An Adenovirus Type 5 Mutant with the Preterminal Protein Gene Deleted Efficiently Provides Helper Functions for the Production of Recombinant Adeno-Associated Virus

Abstract

Production of recombinant adeno-associated virus (rAAV) requires helper functions that have routinely been provided by infection of the producer cells with adenovirus. Complete removal and/or inactivation of progeny adenovirus, present in such rAAV preparations, presents significant difficulty. Here, we report that an adenovirus type 5 (Ad5) mutant with the preterminal protein (pTP) gene deleted can provide helper function for the growth of rAAV. At high multiplicity, Ad5dl308_ΔpTP was as efficient as the phenotypically wild-type Ad5dl309 in permitting growth of rAAV. Use of Ad5dl308_ΔpTP, which is incapable of replication in the absence of complementation for pTP, as a helper avoids the need to remove contaminating adenovirus infectious activity by heat inactivation or by purification. Comparison of the transducing ability of rAAV generated with either Ad5dl308_ΔpTP or Ad5dl309 as a helper demonstrated that the heat inactivation protocol generally used does not remove all of the helper Ad5dl309 function.

The development of adeno-associated virus (AAV) as a gene transfer vector has been hindered by a lack of convenient and efficient methods for generating pure stocks of recombinant AAV (rAAV) (9, 12). A significant difficulty in this regard is the need to supply helper functions (1, 2, 5, 6, 18, 19). Although this is efficiently achieved by adenovirus coinfection of the producer cells, this results in the concomitant production of adenovirus progeny, which are difficult to remove or inactivate completely. Contamination with replication-competent adenovirus is not only unacceptable for gene therapy applications but also significantly influences the transduction efficiency achieved with rAAV vectors, thus complicating assessment of their utility (3, 4). Since adenovirus early functions are sufficient to support AAV replication (19), we investigated the use of a mutant blocked in the late functions as a helper for rAAV production. A deletion mutant in the preterminal protein (pTP) gene of adenovirus type 5 (Ad5), termed Ad5dl308_ΔpTP, was previously described (16), together with complementing, pTP-expressing cell lines that support productive infection by this mutant (10, 15). In noncomplementing cells, Ad5dl308_ΔpTP was shown to be completely defective for DNA synthesis and expression of late functions (16). In contrast, other adenovirus mutants used to provide helper function for rAAV show significant leakiness for replication (18). Here, we report that Ad5dl308_ΔpTP can function as an efficient helper for generating rAAV stocks, free from replication-competent adenovirus.

Stocks of adenovirus helpers, Ad5dl309 and Ad5dl308_ΔpTP, were produced in 293 cells and 293-pTP cells (10, 15), respectively. Ad5dl309 has a deletion and insertion in the E3 region but shows replication equivalent to wild-type Ad5 in cell culture (8). In contrast, Ad5dl308_ΔpTP is incapable of either replicating or directing synthesis of late region products in the absence of complementation. Even with a sensitive RNase protection assay, late transcripts were undetectable in 293 cells infected with this virus (16). Both helpers were purified by isopycnic centrifugation in CsCl. The infectious titers of Ad5dl309 and Ad5dl308_ΔpTP were determined by plaque assay in 293 cells and 293-pTP cells, respectively. Since the plaque assay is technically more difficult in 293-pTP cells, the titer of Ad5dl308_ΔpTP was verified with absorbance measurement to determine the particle concentration. We have consistently observed similar PFU-to-particle ratios (≈1:100) for both Ad5dl309 and Ad5dl308_ΔpTP.

rAAV was produced by transient cotransfection of pAAV.CMV.LUC (11) and pAAV/Ad (13, 14) into the human simian virus 40-transformed cell line 324K by electroporation; we previously reported advantages of these modifications of standard transfection methods (11). pAAV.CMV.LUC contains an rAAV genome with the luciferase reporter driven by the cytomegalovirus immediate-early promoter (11). Immediately after electroporation, the cells were infected either with the phenotypically wild-type Ad5dl309 or with Ad5dl308_ΔpTP. Medium was replaced after 2 to 6 h, and culture media and cell extracts containing AAV.CMV.LUC were harvested at the times indicated (11). Transducing activities of the resulting rAAV preparations were determined by infection of fresh, recipient 324K cells (at 30 to 40% confluence in 12-well plates) and an assay for luciferase activity in cell extracts prepared 2 days after infection by using the Promega luciferase assay kit with a Turner Instruments luminometer.

