PREDICTING THE FATE OF ENGINEERED BACULOVIRUSES
John Buranda,b, Greg Dwyerb, and Joseph S. Elkintonb
aDept. of Microbiology, University of Massachusetts, Amherst, MA 01003; and bDept. of Entomology, University of Massachusetts, Amherst, MA 01003-2410
SUMMARY
Insect baculoviruses have enormous potential for use as microbial control agents against insect pests. The main obstacle to their widespread use has been their slow speed of kill. Although genetic engineering may allow the construction of virus strains that kill more quickly, there is the possibility that such engineered strains may competitively displace wild-type viruses, at least over the short-term. Such competitive displacement may lead to unforeseen environmental problems. Here we present a mathematical model that allows us to assess the relative short-term competitive abilities of engineered and wild-type virus strains. The model suggests that even relatively modest increases in speed of kill can lead to substantial fitness advantages, unless the decreased speed of kill is associated with lower virus transmission rates.
INTRODUCTION
Baculoviruses are naturally-occurring pathogens of insects that have tremendous potential for use as microbial agents for insect pest control. The survival of the virus outside of the insect when the host species is not present is prolonged by the presence of a protective occlusion body protein coat. This OB is thought to provide the virus with protection from inactivation by a variety of environmental factors such as pH, dew and ultraviolet light. Without it, the virus is rapidly inactivated when the infected insect dies (Evans and Harrap, 1982; Entwistle and Evans, 1986).
One problem that has been encountered with the use of baculoviruses for large-scale insect control is their slow speed of kill. Most baculoviruses kill their host 7 to 14 days after ingestion. This is often unacceptable, because the insect is eating and causing damage to crops during much of the time. Several solutions to this problem have been proposed but the one that has gained the most attention is genetic engineering of viruses to carry foreign genes that can cause faster kill and thus less crop damage (Vlak, 1992).
One approach to do this is to arm the virus by substituting a gene coding for an insecticidal protein toxin for the viral polyhedral gene. Part of the reason why viruses have a slow speed of kill is that they must replicate many times to kill the insect, whereas only a few copies of a toxin gene are necessary to kill. Moreover, the removal of the gene for the polyhedral coat (producing what is known as a poly- virus) means that such viruses will not persist for very long in the environment (Miller, 1988). Although this feature reduces the environmental impact of the engineered virus, it also renders the virus ineffective for pest control. In an effort to have the best of both worlds, Hamblin et al. (1990) suggested a strategy of co-occluding engineered and wild-type viruses. This is accomplished by infecting insects in the lab with both the wild-type and engineered strains. The resulting OBs incorporate virions of both strains.
Co-occlusion has the advantage that it causes the engineered virus to disappear from the wild population. This is because there are three possible outcomes of virus replication in insects infected with both the poly- and wild-type viruses. Cells infected with only wild-type virus produce polyhedra containing only wild-type viral capsids, while cells infected with only poly-virus do not produce polyhedra. When the insect dies, viral capsids that are not occluded within polyhedra are rapidly inactivated. Poly- virus can thus survive the death of the insect when cells are infected with both strains, thereby producing more co-occluded polyhedra. Since cells are infected with both strains much more rarely than with only one strain, the poly- strain should eventually go extinct (Hamblin et al., 1990). In cooperation with the Forest Service, we are currently testing the effectiveness of the co-occlusion strategy in limiting the long-term survival of a co-occluded virus.
A second method is to insert a toxin gene, which is under the control of a viral promoter, into a non-essential region of the viral genome. This results in a poly+ virus that expresses the toxin during viral replication. A virus of this type expressing an insect-specific toxin gene was recently field tested in England (Cory et al., 1994). Because of their potential for longterm survival, the use of poly+ viruses as microbial pest control agents has raised important questions about the possibility of environmental risks. These questions include: i. Can the gene carried by the virus be transferred to other baculoviruses that can then replicate in non-target hosts (Altmann, 1992; Vlak, 1992)? ii. Can the released engineered virus competitively displace the wild-type virus (Tiedje et al., 1989)?
Here we present a method for providing at least a preliminary answer to the second of these questions, using protocols that do not require field releases. Our approach is to use a mathematical model for the dynamics of virus epizootics, to predict the outcome of competition between strains. The advantage of the model is that the information that it requires can be generated from lab or greenhouse studies, rather than from field releases. We use the model to ask the question, will the altered properties of the engineered virus allow it to outcompete and replace the wild-type virus?
