INTRODUCTION

Top Abstract Introduction Materials and Methods Results Discussion References

Baculoviruses infect the larvae of many important lepidopteran pest species, both of agriculture and forestry (20). Because of this, they have been seen for many years as potential biopesticides (5). Though widely used against a few species of pest, in general their use as insecticides has been relatively limited. One of the principal factors contributing to this has been their slow speed of action (29). Recently, considerable attention has focused on the possibility of improving their efficiency by genetic manipulation (6, 33, 34). Several reports have described the construction of nuclear polyhedrosis viruses (NPVs) with accelerated action by genetic engineering (4, 6, 10). However, the prospect of releasing genetically engineered virus into the natural environment has given rise to a number of safety concerns, including the possibility of adverse effects of the recombinant virus on nontarget organisms (8) or the risk of spread of the transgenes by recombination with other viruses or even other organisms (12).

A strategy that has been proposed to counter the possible risks of releasing manipulated viruses into the environment is that of co-occlusion (14, 19, 34). Polyhedrin is the major component of occlusion bodies (OBs), which allow the virus to persist in the external environment (11, 24, 30). If a recombinant is engineered such that the novel gene is inserted in place of the polyhedrin gene, the resulting virus will be occlusion negative (occ^-). Such a virus is unsuitable for use as an insecticide by itself because it is inactivated very quickly in the field (4). However, occlusion-positive (occ⁺), wild-type virus and occ^- virus can be grown together in cell culture, resulting in the co-occlusion of both genotypes into OBs when cells are infected by at least a single genome of each genotype. These co-occluded OBs provide a means of delivering the recombinant virus to insects in the field. Within an infected insect, cells infected by only occ^- genomes will not produce any OBs. Virions produced in these cases cannot therefore be passed on to other insects during further rounds of infection and will be lost. This asymmetry means that, over successive rounds of infection in insects, the occ^- genotype should be out-competed by the occ⁺ one.

The phenomenon of co-occlusion is based on an unusual feature of the biology of NPVs, that multiple copies of the genome can be inherited together within the same OB. This has obvious implications for the existence in stable genetic polymorphisms of other types of variants within a baculovirus population. Natural baculovirus isolates are frequently observed to comprise multiple genomic variants (7, 9, 18, 26). However, the dynamics of the interactions between these variants or their significance to the biology of the virus are poorly understood. To provide a conceptual framework for the interpretation of experimental studies addressing these problems, a model of NPV population genetics was developed (13). The model predicts allele frequency changes in mixed NPV genotype infections. It is based on the major components of the NPV replication cycle. The model is expressed as a recurrence equation and can be used to predict the fate of an allele either in an engineered, occ^- genotype or in a naturally occurring genomic variant. The chief value of the model is that it can be used to discover the most critical parameters influencing the rate of change of gene frequency. For the fate of engineered, co-occluded virus, this is the number of viral genomes infecting a cell destined to produce OBs. This parameter is the frequency distribution of the multiplicity of infection (MOI) of cells and is important since it determines the degree to which the disabled genome can "parasitize" wild-type polyhedrin production.

Nothing is known about the frequency distribution of the MOI within naturally infected insects. However, the MOI distribution can be estimated in the laboratory by the rate of loss of an allele of known fitness effects in serial rounds of infection. In this study, we describe experiments, in which Trichoplusia ni (cabbage looper) larvae were coinfected with the L1 strain of wild-type Autographa californica nuclear polyhedrosis virus (AcMNPV) and an occ^- derivative of the same virus (occ^-1). Our data showed considerable persistence of the occ^- strain and pointed to an average MOI of between 4 and 5 virus genomes per cell.

	MATERIALS AND METHODS

Top Abstract Introduction Materials and Methods Results Discussion References

Cells and viruses. T. ni (TN368) (16) and Spodoptera frugiperda IPLB-Sf21-AE cells (28) were used to propagate and assay virus as described previously (23). Cells were maintained in TC100 medium (Life Technologies, Ltd., Paisley, United Kingdom) plus 10% fetal calf serum (M. B. Meldrum, Ltd.). AcMNPV strain L1 (18) was used as the occ⁺ genotype and as the parental strain in the construction of the occ^- virus, AcMNPV.occ^-1 (occ^-1). OBs were purified from cell culture and insects as described previously (23). Nonoccluded virus was disrupted by two washes in 0.5% sodium dodecyl sulfate (SDS) and then one wash in 0.5 M NaCl before resuspendion of the OBs in distilled water. Purified OBs were counted by using a hemocytometer.

