Anticarsia gemmatalis multicapsid nucleopolyhedrovirus v-trex gene encodes a functional 3' to 5' exonuclease

In Brazil, Anticarsia gemmatalis multicapsid nucleopolyhedrovirus (AgMNPV) has been applied for several decades to control the cassava pest, A. gemmatalis (Moscardi, 1999). AgMNPV is considered to be a model for baculovirus-based biocontrol, with annual coverage now exceeding 2 million hectares. Recently, we published a sequence analysis of the 12·5 kbp BamHI D restriction endonuclease fragment from the AgMNPV-2D genome (GenBank accession no. AY542374; Slack et al., 2004). We noted several variations in this region of AgMNPV compared with other group I nucleopolyhedroviruses (NPVs), including the absence of the v-cath and ChiA genes in the gp64 locus. We also noted that AgMNPV shared two gene homologues with Choristoneura fumiferana multicapsid nucleopolyhedrovirus (CfMNPV) that were not found in most other baculoviruses: v-trex (CfMNPV ORF 113; GenBank accession no. NP_848425.1) and p22.2^Cf (CfMNPV ORF 126; GenBank accession no. NP_848438.1). The gene v-trex (viral three-prime repair exonuclease) was named due to its homology with a group of mammalian exonucleases (TREXs) that were described recently by Mazur & Perrino (1999, 2001). Mammalian TREXs are a group of type III exonucleases that have the distinction of functioning independently from the DNA replication complex. A biological role for mammalian TREX proteins has not been defined, but involvement in DNA mismatch repair, DNA UV damage repair and DNA recombination has been proposed (Mazur & Perrino, 1999, 2001). Type III exonucleases have been shown to participate in the protection of Escherichia coli from UV light inactivation (Serafini & Schellhorn, 1999). Baculoviruses are susceptible to inactivation by UV light (Shapiro et al., 2002) and it is important to study the v-trex gene from this standpoint.

In the following investigation, we sought to determine whether the AgMNPV v-trex gene was expressed and whether the V-TREX protein product functions as a 3' to 5' exonuclease. RT-PCR was used to detect v-trex transcripts in the context of AgMNPV infection. The AgMNPV v-trex ORF was also cloned into the baculovirus Autographa californica multicapsid nucleopolyhedrovirus (AcMNPV) under the control of a polyhedrin (polh) promoter. A fluorescence-based assay was then used to examine the exonuclease activity of the overproduced V-TREX protein.

Cell lines and viruses.
Spodoptera frugiperda (Sf-9) cells were used to propagate wild-type (wt) AcMNPV (strain E2), BacPAK6 (Clontech) and recombinant BacPAK6 viruses. For this study, BacPAK6 is referred to as BacPAK-β-gal, with reference to the β-galactosidase gene that replaces the polh gene in this virus (Kitts & Possee, 1993). A. gemmatalis-derived UFL-Ag-286 cells (Ag-286) (Sieburth & Maruniak, 1988) were used to propagate wt AgMNPV-2D virus. Cell lines were cultured at 27 °C in TNM-FH medium (Hink, 1970) containing 10 % (v/v) fetal bovine serum.

Constructs and recombinant baculoviruses.
The 693 bp ORF of v-trex was amplified from AgMNPV genomic DNA by PCR and cloned into the baculovirus transfer vector plasmid pBacPAK8 (Clontech). The PCR primers TREX-LP-XbaI-NcoI (5'-AAACATCTAGAGTTCACCATGGCTGTCGTCAAGAC-3') and TREX-RP-NotI-NcoI del (5'-TAATAAGCGGCCGCTTATTCCCCCATAGGGATGAC-3') were used to engineer 5' XbaI and 3' NotI restriction sites onto the ends of the v-trex ORF such that it could be cloned downstream of the polh promoter of the pBacPAK8 plasmid. The resulting construct, pBacPAK8-v-trex, was co-transfected with Bsu36I-digested BacPAK6 viral DNA into Sf-9 cells, as described by Kitts & Possee (1993). Transfections were facilitated by using the lipid transfection agent Cellfectin (Invitrogen). BacPAK-v-trex virus clones were isolated by plaque purification.