MATERIALS AND METHODS
In the modelling work that we present here, we focus on an engineered gypsy moth virus that kills more quickly than the wild type because it is missing the ecdysteroid glucosyl transferase (egt) gene normally found in the wild type. Egt expression in virus-infected larvae blocks molting and pupation by conjugating sugar molecules to the molting hormone ecdysone (O'Reilly, 1995). With the gypsy moth virus, if the egt gene is deleted, virus-infected larvae die approximately 1 to 3 days sooner then do insects infected with the wild-type virus (Burand et al., 1994).
The model that we use is based on an earlier model of Dwyer and Elkinton (1993), modified to allow for competition between virus strains. The earlier version of the model predicts with fair accuracy the dynamics of the virus in large-scale populations from information obtained in small-scale field trials. In what follows, we use this earlier information, in combination with our estimate of the faster speed of kill of the engineered virus, to sketch out some hypothetical scenarios of competition between the engineered and wild-type strains. In particular, we do not yet have a reliable estimate of the transmission rate of the engineered virus, so we consider a range of transmission rates relative to the wild-type. The model is:
dS/dt = -S(vePe + vwPw) (1)
dPe/dt = evePe(t - e)S(t - e) - µePe (2)
dPw/dt = wvwPw(t - w)S(t - w) - µwPw(3)
where S is the population of uninfected larvae, Pe and Pw are the populations of engineered and wild-type polyhedra outside of any hosts respectively, ve and vw are the engineered and wild-type transmission rates, e and w are the number of OBs produced by the engineered and wild-type strains, e and w are the survival times of larvae infected with the engineered and wild-type strains, and µe and µw are the decay rates of the engineered and wild-type strains.
Equations 1-3 present the most general model; here, we begin by assuming that the decay rates µeand µw are the same. From lab studies, we know that the survival time e of the engineered strain is about 3 days less than the survival time w of the wild-type strain, or about 11 d versus about 14 d. As we will demonstrate, the reduced time to kill of the engineered virus gives it a competitive advantage in terms of numbers of insects infected. The question then becomes, how much greater would the transmission rate of the wild-type have to be for the two strains to be equally competitive? Although strictly speaking we take the term 'transmission rates' to mean the parameters ve and vw, in fact changes in ve and vw have the exact same effect as changes in the numbers of OBs per infected insect, e and w, respectively.
RESULTS
The model output is shown in Figure 1. Here we have plotted the cumulative fraction of secondary infections due to each strain, where this fraction is the number of secondary infections with each strain, divided by the initially uninfected population at the beginning of the season. NPV epizootics in gypsy moth populations are begun when larvae hatch out and become infected as a result of environmental contamination of egg masses (Murray and Elkinton, 1990). The model results in Figure 1 show only the infections arising after the initial wave of infection. That is, each strain, engineered or wild-type, is assumed to infect 15% of the population at the beginning of the season, a value taken from observations of field populations. The initial density of egg masses is about 20,000 per hectare, which is typical of high-density gypsy moth populations.
The differences among the three graphs in Figure 1 are due solely to changes in the wild-type transmission rate vw relative to the engineered transmission rate ve, although as we have mentioned such changes are identical to changes in the number of OBs per infected cadaver. In Figure lA, the engineered and wild-type strains have the same transmission rate, estimated in the field by Dwyer and Elkinton (1993) for a wild-type virus. This graph demonstrates the substantial competitive advantage that the engineered strain has over the wild-type, all else being equal. In fact, however, it is likely that the wild-type strain may have higher transmission, because of tradeoffs between time to kill and infectiousness (Anderson and May, 1981; Ewald, 1994). In Figure 1B, the transmission rate of the wild-type strain has been adjusted so that the fraction of larvae infected by the two strains is the same, a transmission rate which turns out to be about twice (2.1 x) the transmission rate for the engineered strain. In Figure 1C, the transmission rate of the wild-type is 3 x the transmission rate of the engineered strain.
DISCUSSION
Our explorations of the model under different assumptions about the relative transmission rates of the egt- and wild-type strains have generated some useful insights about the impact of changes in survival time on the competitive ability of the engineered virus. Specifically, Figure 1A shows that even a relatively small increase in the speed of kill can lead to a substantial difference in the fraction of larvae infected by the two strains. Over longer time intervals, this difference would almost certainly lead to the competitive displacement of the wild type by the engineered virus, which in turn could have a significant impact upon the frequency and intensity of gypsy moth outbreaks (Anderson and May, 1981; Briggs and Godfray, 1995). Of course, it is possible that tradeoffs between speed of kill and transmission rate could lead to a higher transmission rate for the wild-type virus, leading to at least temporary coexistence between the wild-type and engineered strains. Figure 1B shows that the transmission rate of the wild-type virus would have to be almost twice as high as that of the engineered virus, merely for the two strains to be equally competitive. That would mean, for example, that the wild-type strain would have to produce twice as many OBs as the engineered strain. In fact, however, the most environmentally desirable outcome would be for the wild-type to be more competitive. Figure 1C shows that for the wild-type to have a substantial competitive advantage it might have to be as much as three times as infectious as the engineered virus.
