Could recombinant technology facilitate the realisation of a fertility-control vaccine for possums?

Abstract

While lethal control remains the primary tool for possum control in New Zealand, there has been substantial research effort over the last decade into controlling possum populations by reducing the animals’ breeding potential. In particular, the potential of immunological control to reduce fertility has been investigated, whereby the possums’ immune system is induced to react against its reproductive system, in order to block or destroy key components of the reproductive process. Two of the most targeted key components have been the blocking of circulating hormones (e.g. gonadotrophin-releasing hormone) and the inhibition of functional egg surface proteins (e.g. zona pellucida). While some success has been achieved for each approach, three main obstacles remain to the development of a working fertility-control vaccine for possums: first, ensuring that the vaccine remains efficacious in oral-delivery formulation; second, ensuring sufficient levels of fertility reduction; and third, ensuring that the induced immune response is sustained for a duration sufficient for long-term suppression of reproduction. The use of a genetically modified recombinant organism (parasite, bacterium or virus) to deliver a fertility-control vaccine could satisfy these requirements. A strong precedent for this approach has been set already in wildlife biology, namely the oral rabies vaccine, which is based on a recombinant vaccinia virus and which has been used successfully for the last 25 years as a disease-eliminating vaccine in mesocarnivores in Europe and North America. This review outlines the different forms and examples of recombinant organisms with potential for engineering into recombinant fertility-control vaccines to reduce possum reproduction; non-transmissible agents or fully-transmissible vectors are discussed, with shortcomings and benefits outlined for each.

Keywords:

• possum
• biocontrol
• oral vaccine
• fertility-control
• genetically-modified organism

Background to possum control and immunologically-mediated fertility reduction

Despite many years of sustained effort in lethal control, brushtail possums (Trichosurus vulpecula) remain a major vertebrate threat to New Zealand conservation values and, additionally, to its agricultural industry as the principal wildlife reservoirs of bovine tuberculosis (TB). The combined annual sum spent on possum control by regional authorities, landowners, Department of Conservation (DOC) and the Animal Health Board (AHB) has in recent years exceeded $100 m (Ministerial Report 2000). Currently the only broad-scale tools available are lethal control methods—toxins (e.g. sodium monofluoroacetate/‘1080’) and traps. These tools raise concerns regarding environmental contamination, by-kill of native and game animals, and animal welfare (Cowan 2005). Increasing community pressure has resulted in restrictions on the use of conventional lethal control methods in some areas and by some regional councils (Hellstrom et al. 2008). Biological management of possum numbers, i.e. manipulating a facet of the animals’ physiological processes such that this reduces survival and/or breeding potential, but in a humane fashion, has been recognised as an alternative long-term strategy to poisoning and trapping. This includes both the identification of key biological control targets and the identification and development of the means to deliver a biological control agent (Atkinson & Wright 1993). The biological control method with the highest public acceptability (83%) is reducing or preventing possum breeding (Fitzgerald et al. 2000). Modelling studies have suggested that limiting possum populations by fertility reduction will both effectively reduce the environmental impact of control operations (Bayliss & Choquenot 1999; Ramsey 2007), and reduce TB transmission to similar levels to that observed with conventional lethal control methods (Barlow 1994; Ramsey 2005). The aim of fertility reduction is to suppress the rate at which juvenile recruitment takes place, such that the adult population density declines to, or is held at, a level where unwanted effects remain below acceptable thresholds.

Controlling the reproductive cycle of a wild animal population necessitates some form of chemical or biological intervention. There are chemosterilants available which selectively target components of the male or female reproductive processes (Ericsson 1982; Danilovich & Ram Sairam 2006), but to date none of these have been applied successfully at the population level. One of the most promising alternative approaches is fertility control by immunological means (often termed immunocontraception). This involves immunising a target animal against an essential protein or amino-acid-based molecular component of the reproductive system, to induce an immune response in such a way that the immune system treats that key molecule as foreign (i.e. non-self). If the immune response is of sufficient strength, duration and character it can block the reproductive process and induce infertility.

To date, immunological fertility-control research in wildlife species has focused on two key components of reproduction. The first is circulating hormones that are essential to the reproductive process, particularly the hypothalamus-derived gonadotrophin-releasing hormone (GnRH) that governs the production and release of pituitary gonadotrophins (for review, see Delves & Roitt 2005), as well as the gonadotrophins themselves (luteinising hormone [LH] or follicle-stimulating hormone [FSH]) which play key roles in the maturation and function of the reproductive organs. The second are oocyte surface proteins, particularly the zona pellucida (ZP) proteins of the mature egg, which are sperm-binding sites essential for fertilisation. Fertility-control vaccines that target components of the male reproductive process, most noticeably surface proteins of spermatozoa, have also been investigated, although with lesser emphasis (for review, see Naz 2009). Fertility-control vaccines have been investigated for their use in wild animal populations, feral animals and zoo animals, and efficacy studies of injectable formulations have been published for wild pigs, horses, deer and rodents (GnRH-based vaccines); wild horses, deer and elephants (ZP-based vaccines); and foxes (anti-sperm vaccines). However, it is not the objective of this paper to review the mode of action of these injectable vaccines, nor to evaluate their relative successes and failures in a wildlife context, since this has been covered in detail elsewhere (see 2008 supplement of Wildlife Research for several reviews).

Here we provide a brief overview of research to date on the development of fertility-control vaccines for brushtail possums in New Zealand, focusing particularly on the constraints and limitations facing extension of existing approaches into a practical field vaccine for possums. We review the potential for using genetically modified organisms (GMOs) as a means of improving vaccine delivery and efficacy, and draw attention to the different options available. We also identify gaps in the current knowledge base and highlight the technological limitations inherent in this line of research.

