Introduction

Vaccinia virus (VV) has played a prominent role in one of the greatest achievements in medical history—the eradication of smallpox (caused by variola virus). This occurred in the late 1970s, having claimed the lives of half a billion people worldwide in the preceding three centuries . In 1776, Jenner famously inoculated a small boy with pus obtained from a pustule on the hand of a milkmaid . The active therapeutic responsible for the resulting immunity to smallpox was most likely cowpox virus, which is closely related to but taxonomically different from VV . The origin of VV remains obscure, but by the early 20th century it became the foremost smallpox vaccine .

Although variola is no longer present in any population, threats of bioterrorism from laboratory-preserved derivatives maintain the potential need for rapid widespread vaccination. Vaccine development thus still remains active. In addition, VV has found utility in other roles, including the immunotherapeutic prevention of other infectious diseases as well as in the treatment of cancer . With regard to the latter, the earliest studies (which mainly used replication-attenuated VV recombinants for fear of toxicity) were relatively disappointing in the clinic. Replication-competent VVs retain their ability to lyse tumor cells and spread through tumor tissue. Recent advances in DNA recombinant technology (enabling the rational manipulation of the viral backbone), coupled with the ever-increasing knowledge gains in the fields of molecular virology and cancer cell biology, have aided the development of safe and efficacious replication-competent oncolytic VVs. These are currently at the forefront of the most promising novel anticancer agents.

Different VV strains were used in different areas of the world for mass vaccination. The anticancer potential of the Lister strain (VVL), which was popular in Europe, is the main focus of this chapter.

Positive Biological Features of VV Favoring Use in Cancer Therapy

VV is a member of the poxvirus family, which can be broadly divided into those that infect vertebrates (chordopox) and those that infect insects (entomopox). VV belongs to the group Orthopoxvirae, a genus of the Chordopox subdivision. Structurally, its protein core surrounds a double-stranded linear sequence of DNA that approximates 192 kbp (see http://www-ncbi-nlm-nih-gov.easyaccess2.lib.cuhk.edu.hk/genbank and http://www.poxvirus.org ) with its free ends linked by hairpin loops. This contains approximately 200 largely nonoverlapping genes, some of which are duplicated due to their position in inverted terminal repeat sequences .

The capacity of VV strains to stably accommodate up to 25 kb of exogenous DNA with little hindrance to infectivity makes it an ideal vector for use in vaccination and gene therapy . VVs have several other inherent features that make them particularly suitable for development as oncolytic agents for cancer treatment.

First, VVs have an efficient life cycle, leading to the rapid destruction of infected cells and indeed tissue via local dissemination . Viral replication occurs in juxtanuclear “factories,” commencing 1 or 2 hr following infection. During this time, host cell nucleic acid synthesis completely shuts down as cellular resources are diverted toward viral production . The initial viral replication cycle is usually complete by 8 hr, at which time the first extracellular virions are shed. Infected cells are usually lysed approximately 48 hr postinfection. Up to 10,000 viral genome copies per cell may ultimately be produced, of which half will be packaged into infectious particles .

Second, VV strains have a broad tumor cell tropism. Unlike other oncolytic viruses, cellular entry is not dependent on specific surface receptors (which would thereby limit entry only into tumor cells that express these) but appears to involve a number of nonspecific membrane fusion pathways . During its life cycle, VVs may appear in four infectious forms that differ in the number of layers and composition of their encapsulating lipoprotein membrane(s) . Cellular entry of the two major forms, IMV (intracellular mature virion) and EEV (extracellular enveloped virion), is morphologically different . The single IMV envelope fuses with the cell membrane, releasing its naked core directly into the cytoplasm, whereas the double-layered EEV particle is first engulfed by endocytosis. Its outer membrane is subsequently disrupted by the low endosomal pH following which the inner membrane is able to fuse with the endosomal membrane (and release its nucleoprotein core into the cycytoplasm).