Preparing cell lysates for TREX assays.
T-75 tissue-culture flasks of Sf-9 cells (1x10⁷ cells per flask) were infected at an m.o.i. of 1. At 72 h post-infection (p.i.), cells were collected in 50 ml Corning tubes and counted. Cells were pelleted at 1000 g for 1 min and suspended in 25 ml chilled PBS/EDTA (125 mM NaCl, 10 mM NaH₂PO₄, 5 mM EDTA, 2·5 mM KCl, pH 6·2). Cells were pelleted again at 1000 g for 1 min and suspended in 5 ml chilled PBS/EDTA. Finally, cells were pelleted at 1000 g and suspended at a concentration of 5x10⁴ cells µl¹ in TREX dilution buffer [75 mM NaCl, 50 mM Tris/HCl (pH 8·0), 5 mM NaH₂PO₄, 2·5 mM EDTA, 2 mM DTT, 5 % (v/v) glycerol, 2 % ethanol, 0·25 mM Ac-Leu-Leu-norleucinal (cysteine protease inhibitor)]. Suspended cells were frozen overnight at 20 °C and then thawed on ice and disrupted by sonication for 30 s using a Microson XL ultrasonic cell disrupter (Heat Systems). Cell lysates were centrifuged at 4500 g for 5 min at 10 °C. Supernatants were collected into 1·5 ml Eppendorf tubes and assayed for total protein by using a Coomassie Plus Protein Assay kit (Pierce). Supernatants were diluted in TREX dilution buffer to a protein concentration of 1 mg ml¹. The resulting soluble lysates were used in exonuclease assays.

Budded virus (BV) purification and processing.
Cell-culture supernatant volumes of 33 ml from 4x10⁷ infected Ag-286 cells were collected at 52 h p.i. Cellular debris was removed by centrifuging twice at 1000 g for 5 min. BV-containing cell-culture supernatants were centrifuged for 1·5 h at 100 000 g at 10 °C through a 5 ml cushion of 20 % (w/w) sucrose in PBS (pH 7·4) containing 5 mM iodoacetamide (cysteine protease inhibitor), 5 mM EDTA. BV pellets were suspended in 300 µl TREX dilution buffer and frozen at 20 °C. BV samples were thawed on ice and sonicated for 15 s. BV lysates were assayed for total protein by using a Coomassie Plus Protein Assay kit. BV lysates were diluted in TREX dilution buffer to a protein concentration of 0·9 mg ml¹.

Exonuclease assays.
Exonuclease assays were done in 96-well U-bottomed plates. Plates were placed on ice while reagents were combined. Lysate volumes of 10 µl were combined with 30 µl TREX assay buffer [20 mM Tris/HCl (pH 7·5), 5 mM MgCl₂, 2 mM DTT, 100 µg BSA ml¹]. During assays, plates were covered with aluminium foil and incubated for 1 h at 37 °C. Assays were stopped with 20 µl TREX stop buffer [50 % (v/v) formamide, 3x TBE, 15 % (w/v) sucrose] and assays were stored at 4 °C until analysis. All TBE solutions were made from a 10x TBE stock (890 mM H₃BO₃, 450 mM Tris-base, 20 mM EDTA, pH 8·0). Exonuclease assay sample volumes of 20 µl were fractionated by electrophoresis (3 h, 25 mA, 200300 V) in 13 % (w/v) acrylamide : N,N'-methylene-bis-acrylamide (20 : 1), 1x TBE, 5 M urea gels. Electrophoresis was done by using a Hoeffer SE600 vertical gel unit and 0·7 mmx18 cmx16 cm gels. Gels were scanned in their plates by using a Typhoon fluorescent scanner (Amersham Biosciences) that had been set to 3 mm above the focal plane. For exonuclease assays, two 35 nt, fluorescently labelled DNA oligomers were synthesized at a 25 µmol scale (Integrated DNA Technologies). One oligomer (5HEX-oligo) was covalently linked at its 5' end to hexamethylfluorescein (5'-HEX-GCTCACCACTCCTGCAGCTCTAGATTCCCACCATC-3'). The other oligomer (3FAM-oligo) was covalently linked at its 3' end to 6-carboxymethylfluorescein (5'-AGCAACATAGATCTAGAGCTGCAGGAGTGGTGAGC-FAM-3'). In some experiments, the 5HEX-oligo and the 3FAM-oligo partially annealed to each other such that 25 nt annealed, leaving 10 mismatched nucleotides on the non-labelled ends that did not anneal (see Fig. 7a). Assays containing the 5HEX-oligo were scanned at excitation 532 nm/emission 555 nm BP 20 nm. Assays containing the 3FAM-oligo were scanned at excitation 532 nm/emission 526 nm SP. Assays containing both oligomers were scanned at dual wavelengths and images were separated by using Fluorsep 2.2 software (Amersham Biosciences). All Typhoon-scanned images were analysed on ImageQuant 5.0 (Amersham Biosciences).