Fertility-control vaccine research in possums: progress and constraints

Significant reductions in possum fertility have been achieved by inducing immune responses against the central reproductive hormone GnRH. A single injection of the commercial fertility-control vaccine GonaCon^TM (which comprises a mammalian version of the GnRH molecule in an immune-potentiating adjuvant) into possums induced anti-GnRH antibodies leading to >70% of the immunised animals being rendered infertile for 2 years (Doug C Eckery unpubl. data). Two delivery systems more amenable to oral (or oronasal) delivery have also been trialled—bacterial ghosts (BGs) and virus-like particles (VLPs). Both of these involve (at the point of delivery) non-living/non-replicating agents; BGs comprise the cell membranes of recombinant bacteria that have been engineered to express target reproductive system molecules, while VLPs are lipoprotein nanoparticles artificially constructed to appear immunogenically like viruses but lacking the genomic material necessary for replication. A BG expression system was trialled for nasal delivery using two different recombinant GnRH constructs; one consisted of five GnRH molecules interspersed with multiple known T cell antigenic targets (epitopes) which serve as molecular immune-potentiating adjuvants (Raina et al. 2004; Talwar et al. 2004), and the other a construct which features a heat labile enterotoxin as a potent mucosal immunogen (GP Talwar, unpubl. data). However, neither of these vaccines stimulated a specific immune response to GnRH when delivered nasally to possums (Doug C Eckery pers. comm.). A VLP system, based on the structure of the rabbit haemorrhagic virus (Peacey et al. 2007), has also been trialled, using GnRH chemically conjugated to VLPs (GnRH-VLP). When administered by injection to possums, the GnRH-VLP vaccine elicited high titres of GnRH-specific antibody (with levels higher than those observed in possums injected with GonaCon™), and a significant reduction in blood LH levels was also observed (Tao Zheng unpubl. data). When administered orally the GnRH-VLP vaccine also elicited a GnRH-specific antibody response (Tao Zheng et al. unpubl. data) but the immune response was much weaker than that resulting from the injected formulation. The effect of orally delivered GnRH-VLP on possum fertility has not yet been assessed.

Fertility-control vaccines based on possum ZP antigens have also been investigated, following the cloning and expression of recombinant-form possum ZP2 and ZP3 molecules (Mate et al. 2003) and identification of infertility-associated possum ZP peptide epitopes (Duckworth et al. 2007; Cui et al. 2010). Following injection of possum ZP antigens (either recombinant ZP proteins [ZP2 or ZP3] or ZP3 sub-unit peptides combined with mineral oil immune-potentiating adjuvants such as Freund's adjuvant), 60–70% reductions in egg fertilisation rates were reported in possums (Duckworth et al. 2007; Cui et al. 2010; Janine A Duckworth unpubl. data). The contraceptive effects appeared to be at least marsupial-specific as possum ZP proteins injected into model non-target eutherian and avian species (mice and chickens) had no effect on fertility (Duckworth et al. 2008). Bacterial expression systems (e.g. BG) and VLP constructs were also trialled in possums as non-replicating oral/oronasal delivery systems for ZP antigens. Oronasal administration of BG-ZP2 antigens caused a 36% reduction in egg fertilisation rates (Walcher et al. 2008) and, in a subsequent natural breeding trial, the number of offspring produced was reduced by 61% in the 20 days after treatment and by 33% in the 110 days after treatment (Janine A Duckworth unpubl. data). Similarly, when a ZP3 peptide was conjugated to VLP (ZP3-VLP), and the vaccine was administered intranasally, the number of offspring produced was reduced by 61% in the 20 days after treatment and by 22% in the 110 days after treatment (Janine A Duckworth & Tao Zheng unpubl. data).

There is thus strong proof-of-principle for fertility-control vaccines in wild animals, and some promising indications of efficacy in possums. However, efficacy sufficient for field use has only been shown for such vaccines administered by injection, which is a major constraint. For widely dispersed free-ranging wildlife, an oral-delivery system offers the best prospect to provide sufficient coverage for effective population-level immunisation (Cross et al. 2007; Tompkins & Ramsey 2007); hence an efficacious oral vaccine is required. However, this represents a further constraint, since ensuring immunogenicity and vaccine efficacy represents one of the greatest barriers to the development of oral vaccines in general (for overview and discussion, see Mann et al. 2009). This is because the mucosal immune system (which for the purposes of this review includes the buccal cavity mucosa, external nares and nasal canals, the oropharyngeal mucosal lining and the alimentary tract) is set to a default position of tolerance, a necessary mechanism to prevent immunopathogenesis through over-reaction to the myriad of potential environmental and food-derived antigens to which the immune system is exposed. Hence, immunological strategies to overcome mucosal tolerance (such as employing immune-potentiating adjuvants that act at the mucosal surface to ‘kick-start’ an immune reaction) are often necessary in oral delivery to prime the immune system to recognise the incoming material as ‘foreign’ and to mount a response.

For the development of oral-delivery fertility-control vaccines for possums, two other constraints also exist. First, in order to effect a maximal response, the current orally delivered vaccines are likely to require repeated boosting, which may be difficult and costly in relation to other pest control methods. Second, the immunity that is induced by such non-living vaccines has, to date, been comparatively short-lived (e.g. fertility reduction due to administration of BG-expressed ZP2 antigens or VLP-ZP3 peptide begins to wane 3 months after administration and has lost efficacy by 6 months), and of insufficient magnitude to effectively suppress populations (Tompkins 2007).

Live vaccines: the case for a self-replicating delivery system

The problems inherent with current oral-delivery fertility-control possum vaccines all stem from the fact that the fertility control antigens are delivered in a static form: immunologically speaking, the vaccinee experiences the target antigen only fleetingly. While this is sufficient to mount a short-term effector immune response (such as antibody production), the response is often of low magnitude and insufficient to form long-lasting immunological memory. This can be overcome by employing technology that allows for a longer residence (and hence contact-time) for the vaccine encountering the immune system, for example by using a live organism (usually, but not exclusively, a microorganism or a virus) as a delivery agent for the vaccine. If such agents are capable of even limited self-replication, and/or a long residence time in the host, they have the potential to expose the vaccinee to antigenic material in a sustained fashion. Such exposure makes the vaccinee more likely to mount a prolonged immune response with (importantly) good immunological memory. Live organisms (particularly microorganisms) can also act as immunopotentiating agents in their own right, by providing the adjuvant effect necessary to ‘kick-start’ the immune system in a way that non-living agents may not (principally, by stimulating the immune system through engagement of pattern-recognition receptors; Unterholzner & Bowie 2008).

The means by which a vaccine is ‘delivered’ by a live organism is usually achieved through genetic modification (GM). The genome of a microbial organism (bacterium, yeast, fungus), virus or (potentially) a metazoan parasite can be genetically modified to express heterologous proteins, and form the basis of a vaccine delivery system. For these modified organisms to generate sufficient antigenic material to stimulate a sustained immune response, they must either be capable of some degree of self-replication in the host (such as a microbial or viral agent), or be able to survive and metabolise for a long period of time in the host (e.g. a parasite). Such an agent may be either non-transmissible (i.e. one that establishes an infection site in the vaccinee, but cannot sustain itself long enough for patency) or transmissible (i.e. one that establishes an infection site in the vaccinee sufficient for patency and intra-species transmission). These two systems have different attributes, with associated benefits and detriments for use in a wildlife system, as summarised in Table 1.

