Recombinant adenoviruses (rAV) are among the most efficient vectors for gene therapy purposes and have rapidly become the vector of choice in many human gene therapy protocols. The rAV is used as a vehicle for transport of a therapeutic gene. For most applications, replication of the vector is undesirable (eg for the treatment of genetic disorders). Therefore, the majority of rAVs currently used in gene therapy have a deletion in early region 1 (E1) of the viral genome, where novel genetic information can be introduced. The E1 deletion renders the recombinant virus replication-defective, and thus provides an important safety feature. In specific cases, however, one can choose the use of a replication-competent rAV which is targeted to specific cells (eg tumor cells for the treatment of cancer). E1-deleted rAV can be propagated on dedicated helper cells, specialized cells that provide the E1 functions in trans, such as cell lines 293 and 911. Although encouraging results have been obtained so far with rAV, two major problems are associated with the use of rAVs: viz. the host immune response against the adenovirus particles and the transduced cells, and the generation of replication-competent adenovirus (RCA) during manufacture of rAV lots. RCA are revertant vectors that reacquired the E1 region as a result of homologous recombination with E1 sequences integrated in the helper cells. Recently, we described a new helper cell line, PER.C6, and matched, nonoverlapping E1-deleted vectors that eliminate the generation of RCA by homologous recombination.
Here, we focus on the RCA phenomenon itself. First, we will review current information on RCA. Subsequently, we will discuss the occurrence and properties of various replication-competent or -defective rAV revertants that arose as a result of recombination during propagation of different rAVs on their helper cells.
Cell line 293 has been the most frequently used cell line for the production of adenoviral vectors. This cell line was generated in the 1970s by transfection of diploid human embryonic kidney cells with sheared adenovirus serotype 5 (Ad5) DNA. It was made in the course of a study on the transforming potential of the E1 genes of adenoviruses. Mapping of the Ad5 sequences in the genome of 293 cells indicated the presence of contiguous Ad5 sequences from the left-hand end of the genome up to position 4137.1 Thus, when typical E1 replacement vectors are propagated on 293 cells, there is sequence homology between vector and helper cell DNA of up to 450 basepairs at the left-hand side of the transgene, and approximately 800 basepairs at the right-hand side.
Owing to this sequence overlap, replication of rAV on 293 cells results in the generation of RCA. This was first reported by Lochmüller and co-workers,2 who used an E1+E3-deleted rAV that was passaged multiple times on 293 cells. They detected RCA that contained E1 but lacked E3. This suggested that a small fraction of the rAVs had regained E1 by homologous recombination between overlapping sequences in the rAV DNA and the adenovirus DNA that is present in 293 cells. This was later confirmed by Hehir and colleagues3 by propagation of Ad2-based rAV on the Ad5-transformed 293 cells: they detected RCA carrying the Ad5 E1 region. Despite the presence of the entire left-hand end of the Ad5 genome in 293 cells, all of the studied RCA isolates were found to be generated by two homologous recombination events upstream and downstream of the transgene, resulting in loss of the transgene and re-acquirement by the vector of the E1 region.
The appearance of RCA in rAV batches is a chance event and is therefore unpredictable and difficult to control. This is a significant problem for manufacturing GMP, in particular if large scale batches have to be prepared. A number of reports on the frequency of RCA formation during manufacture of rAVs has been published (Table 1). These data illustrate that with the conventional E1-deleted Ad5 (and Ad2) rAVs, RCA is generated with frequencies that frustrate the large-scale production of clinical lots of rAVs.
It should be noted that homologous recombination is not the only source of RCA. During the generation of rAV, RCAs can also be introduced into the system from outside. The classical method of rAV construction is to cotransfect the large ClaI fragment of Ad5 together with an adapter plasmid that carries the gene of interest into the helper cells. Incomplete restriction-enzyme digestion of the adenovirus DNA can (also) be responsible for RCA production (ie wild-type Ad5, in this example). The use of Ad genomes cloned in bacterial plasmids eliminates this risk. In addition, inadvertent cross-contamination can occur in laboratories where replication-competent adenoviruses are propagated.
