Downregulation of a Chitin Deacetylase-Like Protein in Response to Baculovirus Infection and Its Application for Improving Baculovirus Infectivity

Abstract

Several expressed sequence tags (ESTs) with homology to chitin deacetylase-like protein (CDA) were selected from a group of Helicoverpa armigera genes whose expression changed after infection with H. armigera single nucleopolyhedrovirus (HearNPV). Some of these ESTs coded for a midgut protein containing a chitin deacetylase domain (CDAD). The expressed protein, HaCDA5a, did not show chitin deacetylase activity, but it showed a strong affinity for binding to chitin. Sequence analysis showed the lack of any chitin binding domain, described for all currently known peritrophic membrane (PM) proteins. HaCDA5a has previously been detected in the H. armigera PM. Such localization, together with its downregulation after pathogen infection, led us to hypothesize that this protein might be responsible for the homeostasis of the PM structure and that, by reduction of its expression, the insect may reduce PM permeability, decreasing the entrance of baculovirus. To test this hypothesis, we constructed a recombinant nucleopolyhedrovirus to express HaCDA5a in insect cells and tested its influence on PM permeability as well as the influence of HaCDA5a expression on the performance of the baculovirus. The experiments showed that HaCDA5a increased PM permeability, in a concentration-dependent manner. Bioassays on Spodoptera frugiperda and Spodoptera exigua larvae revealed that NPV expressing HaCDA5a was more infective than its parental virus. However, no difference in virulence was observed when the viruses were injected intrahemocoelically. These findings support the downregulation of a midgut-specific CDA-like protein as a possible mechanism used by H. armigera to reduce susceptibility to baculovirus by decreasing PM permeability.

Baculoviruses are a naturally occurring group of large double-stranded DNA viruses that are specific to arthropods and have potential for widespread use for insect pest management. It has already been proven that they can effectively replace chemical insecticides in the field, for example, in the case of Helicoverpa armigera single nucleopolyhedrovirus (HearNPV) sprayed on cotton fields in Australia (9) and China (38) to control one of the most widely spread polyphagous pests (10). Baculoviruses occur naturally, are nonpathogenic to humans or other vertebrates, and are relatively host specific, and no impact on nontarget organisms has been reported to date. These characteristics make them environmentally safe insecticides. Despite the environmental advantages of baculoviruses, their use as biocontrol agents is limited, mainly due to their slow action compared to that of other pesticides. Naturally occurring baculoviruses, although highly infectious, have adapted to their hosts during their evolution, therefore killing the hosts relatively slowly and achieving maximum viral propagation. It takes up to 10 days for the virus to stop insect feeding or to kill the infected larvae (34). For this reason, reduction in the time of killing has been the main focus of research to improve baculovirus performance, and several strategies have been used, such as coapplying synergistic chemicals or using genetic engineering to introducing foreign genes coding for toxins, hormones, or enzymes into their genomes (18, 19). The strategy of acquiring foreign genes has been used by viruses themselves. Most large cytoplasmic and nuclear DNA viruses have been shown to capture, by horizontal gene transfer, host genes related to ubiquitin signaling, defense against apoptosis, and immune responses (20). The average baculovirus genome contains more than 100 open reading frames (ORFs) encoding predicted proteins of more than 50 amino acids. Phylogenetic analyses suggest that during evolution, several baculovirus genes, such as the inhibitor of apoptosis (iap) and ecdysteroid UDP-glucosyltransferase (egt) genes, were acquired from their insect hosts by horizontal gene transfer (17). Access to the recently available genome of Bombyx mori enabled a survey of B. mori NPV (BmNPV) genes that might have been acquired from the host. The survey identified 35 insect homologs potentially encoded by 37 baculoviruses (22). Knockout studies of insect homologs in baculoviruses have shown that some host homologs are essential for complete in vivo pathogenicity (22). Their functions are maintained or modified in order to control host physiology and cell signaling pathways for better virus multiplication and vertical transmission in nature.