In short, the model suggests that even modest changes in the phenotype of an engineered virus strain can have dramatic effects on the outcome of competition between strains. Our next step is to accurately quantify the relative transmission rates of the egt- and wild-type strains. It may well be that the wild-type strain is considerably more infectious than the egt- strain. Our preliminary results, however, suggest that there is reason to be cautious before releasing egt- into the field. In fact, given that egt- mutants apparently have not appeared in field populations, it seems likely that there are components of the fitness of NPVs that are not captured by the model, and one or more such components would probably lead to the eventual extinction of egt- in the field. Over the short term, however, the environmental impacts of releasing egt- could be significant.
REFERENCES
Anderson, R.M. and R.M. May. 1981. The population dynamics of microparasites and their invertebrate hosts. Phi. T. Roy. Soc. B. 291:451-524.
Altmann, M. 1992. 'Biopesticides' turning into new pests? TREE. 7:10.
Briggs, C.J. and H.C.J. Godfray. 1995. The dynamics of insect-pathogen interactions in stage-structured populations. Am. Nat. 145:855-887.
Burand, J.P., G. Dwyer, V. D'Amico, and J.S. Elkinton. 1994. Methods for monitoring the environmental fate of genetically engineered baculoviruses. In: Biotechnology Risk Assessment: Proceedings of the BIotechnology Risk Assessment Symposium. Univ. Md Biotechnology Institute #9501. pp. 293-306.
Cory, J.S., M.L. Hirst, T. Williams, R.S. Hails, D. Goulson, B. M. Green, T. M. Carty, R.D. Possee, P.J. Cayley and D.H.L. Bishop. 1994. Field trial of a genetically improved baculovirus insecticide. Nature 370: 138-140.
Dwyer, G. and J.S. Elkinton. 1993. Using simple models to predict virus epizootics in gypsy moth populations. J. Animal Ecology. 62:1-11.
Entwistle, P.F. and H.F. Evans. 1986. Viral control. In Kerkut, G.A. and L.I. (eds) Comprehensive Insect Physiology., Biochemistry and Pharmacology. Fol 13. Pergamon Press, Oxford. pp.347-412.
Ewald, P.W. 1994. Evolution of Infectious Disease. Oxford University Press, Oxford, England.
Evans, H.F. and K.A. Harrap. 1982. Persistence of insect viruses. In Minson, A.C. and G.K. Darby (eds) Virus Persistence . SGM Symposium. Cambridge University Press, Cambridge. pp.57-96.
Hamblin, M., van Beek, N.A.M., Hughes, P.R. and Wood, H.A. 1990. Co-occlusion and persistence of a baculovirus mutant lacking the polyheclrin gene. Appl. Environ. Microbiol. 56:3057-3062.
Miller, D.W. 1988. Genetically engineered viral insecticides: Practical considerations. In:Biotechnology for Crop Protection (P.A. Hedin, R.M. Hollingworth and J.J. Mann. Eds) pp. 405-421. American Chemical Society, Washington, D.C.
Murray, K.D. and J. Elkinton. 1990. Transmission of nuclear polyhedrosis virus to gypsy moth eggs via contaminated substrates. Env. Ent. 19:662-665.
Tiedje, J.M., J.R.K. Colwell, Y.L. Grossman, R.E. Hodson, R.E. Lenski, R.N. Mack, and P.J. Regal. 1989. The planned introduction of genetically engineered organisms: ecological considerations and recommendations. Ecology. 70: 298-315.
Vlak, J.M. 1992. Genetic engineering of baculoviruses for insect control. In: Oakeshott, J. and M.J. Whitten (eds) Molecular Approaches to Fundamental and Applied Entomology. Springer-Verlag, New York. pp. 90-127.
Figure 1. Output of model equations 1-3. For all three graphs, parameters are taken from Dwyer and Elkinton (1993), with two exceptions. First, the time between infection and death for the engineered strain is 11 d, as compared to 14 d for the wild type. Second, the transmission rate of the wild type varies among graphs. In A, the transmission rate of the wild type and engineered strains are the same. In B, the transmission rate of the wild type is approximately twice that of the engineered strain, and in C, the transmission rate of the wild type is three times that of the engineered strain. Note that in A, even the relatively slight increase in speed of kill of the engineered strain is sufficient to significantly increase its fitness relative to the wild type, as measured by the fraction of the population dying of each strain. Further, B and C demonstrate that, if there is a tradeoff between speed of kill and infectiousness, the wild type must be drastically more infectious to have a significantly higher fitness.