Third, the relative molecular independence of the life cycle of VVs in comparison to other oncolytic viruses such as adenovirus makes VVs more resistant to alterations in host cell biology . Uniquely, infectious vaccinia virions are prepacked with virally encoded proteins—notably a virus-specific RNA polymerase accompanied by enzymes responsible for further mRNA processing and various early transcription factors . Viral transcription is therefore able to commence almost immediately after cellular entry. This feature ensures minimal reliance on host cellular proteins for replication—a fact that makes it difficult for the host cell to effect antiviral defensive maneuvers based on alterations in host gene transcription.

Fourth, VVs are potentially genetically safer compared with other virus families because the previously mentioned self-sufficiency enables their life cycle to be entirely cytoplasmic, with practically no possibility of host chromosomal integration.

Fifth, VVs have the potential to be efficacious following systemic administration, a feature that would enable the targeting of metastatic disease . VVs have acquired various mechanisms to evade host defenses (discussed later), whilst their different morphological forms are antigenically distinct. Following infection, the relatively stable IMVs form the majority of virion particles in a host. They are brick shaped and measure 300×200×120 nm. IMVs may be released only upon cell lysis. However, upon release, an abundance of exposed viral proteins on their single lipoprotein envelope rapidly mobilize innate and adaptive host defense mechanisms. A cell-associated virion (CEV) resembles a particle just about to bud off from the host cell and affords a means of direct cell-to-cell dissemination. If it does bud off into the systemic circulation, it maintains its additional protective host cell lipoprotein bilayer, which enables the resulting particle (EEV) to be relatively antigenically quiet . This, coupled with virus-derived transmembrane proteins that antagonize aspects of innate (complement) and adaptive (neutralizing antibodies) systemic host defense, permits widespread dissemination . Current mass viral production techniques, which depend on co-culture in a standard cell line, produce mainly IMVs. Strategies to produce proportionately more EEV forms of VV are an active area of research .

Sixth, the efficacy of oncolytic VV is not attenuated and may even be enhanced by cellular hypoxia. Advanced solid tumors often harbor areas of hypoxia secondary to outgrowth from and/or destruction of local vasculature. This contributes to the aggressive, treatment-resistant phenotype of several cancers . Hypoxia is detrimental to the efficacy of other types of oncolytic viruses . In contrast, we have demonstrated that hypoxia does not impair the replication or oncolysis mediated by VVL. Indeed, its oncolytic potency in some tumors may even be enhanced by the presence of hypoxia .

Finally, a wealth of historic clinical data, accrued mainly from its use as a vaccine, confirms the excellent safety profile of VV. Nevertheless, in the extremely unlikely event of unchecked viral replication, effective antiviral treatment options are currently available .

Cancer Selectivity

Several VVs have demonstrated tumor specificity and antitumor efficacy in preclinical models of malignancy . One advantage of using VV as an antitumor therapeutic is that certain molecular drivers of oncogenesis may also be conducive to viral replication. Many tumor cells contain mutations that inactivate common cellular defense mechanisms that act to counter both pathogens and tumorigenesis, such as defects in interferon (IFN) and apoptotic pathways . These mechanisms remain intact in normal cells, in which viral replication can therefore be inhibited. Tysome et al . compared replication of a VVL recombinant in cultured normal human keratinocytes (NHEK) with that in its malignant counterpart (SCC25). There was minimal viral replication in the former, despite the fact that it was infected with five times the dose of virus used in SCC25 cell culture.

Oncogenic mutations that result in excess stimulation of the cell cycle may also promote viral replication, mainly due to the provision of a large nucleotide pool. For example, the epidermal growth factor receptor (EGFR) pathway is overactive in many, if not all, solid malignancies . Interestingly, VV may actively drive this pathway as part of its infectious cycle: VV DNA encodes a secreted EGF homolog, vaccinia growth factor (VGF), which binds to EGFRs on neighboring cells, stimulating their replication and effectively priming them for imminent virus invasion .