Insects and bioassays. T. ni eggs were obtained from J. Cory (NERC Centre for Ecology and Hydrology, Oxford, United Kingdom). The larvae were reared on an artificial diet at 27°C (3). Fifth-instar larvae were infected with virus either by injection with budded virus (BV) or per os with OBs. Injection was subcutaneous, in 5-µl aliquots, administered with a Hamilton syringe. When hemolymph was collected, this was achieved by removal of a hind proleg, followed by bleeding onto parafilm. Infection per os was administered as a suspension of OBs placed on cubes (ca. 2 mm in diameter) of artificial diet. Insects to be infected per os were starved for 6 h prior to being presented with infected diet cubes. When this had been consumed, they were refed with artificial diet as necessary until death. In all insect infection experiments, a number of uninfected insects were retained until pupation as controls. Preliminary bioassays with purified OBs were carried out to assess the dose-response relationship between AcMNPV and fifth-instar T. ni under our experimental conditions (data not shown).

Construction of recombinant virus. occ^- virus was generated by recombination between L1 and a plasmid pEVmXIV (31). This plasmid includes AcMNPV sequences flanking the polyhedrin gene (2.53 to 4.84 map units), but the polyhedrin gene itself is replaced by a short polylinker sequence. This polylinker includes a single EcoRI site, which is introduced into the recombinant genome. After cotransfection of SF21 cells with viral and plasmid DNA, progeny virus was screened by plaque assay for occ^- recombinants. Phenotypically correct clones were verified by restriction enzyme analysis.

Viral replication rates. In order to assess the relative rates of replication of the occ⁺ and occ^- viruses, TN368 cells were infected in triplicate with a mixture of both viruses at a combined MOI of 1, and the cell culture medium was collected at 24, 48, 72, and 96 h postinfection (hpi). The rates of viral replication were also measured in vivo. Twelve insects were infected by injection with a total dose of 10⁴ PFU, and hemolymph was collected from sets of three insects at 24, 48, 72, and 96 hpi (at 96 hpi all insects were dead). In order to confirm that the relative rate of replication of the viruses did not change with serial passage, BV from the in vivo samples taken 72 hpi was passed serially through two more rounds of insect infection, and hemolymph was collected after each round. In all trials, the cell culture medium or hemolymph was screened by plaque assay to assess the relative frequencies of both genotypes.

Southern blot analysis. Viral DNA was digested for 4 h with EcoRI, and fragments were separated by electrophoresis. The two genotypes can be distinguished by analysis of the resulting pattern. A 528-bp DNA sequence corresponding to part of the flanking region upstream of the polyhedrin gene was generated by PCR (baculoL primer [5'-CTGTCGACAAGCTCTGTCCGTT-3'; probeanti primer [5'-CAAAACCGACGATCCCAAATTC-3') and used as a probe. It was labeled with [-³²P]dCTP (3,000 Ci/mmol) by using a High Prime DNA Labeling Kit (Roche Diagnostics, Ltd., Lewes, United Kingdom). Then, 0.8% agarose gels were transferred to Hybond N+ membranes (Amersham Pharmacia Biotech, Inc., Hertshire, United Kingdom) by capillary action as described previously (25). Hybridizations were performed overnight at 65°C in fresh 5× SSC (1× SSC is 0.15 M NaCl plus 0.015 M sodium citrate)-0.5% SDS, 6× Denhardt's solution, and 100 µg of denatured, fragmented salmon sperm DNA/ml plus radiolabeled probe. The blots were washed once in 1× SSC-0.5% SDS for 20 min at 65°C and then twice in 0.5× SSC-0.5% SDS for 20 min at 65°C. The blots were exposed for 1 h to BAS-MP plates, which were then scanned with a BAS 1500 phosphorimager (Fujifilm). The probe extends from positions 3657 to 4185 (1) of the AcMNPV genome, within the EcoRI I fragment, and hybridizes to a 7.3-kb fragment in wild type and a 4.0-kb fragment in the recombinant digest. The relative amount of occ⁺ and occ^- genomes in a preparation was therefore determined by comparing the radiation intensity of these bands with MacBas v.2.5 software. In trial experiments, different amounts of wild-type DNA, quantified by using a Beckman DU 640 spectrophotometer, were digested with EcoRI and analyzed as described above to confirm that a linear response was observed (data not shown).