Replication-competent adenoviruses derived from rAV that are currently known are similar to wild-type adenoviruses, except that in most cases the E3 region is deleted, which has never been observed in wild-type isolates.2,3 Most of the rAV used to date are derived from human adenovirus serotype 5 or 2. Ad2 and Ad5 are mainly associated with mild respiratory infections, and these viruses have a tropism mainly for epithelial cells. RCA derived from such vectors can be expected to result in disease similar to that caused by wild-type Ad5 and Ad2.
The presence of RCA in rAV batches to be used in human patients is clearly undesirable. RCA may replicate in an uncontrolled manner in the patient. Although the replication of the RCA is limited by the recipient's immune system, it is a potential hazard, especially in immuno-compromised patients. In addition, RCA can rescue the vector, increasing the amount of vector shed by the patient. Rescue of the vector by RCA has been observed in cotton rats, a rodent species that is permissive for human adenovirus replication.8 Furthermore, the presence of RCA is associated with inflammatory responses.9 Such inflammatory responses may be caused on the one hand by the fact that multiplication of the adenovirus causes tissue damage, or, on the other hand by the fact that large amounts of adenovirus proteins are synthesized that are toxic for cells (eg hexon, penton), and are immunogenic. Thus, the presence of RCA in rAV batches to be used in clinical trials is undesirable, as it may induce significant pathological side-effects. This is also recognized by regulatory bodies, such as the Food and Drug Administration (FDA). Therefore, labor-intensive and expensive RCA screening tests such as the tissue culture method, the supernatant rescue assay, and PCR assay are required.10 Screening for RCA has significantly increased the manufacturing costs of clinical rAV lots, and has led to delays in the onset of clinical studies.
Currently, intensive research efforts focus on the development of adenoviral vectors that have an altered tissue tropism. This is achieved by changing the genes encoding the capsid proteins such as fiber, hexon and penton. In these cases, the targets may be endothelium or smooth muscle cells, which are refractory to infection by wild-type Ad2 and Ad5. The presence of RCA in preparations of adenoviral vectors with altered tropism constitutes a potential safety risk. In this respect, it is noteworthy that adenoviruses with a tropism for endothelium have been shown to cause lethal infections in deer and mice.11,12 Woods and co-workers11 reported high mortality rates in deer upon infection with adenovirus. Mortality was caused by replication of the virus in endothelium of the animal, causing severe vasculitis. In mice, MAV (mouse adenovirus) can cause lethal infections by targeting the vascular endothelium of the brain.12 Also, in infants with an intact immune system, adenovirus infections can cause severe health problems and even death.13 Therefore, batches of rAV with an altered tropism, to be used in clinical trials, should be free of contaminating RCA.
To reduce the immunogenicity of the rAV, and to increase the insert capacity, several groups are developing strategies to produce rAVs that are deleted of all Ad genes (so-called 'gutless' adenoviruses). Gutless rAVs can be propagated using a helper virus. In the most efficient system to date, an E1-deleted helper virus is used with a packaging signal that is flanked by bacteriophage P1 loxP sites ('floxed'). Infection of the helper cells that express Cre recombinase with the gutless virus together with the helper virus with a floxed packaging signal should only yield gutless rAV, as the packaging signal is deleted from the DNA of the helper virus. However, if 293-based helper cells are used, the helper virus DNA can recombine with the Ad5 DNA that is integrated in the helper cell DNA. As a result, a wild-type packaging signal, as well as the E1 region is regained. Thus, also production of gutless rAV on 293- (or 911)-based helper cells can result in the generation of RCA if an E1-deleted helper virus is used.
Considering the magnitude of the problem, it goes without saying that much effort has been devoted to solving the RCA problem. Strategies to circumvent RCA generation during rAV production have been focused on reducing or eliminating the sequence homology between the vector and the packaging cell line.3,5,7 We have shown that the combination of PER.C6 helper cells and matched vectors that do not share homologous sequences eliminates the generation of RCA by homologous recombination.7 Note that in this system, homology can also be provided by plasmid-derived sequences, as the PER.C6 cell line has been generated by transfection with a cloned adenovirus E1 region. Hehir and colleagues3 demonstrated that deletion or relocation of the gene encoding the minor capsid protein IX resulted in a significant reduction of the frequency of RCA formation.