(108K): Fig. 7. V-TREX exonuclease activity on 3'-labelled substrates and dsDNA. Soluble protein lysates from Sf-9 cells that had been infected with BacPAK-v-trex or BacPAK-β-gal were diluted serially in TREX dilution buffer. Amount of lysate protein is indicated at the top of the panels. (a) Three types of substrates were used, including the 5HEX-oligo (5H), 3FAM-oligo (3F) and the annealed 5HEX-oligo and 3FAM-oligo (5H/3F). (b, c) Lysates from BacPAK-v-trex (b) and BacPAK-β-gal (c) were assayed for exonuclease activity. Exonuclease reactions were fractionated by electrophoresis in a denaturing 13 % acrylamide/urea gel.

RNA preparation and RT-PCR.
T-75 tissue flasks were seeded with 5x10⁶ Ag-286 cells and infected at an m.o.i. of 10 with 4·5 ml viral inoculate. After 2 h rocking, the medium was removed and replaced with 10 ml fresh medium. At various times, cells were harvested by suspension in medium. Cells were pelleted at 1000 g for 1 min and suspended in 11 ml chilled PBS/EDTA. Cells were pelleted at 1000 g for 1 min and processed for total RNA by using an RNeasy Mini kit (Qiagen). RNA samples (100 µl) were stored at 20 °C. Aliquots of 5 µl RNA (500 ng) were used in 100 µl PCRs or RT-PCRs. To eliminate DNA contamination, 5 µl RNA samples were combined with 5 µl 2x restriction endonuclease buffer no. 3 (New England Biolabs) containing the enzymes DNase I (30 mU), NcoI (150 mU) and MluI (150 mU). MluI and NcoI are restriction enzymes that cut the v-trex gene (Fig. 1). After 1 h at 37 °C, 1 µl 25 mM EDTA was added and enzymes were heat-inactivated for 30 min at 65 °C. RT-PCRs and PCRs were then done by using an Access RT-PCR kit (Promega). The DNase/restriction enzyme-treated RNA samples (11 µl) were combined with 100 µl 1x reaction buffer (RB; Promega) containing 1·3 mM MgSO₄ and 200 µM dNTPs. The resultant RNA solution was split into two 0·5 ml Eppendorf tubes. One tube (RT-PCR) received 1 µl (5 U) avian myeloblastosis virus reverse transcriptase and the other control tube (PCR) received no enzyme. The v-trex gene antisense primer TREX-RP (5'-ATATGTAAGCTTTTCCCCCATAGGGATGACGTTTG-3'; Fig. 1) was added (50 pmol per tube). Both RT-PCR and PCR tubes were incubated at 48 °C for 1 h and then chilled on ice. A 5 µl aliquot of 1x RB containing 50 pmol of the sense primer TREX-LP-XbaI-NcoI and 5 U Tfl DNA polymerase was added. Tubes were sealed with 50 µl mineral oil and placed in a 95 °C pre-heated block of a Perkin-Elmer-Cetus thermocycler for 2 min. The PCR was run for 60 cycles of denaturation for 1 min at 94 °C, annealing for 1 min at 47 °C and extension for 1 min at 68 °C.

(48K): Fig. 1. Sequence of the AgMNPV v-trex gene region. The v-trex gene lies between the AgMNPV homologues for Lef-7 and AcMNPV ORF 124 (ORF124Ac). DNA sequence motifs are indicated in upper case: CACGTG and GATATAA sequences in the v-trex promoter region and an ATAATAAA polyadenylation signal at the end of the v-trex gene. The translational start point (tsp) of the v-trex gene is indicated by an arrowhead. Numbering on the right-hand side indicates nucleotide positions relative to the v-trex tsp. Restriction sites for MluI and NcoI that were used to eliminate DNA background from RT-PCRs are labelled and boxed. The location of the primer TREX-RP that was used to generate the initial reverse-transcribed DNA strand in the RT-PCR is also indicated. The protein amino acid sequence is written below the DNA sequence. For V-TREX, the amino acid residues of the exonuclease motifs ExoI, ExoII and ExoIII are shown in bold type. Conserved active-site residues are underlined.