Table 1  Comparative attributes (benefits and detriments) for non-transmissible and transmissible GM organisms as improved delivery technologies for fertility control vaccines.

	Benefits	Drawbacks	Example in wildlife biology
Non-transmissible GMO vector (limited infection)	Some vectors have a proven precedent for safety and efficacy of oral delivery in a wildlife scenario (e.g. poxvirus-vectored wildlife vaccines against zoonotic pathogen-mediated diseases)Some (e.g. pox- and adeno-viruses) are well-characterised systems, offering rapid development time from engineering to efficacy testing	Field use will require large-scale broadcast application (like a toxin)Most likely to have a broad host range, affecting non-target species; application and on-going maintenance costs are incurredBarriers to public acceptability of a living GM organism released to the environment, possibly with a potential to initiate human infection (e.g. recombinant vaccinia-based vaccines)	Oral rabies vaccine (Raboral^TM V-RG) and the prototype oral plague vaccine; both are recombinant poxviruses expressing pathogen glycoproteins, which elicit a protective antibody response in mesocarnivores (rabies vaccine) and prairie dogs (plague vaccine)
Transmissible GMO vector (establishes a patent infection)	Does not require oral delivery (so long as infection foci can be established by some means)Perpetuity of infection—one-off application, most cost effectiveA species- (or group-) specific vector can potentially be developed, limiting cross-species effects	Once established in the environment, may prove impossible to eradicate, unless stringent safeguards are in placeGenerally, not as well characterised as non- transmissibles; unknown technological potential for genetic manipulation. Uncertainty surrounding the behaviour of the GMO in a field situationConcerns from other countries regarding the susceptibility of their native fauna to the introduction of a transmissible agent bearing fertility-control antigens (e.g. see Henderson & Murphy 2007)	Murine cytomegalovirus (mCMV, a fully transmissible rodent-strain of CMV) and rabbit myxoma virus. Both have been researched extensively as potential delivery systems of immunocontraceptive vaccines for mouse and rabbit population control (McLeod et al. 2007)

Consideration of candidate GM delivery agents for use in possums

There is a plethora of organisms which theoretically could form the basis of a vaccine delivery system for possum fertility-control. Hence, as an initial step in conducting this review we first defined a set of criteria to enable a shortlist to be drawn. For inclusion in this shortlist, and to be eligible for more thorough critique in this review, the candidates must have satisfied at least six of the following 10 criteria:

1. The basic biology of the organism should be known (especially: in vitro propagation techniques and in vivo infection dynamics);

2. The ability of the organism to infect possums should be known;

3. It should have already been demonstrated that the genome of the organism can be engineered to express heterologous molecules;

4. There should be some evidence that infection can be established by oral or oronasal delivery (NB: important for non-transmissible agents, less important for transmissible agents);

5. There should be some information regarding the likely effect of the organism on non-target species;

6. There should be evidence that the organism is immunogenic in the target species;

7. There should be some evidence that the organism, in recombinant form, is stable (stable meaning: genomic stability and also viability/robustness of the construct for field use);

8. Ideally, there should be some a priori evidence of the use/study of the organism in wildlife biology;

9. Ideally, there should be some a priori evidence of the engineering of the organism for its use as a fertility control agent;

10. There should be some evidence that the effect of the organism on the immune system is achievable with a single-dose and that its effect is long-lasting (>12 months).

The most promising non-transmissible and transmissible organisms for delivery of possum fertility-control vaccines are discussed on a point-by-point basis below, including: (1) knowledge of the biology of the organism (including its species-specificity and infectivity in possums, its cell/tissue tropism); (2) propagation dynamics (including ability to be propagated in vitro, transmissibility in possums in vivo); (3) amenability to and stability of genetic manipulation; (4) ecological considerations (including the influence of pre-existing immunity to secondary infection, stability in the environment, infectivity in non-target species, epidemiological characteristic in possum populations); (5) safety issues and the likelihood of gaining approval from regulatory authorities.

Potential non-transmissible GM organisms

Non-transmissible organisms offer the benefits of limited in situ replication in a host (and thus sustained exposure of the host to the replication products of the organism, i.e. the antigens), but do not allow the infection to reach patency and spread to other individuals.

Recombinant poxviruses

Ortho- and para-pox viruses naturally infect hosts via dermal or mucosal contact, and are thus amenable to development as oral-delivery agents for vaccines. Individual species or groups commonly have their own unique poxvirus strain, e.g. monkey pox virus in primates, fowl or canary poxvirus in birds, etc. One approach to vaccine delivery is to use a species-matched poxvirus specific for the target host, which has been genetically modified to express a particular vaccine, e.g. raccoon poxvirus was initially characterised and engineered to express rabies glycoproteins precisely so that it could be used to immunise raccoons against sylvatic rabies (Esposito et al. 1988). The use of a species-matched poxvirus to deliver fertility-control molecules was reported by Jackson et al. (1998). They engineered the rodent poxvirus ectromelia to express mouse ZP3 antigen, and delivered this as a single-shot injectable vaccine into mice. Vaccinated mice produced anti-ZP3 antibodies resulting in sterility in 70% of mice treated (with a 75% reduction in mean litter size among those vaccinated mice that did manage to conceive). Some mice remained sterile for 5–9 months, and the development (and subsequent loss) of sterility was strongly correlated with the strength of the anti-ZP3 antibody response; so long as antibody titres were kept high (e.g. by vaccine boosting), sterility was also maintained.

A more common approach for employing a poxvirus as a vaccine-delivery organism is to use an attenuated or species mismatched virus: one that is capable of infecting the target host in only a limited fashion, such that it cannot establish patency. The best example here is the poxvirus vaccinia, which has been in medical and veterinary use for many decades, including its use as a human vaccine to eradicate smallpox. Vaccinia invokes strong antibody and cellular-based immune effector responses, and has been shown to induce long-lasting immunological memory. In a wildlife context, recombinant vaccinia that has been engineered to express rabies virus glycoprotein (known as V-RG) forms the basis of Raboral^TM, one of the frontline oral rabies vaccines (ORVs) used in Europe and North America to successfully eradicate wildlife rabies from several regions (reviewed by Brochier et al. 1996). A further example is the prototype oral-delivery vaccine that has been developed against zoonotic plague in prairie dogs; this is based on a recombinant raccoon poxvirus that expresses protective antigens of the plague-causing bacterium Yersinia pestis (Mencher et al. 2004; Rocke et al. 2008).