Serial rounds of insect infection. In order to track the fate of viral genomes in co-occlusion over serial rounds of insect infection, 10 insects were infected by injection with an equal ratio of both genotypes at a total dose of 10⁴ PFU in 5 µl. OBs were harvested after death, purified to disrupt any nonoccluded virus, and pooled. This was used as the starting inoculum for subsequent rounds of infection, wherein the insects were infected with OBs per os. This was initially performed at high dose (5,000 OBs per insect) for three rounds of infection. Subsequently, a further three rounds of high-dose infection were carried out in parallel with three rounds of low-dose (50 OBs per insect) infection. For each round of infection, insects were divided into 10 sets of three for high-dose infections and 10 sets of five for low-dose infections. From each set, OBs were taken from a single cadaver and used to infect a set of insects for the next round of infection. This allowed 10 independent lines of serial infection to be established, with sufficient replicates within each line to ensure continuation of all of the lines even when less than 100% mortality occurred. In addition, 10 insects were infected by injection with occ^- virus alone. No OBs were observed from these insects after death, and hybridization did not detect any DNA, confirming that nonoccluded virus did not survive the OB purification procedure (data not shown). Viral DNA from OBs from every round of infection was collected (from the insect selected from each set for the next infection round), and the ratio of genotypes in each sample was measured by intensity of hybridization after Southern blotting as described above.

Analysis. We assume the frequency of the two genotypes at the BV stage is the same as that in the OBs that initiated the infection, and we checked to see that neither genotype has a replication advantage. Hence, the rate at which the occ^- genotype is lost depends solely on the probability that it coinfects an OB-producing cell with an occ⁺ genotype. Specifically, let p(t) be the frequency of the occ^- genotype at infection round t and let f(n) be the probability that a total n viral genomes will infect an OB-producing cell. The frequency of occ^- genotype in the next infection round is thus (13):

(1)

occ^- genomes are lost when all n viruses infecting an OB-producing cell are of this genotype.

	RESULTS

Top Abstract Introduction Materials and Methods Results Discussion References

Relative replication rate of L1 and occ^-1. A key parameter in the model that could affect the likelihood of co-occlusion was the relative replication rate of the viruses. For example, if occ^-1 replicated more rapidly than L1, it would accumulate to higher levels within the insect and therefore be co-occluded more frequently. We did not expect any significant difference in replication rates of the virus genotypes but, to confirm this, the replication rates of the viruses, both in cell culture and in insects were compared. TN368 cells were infected in triplicate with a mixture of both viruses at a combined MOI of 1, and the relative frequency of both genotypes measured at various times postinfection by plaque assay. Similarly, T. ni larvae were infected by injection with a mixture of both viruses and hemolymph collected from groups of three insects at various times postinfection for plaque assay. The data in Fig. 1 show that there was no significant difference in the titers of L1 and occ^-1 at any time after infection of either cells or insects.

View larger version (16K): [in this window] [in a new window]   FIG. 1.   Replication rates of AcMNPV strains. The titer of BV was measured during a single round of infection for L1 and occ-1 over 4 days postinfection and compared. Both in cell culture (A) and in insects (B), neither virus was seen to have an advantageous rate of replication. The SE is indicated by bars (n = 3).