This study investigated a recently described baculovirus gene, v-trex (Slack et al., 2004). Our aim was to provide evidence that the v-trex gene is expressed by AgMNPV and that the V-TREX protein has exonuclease activity that would confirm the nomenclature.

Evidence for v-trex expression
From the AgMNPV DNA sequence, it was predicted that the v-trex gene would be expressed at early times during infection. This prediction was based on the presence of an ATCAGT motif 7 bp upstream of the v-trex translation start point (Fig. 1). In addition, the v-trex gene promoter region has a TATAA box and the eukaryotic transcription factor-binding motifs CACGTG and GATA. These elements have been shown to be important for the transcription of early baculovirus genes (Kogan & Blissard, 1994; Shippam-Brett et al., 2001). The GATA element may not be ideal, as it overlaps the TATAA box. With RT-PCR, we were able to detect the presence of v-trex RNA transcripts from 3 to 72 h p.i (Fig. 2). This result confirmed the prediction that v-trex is an early gene. There were no late promoter (A/T)TAAG motifs in the vicinity of v-trex. The presence of v-trex RNA transcripts late in infection may be the result of v-trex transcript stability. The end of the v-trex gene contains a strong polyadenylation signal motif, ATAATAAA, which would promote the production of more stable, polyadenylated transcripts.

(63K): Fig. 2. RT-PCR detection of v-trex gene transcripts. RNA was purified from control uninfected Ag-286 cells (M) and from Ag-286 cells that had been infected with AgMNPV for 3, 6, 8, 24 or 72 h. A portion of each RNA purification product was used for PCR and RT-PCR to detect v-trex gene transcripts. PCR and RT-PCR products were fractionated on a 1·25 % (w/v) agarose/1x TAE gel. A 1 kb Plus Ladder (Invitrogen) DNA size standard (STD) was also run. The gel was stained with ethidium bromide and visualized by using a Typhoon fluorescence imager (532 nm excitation/emission 610 nm BP 30).

Overexpression of v-trex
To characterize the V-TREX protein, we overexpressed the v-trex ORF in an AcMNPV-based expression vector system (AcMNPV does not have a v-trex homologue). The recombinant AcMNPV virus was called BacPAK-v-trex and placed the v-trex ORF under the control of a polh promoter. The recombinant virus produced an abundant protein with an estimated size of 23·7 kDa that represented 8 % of total soluble protein in lysates from BacPAK-v-trex-infected Sf-9 cells. This protein was not present in the parent BacPAK-β-gal virus, wt AcMNPV or in Sf-9 cells (Fig. 3). We concluded from this that the V-TREX protein was overproduced successfully by the BacPAK-v-trex virus. With an apparent molecular mass of 23·7 kDa, V-TREX migrated faster in SDS-PAGE than its predicted size of 25·4 kDa. V-TREX has no strong predictions for post-translational modifications that might affect migration. However, V-TREX has an estimated isoelectric pH of 8·0, which would give this protein a net positive charge in SDS-PAGE buffer at pH 6·8.

(106K): Fig. 3. SDS-PAGE analysis of proteins produced by BacPAK-v-trex virus. Soluble protein lysates from control uninfected Sf-9 cells (lane M) and Sf-9 cells that had been infected with BacPAK-v-trex (lane 1), BacPAK-β-gal (lane 2) or wt AcMNPV (lane 3) were denatured in SDS-PAGE disruption buffer (250 mM Tris/HCl, 2 % SDS, 5 % mercaptoethanol, 20 % glycerol, 0·4 % bromophenol blue) and fractionated by 12 % acrylamide : N,N'-methylene-bis-acrylamide (37 : 1) SDS-PAGE. A Precision Plus protein standard (Bio-Rad) was run alongside samples; protein sizes are indicated on the left. The gel was stained with Coomassie brilliant blue.

Detection of V-TREX-specific 3' to 5' exonuclease activity
We next determined whether the high levels of V-TREX production by the BacPAK-v-trex virus corresponded to increased levels of 3' to 5' exonuclease activity. We developed a fluorescence-based exonuclease assay to replace the conventional radioisotope-based assay. Two fluorescently labelled DNA oligomers were used as substrates in exonuclease assays: 5HEX-oligo and 3FAM-oligo. The 5'-labelled 5HEX-oligo was used for most experiments.