To identify host genes whose expression could be advantageous for the baculovirus to increase its insecticidal characteristics, we checked the change in expression of host genes in response to baculovirus infection. DNA microarray experiments revealed a set of H. armigera midgut genes that were up- and downregulated due to infection with HearNPV (unpublished data). Among them, several expressed sequence tags (ESTs) coding for a chitin deacetylase-like protein (CDA) were found to be downregulated after virus infection, suggesting its possible role in the response to the infection. CDAs have been isolated from various fungi and bacteria, and their biological functions include softening of the insect cuticle to allow easier mycelial penetration (in the case of fungi) and evasion of lysozyme action (in the case of bacteria). They convert chitin, a β-1,4-linked N-acetylglucosamine polymer, into its deacetylated form, chitosan, a natural glucosamine polymer (42). Recently, CDAs were also identified in insects and appear to constitute one of two major classes of proteins recovered from the peritrophic membrane (PM) (4, 30). PM lining the insect midgut represents a major lepidopteran physical barrier against baculovirus infection (14, 35, 46). It consists of chitin and glycoproteins, and its physical role is to protect midgut epithelial cells from food particles, digestive enzymes, and pathogens. It also has a biochemical function, such as the inactivation of ingested toxins and enzyme recycling (3). Disruption of the link between chitin and the protein structure of the PM affects its functions in digestion and also leads to the collapse of the midgut defense against pathogens.

In this work, H. armigera EST sequence analyses allowed the identification and full-length sequencing of three different CDA-like proteins, revealing that only one of them (HaCDA5a) was downregulated during the initial stages of baculovirus infection in larvae. HaCDA5a has been recombinantly expressed in insect cells, and its influence on PM permeability was checked. Given the natural ability of baculovirus to acquire insect host genes in order to improve survival and prevalence, we also analyzed the effects of CDA expression on the performance of baculovirus in insect bioassays. The results revealed that expression of CDA-like proteins by baculovirus may increase its infectivity and speed of kill and thus be applied for better pest control.

MATERIALS AND METHODS

Insects and viruses.

H. armigera, Spodoptera frugiperda, and Spodoptera exigua larvae were reared on an artificial diet at 25°C, 70% humidity, and a 16-h/8-h photoperiod, as described by Shorey and Hale (36), Smits et al. (37), and Zhang et al. (48), respectively. The insect cell line used in this study was Sf21 from S. frugiperda (43). Sf21 cells were cultured at 27°C in Grace's medium (Gibco-Invitrogen Corp., NY) supplemented with 10% fetal bovine serum (FBS).

Viruses used in this study included wild-type HearNPV isolate C1 (47), SeMNPV isolate US1 (11), and a bacmid-derived virus, AcBacΔCC, based on Autographa californica MNPV (21).

DNA microarray analysis of baculovirus-exposed insects.

Based on initial DNA microarray analysis, midgut genes whose expression changed significantly upon virus infection were identified. PCR products derived from a cDNA library from the H. armigera larva midgut, containing around 4,000 ESTs, were used for the microarray printing (1). For each biological replication, around 20 fourth-instar larvae (molted within the last 10 h) were starved overnight and then fed 5 μl of a 5% sucrose-10% methylene blue (MB) solution containing 2.5 × 10⁶ occlusion bodies (OBs) from HearNPV. After 30 min, larvae that consumed the whole drop were selected and used for the subsequent steps. Larvae were transferred to a standard insect diet and maintained individually until tissue collection. Midguts of healthy and virus-treated larvae were collected (at 8 h postinfection [p.i.]) and pooled. Total RNA was extracted using a Perfect RNA purification kit (Eppendorf, North Ryde, Australia) following the manufacturer's specifications. For each sample, 50 μg of total RNA was reverse transcribed and the resulting cDNA sample was split in half for Cy3 and Cy5 labeling. cDNA labeling, microarray hybridization, and data capturing and analysis were performed following standard protocols. Three biological replications were performed for each group of control and virus-treated larvae.