Tumors usually harbor very rich but disorganized and “leaky” blood supplies . Large gaps between endothelial cells readily allow the large VV particles to leave the circulation at these junctures . Vascular permeability may be increased by elevations in temperature. Chang et al. demonstrated a 100-fold increase in uptake of systemically delivered VV by tumor following localized heat treatment. Furthermore, some VV strains have tropism for ovaries , which has been partially attributed to their relatively permeable blood supply; however, recent data obtained by our lab questions this as the sole mechanism (Wang et al. , unpublished data).

Another potential mechanism for the inherent tumor selectivity of VV may relate to the overexpression of vascular endothelial growth factor (VEGF) in tumor microenvironments. VEGF is a major effector of angioneogenesis . We have recently demonstrated enhanced VVL uptake by various solid tumor cell models in the presence of VEGF. This appears to be effected by stimulation of the Akt signaling pathway, which ultimately leads to enhanced viral endocytosis .

Investigators have attempted to improve on the natural tumor selectivity of VVs by genetic manipulation. Traditionally, strategies have relied on disrupting genes that are essential for the virus to replicate in normal cells but that are redundant for replication in cancer cells. The most popular of these has been the disruption in the gene encoding the vaccinia thymidine kinase enzyme (VTK). VTK is essential for the synthesis of deoxyribonucleotides particularly in cells with low nucleotide pools, as is the case in most normal, well-differentiated mammalian cells. Tumor cells, however, usually have an abundance of nucleotides. VTK gene-deleted recombinant VVs have demonstrated enhanced tumor selectivity in a number of in vivo tumor models . For example, disruption of the VTK gene in the neurovirulent laboratory-derived Western Reserve (WR) VV strain reduced viral colonization and replication in the mouse central nervous system following intravenous administration, without affecting tumor tropism . Tumor specificity was enhanced by a further deletion of the VGF gene . This WR double-deleted mutant (WRDD) was more tumor specific and less virulent than either the single VTK-deleted or its wild-type counterparts.

The VVL derivative from Moscow (LIVP) used by our group and others has a premature stop codon in the locus that encodes VTK (J2R) . Interestingly, near complete deletion of the J2R locus, in our case by replacing a significant section of the coding sequence with marker transgenes encoding luciferase and lacZ enzymes, further enhanced the tumor selectivity of LIVP (Hughes et al ., unpublished data).

Other candidate genes that could be disrupted include those involved in combatting host cell defense mechanisms. Numerous tumor types have defective apoptotic pathways and would therefore allow viruses harboring mutations in anti-apoptotic genes to selectively replicate in them. Specific candidates include the anti-apoptotic serpin proteins SP1 and SP2 ; F1L, which inhibits the release of cytochrome c; and N1L, a Bcl-2 homolog with additional putative functions . Although deliberate disruption of these and other genes should theoretically enhance tumor selectivity, viral replication and oncolytic potency often deteriorate in proportion to the extent of genetic manipulation . Thus, a VV in which both SP1 and SP2 coding sequences were disrupted was severely attenuated compared to viruses harboring deletions in either sequence alone . In our experience, a disruption in the J2R locus of LIVP is all that is required to strike a good balance between tumor selectivity and antitumor efficacy. For other strains, however, particularly the more virulent ones such as WR, additional genetic disruption may be required prior to potential clinical use.

Rather than disrupt the coding sequence of a gene in the viral backbone, an alternative strategy to enhance tumor selectivity is one that targets relevant post-transcriptional gene products. Hikichi et al . illustrated this principle for the first time in VV using an attenuated VVL in which the B5R gene (essential for viral trafficking and dissemination) was restored with an additional 3′ sequence that was complementary to the microRNA, Let-7a. Let-7a is commonly downregulated in tumor cells, whereas normal cells contain it in abundance. Translation of the recombinant B5R mRNA (and consequent virus production) therefore only took place in tumor cells.

