Oncolytic virus therapy is a new modality of cancer treatment, which has shown promising results in clinical trials 1. Oncolytic viruses are designed to selectively replicate in and kill cancer cells without harming normal tissue. Since reporting the successful treatment of a brain tumor model by a tk-mutant herpes simplex virus (HSV) vector in 1991 2, we have focused on the development of increasingly efficacious and safe, replication-competent oncolytic HSV vectors. For the generation of new oncolytic vectors, insertion of exogenous genes or promoter sequences is often performed to enhance anti-tumor activity or tumor-selectivity of the vector. An example of the former is insertion of an interleukin gene and of the latter, insertion of a tumor- or tissue-specific promoter 34. Traditionally, recombinant HSV vectors have been generated through homologous recombination by co-transfecting mammalian cells with purified HSV DNA and a plasmid containing exogenous sequences flanked by viral sequences homologous to the insertion site. This method usually requires laborious screening or selection and can take several months. Recent advances in bacterial artificial chromosome (BAC) technology have enabled cloning of the whole HSV genome as a BAC plasmid and subsequent manipulation in E. coli 5678. Thus, we sought a method to generate recombinant oncolytic HSV vectors more easily and quickly using BAC technology.

It is not practical to manipulate BAC-cloned HSV DNA using conventional subcloning methods such as restriction enzyme digestion and ligation, because HSV-BAC DNA is usually more than 150 kilobases (kb) in length. To circumvent this problem and achieve efficient and accurate manipulation in E. coli, site-specific recombination (e.g. Cre/loxP system), homologous recombination/allelic exchange (e.g. RecA, RecE/T, bacteriophage λ red system), and transposon integration have been used for mutagenesis or the introduction of foreign sequences into BAC-cloned virus genomes 68910111213141516. Another factor to consider in BAC-mediated recombinant HSV construction is the packaging capacity of the HSV virion. It has been reported that packaging of DNA fragments larger than 156 kb into HSV virions is not efficient 1718. BAC cloning requires insertion of mini F plasmid sequences and antibiotic resistance genes into the viral genome and the length of these BAC backbone sequences is usually greater than 6 kb in total 19. Insertion of BAC sequences into the wild-type HSV genome (152 kb) will increase the genome length to ~158 kb, and there will be no space left for the insertion of additional sequences. To avoid deleterious effects of the BAC sequences, including growth defects and potential transmission between bacteria and man, some herpesvirus BAC clones have been constructed with loxP site-flanked BAC sequences, which can be removed by Cre recombinase 82021. In this study, we chose to use two independent site-specific recombination systems, Cre/loxP and FLP/FRT, sequentially in two separate steps. These two steps are: (i) integration of the shuttle vector containing a foreign sequence into the HSV-BAC genome and (ii) excision of the BAC sequences from the integrated HSV-BAC-shuttle DNA. We also employed another strategy for efficient generation of correctly recombined HSV vectors. The size of the integrated HSV-BAC-shuttle DNA was designed to exceed the packaging capacity of the HSV virion by inserting a stuffer sequence, so that recombinant HSV genomes are efficiently packaged and propagated only after removal of the BAC and stuffer sequences by FLPe. This strategy enabled us to obtain nearly pure (>99%) preparations of correctly recombined virus lacking the BAC sequences by simply co-transfecting mammalian cells with the integrated HSV-BAC-shuttle vector and an FLPe expression plasmid. We termed this system the Flip-Flop (

FLP

F

lo

P

In this paper, we describe the Flip-Flop HSV-BAC system and its use for the generation of recombinant HSV vectors with the essential ICP4 gene driven by the CMVIE promoter. ICP4 (infected cell protein 4) is the product of the immediate-early α4 or IE-3 gene, and functions as the main transcriptional activator of viral genes 2223. Previously, we reported the generation of transcriptionally-targeted oncolytic HSV vectors using an ICP4-deletion mutant HSV d120 as a backbone 4. We introduced an albumin-promoter-driven ICP4 expression cassette into the tk gene locus of d120 and showed that the resultant virus G92A selectively replicated in and killed hepatoma cells and tumor xenografts in which the albumin promoter was active 24. Using the same strategy, another group generated an oncolytic HSV vector targeting soft tissue and bone tumors with an ICP4 cassette driven by human calponin promoter 25. For the present study, we cloned d120 as a BAC plasmid, with the BAC sequences integrated in the UL39 (ICP6) gene, and then inserted ICP4 expression cassettes into d120-BAC using the Flip-Flop HSV-BAC system. ICP6 is the large subunit of viral ribonucleotide reductase and disruption of this gene limits HSV growth to actively dividing cells 262728. We showed that insertion of an ICP4 expression cassette driven by the CMVIE promoter efficiently drove viral replication, whereas insertion of the ICP4 coding sequence (CDS) alone did not restore viral replication. Use of a tumor-specific promoter instead of the CMVIE promoter, should generate a transcriptionally-targeted oncolytic HSV vector. Furthermore, these studies illustrate the power of the Flip-Flop HSV-BAC system to rapidly generate HSV recombinants.

