The sf32 Unique Gene of Spodoptera frugiperda Multiple Nucleopolyhedrovirus (SfMNPV) Is a Non-Essential Gene That Could Be Involved in Nucleocapsid Organization in Occlusion-Derived Virions

A recombinant virus lacking the sf32 gene (Sf32null), unique to the Spodoptera frugiperda multiple nucleopolyhedrovirus (SfMNPV), was generated by homologous recombination from a bacmid comprising the complete viral genome (Sfbac). Transcriptional analysis revealed that sf32 is an early gene. Occlusion bodies (OBs) of Sf32null contained 62% more genomic DNA than viruses containing the sf32 gene, Sfbac and Sf32null-repair, although Sf32null DNA was three-fold less infective when injected in vivo. Sf32null OBs were 18% larger in diameter and contained 17% more nucleocapsids within ODVs than those of Sfbac. No significant differences were detected in OB pathogenicity (50% lethal concentration), speed-of-kill or budded virus production in vivo. In contrast, the production of OBs/larva was reduced by 39% in insects infected by Sf32null compared to those infected by Sfbac. The SF32 predicted protein sequence showed homology (25% identity, 44% similarity) to two adhesion proteins from Streptococcus pyogenes and a single N-mirystoylation site was predicted. We conclude that SF32 is a non-essential protein that could be involved in nucleocapsid organization during ODV assembly and occlusion, resulting in increased numbers of nucleocapsids within ODVs.


Introduction
The family Baculoviridae comprises a group of viruses characterized by a large double-stranded, circular, supercoiled DNA molecule of 80-180 Kb, enveloped in a rod-shaped nucleocapsid [1]. These viruses are pathogenic to arthropods, particularly Lepidoptera, and have a number of characteristics that favor their development as bioinsecticides, such as restricted host range and high pathogenicity and virulence [2,3]. Baculoviruses are present in the environment, in soil and on foliage, as occlusion bodies (OBs), which are formed by a protein matrix that occludes the occlusion derived virions (ODV). Two phases can be distinguished in their cycle of infection. Primary infection occurs when the larva ingests OBs that degrade in the alkaline midgut and release ODVs that infect epithelial cells. Virus replication and transcription begins in midgut cells and new nucleocapsids are formed in the cell nucleus. Some of these nucleocapsids bud out of the cells acquiring an envelope as they do so, and form budded virions (BV), that disperse within the infected insect to initiate secondary infection. The nucleocapsids that remain in the nucleus of infected cells acquire an external envelope, either singly or in groups, to form the ODVs that are occluded in the polyhedrin matrix to form the occlusion bodies. Upon death the insect disintegrates and liquefies and OBs are released into the environment for transmission to susceptible larvae [1]. As such BVs are specialized for cell-to-cell systemic infection, whereas ODVs are required for insect-to-insect transmission.
The fall armyworm, Spodoptera frugiperda, is an important pest of maize, rice and sorghum in North and South America. This insect can be infected by S. frugiperda multiple nucleopolyhedrovirus (SfMNPV) and natural epizootics of virus disease can spread through high density populations of the pest [4]. Several SfMNPV isolates have been characterized [5][6][7], but development of the virus as a biological insecticide has been limited, mainly due to high production costs and moderate levels of pest control observed following application of viral occlusion bodies (OBs) to infested crops [8,9]. The identification of the genetic factors that determine particular insecticidal properties of the virus may facilitate the selection of particular genotypes with desirable traits for use in bioinsecticidal products, or the development of recombinant viruses, with improved characteristics compared to the wild type [10]. The deletion of certain genes [11] or the insertion of insectspecific toxin genes [12] has been shown to increase the speed of kill of these viruses.
Studies on the genes involved in the insecticidal characteristics of SfMNPV have been facilitated by genome sequencing of three different isolates of this virus [13][14][15]. A total of 62 open reading frames (ORFs) present in these genomes have homologues in all lepidopteran-specific baculoviruses (genus Alphabaculovirus) [16]. Most of these genes encode structural proteins or are involved in DNA replication or transcription [17]. The majority of SfMNPV genes show high sequence homology to genes of Spodoptera exigua multiple nucleopolyhedrovirus (SeMNPV). However, there are twelve open reading frames (ORFs) in the SfMNPV genome that have been identified as unique genes that do not have homologs in other baculoviruses sequenced to date [13,14]. These genes may play a role in the unique characteristics or host specificity of SfMNPV [18,19].
The sf32 gene of SfMNPV is a unique gene located in a hypervariable region of the genome within which mutations and deletions influence viral phenotype [13,20]. In the present study, we examined the role of this gene in the insecticidal properties of this virus. A PCR and a bacmid-based recombination system were used to delete sf32 from the genome and a selection of phenotypic characteristics of the recombinant product was studied. We found that the sf32 gene is not essential for the SfMNPV infection cycle, as infectious viral progeny were obtained following replication of the sf32 deletion bacmid. However, deletion of the gene caused a reduction in OB production per larva, whereas the size of OBs and the number of viral genomes (nucleocapsids) within ODVs increased. We suggest that SF32 may be involved in nucleocapsid organization during ODV assembly and occlusion.