Semiquantitative and real-time quantitative RT-PCR.

The presence and abundance of mRNAs from different CDAs in various larval tissues were first estimated by semiquantitative reverse transcription-PCR (RT-PCR). For this purpose, total RNA was isolated from the anterior, middle, and posterior part of the midgut, from the hindgut, from the fat body, and from Malpighian tubules of fourth-instar H. armigera larvae. RNA was isolated using Tripure (Roche Diagnostics, Mannheim, Germany) and reverse transcribed using oligo(dT) primers with SuperScript II reverse transcriptase (Invitrogen, Carlsbad, CA). For semiquantitative PCR, 5 μl of a 1:50 dilution of cDNA, synthesized from 1 μg of total RNA, was used for amplification with specific primer sets. The amplification reactions were carried out in an Eppendorf thermocycler, using 25 cycles of 94°C (30 s), 50°C (30 s), and 72°C (1 min) and a final extension step of 5 min at 72°C.

Real-time quantitative RT-PCR (qPCR) was used to determine the changes in expression of the Hacda5a gene in virus-infected larvae. Third-instar larvae were infected orally with 1 × 10⁶ HearNPV OBs, by diet contamination. Larvae that consumed the entire disk were transferred to fresh diet and maintained individually until tissue collection. Midguts from infected and control larvae were collected at 6, 24, and 48 h p.i. Total RNA was isolated and cDNA synthesized as described above. qPCR was carried out in an ABI Prism 7000 thermocycler from Applied Biosystems. All reactions were performed using Power SYBR green PCR master mix (Applied Biosystems, Foster City, CA) in a total reaction volume of 25 μl. Five-microliter cDNA templates were added to each reaction mix. Forward and reverse primers designed using Primer Express software (Applied Biosystems, Foster City, CA) were added to a final concentration of 0.3 nM. The reactions were performed in triplicate. All primers used are available upon request. The 2^ΔΔC_T method (25) was used to calculate relative changes in gene expression determined from qPCR experiments. The data are presented as fold changes in target gene expression in infected tissue normalized to the internal control gene (ATP synthase) and relative to the noninfected tissue control.

To confirm the expression pattern of cda genes in another insect-virus combination, Secda5a gene expression was also determined for S. exigua larvae infected with SeMNPV. An Secda5 partial gene sequence (EST) was obtained from an S. exigua midgut cDNA library (15). Third-instar larvae were infected orally with 1 × 10⁶ SeMNPV OBs, by diet contamination, and Secda5 expression was calculated as described above.

Sequencing of H. armigera cda genes.

An EST sequence from an H. armigera gene for CDA (Hacda5a) was used to design specific primer sets to amplify overlapping fragments of the 5′ and 3′ ends of this gene. 5′ and 3′ rapid amplification of cDNA ends (RACE)-ready cDNAs were synthesized using a SMART-RACE kit (Clontech, Saint-Germain-en-Laye, France), and 5′ and 3′ fragments were amplified using two sets of specific primers. The amplified fragments were purified and cloned into the pGEM-T Easy vector (Promega, Madison, WI) for subsequent sequencing. Several clones were sequenced for each cDNA end. Sequences were assembled by using the Seqman program from the DNAstar software package (DNASTAR, Madison, WI). Full sequences for two other cda genes, Hacda5b and Hacda1, were completed based on the information from the H. armigera genomic sequencing project (12). Alignment of all three Hacda genes was performed using ClustalX (39) and visualized in GeneDoc (28). Predicted amino acid sequences were aligned with other lepidopteran CDA sequences available from the NCBI and ButterflyBase databases (29; www.butterflybase.com). Other insect CDA sequences were omitted in our analysis for the sake of clarity. Phylogenetic analyses were performed using the neighbor-joining method with 1,000 bootstraps, using the ClustalX (39) and MEGA 3.1 (24) programs. Distances were corrected for multiple substitutions according to the method of Kimura (23).