Persistence of occ^-1 during serial passage. Ten insects were infected with equal amounts of the two genotypes by injection, and OBs were recovered and pooled. The mean ratio of the genotypes in these was found to be 66% L1 to 34% occ^-1. In order to track the fate of the viral genotypes over serial rounds of infection, 10 insects were infected per os by using this OB preparation. Initially, the insects were infected with a very high dose of virus (>10 × 100% lethal dose [LD₁₀₀]). Six serial rounds of infection were carried out in 10 independent lines of insects, and viral genotype ratios were measured by Southern blotting for all insect lines after each infection round. The data in Fig. 2 demonstrate that the proportion of L1 increases with passage as expected, but only very slowly. The mean ratio after six infection rounds was 81% L1 to 19% occ^-1, and the occ^-1 genotype was still present in all 10 lines of infected insects. We investigated what mean MOI of cells by both viruses within the insect could best account for this rate of loss of occ^-1, assuming a Poisson distribution (using the model described in Materials and Methods). It was found that a mean of 4.3 ± 0.3 (95% confidence interval) genomes per cell provided the best fit to the data.

View larger version (35K): [in this window] [in a new window]   FIG. 2.   Percentage of L1 through serial rounds of co-occlusion. (A) Example (round 3) of a Southern blot used to track the fate of the L1 and occ-1 genotypes over serial rounds of infection. Each lane represents the total virus DNA recovered from OBs from a single insect. The upper band derives from L1, whereas the lower band derives from occ-1. Lines 1 to 10 were independent series of infection. (B) Percentage of the L1 genotype over the course of six rounds of serial infection for 10 independent lines of insects and the best-fitting model prediction if a Poisson distribution of MOI is assumed. In this series, all insects were infected with 5,000 OBs each.

View larger version (17K): [in this window] [in a new window]   FIG. 3.   Percentage of L1 after reduction of infectious dose. The percentage of the L1 genotype was measured over three serial rounds of infection at a dose of 5,000 OBs per insect (the same data as shown in rounds 1 to 3 of Fig. 2) for 10 independent lines of insects. Subsequently, all insects were infected with 50 OBs each (rounds 4 to 6), and the 10 independent lines of insects were continued. Also shown is the mean (SE [error bars], n = 10) for the 10 lines of infection.

	DISCUSSION

Top Abstract Introduction Materials and Methods Results Discussion References

Coinfection of cells by multiple viral genomes can have a significant influence on baculovirus biology, facilitating phenomena such as complementation or recombination between genomes, interference by defective genomes, viral evolution, etc. However, virtually nothing is known about the MOI of cells within an infected insect, largely due to the technical difficulty of directly measuring this parameter. Here, we use an alternative strategy to estimate the distribution of MOI within infected insects. The relative genome frequencies of occ⁺ and occ^- variants of AcMNPV were tracked during serial infections of T. ni larvae. The occ^- virus is only infectious to insects if it is co-occluded within OBs formed by occ⁺ virus. Thus, the rate of loss of the occ^- genome depends on the frequency of coinfection of cells within the insect, since co-occlusion can only occur if both genomes infect the same cell. A mathematical model that describes the relationship between the MOI distribution and changes in genome frequencies during serial passage was described previously (13). Fitting the data obtained here to this model indicated that the MOI within the insect is surprisingly high, with a mean of more than four genomes per cell, if a Poisson distribution is assumed. This assumption provided a very good fit to the data and suggests free mixing of genotypes within the insect, at least by the latter stages of infection, when the majority of OBs are produced. A mean MOI value this high implies that more than 90% of cells in host insects by the time of death are infected by more than a single genome. We have excluded the possibility that the slow rate of disappearance of the occ^- genotype is due to it having a replication advantage over the occ⁺ genotype. There was no detectable difference in the replication rates of both viruses, either in cell culture or in infected insects (Fig. 1).

The model predicts that the most important parameter is the MOI and that the number of infectious units ingested by an insect would only have a small effect on the rate of loss of the occ^- genome, until the number of genomes actually initiating the infection became very small (<4). In the present study, in agreement with the model predictions, the data showed that the mean rate of loss of the occ^- genotype was very similar between the low- and high-dose infections (Fig. 2 and 3). This implies that, even in infections in which the insect receives a single lethal dose, multiple genomes (4) successfully cross from the gut and initiate replication within the insect. It is worth bearing in mind that AcMNPV is a multiply embedded NPV, i.e., virus particles are embedded in groups with the OB. Each group of virus particles is surrounded by a single membrane, so that all virus particles in a group presumably infect a single gut cell simultaneously. It remains to be seen whether the fact that AcMNPV is multiply embedded contributes to the persistence of the occ^- virus, even at low doses. In general, the ramifications of single versus multiple packaging of nucleocapsids remain unclear. Hamblin et al. (14) have suggested that virions containing a single nucleocapsid are several times more infectious than those containing multiple nucleocapsids. On the other hand, van Beek et al. (27) have reported that infection by multiply embedded nucleocapsids results in reduced survival time of insects compared to infection by an equivalent number of singly embedded nucleocapsids. Similarly, Washburn et al. (32) suggested that multiple packaging of nucleocapsids accelerates the onset of systemic infection in insects.