From a safety viewpoint, the use of a recombinant poxvirus as a practical tool for field use in New Zealand is potentially controversial. Selection of an appropriate viral strain with defined host specificity and using a target-specific contraceptive antigen would be critically important factors. For instance, New Zealand's unique avian fauna may preclude consideration of recombinant avian poxviruses as delivery vectors. From the human safety perspective, the proven safety record of the Raboral^TM vaccine sets a precedent for vaccinia, with only two reported cases of human infection (both due to mishandling of the vaccine) from over 100 million doses administered for rabies control across two continents over 20 years (CDC Report 2009). From the veterinary safety perspective, there are no reports of sustained vaccinia infection nor of patency in over 30 non-target animal and bird species tested (reviewed by Brochier et al. 1989; Artois et al. 1990; Hanlon et al. 1997). However, evidence of immune recognition (which implies a self-limiting infection and minimal viral replication) exists in non-target species. Brochier et al. (1989) reported safety-testing of a recombinant vaccinia construct as an oral inoculum in non-target European wildlife, with seroconversion in >50% of all mammals but no reactivity in birds, while Artois et al. (1990) reported seroconversion to the same construct in >75% of North American mammals.

In general, there is a paucity of knowledge regarding the potential infectivity of species mis-matched poxviruses in marsupials. Evidence of at least immune recognition of a eutherian poxvirus would be necessary prior to the organism's consideration as a vaccine-delivery agent; expert opinion is that vaccinia, for example, would most likely cause a limited infection, sufficient to invoke an immune response but insufficient to generate an infectious virus (Tony J Robinson pers. comm.). Rupprecht & Kieny (1988) reported such results from a safety study on the vaccinia-based Raoboral^TM vaccine, in which 100% of North American opossums (Didelphis virginiana) developed immune reactivity after oral exposure, suggesting a self-limiting viral infection. Further, a recent preliminary study indicated virus-specific immune reactivity in over 95% of brushtail possums exposed to vaccinia via oronasal delivery (Martin L Cross et al. unpubl. data) and live virus was also recovered from facial lesions in some of the infected possums up to 15 days post-infection (Stephen B Fleming et al. unpubl. data).

From a technological viewpoint, poxviruses are probably the best-characterised of all viral expression systems with respect to antigen delivery. Poxviruses have a large genome into which several expression sites can be engineered, i.e. the possibility exists for multiple antigen delivery using a single engineered virus. The Raoboral^TM vaccine is the only recombinant vaccine produced at a commercial scale for use in wildlife, and manufacturing techniques for large-scale production of this working wildlife vaccine could have useful application for the development of a fertility control vaccine product for possum control. There is also existing proof-of-principle that fertility-associated glycoproteins can be expressed in an immunogenic fashion by vaccinia (e.g. mouse ZP antigens have been expressed in a recombinant vaccinia construct and the purified antigen was immunogenic when injected into mice; Hardy et al. 2003).

However, there are some technical reservations about the use of recombinant poxviruses for vaccine delivery, since the nature of the antigen being delivered may influence the outcome of the vaccination. A recombinant vaccinia virus expressing fox-ZP and porcine-ZP failed to elicit ZP-specific antibody responses, even with repeated vaccinations intradermally and orally in a permissive species (foxes; Reubel et al. 2005). Reubel suggested this was most likely due to failure of the recombinant forms of the virus to infect mucosal tissues in the fox and/or insufficient transgene expression by their construct to stimulate an adequate immune response. Additionally, research in cats on the potential of using recombinant vaccinia virus expressing porcine-ZP showed that although the vaccine induced an antibody response following oral exposure, the antibody titres were variable and the vaccine did not induce contraception (Boyle 2005). These results are not surprising as porcine-ZP proteins have repeatedly proven to be ineffective as immunocontraceptive antigens in cats even when injected with highly potent immune-potentiating adjuvants (Eade et al. 2009; Gorman et al. 2002).

Recombinant adenoviruses

Adenoviruses naturally infect hosts via mucosal contact, with a common portal of entry being the orobuccal mucosae and tonsils; some strains also infect via the gastrointestinal tract epithelium. In addition, as with poxviruses, adenoviruses tend to be species-restricted, only producing patent infections in definitive hosts; although, they can initiate self-limiting infections in mismatched species and are thus suitable as non-transmissible vaccine-delivery agents.

Adenoviruses have two attributes that are particularly relevant to their use as heterologous antigen-bearing vaccine delivery agents: they invoke markedly strong antibody and cellular-based immune effector responses, and they have been shown to induce long-lasting immunological memory. For example, dogs immunised orally with a recombinant canine adenovirus expressing rabies glycoprotein were still refractory to rabies challenge 2 years later (Zhang et al. 2008). Various species-mismatched adenoviruses have been characterised and used as delivery vectors for oral rabies vaccines, including canine adenovirus in cats and raccoons, human adenovirus in raccoons and skunks, and chimpanzee adenovirus (experimentally in rodents). However, the potential infectivity of a eutherian adenovirus in a marsupial is unknown; the best available evidence is from a wildlife survey by Jamison et al. (1973) who found no seroreactivity in 25 North American opossums captured from an area endemic for wild-type canine adenovirus among carnivores (racoons, coyotes, skunks). From a safety viewpoint, a mismatched adenovirus is unlikely to initiate a persistent infection in non-target species; a 2009 study with recombinant human adenovirus as an oral rabies vaccine vector showed no detectable viral pathology in several vaccinated non-target wildlife species, with only 1% and 6% detection rates for viral particles in the oral mucosa and faeces, respectively, post-vaccination (Knowles et al. 2009).

From technological and achievability viewpoints, several reports have indicated that adenoviruses are amenable to genetic manipulation and can be engineered to express heterologous proteins stably. In fertility-control research, canine adenovirus has been engineered to express lemming (Lagurus lagurus) ZP antigens sufficiently to invoke immune responsiveness and some reduction in fertility when injected into mice (Li et al. 2008). There are, however, no reports of oral delivery of this vaccine. One potential option for fertility control in New Zealand is to search for a cultivable possum adenovirus. Adenovirus has been identified by electron microscopy and molecular biology from possums of New Zealand, and one possum isolate has been characterised on a phylogenetic basis as belonging to the group Atadenoviridae (Thomson et al. 2002). However, the in vitro propagation, in vivo transmission, amenability to genetic modification, possum infectivity, epidemiology, host range and host immune response to this virus have not been reported.