Recombinant baculovirus production.

The full-length Hacda5a ORF was amplified by PCR from H. armigera midgut cDNA. Primers were designed to include the NsiI restriction site to enable further cloning, as well as a polyhistidine (6×His) tag upstream of the stop codon. The obtained amplicon was cloned into the pGEM-T Easy vector for sequence verification. The Hacda5a ORF was then excised from pGEM-T Easy with NsiI and cloned into pFBD-ph (16) downstream of the p10 promoter to obtain pFBD-ph-Hacda5a. pFBD-ph contains the AcMNPV polyhedrin gene under the control of the polyhedrin promoter. In order to generate recombinant viruses, Escherichia coli strain DH10Bac, containing the Δcc bacmid (21) and the pMON7124 helper plasmid (26), was transformed with the pFBD-ph and pFBD-ph-Hacda5a plasmids according to the procedure described for the Bac-to-Bac system (Invitrogen, Carlsbad, CA). Recombinant bacmids were selected based on white-blue screening of DH10Bac colonies, and the positive clones were confirmed by PCR. Bacmid DNA was isolated from bacterial cells according to standard procedure and used to transfect Sf21 cells, using Insect GeneJuice transfection reagent (Novagen, Darmstadt, Germany). Recombinant Δccph and Δccph-Hacda5a bacmid-derived viruses were multiplied in an additional round of infection to produce high-titer stocks for further experiments. Viruses were titrated by end-point dilution assay (EPDA) in Sf21 cells according to the method of Vlak (44), and the titers were expressed as the 50% tissue culture infective dose (TCID₅₀).

HaCDA5a expression in Sf21 cells.

Sf21 cells were infected with Δccph and Δccph-Hacda5a at a multiplicity of infection (MOI) of 5, and cells and medium were collected at 72 h p.i. to analyze protein expression. Cells were lysed in phosphate-buffered saline (PBS)-0.2% Triton X-100. Cell extracts as well as cell culture medium proteins were separated by 12% SDS-PAGE, and the presence of the recombinant protein was detected using antibodies against the 6×His tag (BD Pharmingen, San Diego, CA). For protein production, Sf21 cells were seeded in T75 flasks (Nunc, Rochester, NY) and infected according to previously mentioned conditions.

Chitin binding assay.

The chitin binding activity of HaCDA5a was checked according to a previously described method (13). Chitin was regenerated as follows. Two hundred fifty milligrams of chitin from crab shells (Sigma) was dissolved in 5 ml of 85% phosphoric acid and incubated for 16 h at 4°C, with stirring. After incubation, the suspension was centrifuged for 5 min at 10,000 rpm, and the gelatinous chitin pellet was washed several times with water, until a pH of about 7.0 was obtained. Finally, the chitin pellet was broken in a mortar. For the binding assay, 4 ml of cell culture medium containing recombinant HaCDA5a was incubated with 100 mg (wet weight) of regenerated chitin in the presence of a protease inhibitor cocktail (Roche Diagnostics, Mannheim, Germany) for 4 h at 4°C, with continuous stirring. The mixture was centrifuged, and the chitin pellet was washed three times with PBS, followed by centrifugation. After the last centrifugation step, chitin pellets were suspended in 0.2 ml of PBS, 0.5 M NaCl, 20 mM acetic acid, 6 M urea, and 2% SDS with the addition of 5% β-mercaptoethanol or in 1% calcofluor white (fluorescent brightener 28; Sigma) at room temperature in the presence of the protease inhibitor cocktail. After 15 min of incubation, suspensions were centrifuged for 10 min at 10,000 rpm, and the supernatants were checked for the presence of HaCDA5a released from chitin by Western analysis.

PM permeability assays.

