV5 (Fig. 1A, compare lanes 1 through 8, 11, and 12 to lanes 9 and 10), both as measured directly (Fig. 1C) and as a percentage of total RNA (Fig. 1D, compare white bars to black bars). Identical amounts of RNA were used for each protection. Because no suitable internal transfection control whose expression was independent of Ad5 was identified, direct activation levels (Fig. 1C) were calculated as the averages for at least three separate complete experiments that included all test plasmids, and various plasmid preparations were tested. Southern analysis was initially used to detect and standardize for cell-associated plasmid DNA after transfection as previously reported (11). However, for our experiments, this was not proportional to DNA that had entered the nucleus (data not shown) and so was also deemed an unsuitable control for these purposes. As previously reported (14), in 293 cells the activity of the P40 promoters in RepCap constructs remained relatively low compared to the activities of the P5 and P19 promoters, even in the presence of Ad5, and was significantly lower than that seen during AAV2 infection.
That the expression of the AAV5 P41 promoter was high in these assays was surprising because the AAV5 RepCap plasmid is not predicted to contain a Rep-binding site. This suggested that either AAV5 Rep activation of P41 did not require binding to the transcription template or P41 activity was independent of Rep under these conditions. The latter seemed to have been the case, because in contrast to what was observed for the other AAV serotypes, the activity of the P41 promoter also remained high in similar constructs in which expression of the Rep protein was abolished by the introduction of a translation termination signal at nt 480 in the region of the gene encoding the amino terminus of its product (RepStopCap plasmids), both in the presence and in the absence of Ad5 (Fig. 1B, compare lanes 13 through 20 and 23 through 24 to lanes 21 and 22; also Fig. 1D). The addition of homologous Rep proteins in trans to AAV1, AAV2, AAV3, AAV4, and AAV6 RepStopCap constructs enhanced their P40 promoters to various extents; however, AAV5 P41 was not further activated (data not shown). These results suggested that basal activity of P41 in 293 cells was constitutively high and not further enhanced by either Ad5 infection or Rep.
Expression of AAV5 P41 was also independent of the AAV5 Rep protein in HeLa cells, and in these cells, the difference between AAV5 P41 and AAV2 P40 was even more marked. In HeLa cells, the basal levels of both the upstream promoters and the capsid gene promoters of both AAV2 and AAV5 RepCap constructs were considerably less than those seen in 293 cells (Fig. 1E, lanes 1 and 5, respectively), and the levels of both of these promoters were thus dramatically enhanced by adenovirus (Fig. 1E, lanes 2 and 6, respectively). Full levels of activation for the AAV2 promoters in HeLa cells were dependent upon the Rep protein (Fig. 1E, compare lanes 1 and 2 to lanes 3 and 4); however, as in 293 cells, the activity of the AAV5 constructs was independent of Rep expression (Fig. 1E, compare lanes 5 and 6 to lanes 7 and 8). Furthermore, these results suggested that the higher activity of AAV5 P41 in 293 cells shown in Fig. 1B was likely due to the resident Ad E1A and/or E1B gene products. Thus, while expression of the AAV2 P40 promoter requires Ad5 infection plus Rep in 293 cells, expression of AAV5 P41 can be sustained by E1A and E1B alone.
Compared at optimal alignment, there are significant differences between the two promoters that may account for the higher basal activity of P41 (Fig. 1F). Interestingly, the Sp1-50 binding site, which is important for AAV2 Rep activation of P40 (11), is significantly altered in P41. A comprehensive comparison of the two promoters is in progress.
Although the P41 promoter did not require Rep for activation, because the AAV5 RepCap constructs are predicted to lack an AAV5 Rep-binding site, the intrinsic transcription-transactivating potential of the AAV5 Rep protein needed to be tested in another way.