Unique Characteristics of the Lister Strain VV

Different strains of VV were used in different areas of the world in the days of mass vaccination. The New York City Board of Health (NYCBOH) strain and its derivative, Wyeth, were popular in the United States, whereas Copenhagen (CPN) and Lister found prominence in Europe. Historical records trace the origin of the Lister strain to Elstree, UK (1961), where the original master stock was prepared from calf lymph and distributed to centers in France, Moscow, Tokyo, and Atlanta. Early animal studies with VVL suggested a superior safety profile, particularly with regard to neurological sequelae . During the 1970s, in Japan, an attenuated Lister strain mutant LC16m8, selected for its extremely low neurovirulence, was used to safely vaccinate more than 50,000 infants against smallpox .

During approximately the past decade, the WR strain of VV, a virulent laboratory-derived NYCBOH derivative with potent tumor lytic ability, has been the most extensively studied and genetically manipulated oncolytic VV. It is currently thought to be the most superior oncolytic strain, although this assumption was made on the basis of its replication and oncolysis in only two human cancer cell lines in the only published comparative study to date . In that study, the WR strain was selected to be rationally engineered to be more tumor selective, with preservation of its oncolytic potency (WRDD). Despite this, concerns remain regarding its safety, and it obviously lacks the advantage of previous clinical usage. Furthermore, we recently discovered that VVL15 ( Table 16.1 ) actually replicated in and lysed tumor cells significantly better than the WRDD in 11 of 14 human cancer lines tested. However, WRDD did perform better in similar in vitro analyses with murine cancer lines (Hughes et al ., unpublished data). Despite the latter, VVL15 surprisingly demonstrated significantly better antitumor potency (compared to WRDD) in several immune-competent murine cancer models (Hughes et al. , unpublished data). Given the potency of WRDD against murine cancer lines in vitro , the latter effect must be mediated by indirect mechanisms—for example, through modulation of host immunity or vascular compromise. We are currently investigating putative underlying mechanisms.

Many of the open reading frames in VV are involved in viral genome transcription, replication, structure, and assembly. These are located centrally in the viral genome and are highly conserved among all orthopoxviruses. Peripherally located genes are responsible for modulation of host antiviral defense and are strain specific. The function and indeed expression status of many of these have yet to be delineated, but genes encoding proteins that interfere with IFN pathways seem to be overrepresented, indicating their significance in host defense. Although there is only an 8% difference between the genomes of Lister and other strains , phenotypic and functional differences between strains must be a consequence of these differences.

Garcel et al. sequenced a clonal isolate of a VVL substrain . The vast majority of the 201 putative VVL open reading frames (192) had more than 98% sequence homology to those previously identified in other strains, including CPN and WR strains. L172 (A53R) encodes a tumor necrosis factor (TNF) receptor homolog found in VVL but absent in WR. An extra interleukin (IL)-18 binding protein gene (L013) is again present in VVL but missing in the WR strain.

A genomic comparison between Lister, CPN, and WR strains conspicuously revealed a short sequence encoding completely different proteins in the Lister genome . Part of this sequence (L196) bears resemblance to that encoding a transmembrane Bax inhibitor motif containing protein found in various mammalian species and therefore implicates a historic recombination event with mammalian DNA . L196 encodes an apoptosis inhibitor . Another unique gene in this region that is missing CPN and WR strains encodes a TNF binding protein (L195) . Furthermore, its neighboring gene, L194 encodes a truncated version of a type 1 IFN binding protein that is intact in the WR strain. The alternative sequence in the corresponding section of CPN and WR genomes includes an open reading frame encoding an IFN-α/β binding protein (B19R/WR200). There is also a different anti-apoptotic gene, WR195, which encodes the serpin protein SPI-2 (an IL-1β converting enzyme inhibitor leading to the inhibition of FAS-mediated apoptosis).

VVL therefore codes for at least two extra TNF binding proteins (L172 and L195) in comparison to WR and CPN strains and could thus be better armed against the host TNF response. On the other hand, it lacks at least one IFN binding protein gene (WR200). There are also qualitative differences in genes encoding anti-apoptotic proteins.

Interestingly, the gene encoding the semaphorin-like protein L156 (A39R), which appears to stimulate monocytes and enhance production of proinflammatory cytokines , is frame-shift mutated in the WR strain. That this plays a role in the immunogenicity of VV is dramatically demonstrated by the induction of systemic inflammation and pulmonary edema in mice administered a WR recombinant expressing this transgene.