Changes in permeability of the PM to methylene blue (MB) due to HaCDA5a exposure were measured with an Ussing chamber (CHM8; World Precision Instruments, Stevenage, United Kingdom). Preliminary experiments showed a significant loss of HaCDA5a protein when it was purified through HiTrap chelating HP columns. For this reason, permeability assays were performed with cell culture medium, with or without HaCDA5a, partially purified by removal of budded viruses (BVs) by 1 h of centrifugation at 20,000 × g at 4°C, and then concentrated by incubation on dry sucrose in a dialysis tube.

S. frugiperda PMs were isolated from actively feeding last-instar larvae. Briefly, each larva was anesthetized on ice for 2 min, and the midgut was then extracted and opened longitudinally to expose the PM. The PM was isolated, carefully laid on a large mesh cotton film, cut lengthwise, and gently rinsed with PBS to remove the food content. A properly cut portion of the cotton film with the attached PM was mounted on the flow chamber, with the result of a 12.6-mm² portion of PM exposed to the two compartments. The compartment with the ectoperitrophic side (the ectoperitrophic compartment) was filled with 450 μl of PBS, and the one with the endoperitrophic side (the endoperitrophic compartment) was filled with 450 μl of a 0.2-mg/ml MB solution in PBS. Prior to the start of the assay, the integrity of isolated PMs was checked by measuring the flux of MB through the PM after the first 30 min of incubation at room temperature. Next, 100 μl of the buffer solution in both compartments was replaced with an equivalent volume of cell culture medium containing HaCDA5a from Sf21 cells infected with Δccph-Hacda5a. Culture medium from Sf21 cells infected with Δccph was used as a control. After 1 h of incubation, the solutions in both compartments were recovered, and the concentration of MB was calculated based on the absorbance measured at 661 nm. The MB flux was expressed as μg of dye (calculated by means of a calibration curve after subtracting the flux at 30 min from the final flux) that passed through the 12.6-mm² portion of mounted PM in 1 h. Three independent replicates were performed for each data collection.

Biological activity assays.

The effect of heterologous expression of HaCDA5a on baculovirus performance was determined both ex vivo (with Sf21 cells) and in vivo (with S. frugiperda and S. exigua larvae).

Sf21 cells were infected for 1 h with Δccph and Δccph-Hacda5a at an MOI of 1. After infection, cells were washed and incubated in fresh medium. At 0, 24, 48, 72, and 96 h p.i., a 10-μl aliquot of medium was harvested and the quantity of BVs in each sample was determined by an end-point dilution assay on Sf21 cells. Experiments were duplicated for each time point and each virus.

For oral in vivo assays, fourth-instar S. frugiperda and S. exigua larvae were fed diet disks contaminated with 5 × 10⁴ OBs of a virus. Δccph and Δccph-Hacda5a OBs were purified from cell culture according to standard procedure. Briefly, the cells were incubated in 0.5% SDS for 30 min at 25°C and then washed by a series of washes in water and 0.5 M NaCl. Finally, the OBs were pelleted through a cushion of 30% (wt/wt) sucrose for 30 min at 9,000 rpm. To discard any effect on the number of nucleocapsids in the OBs due to the expression of Hacda5a protein, the numbers of DNA copies in OBs from both viruses (as an estimation of nucleocapsid numbers) were compared by means of qPCR, using specific primers for the DNA polymerase gene.

Larvae that consumed the entire disk were transferred to fresh diet and maintained individually until death or pupation at 27°C. Larval mortality was recorded every 12 h. For each virus and insect species, 20 to 48 larvae were tested, and the bioassays were conducted in duplicate. The mean time to death was calculated for each virus and compared using two-way analysis of variance (ANOVA) (GraphPad Prism) (27).

For in vivo intrahemocoelic infection assays, fourth-instar S. frugiperda and S. exigua larvae were injected with 10 μl of the corresponding suspension of BVs at a concentration of 10⁷ TCID₅₀/ml in Grace's cell culture medium containing phenol red to monitor the injections. At least 20 larvae were used for each virus, and the assay was conducted in duplicate for each insect species. Larval mortality was recorded every 12 h. The mean time to death was calculated for each virus-insect combination and compared using two-way ANOVA (GraphPad Prism) (27). The bioassays were conducted at 27°C.