In contrast to the AAV2 Rep protein, the AAV5 Rep protein is a poor transactivator of the inducible AAV2 P40 promoter. To examine the activation potential of AAV5 Rep further, we chose to examine the transactivation of the AAV2 P40 promoter resident within a series of minimal constructs to which various naturally occurring potential Rep-binding sites were attached. As expected, the AAV5 P7 promoter region, which lacks a consensus binding site for either the AAV2 or the AAV5 Rep proteins, conferred to P40 only a modest amount of activation by either the AAV2 or the AAV5 Rep proteins (Fig. 2A, lanes 7 to 9) compared to the background parent P40VP (Fig. 2A, lanes 1 to 3). When a Rep-binding element was supplied by the AAV2 P5 promoter (P5P40VP), activation by AAV2 Rep increased approximately sevenfold over background levels (Fig. 2A, compare lanes 4 and 5); however, activation by AAV5 Rep remained approximately twofold over background levels (Fig. 2A, compare lanes 4 and 6). When the AAV2 ITR was used to supply the RBE (V2ITRP40VP), transactivation by AAV2 Rep, but not by AAV5 Rep, was enhanced further yet (Fig. 2A, compare lane 10 to lanes 11 and 12, respectively). The AAV5 Rep protein exhibited somewhat higher activation of AAV2 P40 when the AAV5 ITR was linked in cis (V5ITRP40VP), approximately three- to fourfold over background levels, similar to levels achieved by AAV2 Rep with this construct (Fig. 2A, compare lane 13 to lanes 14 and 15). The AAV5 and AAV2 Rep proteins accumulated to similarly high amounts following transfection of 293 cells as detected by Western blot analysis (data not shown), and the AAV5 Rep protein was also shown to bind these elements as well as did the AAV2 Rep protein, as measured by the ability to target the VP16 activator to a test template (see below).
These results suggested that the AAV5 Rep protein was a significantly less potent activator of the inducible P40 promoter than was AAV2, even when targeted to the transcription template. As expected, the AAV5 P41 promoter was only poorly inducible by either AAV2 or AAV5 Rep, regardless of the nature of the RBE linked in cis, primarily because of its higher basal activity (Fig. 2B).
A difference in activation potential between AAV2 and AAV5 Rep can be mapped to the N-terminal domains of the molecules. Although AAV2 and AAV5 Rep proteins share significant sequence homologies and can both bind to a (GAGY)3 consensus binding site with similar affinities in vitro (2), they activated the inducible AAV2 P40 promoter to significantly different levels when targeted to the transcription template by either the AAV2 P5 or the ITR RBE. To determine regions of the two Rep proteins that might distinguish their transactivation properties, we made a series of chimeras between the two (diagramed in Fig. 3A) and tested them for their ability to activate the inducible AAV2 P40 promoter. Constructs expressing luciferase from a P40 promoter with an RBE-containing AAV2 P5 element linked upstream in an orientation inverted relative to the direction of transcription of P40 (Fig. 3B, top) were cotransfected in 293 cells with the pHelper plasmid (Invitrogen), which supplies Ad5 E2A, E4orf6, and virus-associated (VA) RNA. All of the wild-type and chimeric Rep proteins assayed in the following experiments accumulated to similar levels as assayed by Western blotting (data not shown).
Consistent with results presented in the previous sections, the full-length AAV2 Rep protein was a strong activator in this system (Fig. 3B, lane V2). The full-length AAV5 Rep protein, however, which bound the AAV2 P5 RBE to an extent similar to that for AAV2 Rep in the heterologous binding assay described below, showed dramatically reduced activation potential (Fig. 3B, lane V5). Analysis of chimeras in which carboxyl-terminal residues of two proteins were exchanged demonstrated that the amino-terminal regions of these Rep proteins governed their transactivation properties in these assays. Both V2 362 and V2 211, which contained the AAV2 Rep amino-terminal domain, retained transactivation activities similar to that of AAV2 Rep, while V5 362 and V5 211, which contained the amino-terminal region of AAV5 Rep, were as inactive in this assay as they were for the wild-type AAV5 Rep protein (Fig. 3B). Based on these results, chimeras within the first 211 amino acids were generated within the AAV5 Rep background (within the V2 211 parent) to attempt to identify regions which were responsible for differences between the activation properties of these proteins. These chimeras exhibited variability in their ability to transactivate the test template. V2 211-generated Rep proteins, in which either aa 181 to 211 or aa 161 to 181 were replaced with residues of AAV5 Rep, retained approximately half of the transactivation activity of the parent protein [Fig. 3B, lanes V2 211 (V5 181-211) and V2 211 (V5 161-181), respectively]. Replacement of V2 211 amino acids 101 to 121 with those of AAV5 reduced the activation activity of the molecule further [Fig. 3B, lanes V2 211 (V5 101-121)], while the replacement of amino acids 1 to 57, 57 to 101, or 121 to 161 reduced activity to the very low levels seen for the AAV5 wild-type Rep molecule [Fig. 3B, lanes V2 211 (V5 1-57), V2 211 (V5 57-101), and V2 211 (V5 121-161), respectively]. These results suggested that a region that lies within the previously identified DNA-binding region of AAV2 Rep governs its activation properties. While the full 211 amino termini of these Rep proteins are important, these results generally suggest that the amino-terminal 100 amino acids are most critical for activity.