Transcriptionally-targeted oncolytic virus vectors are promising cancer therapeutic agents and some of them have already been applied to cancer patients in clinical trials. For example, an oncolytic adenovirus vector CV706 whose replication is driven by the prostate-specific antigen (PSA) promoter/enhancer element has completed Phase I clinical trials and some of the patients showed a significant decrease in serum PSA levels 30. Although there are many potentially promising tumor-specific promoters reported in the literature and many of them have been used for generation of oncolytic adenoviruses 31, only a few have been tested in the context of transcriptionally-targeted HSV vectors due to the extensive labor involved in constructing recombinant HSV vectors 42532.

Traditionally, recombinant HSV vectors have been generated via homologous recombination between purified HSV DNA and recombination plasmid in co-transfected cells. In addition to the inefficiency of recombination, there is the need to screen or select plaques for the correct recombinant. This has hampered the development of new recombinant HSV vectors. Recent advances in BAC technology have enabled the cloning of herpesvirus genomes as BACs and their manipulation in E. coli 567891011121314162021. Using our Flip-Flop HSV-BAC system, once a shuttle vector containing a promoter of interest is constructed, nearly pure (>99%) preparations of recombinant vector can be obtained typically within two weeks, which can be easily plaque-purified subsequently. It should be noted that this system can also be applied to the generation of transgene-expressing HSV vectors if replication-competent HSV-BAC is used as a prototype BAC 33. A similar strategy for the generation of oncolytic HSV vectors, termed the HSVQuik system, has recently been reported 34.

In Step1 of the Flip-Flop HSV-BAC system, Cre-mediated recombination was performed biochemically and then integrated BAC DNA electroporated into E. coli. As high-molecular-weight BAC DNA is vulnerable to mechanical shearing, these procedures could cause double-strand breaks, lowering the proportion of correctly integrated HSV-BACs. When we generated pM24-BAC-null clones, more than half of the clones (7 out of 13) had suffered large deletions occurring at random locations (Fig 4b), while for pM24-BAC-CMV and pM24-BAC-null cloning, 90% (9 out of 10) of the BAC clones contained a full-length HSV-BAC genome, with either single or multiple shuttle inserts (Fig 4a and data not shown). To date, we have generated several different transcriptionally-targeted HSV vectors using this system and found that the efficiency of full-length HSV-BAC cloning after Cre recombination varied between 40 to 100% (average 69%; 67 complete clones out of 97 clones) (TK, unpublished data). An alternate way to avoid BAC DNA shearing in Step1 would be to perform site-specific recombination in bacteria by introducing the shuttle and Cre-expressing plasmids into the E. coli carrying the prototype HSV-BAC plasmid, as reported previously 35. Terada et al. employed, in their HSVQuik system, a similar strategy using an FLP-expressing plasmid for the integration of their HSV-BAC and transfer plasmids, and reported that ~80% of HSV-BAC clones contained correct co-integrants 34.

The efficiency of FLPe-mediated excision of the BAC and stuffer sequences from the integrated HSV-BAC-shuttle DNA was high, with over 99% of recombinant virus harvested after Step2 LacZ-positive and EGFP-negative. In order for recombined virus to be generated, several events need to occur: (i) both the integrated HSV-BAC-shuttle and the FLPe plasmid is transfected into a same cell, (ii) FLPe is expressed, (iii) the BAC sequences are excised from the HSV-BAC DNA, (iv) the recombined viral genome is replicated and (v) packaged. There are multiple factors affecting these processes, including the rate and level of FLPe expression and the timing of recombination. In fact, we do not know at which stage the FLPe recombination occurs, i.e. in the original HSV-BAC DNA or in the replicated virus DNA concatemer. In the former case, all the replicated viral DNA are correct recombinants and readily packaged, whereas in the latter case, only a fraction of the viral DNA concatemer may be packaged. Therefore, the efficiency of viral production through all these events could be lower than that after transfection of pM24-BAC alone, from which bM24-BAC virus can be generated directly without recombination. In our experiment, the yield of FLPe-recombined virus was sufficient, albeit a log lower than that of bM24-BAC (Fig 5b). For the generation of bM24-empty virus, we used pM24-BAC-empty clones that contained multiple integrated copies of shuttle vector. Even in these multiple inserts, FLPe recombination between the most distal pair of FRT sites results in the generation of correct recombinant virus, while recombination between other pairs of FRT sites generates viral genomes, which are still oversized and cannot be efficiently packaged. Thus, multiple-insert BAC clones can be used for recombinant virus generation.