A second experiment was performed with third-, fourth-, and fifth-instar S. exigua larvae. Larvae (32 to 48 for each instar) were fed diet disks contaminated with 1 × 10³ OBs (L3 larvae), 5 × 10³ OBs (L4 larvae), and 1 × 10⁴ OBs (L5 larvae). Larvae that consumed the entire disk were transferred to fresh diet and maintained individually until death or pupation. Bioassays were conducted in duplicate. Mortality was recorded every 12 h, with the exception of L5 larvae, where the mortality was recorded every 24 h. The percentage of mortality was calculated for each instar, and the results were analyzed as before. The bioassays were conducted at 25°C.

Nucleotide sequence accession numbers.

Sequences were deposited in GenBank under the following accession numbers: GQ411189 (Hacda1), GQ411190 (Hacda5a), and GQ411191 (Hacda5b).

RESULTS

Identification and expression analysis of chitin deacetylase-like proteins from H. armigera midgut.

Among the midgut genes that were differentially expressed after baculovirus ingestion, 8 ESTs with homology to cda genes were detected. These ESTs were downregulated, with an average change in the expression ratio of around threefold. In contrast, another 3 ESTs with homology to cda genes did not show changes in gene expression after baculovirus ingestion. Sequence analysis revealed that the eight downregulated ESTs belonged to the same gene (Hacda5a), while among the other three ESTs, two belonged to a second cda gene (Hacda5b) and the other belonged to a third one (Hacda1). These results prompted us to study the role of CDA proteins in the response to baculovirus.

In order to confirm the downregulation of Hacda5a expression, qPCR was performed on total midgut cDNA from H. armigera third-instar larvae infected with HearNPV, using primers specific for Hacda5a. Experiments showed a 4.8-fold decrease in expression of Hacda5a at 6 h p.i. and a 2.6-fold decrease at 24 h p.i. At 48 h p.i., Hacda5a expression had returned to the regular level (Fig. 1A). Downregulation of the cda gene in the first hours after baculovirus infection was also observed in S. exigua larvae infected with SeMNPV. Experiments showed a 3.9-fold decrease in expression of Secda5a at 6 h p.i. At later times p.i. (24 and 48 h p.i.), Secda5a expression in the midguts of infected larvae did not significantly differ from the expression in the midguts of control larvae (Fig. 1A).

FIG. 1.

Expression of different CDAs in H. armigera larvae. (A) Changes in expression of Hacda5a and Secda5a after baculovirus infection in the midguts of third-instar H. armigera (L3) and S. exigua (L3) larvae, respectively. Total RNAs were isolated from the midguts of infected larvae at different time postinfection and then reverse transcribed, and qPCR was performed using specific primers. The results are the means ± standard deviations for three independent infection experiments. Asterisks demark changes in expression that are significantly different from expression in the control larvae (by the t test). Dotted lines demark onefold changes in expression (no change in expression compared to control, noninfected larvae). (B) Expression of CDAs in different H. armigera (L4) tissues by semi-qPCR, using specific primers for Hacda5a, Hacda5b, and Hacda1. A, anterior midgut; M, middle part of the midgut; P, posterior midgut; H, hindgut; FB, fat body; MT, Malpighian tubules.

Because Hacda5a expression was first detected in the EST microarray of midgut tissues and confirmed by qPCR to occur in the larval midgut as well, the tissue specificity of expression of Hacda5a and the two additional genes coding for chitin deacetylase-like proteins (HaCDA5b and HaCDA1) was determined. Among the tissues tested, Hacda5a was expressed mainly in the gut (in the anterior, middle, and posterior parts of the midgut and in the hindgut), with comparatively very little expression in the Malpighian tubules and fat body. Hacda5b and Hacda1 were expressed mainly in the fat body and Malpighian tubules, although the former also showed limited expression in the midgut (Fig. 1B).