Although the chimeras were made from two parents, each competent for binding to an RBE, three amino-terminal chimeras were further analyzed to reveal any potential differences in their ability to bind to the P5 RBE, which might account for differences in their activation activities (Fig. 4). This was done using a heterologous reporter assay, in which DNA binding was measured indirectly via targeting of the VP16 activation domain. The amino-terminal 244-aa Rep-binding domains from the wild-type or chimeric Rep proteins assayed above, fused with a modified leucine zipper dimerization domain from GCN4 (TZ) (1) and the VP16 activation domain, were assayed for levels of VP16-dependent activation of a P5P40 test plasmid (diagramed in Fig. 4A, top). As expected from previously published reports (2), the AAV2 and AAV5 amino-terminal binding domains (V2 BD.TZ.AD and V5 BD.TZ.AD, respectively) exhibited similar levels of binding to the P5 RBE (as measured by VP16 activation of P40 [Fig. 4A, compare lanes 3 and 4 to lanes 5 and 6]). Three of the chimeras tested for activation as described above, V5 161-181 BD.TZ.AD, which exhibited the highest level of activation, V5 101-121 BD.TZ.AD, which exhibited an intermediate level of activation, and V5 121-161 BD.TZ.AD, which exhibited little if any activation, all exhibited similar levels of binding to the P5 RBE as measured in this assay (Fig. 4A, compare lanes 7 and 8, 9 and 10, and 11 and 12, respectively). Thus, the chimeras bound to the AAV2 P5 RBE as well as the two parents, and gross differences in binding could not explain differences in their activation properties. Activity over vehicle transfection alone (SK) was not detected following transfection of a Gal4VP16 chimera (Gal-VP) (Fig. 4A, compare lanes 1 and 2) or when the VP16 activation domain was not present on the Rep-binding domain/GCN4 leucine zipper domain fusions (Fig. 4A, compare lane 2 to lanes 3, 5, 7, 9, and 11). Activation of the P5 promoter by VP16 showed a similar profile (Fig. 4A, arrowhead). All Rep fusion proteins were expressed at similar levels in these assays (data not shown). These results suggested that the transactivation potentials of the two Rep proteins could be separated from their binding activities, at least in these simple binding assays.
Since both the AAV2 and the AAV5 Rep proteins bound to the AAV2 P5 RBE yet show significant differences in activation activity, we tested whether the AAV5 Rep protein could compete for, and thus inhibit, AAV2 Rep activation of a P5 RBE-linked reporter template. As seen in Fig. 4B, the addition of excess AAV5 Rep-expressing plasmid, but not the plasmid vehicle, inhibited activation of these reporter templates in response to a broad range of AAV2 Rep-expressing plasmids.
DISCUSSION
Parvoviruses generate capsid-encoding mRNAs either from an internal promoter (6, 14) or by RNA processing of a large pre-mRNA generated from a single promoter (12, 19). In previously characterized examples, expression from internal capsid gene promoters is low in all cell types examined prior to activation either by a viral protein alone (the large nonstructural viral protein for the Parvovirus genera and goose parvovirus) or by the viral Rep protein plus helper virus (for the AAV group of viruses). In this work, we show that within RepCap constructs lacking ITR sequences, expression in 293 cells of the AAV5 P41 capsid gene promoter is high and is not further activated by either its large Rep protein or Ad5 infection. In addition, the AAV5 Rep protein is itself a poor activator of the inducible AAV2 P40 promoter. The sequences immediately upstream of the P41 and P40 promoters are quite different, and this may play a role in the different activities of these promoters in minimal constructs.