Insects, Cells and Viruses
Spodoptera frugiperda larvae were obtained from a colony that was started using pupae from a laboratory population maintained in Honduras and refreshed periodically with pupae from southern Mexico. The colony was maintained at 2561uC, 75% relative humidity (RH) and 16 h light: 8 h dark photoperiod on a wheatgerm-based semisynthetic diet [21]. Spodoptera frugiperda Sf9 cells were maintained in TC100 medium containing 10% fetal calf serum at 28uC [22]. A wild-type isolate of SfMNPV was collected in Nicaragua and characterized genotypically by Simón et al. [23]. The B genotype (SfMNPV-B) includes the largest genome of the virus. This genotype was selected for the SfMNPV bacmid construction (Sfbac). The complete genotype was cloned into a pBACe3.6 vector modified by replacing the pUC19 element with pBluescriptKS I containing AscI restriction sites, which cuts the SfMNPV-B genome once [24].

Construction of Sf32null and Sf32null-repair Viruses
The Sf32null bacmid was constructed by deleting sf32 from Sfbac by homologous recombination using Red/ET recombination (Gene Bridges GmbH). A kanamycin resistance gene was amplified twice using the Tn5-neo PCR template and primers that added 50 nucleotide (nt) terminal sequences corresponding to either 39 or 59 untranslated regions of sf32. First, a PCR fragment was amplified using Sf32del.1 and Sf32del.2 primers (Table 1) and the Tn5-neo template. Then, in a second PCR, 25 nt terminal sequences were added using Sf32del.3 and Sf32del.4 primers (Table 1) and the PCR product of the first amplification. The bacteria containing Sfbac were made electrocompetent and transformed with the Red/ET plasmid pSC101-BAD-gbaA (Gene Bridges GmbH). The PCR product containing the terminal sequences of the sf32 gene was used to transform the electrocompetent cells containing Sfbac and pSC101-BAD-gbaA. These cells were also made electrocompetent and induced with arabinose (0.1-0.2% w/v) to express the recombination protein (gbaA). Recombinants were selected as resistant colonies on medium containing chloramphenicol and kanamycin. To confirm deletion of sf32, restriction PstI profiles of the bacmid DNA were examined and PCR amplifications with Sf32del.3 and Sf32del.4 primers were sequenced.
For the construction of the repair virus, the sf32 coding region was amplified by PCR using primers amplifying outside the coding region, Sf32rep.1 and Sf32rep.2 (Table 1), and the Sfbac DNA as template. Fourth-instar S. frugiperda larvae were injected with 10 ml from a mixture containing 50 ml of Sf32null bacmid DNA (100 ng/ ml), 50 ml of the PCR product that covered the sf32 region (500 ng/ ml) and 50 ml of Lipofectin reagent (Invitrogen). Inoculated larvae were transferred to diet and reared individually at 25uC. Virus-induced mortality was recorded daily. The OBs were purified from cadavers and virus DNA extracted as described below. A PCR was performed with Sf32rep.1 and Sf32rep.2 primers to determine whether recombination had replaced the kanamycin cassette with the sf32 gene in the Sf32null bacmid. DNA was transfected into DH5a electrocompetent cells. In order to select colonies containing the sf32 gene, bacmid DNAs were purified by alkaline lysis and restriction PstI profiles and PCR amplifications with Sf32del.3 and Sf32del.4 were examined. PCR amplification products generated using Sf32del.3 and Sf32del.4 primers of the selected bacmid, were sequenced to confirm the correct insertion of the gene.