As shown in Fig. 1, Ad5dl308_ΔpTP was effective as a helper for rAAV production, although a higher multiplicity of infection (MOI) of this virus than of Ad5dl309 was required. Thus, similar yields of transducing AAV.CMV.LUC were obtained by using Ad5dl308_ΔpTP at 100 PFU per cell or Ad5dl309 at 10 PFU/cell. In several experiments performed under these conditions, the apparent yield of AAV.CMV.LUC obtained with Ad5dl308_ΔpTP ranged from 64 to 104% of that with Ad5dl309. Since these measurements were made with heat-treated virus preparations (see below), these figures reflect a slight underestimate of the relative yield of rAAV obtained by using Ad5dl308_ΔpTP as a helper. Yields were decreased 10- to 100-fold if the MOI of Ad5dl308_ΔpTP was lowered to 10 PFU/cell (Fig. 1). The higher MOI required for Ad5dl308_ΔpTP probably reflects the need to express helper functions at levels comparable to those provided by Ad5dl309 after replication. These data demonstrate that Ad5dl308_ΔpTP efficiently provides helper function for rAAV and, further, that pTP expression is not required when helper function is provided by adenovirus infection.

FIG. 1.

Efficient production of luciferase-transducing rAAV with Ad5dl308_ΔpTP as a helper. 324K cells were suspended and electroporated with pAAV.CMV.LUC plus pAAV/Ad, as previously described (11), and were infected by dilution into medium containing either Ad5dl309 or Ad5dl308_ΔpTP at the estimated MOI indicated, expressed as PFU per surviving cell (cell survival from electroporation was ≈50%). After 3.5 h, the inoculum was aspirated and the attached cells were rinsed gently and given fresh medium. After 4 days, culture supernatants and frozen-thawed cell extracts were collected (11). Samples were incubated at 56°C for 30 min before being assayed for transducing activity in fresh recipient 324K cells (11). Values on the ordinate are luciferase transducing activity (total light units [LU]) calculated for supernatant plus cell extract; measurements ranged from 3 to 500 LU above the luminometer background of 0.01 LU. The data shown are from a representative experiment. In repeated experiments with some variations in time of exposure to the adenovirus inoculum and time of harvest, the transducing activities generated (total LU) were (1.15 ± 0.92) × 10⁴ and (2.17 ± 0.26) × 10⁵ with Ad5dl308_ΔpTP at 10 and 100 PFU/cell, respectively, and (2.77 ± 0.51) × 10⁵ with Ad5dl309 (mean ± standard deviation for three experiments in each case).

In the experiment represented in Fig. 1, rAAV was harvested at 4 days after transfection/infection of the producer cells. In further experiments, similar yields of AAV.CMV.LUC transducing activity were obtained whether harvesting was performed after 4 or 5 days, although the yield was substantially lower after only 3 days. Infection with Ad5dl308_ΔpTP at high MOI (100 PFU/cell) resulted in considerable cell detachment at later times. However, the proportion of transducing activity found in cell extract versus supernatant medium was similar to that reported previously for rAAV production from 324K cells with Ad5dl309 as a helper (11). (Approximately 60% of the total activity was recovered in the extract from combined attached plus detached cells [data not shown].)

The results presented in Fig. 1 were obtained by using samples of rAAV preparations that had been heated for 30 min at 56°C, prior to transduction of the recipient cells. A similar heat treatment is used routinely in many laboratories to inactivate the adenovirus present in rAAV preparations. Figure 2A shows the effects of omission of this heating step on the transducing activity of AAV.CMV.LUC generated with either helper. This omission increased the apparent transducing activity of the Ad5dl309-helped rAAV markedly (50- to 100-fold), presumably due to the well-established stimulatory effect of adenovirus coinfection on rAAV transduction, probably mediated via enhancement of second-strand synthesis (3, 4). In contrast, omission of heating had little effect (approximately twofold) on transduction by Ad5dl308_ΔpTP-helped AAV.CMV.LUC (Fig. 2A), consistent with the expected lack of replication of the Ad5dl308_ΔpTP virus in the producer cells. The minor effect seen was most probably due to thermal inactivation of a proportion of the rAAV itself.

FIG. 2.

Absence of contaminating adenovirus helper activity in AAV.CMV.LUC stock, generated with Ad5dl308_ΔpTP as a helper, as indicated by the lack of effect of incubation at 56°C (A) and strong stimulation of transduction by added adenovirus Ad5dl309 (B). (A) Samples of the same rAAV preparations as in Fig. 1 were assayed for transducing activity with (+56°) or without prior incubation at 56°C. Note that this treatment had little effect on the Ad5dl308_ΔpTP-helped AAV.CMV.LUC, whereas the apparent transducing activity of the Ad5dl309-helped rAAV was ≈100-fold higher when heat treatment was omitted. (B) Recipient 324K cells were coinfected with heated cell extracts (56°C, 30 min) containing AAV.CMV.LUC, generated with either Ad5dl308_ΔpTP or Ad5dl309 helper virus, together with Ad5dl309 added at the MOI indicated. Luciferase activity, assayed after 2 days, is expressed as percentage of the activity observed with the maximum amount of added Ad5dl309 (MOI, 3 PFU per cell). Note the differential enhancement of transducing activity of the Ad5dl308_ΔpTP-helped rAAV at low MOI of added Ad5dl309.