Does this imply that, unlike for other parvoviruses, expression of the AAV5 capsid proteins is unregulated during infection It may be that even though P41 basal activity is high within minimal expression constructs in 293 cells, capsid expression from the viral genome may be regulated during infection in response to Rep and the presence of helper virus. This is currently being examined. High basal activity of P41 requires Ad E1A and/or E1B, suggesting that helper virus gene products, and perhaps Rep, play a role in P41 expression during infection.
Our results also suggest that efficient transcriptional activation by AAV2 Rep is likely dependent on the nature of its interaction with the RBE-binding site in vivo. Surprisingly, AAV2 Rep activation of the V5ITRP40VP construct was less than for those containing an authentic AAV2 RBE. Although AAV2 Rep binds both of these sites with similar affinities in in vitro gel shift assays (2), AAV2 Rep binding to the AAV5 ITR is not sufficient to fully activate the AAV2 P40 promoter. Similarly, a synthetic minimal RBE site alone is not sufficient to allow Rep-mediated transactivation of the AAV2 P40 promoter; flanking sequences including the TRS are required for optimal transactivation (data not shown) (14). In addition, AAV2 and AAV5 Rep proteins cannot substitute for one another during DNA replication, because the TRS sequences of the AAV2 and AAV5 ITRs are functionally different (2). Thus, the natures of AAV2 Rep and AAV5 Rep binding to their respective ITRs in vivo are different, and this may account for their differences in activation potential. Alternatively, these binding sites may themselves differ in their ability to bind additional cellular factors required for the activation process.
Our examination of chimeras between the AAV2 and AAV5 Rep proteins has revealed a separation between the RBE-binding and transactivation capabilities of Rep. Amino-terminal chimeras which transactivated strongly, intermediately, or weakly showed indistinguishable binding to the AAV2 P5 RBE in heterologous binding assays. Thus, while the amino termini of both AAV2 and AAV5 Rep are interchangeable in regard to binding of the RBE in our assay, the amino terminus of AAV2 Rep contains regions required for transactivation that are lacking in AAV5 Rep.
An independent activation domain of Rep has not yet been identified; however, the DNA-binding domain of Rep has been localized to the amino-terminal region (7, 10, 17, 18), and the structure of the AAV5 Rep DNA-binding domain (aa 1 to 197) has been determined (4). This region of AAV5 Rep protein displays a ferredoxin-like fold, which is topologically homologous to numerous other DNA-binding proteins, most notably simian virus 40 T antigen and the DNA-binding domain of the replication initiation protein E1 of bovine papillomavirus. Interestingly, the core elements responsible for this high homology (4, 1, 3, 2, B, and C) (4) reside in the region of AAV5 Rep that, when transferred to the analogous region of AAV2 Rep, most significantly impaired its ability to activate in our assays (1 to 57: 1; 57 to 101: B, 2, and 3; 101 to 121: C; 121 to 161: 4).
Furthermore, the structure of AAV5 Rep aa 1 to 197 complexed with the AAV5 ITR RBE has also been determined (5). Seven amino acids that make contact with the RBE are conserved between the two proteins (5), yet they fall into the same region (aa 121 to 161) which our assays identify as functionally different between the two with regard to transactivation. However, differences between the two proteins in this region—specifically, that AAV2 Rep has a single Arg at residue 138 rather than the Lys137 and Lys138 residues of AAV5 Rep and that for AAV5, the downstream Gly139 and Gly140 are followed by Ala141 rather than the three sequential glycines found for AAV2 Rep—have led to the prediction that the interaction of AAV5 with its RBE may be qualitatively different from the interaction of the other AAV Rep proteins with their RBEs (5). Thus, while the conserved similarities between the AAV2 and AAV5 Rep-binding domains may help explain why the two proteins behave comparably in our simple binding assays, it will be important to determine whether differences within the region required for transactivation which have been predicted to result in subtly different DNA-binding properties account for differences in the transactivating activities of the two Rep proteins.
ACKNOWLEDGMENTS
We thank Lisa Burger and Fang Cheng for excellent technical assistance and Greg Tullis for critical reading of the manuscript.
This work was supported by PHS grants RO1 AI46458, RO1 AI56310, and RO1 AI21302 from NIAID to D.J.P.
