Viruses that infect vertebrate cells are capable of recognizing host cells and executing gene expression, genome replication and virion formation. The ability to use and control these processes is at the origin of virus-based gene-technology applications, such as the delivery of transgenes or the use of viruses for vaccinal or vaccine-carrier purposes. Based on their inherent ability to kill cancer cells, a number of viruses have been tested as anticancer agents, including those with double- [1, 2] and single-stranded DNA genomes  and viruses with double- , negative-  and positive-stranded  RNA genomes. Typically, these viruses can also infect normal cells; engineering specificity for cancer cells is therefore crucial. Several strategies have been tested for this purpose, e.g., the creation of p53-sensitive mutants , the alteration of viral receptor specificity [8, 9], the use of cancer-specific promoters in recombinant viruses [10, 11] and the regulation of gene expression in oncolytic viruses using cellular microRNAs [12, 13]. The general shortcomings of such viral vectors are their specificity to only certain cancer types and an attenuated oncolytic potential. In addition, genetically modified viruses, especially those with RNA genomes, tend to revert, or compensate for the introduced changes, which are usually unfavorable for the infection cycle. Therefore, new approaches are needed that are applicable to different viral systems and allow for the possibility of combination with other regulation strategies without compromising the anticancer properties of the vector.
Adenoviruses (family Adenoviridae) are currently the most extensively used oncolytic DNA viruses. They have non-enveloped virions, linear double-stranded genomes and replicate in the nucleus of infected cells . Adenovirus gene expression, replication, virion formation and cytotoxicity are dependent on the E1A proteins expressed via a constitutively active promoter [15, 16]. In many engineered adenovirus vectors, including the human adenovirus 5-based Adeasy system , the E1A region has been deleted and, therefore, these vectors are not capable of replication unless the E1A proteins are co-expressed.
Among positive-strand RNA viruses, the alphaviruses (family Togaviridae), including Semliki Forest virus (SFV), represent one of the most promising candidates for vaccine and anticancer vectors . Alphaviruses form cytoplasmic, membrane-associated replicase complexes [18, 19] and encode for two polyprotein precursors: a nonstructural (ns) polyprotein (replicase), translated directly from genomic RNA, and a structural polyprotein, from subgenomic (SG) mRNA . The alphavirus-based replication-competent vectors contain a complete viral genome and one or more foreign-protein-encoding sequences [6, 21], whereas replicon vectors lack the region coding for viral structural proteins and are unable to spread . Typically, the RNA genomes of alphavirus-based vectors are rescued from cloned infectious cDNA by in vitro transcription . Alternatively, cDNA can be flanked with eukaryotic transcription sequences; here, rescue involves cellular transcription by RNA polymerase II and nuclear exit of the RNA product followed by translation and subsequent replication. These vectors are commonly known as DNA/RNA layered vectors .
With the exception of poxviruses, all DNA viruses use splicing for the expression of some of their mRNAs. Therefore, the discovery that splicing patterns can be regulated by antisense splice-switching oligonucleotides (SSOs) [25–28] created novel possibilities for regulating viral gene expression and replication. In contrast, positive-strand RNA viruses never utilize splicing; however, the insertion of introns into their cDNA sequences is often used for the construction of corresponding DNA/RNA layered vectors [29–34]. Thus, all these viral vectors can be engineered to contain aberrantly spliced introns that block the expression of correct gene products (DNA viruses) or the rescue of infectious RNAs (DNA/RNA layered vectors). Such introns frequently occur in nature; for example, human beta thalassemia is often caused by a single-nucleotide mutation within an intronic segment of the human beta-globin gene, creating an aberrant splicing site [35, 36]. Targeting aberrant splicing sites with antisense SSOs has resulted in efficient correction of splicing in both in vitro [37, 38] and in vivo models [39, 40].
Here, we present evidence that insertion of aberrantly spliced introns into the genome of a recombinant adenovirus vector results in severely defective marker-gene expression. Similar manipulations of DNA/RNA layered SFV vectors also resulted in a reduction of the rescue of infectious transcripts. Both of these defects were reversed by SSOs. In essence, the presence of such introns in crucial regions of the viral vector represents a lethal mutation and SSOs represent an artificial and efficient cofactor required by the constructed vector. This approach represents a novel, universal and powerful method for controlling gene expression, replication, viral spread and, by extension, virus-induced cytotoxic effects.