Soluble protein lysates from insect cells were diluted serially and incubated at 37 °C with the 5'-fluorescently labelled 5HEX-oligo. Exonuclease assays included lysates from control uninfected Sf-9 cells and Sf-9 cells that had been infected with wt AcMNPV, BacPAK-β-gal or BacPAK-v-trex. Exonuclease assays were analysed in denaturing acrylamide/urea gels. Lysates from Sf-9 cells produced a gradient of faster-migrating 5HEX-oligonucleotides when protein amounts were 11003300 ng (Fig. 4, top panel). Lysates from wt AcMNPV and BacPAK-β-gal produced faster-migrating 5HEX-oligonucleotides when protein amounts were 3703300 ng (Fig. 4, middle panels). Our interpretation of these results was that the increased mobility of 5HEX-oligo was the result of exonuclease activity decreasing the size of the oligomers. Lysates from BacPAK-v-trex caused the 5HEX-oligo to shift to a faster-migrating species when protein amounts were 4·53300 ng (Fig. 4, bottom panel). At protein amounts between 0·17 and 1·5 ng, there was a gradient of 5HEX-oligo sizes.

(63K): Fig. 4. Exonuclease activity assay of the BacPAK-v-trex virus. Soluble protein lysates from control uninfected Sf-9 cells and Sf-9 cells that had been infected with wt AcMNPV, BacPAK-β-gal or BacPAK-v-trex were assayed for exonuclease activity (see Methods). A one-third dilution series was made of each lysate group in TREX dilution buffer. Exonuclease reactions were fractionated by electrophoresis in a denaturing 13 % acrylamide/urea gel. Total amount (ng) of protein lysate used in each exonuclease reaction is indicated above each lane.

We estimated that the BacPAK-v-trex virus lysates contained exonuclease activity that was 6000-fold greater than the activity in lysates from uninfected Sf-9 cells. In addition, BacPAK-v-trex lysates had at least 2000-fold greater activity than lysates from BacPAK-β-gal or wt AcMNPV-infected Sf-9 cells. We concluded that the greatly increased exonuclease activity associated with the BacPAK-v-trex virus was due to the presence of V-TREX. The large increase in exonuclease activity suggested that V-TREX does not require the proportional presence of other viral proteins for activity. This supports the prediction that V-TREX is an independently active exonuclease (Slack et al., 2004).

Effects of pH, oligomer competitors, EDTA and divalent cations on V-TREX
V-TREX-associated exonuclease activity on the 5HEX-oligo substrate was inhibited when unlabelled oligomers were added in molar excess (Fig. 5a). Activity was also inhibited in the presence of EDTA (Fig. 5b), indicating that V-TREX is a metalloenzyme that requires the presence of divalent cations.

(96K): Fig. 5. Effect of unlabelled competitor, EDTA and divalent cations on V-TREX exonuclease activity. Soluble protein lysates from Sf-9 cells that had been infected with BacPAK-v-trex or BacPAK-β-gal were diluted in TREX dilution buffer to give 80 ng lysate protein per exonuclease assay. Exonuclease reactions were fractionated by electrophoresis in a denaturing 13 % acrylamide/urea gel. (a) Unlabelled oligomers were added at increasing concentrations to compete with 0·25 mM 5HEX-oligo. Concentration (mM) of unlabelled competitor oligomers in each reaction is indicated above each lane. (b) Increasing concentrations (mM) of EDTA were added to the exonuclease assays as indicated. These exonuclease assays also included 5 mM MgCl2. (c) Increasing concentrations (µM) of different divalent cations (in the form of MgCl2, MnCl2, ZnCl2 and CaCl2) were added to the exonuclease assays as indicated. The type of lysate present is indicated at the bottom: BacPAK-v-trex (V-TREX or VT) or BacPAK-β-gal (β).