Sequence analysis of Hacda genes.

A partial sequence of Hacda5a (EST) was used to design two sets of primers for amplification, cloning, and sequencing of 5′ and 3′ cDNA ends of this gene. The full sequence has been obtained and deposited in the NCBI GenBank under accession number GQ411190. The full assembled sequence of Hacda5a cDNA was 1,277 bp long and contained an ORF of 1,173 bp, followed by an AT-rich untranslated region (UTR) and three putative polyadenylation signals (AATAAA), located 5, 17, and 22 bp upstream of the poly(A) tail. The protein encoded by this ORF contains 391 amino acid residues, with a 17-amino-acid putative signal peptide. The predicted molecular mass of the secreted protein without the signal peptide is around 42 kDa. A search of the Conserved Domain Database (NCBI) identified a putative polysaccharide deacetylase domain in amino acid residues 63 to 193. Fifteen cysteine residues are present along the HaCDA5a sequence; however, their distribution does not resemble that for any peritrophin motif described for known PM proteins.

In a comparison with the other two chitin deacetylase-like proteins identified in H. armigera, HaCDA5a shows a high amino acid sequence similarity (77%) to HaCDA5b and a similarity of only 35% to HaCDA1. All three HaCDAs contain a putative chitin deacetylase domain (CDAD), while only one of them, HaCDA1, codes additionally for a putative chitin binding domain (ChBD) and a lipoprotein receptor class A domain (LDLa). HaCDA1 is significantly longer than HaCDA5a and HaCDA5b and probably belongs to a distant group of deacetylases (Fig. 2).

FIG. 2.

Alignment of deduced amino acid sequences from H. armigera CDAs. The sequences have been deposited in GenBank under accession numbers GQ411189 (HaCDA1), GQ411190 (HaCDA5a), and GQ411191 (HaCDA5b).

A phylogenetic analysis was performed with all available chitin deacetylase-like protein sequences from the Lepidoptera, most of them from Bombyx mori. Lepidopteran CDAs form three main branches that correlate with their domain contents (Fig. 3). Two of the branches are further subdivided, according to Campbell et al. (4), rendering five CDA groups altogether (groups I to V). HaCDA5a and HaCDA5b form a clade, together with Trichoplusia ni CDA, S. frugiperda CDA, Mamestra configurata CDA, Epiphyas postvittana CDA, and three B. mori CDAs, and belong to CDA group V, whose members contain the CDAD but not ChBD or LDLa (Fig. 3). HaCDA1 is clearly distant from the other two HaCDAs and, together with Agrotis ipsilon CDA and one B. mori CDA (Bmb030914), falls into CDA group I, whose members contain all three domains (CDAD, ChBD, and LDLa).

FIG. 3.

Unrooted phylogenetic tree obtained from alignment of lepidopteran CDAs. CDA groups I to V and their domain contents are indicated. CDA, chitin deacetylase domain; ChBD, chitin binding domain; LDLa, low-density lipoprotein receptor class A domain. Sequences used in the alignment were as follows (with GenBank accession numbers in parentheses): H. armigera CDAs HaCDA1, HaCDA5a, and HaCDA5b (this publication), Agrotis ipsilon CDA (FJ899541), Epiphyas postvittana CDA (EV809282), Mamestra configurata CDA (EU660852), Spodoptera frugiperda CDA (ButterflyBase cluster SFP00364), and Trichoplusia ni CDA (AY966402). Bombyx mori sequences were obtained from SilkDB (http://silkworm.genomics.org.cn), and their accession numbers are listed in the figure. Sequences were aligned using ClustalX, and a neighbor-joining tree with 1,000 replicates was generated. Numbers on the branches indicate neighbor-joining bootstrap percentages. The scale bar indicates an evolutionary distance of 0.1 amino acid substitution per position in the sequence.