Recombinant macropodid herpesvirus (MaHV)

Three types of marsupial herpesvirus (MaHV-1, MaHV-2 and MaHV-3) have been isolated from sick macropods in captivity. MaHV-1 was isolated from a sick parma wallaby imported from New Zealand to Australia (Finnie et al. 1976), MaHV-2 from a dead dorcopsis wallaby in Australia (Wilks et al. 1981), and MaHV-3 from an eastern grey kangaroo imported from Australia to the USA (Smith et al. 2008). MaHV-1 replicates preferentially in marsupial cells over eutherian cells in vitro (Whalley & Webber 1979), although its infectivity in eutherian mammals in vivo has yet to be investigated. The clinical manifestations of infection in macropods range from herpes-like vesicles around the mouth, nose and anogenital regions to pneumonia, hepatitis and death (Finnie et al. 1976; Dickson et al. 1980; Callinan & Kefford 1981; Smith et al. 2008). Transmission in naturally occurring infections is through direct social contact between infected and susceptible animals.

Serological surveys indicated that MaHV-1, or a closely related herpesvirus, was endemic in the parma wallabies of Kawau Island in New Zealand (Webber & Whalley 1978), but the virus itself was not isolated. Survey data from Australian sites indicate that macropods are generally immune-reactive to MaHV-1, suggesting that MaHV-1 (or a close relative) is widespread. MaHV-1 is infective to macropods in captivity and can produce latent infections that reactivate under adverse conditions (e.g. handling of the animals, capture, malnutrition or captivity stress). Although MaHV-1 is infectious to possums when dosed experimentally via ocular-nasal-route exposure (Zheng et al. 2004), latency and reactivation of infection have not been demonstrated; only mild conjunctivitis was observed in some individuals, with the virus recovered from the inoculation site for up to 7 day post-infection. However, humoral immune responses were detected from all inoculated animals, indicating that the virus had replicated and viral antigens were presented to the possum immune system (Zheng et al. 2004). A preliminary transmission study indicated that MaHV-1 was not transmitted to naïve possums through contact from inoculated possums (Tao Zheng et al. unpubl. data), indicating that it could be a good candidate for development as a non-transmissible organism for delivery of a fertility-control vaccine to possums.

In terms of the feasibility of genetically modifying MaHV-1, a novel technology has been described for engineering its genome to express heterologous molecules (Thomson & Smith 2001); the gene for green fluorescent protein (GFP) was inserted into the gD region of MaHV-1 and expression of protein was detected in several isolates (Darelle M Thomson unpubl. data). Whether or not a recombinant construct of MaHV-1 would be infective or stimulate an immune response in possums in vivo is unknown. In general terms, however, other animal herpesviruses (including bovine, canine and murine strains) have been shown to be amenable to stable genetic modification, and to accommodate and express a range of foreign antigens (reviewed in Brun et al. 2008). In addition, modification of the herpesvirus murine cytomegalovirus to express mouse ZP antigens has been shown to cause sterility in infected mice (reviewed in Hardy et al. 2006).

Recombinant tissue-dwelling bacteria

Bacteria can also be genetically modified to express heterologous proteins. Several bacteria that are capable of infecting a host via oral delivery, such as salmonellae and mycobacteria, have intra-tissue replication sites and are thus amenable to delivering their heterologous load in an immunogenic fashion.

An attenuated strain of Salmonella typhi forms the basis of the current human oral typhoid vaccine (Kopecko et al. 2009). In addition, the so-called ‘mouse typhoid’ organism (S. typhimurium) has been engineered to express fox sperm antigens (LDH-C4 or Sperm Antigen 10) and was trialled as a fertility-control oral-delivery vaccine in male foxes. Results were mixed, with one study showing systemic and reproductive tract antibody responses against sperm (Bird et al. 1998), but a follow-up study failed to demonstrate immune recognition (de Jersey et al. 1999).

Nothing is known about infection dynamics, persistence and antigen expression of attenuated vaccine strains of Salmonella in possums (although pathogenic Salmonella infections are frequently recorded in captive possums recently captured from the wild; Presidente 1984). However, we do have a good understanding of mycobacterial infections through TB vaccine research. Following oral delivery, BCG (an attenuated strain of Mycobacterium bovis, the causative agent of bovine TB) colonises the lymphatic system draining the possum alimentary tract, with a minimum residence time of 2 months (Wedlock et al. 2005). In a wildlife context, orally-delivered BCG forms the basis of the Liporale^TM oral vaccine that is currently under field trials as an anti-TB vaccine for possums (Tompkins et al. 2009). Oral vaccination with recombinant BCG expressing an outer surface protein from the bacteria Borrelia burgdorferi (casual agent of Lyme disease) elicited immune responses against the expressed antigen in white-tailed deer (Miller et al 1999). Although BCG's ability to express heterologous antigens associated with fertility control has not been reported, the genome of BCG is amenable to insertion and expression of a wide range of heterologous antigens (reviewed in detail by Bastos et al. 2009).

One distinguishing feature of BCG is that it is a very slow growing bacterium, both during its intralymphatic residence in the host and in any laboratory-based in vitro studies. The former need not necessarily impair the immunogenicity of heterologous molecules expressed by BCG, but it may require that there is no physiological perturbation to either the lymphatic replication site or the host's immune system during the extended phase of in situ immune-sensitisation. It may also be important that the vaccinee has had no prior exposure to mycobacteria as this would inhibit establishment and replication of newly delivered BCG, although this potential limitation is contentious. However, notwithstanding the technological difficulties of producing and testing a recombinant BCG, one point in favour of this vector is its current development as an oral-delivery TB vaccine; a recombinant BCG engineered to co-express possum-specific fertility control antigens could be developed, to fulfil dual purposes of disease eradication and pest population reduction. Wu et al. (2009) described a similar multiple-utility approach with a rabies vaccine, with a live attenuated vaccine-strain of the rabies virus engineered to co-express GnRH; the virus was immunogenic for both hormone and viral epitopes when injected into mice, raising the possibility of a combined rabies and fertility-reduction vaccine.

Overview of the suitability/applicability of non-transmissible GM organisms for delivery of a possum fertility-control vaccine

Four candidates have been discussed in detail in this review: vaccinia virus, adenovirus, macropodid herpesvirus-1, and M. bovis BCG. Each has attractions and reservations, and so far three of these agents (vaccinia virus, MaHV-1 and BCG) have been shown to infect possums (with species-mismatched adenoviruses yet to be tested).

A recombinant vaccinia virus would seem to be a good candidate to study further for vaccine delivery in possums, with its history of use, extensive safety profile, proven immunological activity and amenability to genetic modification. Furthermore, preliminary evidence indicates that vaccinia can replicate sufficiently to induce immune recognition in marsupials, including brushtail possums. Although, ZP-expressing vaccinia recombinant constructs have not been shown to be effective when delivered orally in two species tested (foxes and cats), results were compromised by inappropriate antigen choice and poor transgene construction and expression.