When pM24-BAC-null was co-transfected with the FLPe plasmid to excise the BAC sequences, the ratio of bM24-null to bM24-BAC-null was over 300 (Fig 5b). There are two factors affecting the final yield ratio between the recombined and non-recombined virus: (i) efficiency of recombination, and (ii) difference in growth kinetics between the two viruses. Previously, Smith and Enquist, reported that co-transfection of loxP-carrying PRV-BAC DNA and a Cre expression plasmid resulted in a virus preparation containing 10–15% non-recombined viruses 21. In order to improve the yield of correctly recombined virus, they inserted a Cre expression cassette into the loxP-flanked BAC cassette so that Cre was expressed from non-recombined BAC-virus. This self-excision strategy worked very well and the yield of correctly recombined virus was more than 99.9%. In the Flip-Flop HSV-BAC system, we took a different approach to maximize the yield of correct recombinant virus; a stuffer sequence was included in the shuttle vector so that the integrated HSV-BAC clones are oversized and produce virus efficiently only after the BAC and stuffer sequences are removed. Using this strategy, ~99.7% of harvested viruses were correct recombinants (Fig 5b). As shown in Fig 5c, the growth of oversized bM24-BAC-null virus (~166 kb) was much reduced compared to that of recombined virus bM24-null. As HSV-BAC can accommodate larger DNA, an integrated HSV-BAC with a much larger genome size (e.g. >180 kb) using a longer stuffer sequence, should further reduce the production of non-recombined virus.

We previously reported the generation of albumin-promoter-driven HSV vector G92A in which the ICP4 expression cassette was inserted into the thymidine kinase (tk) gene 4. Unfortunately, tk-mutant viruses are resistant to acyclovir treatment, which is not desirable for clinical application 36. To avoid this problem, we chose to insert the BAC vector sequences and ICP4 expression cassette into the UL39 gene, preserving the tk gene. UL39 encodes ICP6, the large subunit of the viral ribonucleotide reductase, and its inactivation limits vector replication to actively dividing cells 262728. Additionally, ICP6 inactivation increases virus sensitivity to acyclovir 3738. It is worth noting that the initial HSV-BAC construct can be generated by insertion of the BAC cassette into any desired sequence or gene in the viral genome, other than ICP6.

In the current study, it was important to first demonstrate that there were not cryptic regulatory sequences that would drive ICP4 expression in the absence of promoters when inserted into the ICP6 region of HSV. bM24-null is such a construct and was highly defective in replication, only minimally better than bM24-empty, lacking any ICP4 sequences. These results show that pM24-BAC is suitable as a prototype construct for the generation of transcriptionally-targeted HSV vectors. In a plaquing assay, bM24-null formed a few plaques on Vero cells when infected at 10000 pfu/well, while no plaques were formed by the same amount of bM24-empty (Fig 7b). These plaques might have arisen from bM24-null mutants with genome rearrangements or deletions that activated ICP4 expression.

When ICP4 was expressed under the control of the CMVIE promoter, its level of expression was much lower than KOS and hrR3, which have diploid copies of wild-type ICP4 gene (Fig 7a). Recently, Terada et al. reported that HSV promoters, ICP6, IE2 and IE4/5, induced a higher level of luciferase expression compared to the CMVIE promoter when inserted in the ICP6 gene 34, the same location as in bM24-CMV. Regulation of HSV transcription and translation is complex, with ICP4 playing a major role, both as a trans-activator and repressor 23. Upon infection, the tegument protein VP16 activates HSV immediate-early gene expression, however it does not similarly activate the CMVIE promoter 39, which could account for the delayed kinetics. In a study of CMVIE promoter-GFP expression in HSV immediate-early gene mutants, it was found that CMVIE promoter activity was dependent upon ICP0 expression levels 40 and that the CMVIE promoter in the HSV genome was actively repressed in the absence of viral activators 41.

It is not clear what levels of ICP4 expression are necessary for viral replication. In a single-step growth assay (Fig 8a,b), bM24-CMV grew efficiently in Vero cells, with slower kinetics than hrR3 or KOS, presumably reflecting the lower and slower induction of ICP4 by the CMVIE promoter. Furthermore, the levels of bM24-CMV replication, as opposed to hrR3, varied considerably in different tumor cell lines. This could be due to variable activity of the CMVIE promoter or different levels of ICP4 required for replication in different cell lines. In fact, we found that the transcriptional activity of the CMVIE promoter in SW480 was 3 to 8 fold lower than in HepG2 or A549, using a transient luciferase reporter assay 42. While considered constitutive, CMVIE promoter activity has been shown to vary greatly in different cells; in transgenic mice, or with transiently transfected plasmids or adenovirus vectors 43444546.

The Flip-Flop HSV-BAC system we describe enables rapid generation of HSV vectors carrying transgene inserts. By introducing a tumor-specific-promoter-driven ICP4 cassette into pM24-BAC using this system, one should be able to generate transcriptionally-targeted oncolytic HSV vectors. We believe this system will greatly facilitate the screening of a plethora of clinically useful tumor-specific promoters in the context of oncolytic HSV vectors.

TK participated in the design of the study, designed and constructed the vectors, carried out molecular biological and virological experiments, and drafted the manuscript. TT participated in the design of the experiments; RLM and SR participated in the design and interpretation of the experiments, and in drafting the manuscript. All the authors read and approved the final manuscript.