The effect of different divalent cations on exonuclease activity was examined. Initially, we screened Mg²⁺, Mn²⁺, Zn²⁺, Ca²⁺ and Cu²⁺ over concentration ranges of 150 mM. Exonuclease activity occurred in the presence of Mg²⁺ at all concentrations tested (data not shown). We observed residual exonuclease activity in the presence of 1 mM Mn²⁺. Other cations did not catalyse any exonuclease activity. We carried out further exonuclease assays with Mg²⁺, Mn²⁺, Zn²⁺ and Ca²⁺ at concentration ranges of 602000 µM. At this lower concentration range, it was possible to titrate out Mg²⁺-catalysed V-TREX exonuclease activity to 250 µM (Fig. 5c). It was surprising to observe that Mn²⁺ catalysed some exonuclease activity between 250 and 500 µM. A similar window of Mn²⁺-catalysed exonuclease activity has been reported for the baculovirus alkaline nuclease (AN) protein (Li & Rohrmann, 2000). In that study, Mn²⁺-catalysed AN activity was lower than Mg²⁺-catalysed activity. It was later reported that Mn²⁺ produced lower levels of AN activity, but did not inhibit activity (Mikhailov et al., 2004). In the present study, 5HEX-oligos were only partially digested in the presence of Mn²⁺, compared with complete digestion in the presence of Mg²⁺. Based on the homology of V-TREX with mammalian TREX proteins (Mazur & Perrino, 2001), it had been predicted that V-TREX would be activated to similar levels in the presence of either Mn²⁺ or Mg²⁺. There may a physiological difference between mammalian cells and baculovirus-infected insect cells that has driven the evolution of V-TREX to become more selective towards Mg²⁺.

Exonuclease assays were carried out over a range of pH values (Fig. 6). V-TREX activity was optimal between pH 6·1 and 7·4. This differentiates V-TREX from the more alkaline-active baculovirus exonuclease AN (Li & Rohrmann, 2000). V-TREX also has a more acidic activity profile than mammalian TREX proteins (Mazur & Perrino, 2001).

(117K): Fig. 6. Effect of pH on V-TREX exonuclease activity. Soluble protein lysates from Sf-9 cells that had been infected with BacPAK-v-trex or BacPAK-β-gal were diluted in TREX dilution buffer to a concentration of 10 ng lysate protein per exonuclease assay. Extract volumes of 5 µl were combined with 45 µl pH-adjusted buffer (5 mM MgCl2, 50 mM Tris/HCl, 50 mM Na2CO3, 2 mM DTT, 100 µg BSA ml1). Exonuclease reactions were fractionated by electrophoresis in a denaturing 13 % acrylamide/urea gel.

V-TREX activity on 3'-labelled oligomers and dsDNA
Further exonuclease experiments were done with different oligomer substrate combinations that included the 3'-labelled 3FAM-oligo (Fig. 7a). In this set of exonuclease experiments, only residual exonuclease activity was detected in cell lysates from BacPAK6-β-gal virus-infected Sf-9 cells (Fig. 7c). We assumed that the exonuclease activity that was detected in cell lysates from BacPAK-v-trex virus-infected Sf-9 cells was mostly V-TREX activity (Fig. 7b).

V-TREX produced different results when acting on 5'-labelled and 3'-labelled ssDNA substrates. As in earlier assays, increasing amounts of V-TREX extracts generated a gradient of smaller 5HEX-oligo fragments (Fig. 7b, panel 1). In contrast, the 3FAM-oligo abruptly dropped to a very small size when treated with V-TREX extracts (Fig. 7b, panel 3). This was interpreted to be the result of V-TREX cleaving off the labelled terminal nucleotide on the 3' end of the 3FAM-oligo. These results are as would be predicted for a 3' to 5' exonuclease and are the converse of what others have observed for the baculovirus 5' to 3' exonuclease AN (Mikhailov et al., 2003).

To examine the effects of dsDNA on V-TREX exonuclease activity, the 5HEX-oligo and 3FAM-oligo were annealed. The HEX and FAM fluorescent labels could be seen separately in the same gel, due to different emission spectra (see Methods). The annealed 5HEX-oligo and 3FAM-oligo substrates required more TREX extract in order to be digested (Fig. 7b, panels 2 and 4). This indicated that V-TREX exonuclease activity has some ssDNA specificity. The 5HEX-oligo and 3FAM-oligo design was such that when these 35 nt oligomers were annealed, 10 bp mismatched ends would be present (Fig. 7a, 5H/3F). An intermediate-sized 5HEX-oligo product was generated at protein extract concentrations of 41123 ng (Fig. 7b, panel 2). No such intermediate-sized products were generated from the 3FAM-oligo (Fig. 7b, panel 4). V-TREX thus exhibited characteristics of a 3' repair exonuclease by targeting the misannealed 3' end. This type of activity has been observed for mammalian TREX proteins (Mazur & Perrino, 2001).