The development of adenovirus as a vector for vaccine delivery can probably be discounted as this approach has the most uncertainty associated to it. Although adenoviruses sensu lato have an excellent safety record (in attenuated form, e.g. as the basis of the commercial canine hepatitis vaccine), and despite the fact that in recombinant form the adenovirus genome has been shown capable of accommodating and expressing fertility-associated molecules, the potential infectivity of this vector in possums is unknown.

Macropodid herpesvirus-1 is also a possible candidate for further study, but far less is known about its basic biology than for vaccinia and BCG. Hence, more fundamental research would likely be required for development of a MaHV-1-based vaccine (e.g. extensive safety testing in non-target species). On the plus side, herpesviruses sensu lato have been shown to accommodate heterologous genes and to express these stably in recombinant form. However, a further disadvantage of MaHV-1 is its known pathogenicity in some marsupials, leading to ethical issues regarding its use in possums.

Like vaccinia, BCG has an extensive safety profile, although formal studies of its safety profile in New Zealand wildlife have yet to be conducted. In addition, to its advantage, its mode of action has been researched extensively in possums for nearly 20 years, with the weight of evidence suggesting that a single oral dose is sufficient to invoke a long-lasting immune response (>1 year; Buddle et al. 2006). Although the insertion of fertility-control molecules into BCG has yet to be undertaken, that BCG is already being developed as an oral-delivery vaccine for possums is a benefit to its parallel development for fertility control.

Transmissible GM organisms

Immunocontraceptive vaccines could also be delivered to possums using a transmissible vector. This would necessitate the identification and genetic exploitation of a possum-specific organism that could establish self-sustaining infections in possum populations post-release. Such an approach would likely facilitate the most cost effective fertility-control method of all, as it would not incur ongoing application costs. As with the preceding discussion on non-transmissible GM vaccine-delivery agents, this section will outline and discuss aspects of environmental safety, technical achievability and feasibility, and infection dynamics and modelling relevant to the consideration of candidate vectors capable of intraspecies transmission within possum populations.

Species-specific infectious agents as transmissible vectors: examples

The most effective transmissible vector for possums would be one that is highly infectious for the target species yet monospecific in its host range. As a comparative example in another pest species, the mouse-specific herpesvirus mCMV (murine cytomegalovirus) has been studied for its potential as a transmissible viral vector to control wild mouse populations. In its wild-type form, mCMV is highly infective to mice and apparently suitable as a transmissible vector system to control wild mouse populations. Murine cytomegalovirus was engineered to express murine ZP antigens; when injected into mice under laboratory conditions, this recombinant virus provided high levels of permanent sterility (Smith et al. 2005; O'Leary et al. 2008). However, upon further study researchers encountered a major problem, which subsequently curtailed further development of the GM virus into a working field tool for population-level mouse control: recombinant mCMV showed reduced infectivity (compared with the wild-type virus) due to its attenuation during genetic manipulation (Hardy et al. 2006). In wild mouse populations, which can express varying degrees of pre-existing anti-CMV immunity, any reduced fitness of the virus would correspond to reduced transmissibility, which would hamper its natural dissemination through populations (Gorman et al. 2008). Similar problems with pre-existing immunity against, and competition from, wild-type viruses have also affected the potential effectiveness of a GM virus to control reproduction in rabbits. Myxoma virus (which was used earlier as a lethal biocontrol for rabbits; Burnett 1952) was engineered to express rabbit-ZP antigens, and investigated for use as a species-specific viral vector for fertility-control in Oryctolagus cuniculus (Robinson et al. 1997). The vaccine induced up to 70% reduction in fertility upon first breeding when injected into female laboratory rabbits (Mackenzie et al. 2006), and preliminary field-release studies showed that such a virus could disseminate well in wild rabbit populations (Robinson et al. 1997). However, subsequent simulation models indicated that in a field situation, competition from wild-type myxoma virus would likely reduce vaccine efficacy to a level below that required for effective population control (reviewed in McLeod et al. 2007).

The search for possum-specific viruses

Identification and development of a possum-specific transmissible vector for biological control was regarded as a key component of the possum biological management strategy in the 1990s (Atkinson & Wright 1993). No suitable candidates for development as possum-infective agents were isolated from possums in Australia, but adenovirus, herpesvirus, coronavirus-like virus, papillomavirus, enterovirus and an indigenous type D retrovirus isolates were recorded from possums in New Zealand (Rice & Wilks 1996; Perrott et al. 2000b; Baillie & Wilkins 2001; Zheng 2007). Additionally, a novel possum-infective agent was identified to cause wobbly possum disease (WPD), a disease recorded in both captivity and the wild in New Zealand (Mackintosh et al. 1995; Perrott et al. 2000a). Wobbly possum disease is believed to be a viral aetiology (O'Keefe et al. 1997) and has been proposed to be a Borna-like virus (MAF 2006). In captivity WPD is highly pathogenic and the disease is transmissible from possum to possum via various body samples (blood, urine, tissue homogenates and ectoparasitic homogenates; O'Keefe et al. 1997; Perrott et al. 2000b). Viruses were observed under electron microscopy in samples prepared from diseased tissue samples (O'Keefe et al. 1997) and the disease was reproducible in possums inoculated with tissue filtrate (0.22 µm) of diseased animals (O'Keefe et al. 1997). However, definitive genomic information remains to be presented, which has significantly hampered further characterisation of the disease and the development of tools to investigate its status in wild possums.

Possum enteroviruses were isolated by cell culture from intestinal contents of healthy possums in New Zealand (Zheng 2007). The viruses replicated well in vitro in the laboratory, and belong to a group of positive single-stranded RNA viruses, with a genome of approximately 7.4 × 10³ base pairs (Zheng 2007). Possum enterovirus has been shown to establish infection in experimentally challenged possums, the majority of the inoculated animals mounting a detectable antibody response (Zheng & Chiang 2007) with no apparent associated illness. A serological survey indicated a patchy distribution of enterovirus in possums in New Zealand with the highest seroprevalence detected in the Manawatu and Wanganui regions where the viruses were originally isolated (Zheng et al. 2010). In general viral vaccine research, the Sabin strain of the enterovirus poliovirus has been successfully engineered into a replication-competent vaccine delivery agent for human vaccination (Evans et al. 1989; Andino et al. 1994; Halim et al. 2000). Hence, enteroviruses have good potential for development as transmissible vectors of fertility control antigens to possums.