Sfbac and Sf32null DNA Infectivity and Production of the OBs
Sfbac, Sf32null and Sf32null-repair bacmid DNAs were purified by alkaline lysis and caesium chloride gradient centrifugation [22]. To determine DNA infectivity and produce Sfbac, Sf32null and Sf32null-repair OBs, S. frugiperda fourth instars were injected with a DNA suspension including bacmid DNAs and Lipofectin reagent (Invitrogene) in a 2:1 proportion [24][25]. A 100 mL volume of each bacmid DNA containing 100 ng/ mL was mixed with 50 mL of Lipofectin. After 10 minutes, 10 mL of this suspension was injected into individual larvae (666 ng/larva). Inoculated larvae were transferred to diet, reared individually at 25uC and virus mortality was recorded daily until death or pupation. Groups of 24 larvae were injected with DNA from each virus and the experiment was performed three times.
OBs obtained from dead larvae were filtered through cheesecloth, washed twice with 0.1% (w/v) sodium dodecyl sulphate (SDS) and twice with double-distilled water, and resuspended in double-distilled water. The resulting OB suspensions were counted in a Neubauer chamber and were stored at 4uC. To confirm the authenticity of the recombinant OBs, DNA was extracted from OBs as described in the following section and verified by restriction endonuclease and PCR analyses.