Another strategy that could prevent the formation of RCA is to delete additional essential genes from the vector backbone. Several of such strategies have been developed, aimed at reducing the immunogenicity of the rAV. In most cases, rAVs are constructed with an additional deletion in the adenoviral E2 or E4 region. These rAVs are propagated on cell lines that complement both E1 as well as the other gene. Production of such rAVs on appropiate helper cell lines is expected to reduce or eliminate the risk of generating RCA, as multiple recombinations would be required. However, a potential problem associated with the use of 293-based cell lines is that homologous recombination in the E1 region of adenovirus will generate adenoviruses which have reacquired the E1 region but still have defects in their E2 or E4 genes. Such an adenovirus revertant is not an RCA in the strict sense, as it is not able to replicate independently in human cells. However, the presence of the E1 region in such E1 revertants (designated REA: revertant E1 adenoviruses) poses another risk: the Ad E1 region has the potential to transform and immortalize rodent cells, and, albeit with much lower frequency, some human cell types. E1-containing adenoviruses that are deleted in either E2A or E4, are able to transform primary baby-rat kidney (BRK) cells (Table 2). In contrast, none of the vectors that are deleted in E1 were able to transform such primary cells (Table 2). Ads that carry lethal deletions have in fact been shown to transform cells more efficiently than wild-type Ad5. For example, H5ts125 encodes temperature-sensitive DNA-binding proteins, due to a defect in the E2A region. This adenovirus mutant exhibits a higher transformation frequency at the non-permissive than at the permissive temperature. Probably, E2- or E4-deleted Ads, in contrast to wild-type Ad, do not contain sequences that are toxic for BRK cells. Although the number of foci obtained by infection with E1-containing Ads was slightly lower compared to the number of foci that arose upon transfection with an Ad5 E1 plasmid (Table 2), one should bear in mind that 5 ´ 107 virus particles carry approximately 2 ng DNA, whereas we used 5  g plasmid DNA for transfection.
Whether REAs are able to induce tumors in humans is unknown. On the one hand, given the fact that the E1A and E1B proteins contain strong CTL epitopes, the risk may only be theoretical for immuno-competent individuals. On the other hand, REAs may be harmful for immuno-compromised patients. In this respect, we should not forget that retroviral vectors, derived from Moloney murine leukemia virus, were considered to be relatively harmless in primates until Donahue and colleagues14 reported the development of T cell lymphomas induced by replication-competent retroviruses in rhesus monkeys.
The use of conventional helper cell lines such as 293 and 911, and their derivatives, in conjunction with new generations of rAVs, has made the RCA problem more complex. A rAV revertant can be the classical RCA (viz. which lost the transgene, regained E1, and is replication-competent), or REA (viz. reacquired E1, but is still replication-defective). First, it would be worthwhile to screen rAV lots, especially those intended for clinical use, for the presence of adenovirus E1 sequences, as this will reveal RCAs as well as REAs. It would also be worthwhile to characterize the revertants that are generated in the newer helper/vector combinations. Second, vector systems that prevent RCA-formation altogether should be employed.
Currently, adenoviral vectors are the most efficient vectors for gene therapy applications. Adenoviral vectors are therefore being extensively manipulated to make them suitable for specific applications. Such developments should be accompanied by the parallel development of procedures to make the rAV a safe pharmaceutical product: a manufacturing process that prevents contamination of the viral preparations with either RCA or replication-defective revertants. Despite the fact that, to our knowledge, no accidents have happened to date with RCA-contaminated rAV preparations in clinical trials, we feel that if we intend to improve the Ad vector system for gene therapy purposes, we should focus on enhancing the therapeutic potential, as well as the safety features. The use of helper cells and nonoverlapping vectors eliminates this problem, and allows production of safe, clinical grade batches of rAVs. Only safe production systems, developed in parallel with appropriate testing methods, will warrant safe clinical application of rAVs. This is an essential prerequisite to prevent accidents that may not only have a deleterious impact on the patient, but may also damage the reputation of the gene therapy field.
|
E1 + 