The possum nematode parasite, Parastrongyloides trichosuri

Parastrongyloides trichosuri has received considerable attention as a potential transmissible vector for delivering fertility control agents to possums (Cowan et al. 2008). This intestinal parasite is thought to be monospecific for Trichosurus, although infection has been demonstrated in experimentally inoculated sugar gliders (Petaurus breviceps). The parasite is lumen-dwelling with no resident intratissue stages, thus exposure of potentially immunogenic parasite-derived material to the host is primarily via the gut mucosal surface. Parasitic larvae infect via dermal penetration followed by tissue migration through the lungs and/or oronasal region before reaching their residence site in the gut, so there is potential for transient introduction of parasite-derived material into somatic tissues by this route. Under experimental conditions, infection doses >10³ infective larvae per possum are used to initiate infection, with subsequent adult worm burdens recoverable from the gut in the low 10² range (i.e. 10–20% successful establishment following skin contact; Crook et al. 2005). It is uncertain whether, under conditions of natural exposure, T. vulpecula develops any degree of immune resistance to reinfection with P. trichosuri. However, acquired immunity and reduction of worm burdens, which is determined by inheritable MHC genetics, is a noticeable feature of strongyloid infections in most mammals (Hohenhaus & Outteridge 1995).

One feature of P. trichosuri that makes it particularly attractive as a potential transmissible vector for reproduction control is its free-living life cycle stage that, unlike other strongyloid worms, can be autogenous for several generations without the need for parasitism (Grant et al. 2006). Since the triggers that initiate conversion from free-living to infective-stage parasitic larvae have been identified (Grant et al. 2006) it is feasible to mass-produce infective-stage larvae in vitro, suitable for wide-scale field application. The genome of infectious-stage P. trichosuri is amenable to manipulation, and has been engineered to express heterologous antigens under the control of a heat-shock promoter gene (Grant et al. 2006). However, further research is needed to address three key issues. First, although transgenesis has been demonstrated, evidence that an expressed recombinant molecule can be produced and presented in an immunogenic fashion in vivo needs to be confirmed (two in vivo trials of transgenic worms expressing the highly immunogenic mycobacterial antigens Ag85 and ESAT-6 showed that possums produced antibody responses in one trial but failed to react in a repeat study; Warwick N Grant unpubl. data). Second, further research on the stability and inheritance of the transgene is needed. Worms expressing the model heterologous gene beta-galactosidase showed variable inheritance between different transgenic lines, although all lines were shown able to establish patent infections and the model transgene was transmitted effectively from the parasitic adult to the free-living progeny. In addition, however, modelling has indicated that parasite aggregation in hosts could result in many possums bearing low worm burdens, which may not result in sufficient antigenic exposure to invoke a sterility-inducing immune response (Tompkins 2007). Third, the potential for long-term persistence of a transgenic strain in the field is unknown, since any fitness cost of transgenesis would put recombinant worms at a competitive disadvantage with wild-type strains.

Studies on P. trichosuri dynamics in the field have demonstrated dispersal from a single artificial infection site in the Upper South Island to an area of approximately 6000 ha within 3 years (Cowan et al. 2006). By releasing a second genetically distinguishable worm strain at the same site 4 years after the initial release, it was further demonstrated that a new strain could be established and persist (at least in the short term) alongside existing strains in the field (Warwick N Grant & Phil E Cowan unpubl. data). The ability of the worm to survive for extended periods as a free-living/non-parasitic form also makes it attractive for field dispersal, since P. trichosuri can persist as a potentially infective agent in the field even at low possum density.

In summary, the possible use of a transgenic Parastrongyloides expressing fertility-control antigens has been modelled at both the laboratory and field levels. Some technological hurdles remain (e.g. increasing the rate of heritability of the transgene, producing a transgenic worm expressing fertility-associated molecules, and proving that this construct is immunogenic); these would require several more years of fundamental research before a transgenic worm bearing fertility control-associated molecules would be available for in vivo trials.

Modified possum-specific bacterial flora (gut commensals, pathogens)

The manipulation of possum-specific bacterial flora is a further option for the development of a transmissible vector. Key examples here would include gut-dwelling commensal bacteria (e.g. lactobacilli and bifidobacteria) or mucosal-surface pathogenic bacteria (an example would be the venereal pathogen Chlamydia). Colonisation of a mammal's intestinal tract with commensal bacteria usually occurs during neonatal development. Strains of lactobacilli and bifidobacteria are frequent colonisers, being resistant to gastric hydrolysis and bile salt emulsification. Usually part of the mother's intestinal tract flora, these bacteria naturally colonise the neonate during the prolonged period of contact with its parent, but if necessary could be artificially introduced to juvenile possums as an orally administered probiotic. Large numbers of live commensal bacteria are required to provide sufficient colonisation of the gastrointestinal tract (NB: human probiotics are generally administered at >10⁹ live bacteria/dose) but lactobacilli and bifidobacteria are easily mass-produced in in vitro culture, and so these numbers are achievable.

The genomes of lactobacilli and bifidobacteria are easily manipulated to express heterologous antigens, and many have been proposed as vaccine-delivery vectors for use in human and veterinary health (Ahmed 2003). However, because the bacteria are lumen-dwelling (with no intratissue contact except in the case of intestinal tract disturbance) it is difficult to raise an immune response in an animal that has received such a recombinant. Some success has been reported in laboratory animal models using recombinant Lactobacillus spp. to deliver highly immunogenic antigens (reviewed by Seegers et al. 2002); however, there is less evidence that heterologous antigens of low immunogenicity can be effective when delivered by gut commensal bacteria. Moreover, the ability of a recombinant commensal bacterium to induce an immune response depends mainly on the immunomodulatory nature of the carrier bacterium itself; all lactobacilli can potentially be engineered to express heterologous molecules, but relatively few strains colonise the gut while imparting a necessary stimulatory effect of the immune system. In this regard, the search for a suitable gut-colonising possum-specific commensal bacterium would be a long-term research proposition, not only for first discovering, identifying and characterising the bacterium (i.e. bioprospecting of possum gut flora contents), but also for selecting only those strains which readily recolonise the gut on oral delivery and which readily impart an effect on the immune system.

As an alternative to using non-pathogenic bacteria as vaccine-delivery vectors, it may be possible to engineer possum-specific mucosal pathogens to serve as delivery vectors. One such example is Chlamydia, a genital tract pathogen. There are two advantages of using Chlamydia as a live-vaccine delivery agent: first, Chlamydia strains are normally species-restricted (so theoretically a possum-specific strain could be used); and second, under natural conditions, Chlamydia is highly infective and readily disseminates throughout mammalian populations via venereal transmission. However, this line of research is yet to be tested and it is unknown whether any putative possum-specific Chlamydia would be amenable to genetic modification to express fertility-control molecules, nor whether such a construct would be stable.