DNA and ODV Content Within OBs
Genomic DNA was extracted from samples of 1610 6 OBs of Sfbac, Sf32null and Sf32null-repair bacmids. ODVs were released from OBs by mixing OB suspensions with 100 mL of 0.5 M Na 2 CO 3 and 50 mL of 10% (w/v) SDS in a final volume of 500 mL and incubating at 60uC during 10 minutes. Undissolved OBs were removed by low speed centrifugation (3,8006g, 5 minutes). The supernatant fraction containing virions was treated with 25 mL of proteinase K (20 mg/mL) and incubated at 50uC for one hour. Viral DNA was extracted twice with 500 mL of phenol and once with chloroform and isolated by alcohol precipitation. The resulting pellet was resuspended in 50 mL of 0.16TE buffer by incubation at 60uC during 10 minutes. DNA samples were diluted 1:100 and quantified using qPCR based on SYBR green fluorescence in an ABI PRISM 7900HT Sequence Detection System (Applied Biosystems). The reaction mixture (20 mL) contained 10 mL SYBR Premix Ex Taq (2x), 0.4 mL of ROX Reference Dye (50x), 0.2 mL of each SfMNPV primer (10 pmol/ mL) ( Table 1) and 1 mL of DNA template. qPCR was performed under the following conditions: 95uC for 30 s, followed by 45 amplification cycles of 95uC for 5 s and 60uC for 30 s and finally a dissociation stage of 95uC for 15 s, 60uC for 15 s and 95uC for 15 s. Data acquisition and analysis were handled by Sequence Detector Version 2.2.2. software (Applied Biosystems). Known dilutions of SfMNPV CsCl-purified DNA (10 25 -1 ng/ ml) were used as internal standards for each qPCR reaction. Melting-curve analysis was performed to confirm specific replicon formation in qPCR.
Mean ODV content within OBs was determined by end point dilution in Sf9 cells [25]. ODVs were released from 5610 8 OBs in a volume of 500 mL by incubation with Na 2 CO 3 0.1 M at 28uC for 30 minutes. This suspension was filtered through a 0.45 mm filter and serially diluted 1:5 in TC100 medium. Sf9 cells were infected with 10 mL of each ODV dilution in 96-well plates. Twenty-four wells containing 1610 4 cells/well were inoculated with each dilution in triplicate. Dishes were incubated at 28uC for 7 days and then examined for signs of virus infection. Results were analyzed by the Spearman-Kä rber method [26] to estimate 50% tissue culture infectious dose (TCID 50 ). TCID 50 values were subsequently converted to infectious units per 5 10 8 OBs and compared by t-test in SPSS 15.0 (SPSS Inc, Chicago, IL).  Biological Activity of OBs The 50% lethal concentration (LC 50 ), mean time to death (MTD) and OB production were determined in second instars by per os inoculation following the droplet feeding technique [27]. To estimate LC 50 values, larvae were starved overnight and allowed to drink viral suspensions in 10% (w/v) sucrose solution containing 0.001% (w/v) blue food dye and one of the following concentrations of OBs of each virus: 1.9610 3 , 9.6610 3 , 4.8610 4 , 2.4610 5 and 1.2610 6 OB/ml. This range of concentrations was previously estimated to kill between 5 and 95% of the insects. Larvae that drank OB suspension in a 10 min period were reared individually on diet at 25uC and virus mortality was recorded daily until death or pupation. Bioassays were performed on four occasions using groups of 24 larvae per virus concentration and 24 control larvae that consumed sucrose and food dye solution without OBs. Virus induced mortality data were subjected to probit analysis with the POLO statistical program [28].
In order to determine speed-of-kill, groups of 24 recently-molted second instars were starved overnight and allowed to drink an OB concentration estimated to result in 90% mortality, namely, 2.19610 5 OB/mL for Sfbac and 2.24610 5 OB/mL for Sf32null. Larvae that drank OB suspensions within 10 min were reared individually on diet at 25uC and mortality was recorded at 8 hour intervals until larvae had died or pupated. The experiment was performed on four occasions. Time-mortality results were subjected to Weibull analysis using the generalized linear interactive modelling (GLIM) program [29] and to Kaplan-Meier survival analysis using SPSS 15.0.
OB production was determined in larvae that died in the MTD assays. Each corpse was homogenized in 100 mL distilled water and OBs were counted in a Neubauer hemocytometer. OB counts from each larva were performed in duplicate using three of the four replicates from the MTD experiment. Results were subjected to t-test in SPSS 15.0.

Budded Virus Production In Vivo
Budded virus (BV) production was studied in newly-molted fourth instars that had drunk a suspension of 10 8 OB/ml, estimated to result in ,95% mortality for both viruses. The number of viral genome copies present in larval hemolymph was estimated by qPCR. Hemolymph samples taken from groups of 20 larvae at 0, 12, 24, 48, 72, 96, 120 hpi were centrifuged at 2,0006g for 10 minutes at 4uC to pellet cells. DNA extraction was performed on the supernatant using the MasterPure Complete DNA Purification kit (Epicentre Biotechnologies) and DNA concentrations were measured by qPCR as described for DNA content quantification.