Exonuclease activity associated with AgMNPV BV
Experiments were done using the 5'-labelled 5HEX-oligo to determine whether there was 3' to 5' exonuclease activity associated with AgMNPV infection. Soluble protein lysates from Ag-286 cells that had been infected with AgMNPV were compared with lysates from uninfected Ag-286 cells. The relative amount of 3' to 5' exonuclease activity associated with AgMNPV-infected Ag-286 cells was not significantly different from that of uninfected Ag-286 cells (data not shown).

We also looked for exonuclease activity associated with AgMNPV BV. Ag-286 cells were infected with AgMNPV or AcMNPV. It was ensured that similar levels of infection had been achieved (Fig. 8a) and that sucrose-cushion ultracentrifugation-purified virion preparations were diluted to contain similar amounts of total protein (Fig. 8b). The 5HEX-oligo substrate was incubated with sonicated BV preparations from AgMNPV and AcMNPV. Significantly more exonuclease activity was present in AgMNPV BVs than in AcMNPV BVs (Fig. 8c).

(123K): Fig. 8. Exonuclease activity associated with AgMNPV. Ag-286 cells were infected with AgMNPV or AcMNPV. Cells were harvested at 52 h p.i. prior to lytic release of occlusion bodies. (a) Infected cells were examined by phase-contrast microscopy just prior to harvesting of cell-culture supernatants. Bars, 10 µm. (b) BVs were purified by sucrose-cushion ultracentrifugation, disrupted by sonication and diluted to the same relative protein concentration in TREX dilution buffer. A portion of each BV sample was fractionated by 12 % acrylamide : N,N'-methylene-bis-acrylamide (37 : 1) SDS-PAGE. (c) A one-third serial dilution series of both lysates was assayed for exonuclease activity on a HEX-labelled oligomer. Exonuclease reactions were fractionated by electrophoresis in a denaturing acrylamide/urea gel. Total amount of protein lysate (ng) used in each exonuclease reaction is indicated above each lane.

Insect cells already contain significant levels of 3' to 5' exonuclease activity such that wt AgMNPV-infected cells do not have detectably elevated exonuclease levels due to V-TREX. Our data point to the possibility that V-TREX is specialized to associate with BVs of AgMNPV. However, we acknowledge that the present data are inconclusive and that more direct evidence is needed to confirm that V-TREX is the origin of 3' to 5' exonuclease activity in AgMNPV BVs.

Conclusions
The evidence presented in this study leads to the conclusion that the v-trex gene product is a functional 3' to 5' exonuclease and that V-TREX belongs to the TREX family of exonucleases. At 23·7 kDa, V-TREX is one of the smallest functional 3' to 5' exonucleases to be described. V-TREX showed remarkable stability throughout this study, with activity varying little over several months of repeated freezing and thawing. Recently, a v-trex gene homologue appeared in GenBank as ORF 119 of the C. fumiferana defective NPV (CfDEFNPV) baculovirus genome (accession no. AY327402.1). The CfDEFNPV v-trex homologue is predicted to encode a protein that is 148 aa in size and appears to be missing one-third of its C-terminus.

The v-trex gene has not been identified in the genomes of most other sequenced baculovirus species. There are no v-trex homologues in other virus families and v-trex is most similar to eukaryotic genes. This suggests that the v-trex gene was probably acquired recently in baculovirus evolution. Studies are currently being done to determine whether the v-trex gene is essential for the replication of AgMNPV and what biological function v-trex may have. We anticipate that v-trex will be classified as a baculovirus auxiliary gene, along with such genes as v-cath (Slack et al., 1995) and ChiA (Hawtin et al., 1995).

The authors thank Drs Matthew Greenstone, Dawn Gundersen-Rindal and Mike Blackburn for their advice on this manuscript. We also thank Dr Dwight Lynn for providing cell lines. Mention of a trademark or proprietary product does not constitute a guarantee of warranty of the product by the USDA and does not imply its approval to the exclusion of other products that may also be suitable.

Footnotes

†Present address: Department of Entomology, Soils and Plant Sciences, Clemson University, Clemson, SC 29634-0315, USA.