During the past fifteen years, the most commonly reconstructed Lister variant for the purposes of antitumor therapy has been the Moscow derivative, LIVP. The WR VV is particularly neurovirulent, reflecting its origin from multiple passages through the murine brain . Although systemic inoculation with a VTK gene-deleted WR virus reduced viral recovery from off-target organs, titers in these organs were still significantly higher in comparison to that with LIVP . In one study, there was no viral recovery from the brain or ovaries of nude mice following an intravenous dose of 1×10 ⁷ plaque-forming units (PFU) of VVL . Other organs also had a higher affinity for WR VV compared to VVL. Pulmonary titers were consistently higher at any given time-point following intranasal inoculation of similar doses of viruses . Indeed, as little as 5×10 ⁵ PFU of intranasally inoculated WR caused severe signs of toxicity (including significant weight loss) and 50% death in immune-competent BALB/c mice . Similar toxicity was only observed with VVL if administered in doses 100-fold higher .

Although difficult to generalize, the reduced persistence and virulence of VVL in immune-competent animal hosts implies the stimulation of a more potent “primary” antiviral immune response in comparison to WR VV. The implications of this enhanced immunogenicity with regard to antitumor efficacy have yet to be delineated. Further characterization of the expression and functional aspects of individual VV genes will enable the rational engineering of more selective, safer, and potentially more effective recombinants. Table 16.1 outlines the range of VVL constructs that we and other groups have used as putative anticancer therapeutics. Details of these will follow in the remainder of this chapter.

Lister Strain Derivatives with Demonstrated Antitumor Efficacy

In general, VV and other oncolytic viruses mediate their antitumor effects via direct and indirect mechanisms. These may be broadly classified as a consequence of (1) direct viral oncolysis, (2) local vascular destruction, and/or (3) focusing of the effector host immune defenses (both innate and adaptive) into the tumor microenvironment. All of these mechanisms may be augmented by incorporation of relevant therapeutic transgenes into the viral backbone.

As with other strains, VVL derivatives have demonstrated tropism for a variety of different tumor types both in cell culture and in animal models. The latter have predominantly been xenograft models of human malignant solid tumors, including breast , liver , prostate , bladder , squamous cell , pancreatic , thyroid (anaplastic) , and mesothelioma .

Ascierto et al . investigated whether molecular signatures within a particular tumor cell could predict permissivity of infection by the Lister strain recombinant GLV-1h68 ( Table 16.1 ) . They screened the NCI 60 panel of human cancer cell lines , which covers a broad spectrum of human malignant cancers. Three genes were found to be consistently upregulated in permissive versus nonpermissive lines: GDF15, a member of the TGF superfamily that regulates inflammatory and apoptotic pathways; CD9, a transmembrane protein that regulates cell growth, activation, and mobility; and integrin β ₅ , which plays a role in cell movement and adhesion. In contrast, far more genes were conspicuously downregulated and could be grouped into those involved in NF-κB, TGF-β, or IFN-α/β signaling and/or the activation of the RNA polymerase complex. In other words, the most permissive cell lines to GLV-1h68 had predampened host cell antiviral defense mechanisms and host gene transcription.

Chen et al . demonstrated a positive correlation between the replication efficiency of VVL derivatives in tumor cell culture and antitumor efficacy in the corresponding xenograft model. Although the latter largely reflected higher titers of replicating virus and thus oncolysis, they also postulated a significant contribution from the innate immune system. Indeed, expression profile analysis revealed upregulation of characteristic “innate” cytokines, chemokines, and growth factors in tumor tissue associated with a cellular infiltrate consisting of dendritic cells, monocytes, and neutrophils . Thus, even in these immune deficient models, mechanisms additional to viral oncolysis contributed to antitumor efficacy. It is these latter mechanisms that will require maximizing as viral oncolysis is likely to be curtailed in immune competent models and human patients.