Electron Microscopy
Scanning electron microscopy (SEM) was used to determine OB diameter. OBs in suspension were fixed overnight by mixing with an equal volume of fixative (4% formaldehyde and 1% glutaraldehyde in 0.1 M phosphate buffer, pH 7.4) and then washed twice with 0.1 M phosphate buffer. Samples were then partially dehydrated with ethanol 70%, dried, placed on aluminum mounts using carbon tags, sputter-coated with gold-palladium and photographed at magnifications of 6,0006 and 25,0006 using a scanning electron microscope (Philips SEM 550). Images were analyzed with the ImageJ software (National Institutes of Health) and Feret's diameter (the longest distance between two parallel tangents) was taken as a measure of OB diameter. A total of 500 OBs were analyzed for the Sfbac virus and 345 were analyzed for the Sf32null virus. OB measurements were normalized by squareroot transformation and compared by t-test using SPSS 15.0.
The number of ODVs occluded within OBs and the number of nucleocapsids per ODV were determined by examination of OB sections by transmission electron microscopy (TEM). OBs in suspension were fixed for 2 h at 4uC with 1.5% glutaraldehyde. Samples were then concentrated in 0.4% agar, washed with phosphate buffer (0.2 M, pH 7.3), post fixed with 2% osmium tetroxide for 2 h, followed by 1 h treatment with 2% uranyl acetate. Samples were then embedded in epoxy resin, sectioned, stained with lead acetate and observed under TEM at 100 KV (JEOL JEM 1010). Different fields of each sample were photographed at a magnification of 640,000. Images were analysed using ImageJ software for each sample, the number of ODVs was counted in 30 OBs. Similarly, the number of nucleocapsids was counted in 300 ODVs. The mean numbers of ODVs and nucleocapsids for each sample were compared by ttest. Feret's diameter was also measured in approximately 100 OBs of each virus (89 for Sfbac and 100 for Sf32null) and compared by t-test using SPSS 15.0.

Gene and Protein Sequence Analysis
To determine the nature of the putative SF32 protein, nucleotide and amino acid sequences homologs were searched in the updated Genbank and EMBL databases using BLAST [30]. Protein properties were studied using the Peptide Property Calculator (Center for Biotechnology, Northwestern University). PSIPRED and JPRED3 tools were used to predict protein secondary structure [31] and signal sequences were screened using SIGNALP v3.0 [32]. Cellular location was predicted by TargetP 1.1 server [33]. The presence of transmembrane domains was detected using TMHMM v2.0 and MEMSAT3 [34]. Finally, post-translational modifications were predicted using PROSITE [35].

Generation of Sf32null and Sf32null-repair Viruses
The selected Sf32null bacmid was expected to contain a deletion of the sf32 gene, located between the nucleotides 30,955 and 31,338 in the SfMNPV-B genome [14]. Replacement of the sf32 gene by the kanamycin cassette in Sf32null was confirmed by restriction endonuclease analysis and PCR with specific primers targeted at the predicted recombinant junction regions. The genomic arrangement of the recombinant virus was verified by sequencing. The same method was performed to confirm the correct insertion of sf32 gene in the Sf32null-repair bacmid.

sf32 is an Early Transcribed Gene
Temporal regulation of the sf32 transcript was examined by RT-PCR using total RNA isolated from S. frugiperda larvae orally infected with Sfbac OBs. Control amplifications performed to ensure the absence of contaminant DNA in the samples were consistently negative. Equal volumes of the treated RNA were used for the sf32, polh and egt transcript amplifications. Single RT-PCR products of the expected sizes were obtained following amplification of sf32, polh and egt. The sf32 amplification product was detected at a very low level at 2 hpi, increased at 4 hpi and remained at a steady-state level up to 144 hpi. In contrast, an amplification product from the early transcribed egt gene was detected from 4 hpi to 144 hpi. The late transcribed gene polh amplification product was detected at a very low level at 24 hpi, increased at 48 hpi and remained at a steady-state level up to 144 hpi (Figure 1). A diffuse band was observed below the expected amplification product due to excess primer.