To examine further the stimulatory effect of coinfecting adenovirus on transduction, 324K cells were transduced with AAV.CMV.LUC generated with either helper adenovirus, together with increasing MOI of added Ad5dl309, and transduced luciferase activity was assayed after 2 days. As shown in Fig. 2B, although Ad5dl309 stimulated transduction by both virus preparations, this effect was considerably stronger for the Ad5dl308_ΔpTP-helped rAAV than for the Ad5dl309-helped rAAV, particularly at lower MOI of added Ad5dl309. We interpret this result to indicate that the Ad5dl309-helped rAAV preparation retained a low but significant level of adenovirus that survived the 56°C treatment and stimulated transduction, so that low levels of added Ad5dl309 showed little additional stimulatory effect. These results provide further support for the absence of contaminating adenovirus from rAAV generated with Ad5dl308_ΔpTP as helper.

As a further comparison of the yield of rAAV obtained with Ad5dl308_ΔpTP and Ad5dl309 helpers, we performed semiquantitative PCR on viral DNA with luciferase primers. Virus samples were first treated extensively with DNase I to degrade any contaminating plasmid or other DNA not protected by packaging in virions. As shown in Fig. 3, similar signal intensities were observed regardless of which adenovirus helper had been used to generate the AAV.CMV.LUC. These results further substantiate the conclusion, from relative transducing activity (Fig. 1), that the Ad5dl308_ΔpTP helper, used at high MOI, gives yields of rAAV comparable to those obtained with a conventional, replication-competent adenovirus helper.

FIG. 3.

Comparison of rAAV yield obtained with Ad5dl308_ΔpTP (versus Ad5dl309) helper virus by PCR analysis of packaged AAV.CMV.LUC DNA. Viral supernatants containing rAAV, generated with either helper, were incubated with DNase I (200 μg/ml, 37°C, 4 h; control experiments showed complete digestion of plasmid DNA under these conditions), and the enzyme was inactivated at 64°C for 15 min. PCR was performed with luciferase primers for 30 cycles, as described previously (17), on diluted samples of the DNase-treated supernatants. Under these conditions, packaged DNA is rendered available as a template for PCR by disruption of the virions in the initial incubation at 94°C. The products were analyzed by electrophoresis in a 1.5% agarose gel and staining with ethidium bromide. As standards, samples containing the indicated amounts of a linearized luciferase plasmid were amplified in parallel. Lanes 1 to 4, plasmid standards (0, 3, 30, and 300 fg, respectively); lane 5, marker DNA (123-bp ladder [GIBCO BRL]); lanes 6 to 8, rAAV generated with Ad5dl308_ΔpTP helper (samples equivalent to 1, 10, and 100 nl of supernatant, respectively); lanes 9 to 11, rAAV generated with Ad5dl309 helper (samples equivalent to 1, 10, and 100 nl of supernatant, respectively).

In most laboratories, rAAV is routinely produced by transient cotransfection of a plasmid containing the recombinant genome, together with a second plasmid expressing AAV Rep and Cap, with concomitant adenovirus infection of the producer cells to supply the necessary helper functions (E1A, E1B, E2A, E4 and VA RNA) (9, 12, 19). Efficient introduction of these functions by virus infection has heretofore been at the cost of contamination of rAAV by progeny adenovirus. Use of Ad5dl308_ΔpTP as a helper, as described here, removes this limitation. An alternative means of supplying relevant adenovirus functions from a cotransfected expression plasmid has recently also been shown to give efficient production of rAAV, free from adenovirus (19). Direct comparison of the yield of rAAV from this system with those we have obtained using Ad5dl308_ΔpTP is difficult for several reasons, relating to the use of different assays and to the fact that the transfection system also used a novel, replication-deficient AAV helper plasmid (19). The latter system requires the use of the 293 cell line to provide E1A and E1B functions which are not encoded by the Ad5-derived plasmid. In contrast, Ad5dl308_ΔpTP can be used as helper for rAAV production in other cells, such as the 324K line, which offers certain practical advantages as described previously (11). Other systems under development for achieving higher production of rAAV seek to avoid transfection procedures by incorporating conditionally expressed AAV helper sequences (rep and cap) into an adenovirus helper or into herpes simplex virus-based helpers (2, 7). Such an approach based on pTP-deleted adenovirus might enable high efficiency of rAAV production without generating adenovirus progeny.