REFERENCES
Cathomen, T., D. Collete, and M. D. Weitzman. 2000. A chimeric protein containing the N terminus of the adeno-associated virus Rep protein recognizes its target site in an in vivo assay. J. Virol. 74:2372-2382.
Chiorini, J. A., S. Afione, and R. M. Kotin. 1999. Adeno-associated virus (AAV) type 5 Rep protein cleaves a unique terminal resolution site compared with other AAV serotypes. J. Virol. 73:4293-4298.
Chiorini, J. A., L. Yang, Y. Liu, B. Safer, and R. M. Kotin. 1997. Cloning of adeno-associated virus type 4 (AAV4) and generation of recombinant AAV4 particles. J. Virol. 71:6823-6833.
Hickman, A. B., D. R. Ronning, R. M. Kotin, and F. Dyda. 2002. Structural unity among viral origin binding proteins: crystal structure of the nuclease domain of adeno-associated virus Rep. Mol. Cell 10:327-337.
Hickman, A. B., D. R. Ronning, Z. N. Perez, R. M. Kotin, and F. Dyda. 2004. The nuclease domain of adeno-associated virus rep coordinates replication initiation using two distinct DNA recognition interfaces. Mol. Cell 13:403-414.
Lorson, C., and D. J. Pintel. 1997. Characterization of the minute virus of mice P38 core promoter elements. J. Virol. 71:6568-6575.
McCarty, D. M., T. H. Ni, and N. Muzyczka. 1992. Analysis of mutations in adeno-associated virus Rep protein in vivo and in vitro. J. Virol. 66:4050-4057.
Mouw, M. B., and D. J. Pintel. 2000. Adeno-associated virus RNAs appear in a temporal order and their splicing is stimulated during coinfection with adenovirus. J. Virol. 74:9878-9888.
Naeger, L. K., R. V. Schoborg, Q. Zhao, G. E. Tullis, and D. J. Pintel. 1992. Nonsense mutations inhibit splicing of MVM RNA in cis when they interrupt the reading frame of either exon of the final spliced product. Genes Dev. 6:1107-1119.
Owens, R. A., M. D. Weitzman, S. R. Kyostio, and B. J. Carter. 1993. Identification of a DNA-binding domain in the amino terminus of adeno-associated virus Rep proteins. J. Virol. 67:997-1005.
Pereira, D. J., and N. Muzyczka. 1997. The adeno-associated virus type 2 p40 promoter requires a proximal Sp1 interaction and a p19 CArG-like element to facilitate Rep transactivation. J. Virol. 71:4300-4309.
Qiu, J., F. Cheng, L. R. Burger, and D. Pintel. 2006. The transcription profile of Aleutian mink disease virus in CRFK cells is generated by alternative processing of pre-mRNAs produced from a single promoter. J. Virol. 80:654-662.
Qiu, J., R. Nayak, G. E. Tullis, and D. J. Pintel. 2002. Characterization of the transcription profile of adeno-associated virus type 5 reveals a number of unique features compared to previously characterized adeno-associated viruses. J. Virol. 76:12435-12447.
Qiu, J., and D. J. Pintel. 2002. The adeno-associated virus type 2 Rep protein regulates RNA processing via interaction with the transcription template. Mol. Cell. Biol. 22:3639-3652.
Rutledge, E. A., C. L. Halbert, and D. W. Russell. 1998. Infectious clones and vectors derived from adeno-associated virus (AAV) serotypes other than AAV type 2. J. Virol. 72:309-319.
Schoborg, R. V., and D. J. Pintel. 1991. Accumulation of MVM gene products is differentially regulated by transcription initiation, RNA processing and protein stability. Virology 181:22-34.
Weitzman, M. D., S. R. Kyostio, B. J. Carter, and R. A. Owens. 1996. Interaction of wild-type and mutant adeno-associated virus (AAV) Rep proteins on AAV hairpin DNA. J. Virol. 70:2440-2448.
Yang, Q., and J. P. Trempe. 1993. Analysis of the terminal repeat binding abilities of mutant adeno-associated virus replication proteins. J. Virol. 67:4442-4447.
Yoto, Y., J. Qiu, and D. J. Pintel. 2006. Identification and characterization of two internal cleavage and polyadenylation sites of parvovirus B19 RNA. J. Virol. 80:1604-1609.(Chaoyang Ye, Jianming Qiu)