Reduced Infectivity of Sf32null DNA
Intrahemocelic injection of Sf32null bacmid DNA resulted in significantly lower mortality of larvae than observed following injection of Sfbac or Sf32null-repair DNA (F 2,8 = 25.878, p = 0.001). Mean (6SE) mortality was 15.761.3% for insects injected with Sf32null bacmid DNA, whereas 49.164.5% and 53.865.3% was recorded for those injected with Sfbac and Sf32null-repair DNA respectively (Figure 2a), indicating that Sf32null DNA was approximately three fold less infectious than viral DNA that included the gene. DNA extracted from the resulting OBs was subjected to restriction endonuclease analysis and PCR which confirmed that the three viruses had similar DNA restriction profiles and amplification products to those obtained with the parental DNA.

Deletion of sf32 had no Significant Effects on OB Pathogenicity or Speed of Kill but Increased OB Production
The estimated LC 50 values of Sfbac and Sf32null bacmid OBs were almost identical at 1.76610 4 and 1.77610 4 OBs/mL, respectively. The 95% confidence levels of the relative potencies, representing the ratio of effective concentrations [36], overlapped broadly in both viruses indicating no significant differences in OB infectivity between these viruses in S. frugiperda larvae ( Table 2).
Following inoculation with an estimated LC 90 concentration of OBs the mean (6SE) observed mortality of second instars was 86.764.8% for Sfbac and 86.563.6% for Sf32null. The mean time to death (MTD) value was 161 hours post infection (hpi), which was identical for both viruses ( Table 2). Mortality results were also subjected to survival analysis in SPSS 15.0. Kaplan-Meier curves (Figure 2d) and Log Rank test (x 2 = 0.210, df = 1, p = 0.647), which confirmed that the speed-of-kill was similar in both viruses.
OB production differed significantly between Sfbac and Sf32null viruses (t = 6.6; df = 4; p = 0.003). Sfbac infections produced 8.68610 7 60.57610 7 OB/larva (mean 6 SE) whereas Sf32null produced 5.29610 7 60.42610 7 OB/larva (Figure 2e), which represents 39% fewer Sf32null OBs/larva compared to the production observed in insects infected by Sfbac.  Sf32 had no pronounced effects on budded virus production (Figure 2f). The amounts of viral genomic DNA in hemolymph for both viruses at different times post-infection were very similar as determined by qPCR (efficiency = 108%, r 2 = 0.998), i.e., sf32 deletion appeared to have no clear influence on BV production at any stage of the infection. The presence of non-specific amplification was considered improbable as only one peak was observed in the melting curve.

Sf32null Virus OBs are Larger than those of Sfbac Virus and Occlude ODVs Containing a Higher Number of Nucleocapsids
Sf32null OBs had diameters between 1.40 and 3.37 mm with an average (6 SE) value of 2.2560.02 mm (N = 345), which was significantly larger than the diameter of Sfbac OBs, which ranged from 1.00 to 3.16 mm with an average value of 1.9160.02 mm (N = 500) (t = 12.3, df = 843, p,0.001). Although Sf32null OBs were approximately 18% larger than those of Sfbac, no additional gross morphological differences were observed in OB structure between these viruses when examined by SEM ( Figure 3A).
TEM observation revealed that the number of ODVs present within the OBs was similar in Sfbac and Sf32null OBs (t = 1.367, df = 26, p = 0.183) ( Figure 3B, C). However, the number of nucleocapsids per ODV was approximately 17% higher in Sf32null ODVs than in Sfbac ODVs (t = 2.513, df = 311, p = 0.013) ( Figure 3D). Significant differences in the size of the OBs were confirmed by measuring the Feret's diameter in the TEM images (t = 3.47, df = 184, p = 0.001). The mean diameter of the Sfbac OBs was 1.6460.04 mm, whereas the mean diameter of the Sf32null OBs was 1.8460.04 mm.