Overview of the suitability/applicability of transmissible GM organisms as delivery systems for a possum fertility-control vaccine

The few candidate transmissible organisms suggested for development as potential fertility-control-delivery vectors reflects the relative paucity of relevant literature on the possum's pathogenic and commensal fauna, and the lack of in-depth characterisation of the potential vectors themselves. While possum enteroviruses have been isolated, little is known of their infectivity or transmission potential; at an even more rudimentary stage is research into the aetiological agent of WPD, over which there still remains debate as to whether this is a virus and, if so, to which taxonomic group it belongs. However, substantial progress has been made in understanding the basic biology, life cycle and transmission dynamics of the nematode P. trichosuri, a parasite with several characteristics that make it a good candidate for further development. In the short-to-medium term, the major hurdles for further progress with this approach are technological, especially at the genomic level: improving the stability of an inserted transgene, ensuring high-level expression of the transgene in vivo, improving generational heritability of transgene expression, and finally, ensuring immunological recognition and activation of a host antibody response upon infection. The long-term persistence of transgenic strains in the field is also currently unknown.

Overview and summary of the future for a possum fertility-control vaccine

The aim of this review has not been to propose one definitive delivery strategy as the preferred methodology for the future development of a fertility-control approach for possums. Rather, it has been the intention to first overview the research that has been conducted to date, then to present and debate the relative merits of the most promising organisms and systems that may be suitable for further research; and in doing so, to provide a critique of each organism in order to identify shortcomings and barriers to future use.

Key issues for a GM organism-based vaccine-delivery system include stability and achievability of the desired levels of expression of the fertility control molecule in the organism. In that regard, it should be considered a priority to determine (via captive possum studies) whether delivery of fertility-control molecules using a GM organism does indeed offer highly effective and sustained fertility reduction in possums. If the responses are not substantially improved beyond those recorded with existing oral-delivery systems for GnRH and ZP antigens, or unless alternative and more potent-effect sterilising antigens are identified, then the further development of GM technologies for possum fertility control would be precluded early-on. However, early indications in other species suggest that, developed and used optimally, vaccines delivered by GM organisms can provide pronounced and sustained fertility reduction (e.g. Hardy et al. 2006; O'Leary et al. 2008).

Clearly, given New Zealand's unique indigenous fauna, safety of any GM organism in the environment will be a further issue of importance. Some of the organisms reviewed (e.g. vaccinia) have been extensively safety-tested in birds and eutherian mammals in Europe and North America, but still require a proven safety profile for field use in New Zealand. The selection of an appropriate species-restricted contraceptive antigen will also be key; this is particularly true for those vaccine-delivery agents (e.g. vaccinia virus and BCG) that have a broad host range, especially if that includes humans. In particular, the GnRH molecule is phylogenetically-conserved in mammalia, a property which is utilised in the injectable GonaCon^TM vaccine to enable cross-reactivity of the peptide-sized molecule among many species; but this very cross-reactivity would likely preclude use of GnRH as a fertility-control antigen in a GMO-delivered vaccine, no matter how small the risk of non-target species infection. Much more likely for a GMO-delivered fertility-control vaccine would be the use of larger, polypeptide-sized target molecules, since these would allow for the selection of epitopes restricted to the target species. For example, Duckworth et al. (2008) demonstrated that a vaccine based on possum ZP3 epitopes failed to affect reproductive performance when injected into non-target mammals and birds, indicating that it would be feasible to develop a GMO-delivered possum vaccine so long as this was based on species-restricted epitope recognition.

Fertility control remains feasible for possums provided the vaccines can be demonstrated, experimentally, to be effective and target specific. In this respect, public concerns and regulatory arrangements for GM organisms in New Zealand will need to be addressed so that all risks have been thoroughly explored before field-testing could be considered. In some cases, this condition may be partially alleviated by historical epidemiological observations. For example, the nematode P. trichosuri and the herpesvirus MaHV-1 have both been documented in New Zealand for several decades but have had no reported impact on native fauna or human health. Of importance with the latter case, however, is safety with respect to indigenous Australian fauna; given that a vector that was possum-specific in New Zealand could potentially infect and cause reduced fertility, as well as disease in endangered Australian species, safety considerations place even greater constraints on possum fertility control development in New Zealand than just the New Zealand safety considerations. This issue has been debated in detail elsewhere (e.g. Gilna et al. 2005; Henderson & Murphy 2007; Cowan et al. 2008) and, particularly in the case of transmissible GM vectors, the concerns raised may be of sufficient magnitude to discourage development of a fertility-control vaccine based on such a vector. Hence, organisms that are either non-transmissible or have imperfect transmission or transgene inheritance (such that the organism and/or transgene persists long enough to cause possum population suppression, but fails to persist in the long term) are likely to be more acceptable to Australian regulatory authorities (Tompkins 2007).

Further research will be required before a fertility-control vaccine for possums becomes a reality. Given recent advances in molecular technology, engineering of any of the organisms discussed here should be achievable. Engineering of a well characterised organism to express fertility-control molecules may take as little as a year in the laboratory; and proof-of-effect of such a construct in captive possums may take several years. In contrast, expansion of preliminary findings into a field-ready vaccine for testing will require substantial technological development, safety testing and regulatory approval, which will take many years and substantial R&D investment, making it a long-term proposition for possum control. Although beyond the scope of this review, there are alternative genetic technologies available for control of reproduction through non-immunological means (e.g. daughterless offspring technology, or interference RNA techniques for single-gender selection) although these represent an even longer time-horizon for return on R&D investment (in excess of 20 years; Thresher 2007). However, possums remain an ongoing threat to New Zealand's fauna, flora, agriculture and environment, and so long as there are public concerns over existing lethal control technologies, it is suggested that at least consideration should continue to be given to long-term biocontrol strategies. These strategies may well include fertility control objectives as tools in an integrated pest management programme for possums in order to maintain sustainable, long-term reductions in possum abundance throughout the country.

Acknowledgements

This is a publication of the National Research Centre for Possum Biocontrol, which was funded by the NZ Foundation for Research, Science and Technology (Possum Biocontrol OBI C10X0501). The authors thank the personal correspondents for their advice and provision of unpublished result findings (Dr Doug Eckery, Victoria University; Dr Warwick Grant, La Trobe University; Dr Darelle Thomson, Massey University; and Dr Tony Robinson, formerly CSIRO Australia). The authors also thank the comments and advice provided on an earlier draft of this manuscript by Dr Dan Tompkins and Dr Brian Hopkins (Landcare Research).