Co-occlusion was proposed as an attractive method for delivering a recombinant baculovirus insecticide to the field because it should provide a degree of biological containment. The occ^- recombinant should be outcompeted by the wild-type virus and should therefore not persist in the environment. Our data demonstrate that, as a consequence of the virus high MOI within the insect, even after low-dose infections, co-occlusion is unlikely to provide significant biological containment. The vast majority of cells in an insect will be infected by more than one genome, and the occ^- recombinant will only disappear very slowly. Indeed, model predictions indicate that it would take more than 40 rounds of infection for the frequency of the occ^- genotype to drop from 50 to 1%.

Our results seem to conflict with those of Miller (19), who presented data suggesting an occ^- recombinant would disappear in three to four generations. However, the precise nature of the occ^- virus used in those studies is not clear. In particular, it was not specified whether the virus was a simple polyhedrin deletion mutant or whether it was a recombinant expressing a foreign gene. Thus, comparison with our data is difficult. Hamblin et al. (14) and Wood et al. (34) have also presented studies evaluating co-occlusion as a biological containment strategy. The first of these studies only involved a single round of infection in insects (14). The data of Hamblin et al. are broadly consistent with ours. The second study was a field trial designed to monitor the loss of an occ^- recombinant after field application (34). These authors observed persistence of the occ^- virus for at least 2 years. However, because of a lack of larval infestations at the test site, Wood et al. were observing environmental persistence of OBs released at the start of the trial and not persistence of the virus after multiple infection rounds in insect hosts.

Co-occlusion of a polyhedrin deletion mutant is a relatively artificial example of the coexistence of baculovirus variants within an insect population due to complementation. Our data are also relevant to other, naturally occurring variants. A particularly good example is the FP (few polyhedra) mutant phenotype frequently observed in baculovirus stocks (17). FP mutants are often associated with interruption of the FP25 gene by deletions, or insertion of exogenous DNA within this gene (2) and are characterized by the production of small numbers of OBs and poor occlusion of virions (15). However, unlike the simple polyhedrin deletion mutants discussed above, FP mutants have a replication advantage over wild-type virus (15). Given the high MOI indicated here, it is predicted that FP mutant genomes could persist and reach an equilibrium within baculovirus populations with only a relatively moderate replication advantage. If we assume a Poisson distribution with a mean MOI of 4.3, an FP mutant with a replication rate of only 1.1 times wild-type could reach an equilibrium value of 10% of the whole population. In this case, the FP genotype could be described as parasitic on the wild-type population. We are currently carrying out experiments to determine whether FP mutants can in fact be parasitic on wild-type populations, although we note that there have been no clear reports of FP mutants segregating in field populations. However, one case of a parasitic baculovirus mutant has been reported (21) that appears to maintain a stable relative frequency in test populations over at least four rounds of infection (22). Also, it is known that baculovirus populations in the field often comprise multiple genomic variants (7, 9, 18, 26). The high MOI indicated here would provide ample opportunity for complementation between coinfecting viral genomes and probably contributes significantly to the evolution of the heterogeneity observed.

A concern about the use of genetically engineered viruses as insecticides is the risk of spread of the transgene to endogenous viruses by recombination. There is very little information available on the likelihood of such events taking place. However, we note that one of the conditions required for recombination is coinfection of a cell by multiple genomes. It is important to bear in mind that, although our data indicate that coinfection is highly prevalent within an infected insect, this refers to infection of the insect by genomic variants of the same virus that are identical to each other in most respects. It is entirely possible that similar high levels of coinfection would not be observed after coinfection of an insect by distinct baculovirus species.