Sequence Analysis
Sequence analysis revealed that sf32 is a reverse gene located between nucleotides 30,957 and 31,475 in the SfMNPV Nicaraguan isolate genome [14]. This gene is positioned between sf31, which encodes a putative protein-kinase interacting protein similar to ac24, and sf33, a putative actin rearrangement-inducing factor similar to ac20 and ac21. A TATA box and a baculovirus consensus early promoter motif CAGT were present at 277 and 292 nt upstream from the initiation codon (ATG), respectively, suggesting that sf32 might be an early gene, which was confirmed by RT-PCR temporal expression analysis (Figure 1).
SF32 is a unique small protein of 172 amino acids (Aa), present in all sequenced SfMNPV genotypes; no homologs were detected in other baculoviruses. The SF32 protein showed 26% identity and 44% similarity to the trypsin-resistant T6 surface protein of Streptococcus pyogenes serotype M6 and also showed sequence similarity with the fimbrial structural subunit of the same bacterial species (26% identity, 44% similarity). No signal peptide or transmembrane domains were detected in the putative protein.
The estimated volume of this protein was 24,688 A 3 using the Peptide Property Calculator (Chazan). Secondary structure prediction revealed nine strands and five helices, but larger structures could not be predicted from these data. The PROSITE tool predicted some post-translational modifications and functional motifs of SF32, including a protein kinase C phosphorylation site (Aa7-9), two N-glycosylation sites (Aa60-63, 156-159), a tyrosinekinase phosphorylation site (Aa89), a casein-kinase II phosphorylation site (Aa158-161) and an N-myristoylation site (Aa166-171) ( Figure 4).

Discussion
Baculoviruses may acquire genes from their hosts or from other coinfecting agents resulting in viruses with novel phenotypes [18]. Such modified phenotypic characteristics may confer a selective advantage related to the host range of the virus, its insecticidal activity or its transmissibility [17,19,37]. SfMNPV has twelve ORFs which have no homologs in other sequenced baculoviruses [13]. The number of unique genes varies between baculoviruses: 24 unique ORFs were identified in the genome of Chrysodeixis chalcites NPV (ChchNPV) [19], 20 unique ORFs were identified in the Helicoverpa armigera NPV (HaNPV) genome [38], whereas only three genes are unique to the AcMNPV genome [1,39].
In order to increase our understanding of SfMNPV as a fall armyworm pathogen, the role of the sf32 unique gene in the replication and transmission of this virus was studied by producing bacmid-based mutants that lacked the gene. Transcription of sf32 starts very early during infection and continues for at least 144 hours. The temporal expression observations are in agreement with the presence of the early promoter detected by sequence analysis. The products of baculovirus early genes are often involved in DNA replication, regulation of late gene expression and host-modification processes [40]. Early gene transcription is mediated by the host RNA polymerase II and is strongly influenced by the immediate-early IE-1 and IE-2 factors [40]. Deletion of sf32 resulted in a three-fold decrease in DNA infectivity of the bacmid virus, for reasons that remain unclear. Nonetheless, this early gene is clearly not essential for virus replication, as infectious OB progeny were obtained from the sf32 deletion bacmid.
A 62% increase in the average amount of DNA per OB was observed in the Sf32null recombinant, but a similar number of ODVs were present in both Sfbac and Sf32null OBs. Other early genes, such as ac23, modify the number of nucleocapsids per OB without affecting the number of occluded ODVs [41]. In this case it can be assumed that gene deletion decreased total DNA content, in contrast to that observed with the sf32 deletion mutant. The role Probit analysis was performed using the PoloPlus program. The hypothesis of equality was not rejected (x 2 = 0.01, df = 2, p = 0.997) and a test for nonparallelism was not significant (x 2 = 0.01, df = 1, p = 0.947) such that regressions were fitted with a common slope of 1.16360.092 (mean 6 S.E.). Relative potency was calculated as the ratio of effective concentrations relative to Sfbac OBs. Mean time to death values (MTDs) were estimated by Weibull analysis [29]. MTDs labeled with same letter did not differ significantly (p.0.05). doi:10.1371/journal.pone.0077683.t002 of sf32 in the SfMNPV virus therefore differs from that of ac23 in AcMNPV.
Despite the increased DNA content of Sf32null OBs, sf32 deletion had a negligible effect on OB pathogenicity or speed-ofkill. This was an unexpected result given that when a larva ingests an Sf32null OB it will be exposed to a greater number of viral genomes than occurs following ingestion of an Sfbac OB. However, deletion of the previously mentioned ac23, which also modifies the DNA content within OBs, had no significant influence on OB pathogenicity but increased the mean time to death of infected insects [42], so that changes in the DNA content of OBs are not neccessarilly linked to changes in their pathogenicity.
Interestingly, Sf32null OBs were 18% larger in diameter than those of the complete virus. This difference may not seem particularly important, but an 18% increase in diameter represents a 60% increase in volume of the OB (assuming that an OB approximates to the shape of a sphere), which is very similar to the observed increase in the DNA content of Sf32null OBs.
The total production of OBs/larva was ,39% lower in Sf32null infected insects compared with those infected with the complete virus, whereas BV production was similar in both viruses. Increased OB productivity is often linked to an extended period of infection and an increase in the mean time to death [24,43,44], but this was not observed in insects infected by the Sf32null virus. Altered OB production could also be the consequence of alterations in the packaging and occlusion process. Clear similarities in the dynamics of BV production between viruses suggested that DNA synthesis was not affected by sf32 deletion, yet Sf32null OBs contained approximately 62% more DNA than Sfbac OBs. This led us to suspect that more ODVs would be occluded in Sf32null OBs compared to Sfbac OBs, or a greater number of nucleocapsids would be enveloped in the Sf32null ODVs. The number of virions occluded within OBs was determined in cell culture and by direct TEM observation of OB sections, and was not affected significantly by deletion of sf32. However, deletion of this gene resulted in a significant increase in the number of viral nucleocapsids enveloped within each ODV, which reflects the higher DNA content of Sf32null OBs. Therefore, it appears that sf32 likely affects nucleocapsid organization during ODV assembly and occlusion. The higher nucleocapsid content of Sf32null OBs compared to that of Sfbac may be responsible for the reduced OB production, as more genomic DNA is required for production of Sf32null OBs than in Sfbac OBs. Whether or not the increase in the number of nucleocapsids per ODV has a direct influence on the occlusion process, leading to the physically larger OBs that we observed, is unclear at present.
Analysis of the putative SF32 protein revealed that this protein does not contain any signal peptide or transmembrane domains, suggesting that is likely to be an intracellular polypeptide. Nevertheless, this protein is homologous to the trypsin-resistant T6 surface protein of Streptococcus pyogenes serotype M6 and the fimbrial structural subunit of the same species, both of which have adhesion functions [45]. The idea that SF32 could be involved in adhesion processes is an appealing concept that appears consistent with its hypothesized role in nucleocapsid organization during ODV assembly and occlusion. Some strains of Streptococcus are pathogenic to insects [46,47] and horizontal transfer between viruses and bacteria has been proposed for other baculovirus genes, notably chitinase [48,49]. The idea that SF32 may be involved in adhesion functions is reinforced by the presence of an N-myristoylation site in the amino acid sequence as this modification is usually related to weak and reversible proteinmembrane and protein-protein interactions [50,51]. In other viruses, myristoylated proteins are involved in assembly, structure, budding, intracellular host interactions and viral entry [52].
We conclude that the SfMNPV unique SF32 protein is a nonessential protein, as viral replication was not compromised by gene deletion. This gene might be directly or indirectly involved in mediating nucleocapsid organization during ODV assembly and occlusion. Deletion of sf32 resulted in a reduction in OB production per insect and substantial increases in the size of OBs and an increase in the average number of nucleocapsids present within ODVs. Gene deletion did not affect OB pathogenicity, speed of kill, ODV content within OBs or the dynamics of BV production. Homology with bacterial adhesion proteins and the presence of an N-myristoylation site suggests that SF32 may affect nucleocapsid packaging in ODVs via interactions with other proteins.