According to Dixit et al. (7) and Campbell et al. (4), insect CDAs are divided into five major classes, classes I to V. Classes I and II contain three recognized domains, CDAD, ChBD, and LdLa. Class III and IV CDAs contain CDAD and ChBD, and class V CDAs contain only CDAD. To prevent nomenclature confusion, we propose to follow these divisions and to name CDAs from a given insect species with the Arabic numeral corresponding to the CDA class to which they belong followed by a lowercase letter if more than one CDA from a particular class is identified. For this reason, we have named the CDAs found in H. armigera HaCDA1, HaCDA5a, and HaCDA5b (see Fig. 3 for the phylogenetic class grouping).

Heterologous expression of HaCDA5a in insect cell culture.

In order to study the possible role of HaCDA5a in baculovirus infection, a recombinant Δcc bacmid expressing 6×His-tagged HaCDA5a was constructed. Δcc is a deletion mutant of Ac-bacmid in which the chitinase and v-cathepsin ORFs have been removed, and it was chosen due to the lack of possible interference of chitinase activity.

HaCDA5a was successfully expressed in Sf21 cells, and its expression was monitored by Western analysis using anti-6×His antibody. HaCDA5a expression was detected already at 24 h p.i. (not shown), but the maximum yield was observed at 72 h p.i. HaCDA5a was present mainly in the cell extracts, but part of it was also secreted into the cell culture medium (Fig. 4A). The mobility of HaCDA5a in 12% SDS-PAGE gels was estimated to be around that of a 42-kDa protein, as predicted from the amino acid sequence without the signal peptide.

FIG. 4.

Recombinant expression and chitin binding of HaCDA5a. (A) Detection of 6×His tag from HaCDA5a by Western blot analysis of the medium and cell extract (72 h p.i.) of Sf21 cells infected with the Δccph-HaCDA5a baculovirus or infected with the control Δccph baculovirus. The total amount of protein loaded was 10 μg for the cell extract lanes and 30 μg for the medium lanes. (B) Detection of 6×His tag of HaCDA5a released from chitin after incubation in PBS, 1% calcofluor white (fluorescent brightener 28; Sigma), 0.5 M NaCl, 6 M urea, 2% SDS-5% β-mercaptoethanol (SDS-βME), or 20 mM acetic acid, using 6×His antibody.

Chitin binding activity assay.

Due to the facts that the HaCDA5a homolog in Trichoplusia ni was found in the PM (13) and that the PM consists of proteins and chitin, the ability of HaCDA5a to bind chitin was checked. Recombinant HaCDA5a expressed in Sf21 cells and secreted into the cell culture medium exhibited a strong binding affinity for regenerated chitin. After incubation with an excess of regenerated chitin, no HaCDA5a protein remained in the cell culture medium (Fig. 4B, PBS lane). Most of the protein bound to chitin could be released by treatment with either 6 M urea, 1% calcofluor white, or 2% SDS-5% β-mercaptoethanol (Fig. 4B). Chitin-bound HaCDA5a could not be released by treatment with 0.5 M NaCl or 20 mM acetic acid.

Effect of HaCDA5a on PM permeability.

According to our hypothesis that HaCDA5a protein might be responsible for PM structure homeostasis, we decided to determine if HaCDA5a had any influence on PM permeability. To check the influence of HaCDA5a on PM permeability, we intended to isolate PMs from H. armigera larvae. But H. armigera PMs appeared unsuitable for these experiments due to their extremely fragile structure. Instead, we used S. frugiperda larvae, whose isolated PMs have properties that enable them to be mounted more easily in the experimental chamber.

Incubation of S. frugiperda PM in the presence of HaCDA5a clearly permeabilized the PM, as demonstrated by the increase of the flux of the MB dye, in a concentration-dependent manner (Fig. 5).

FIG. 5.

Methylene blue flux across isolated S. frugiperda peritrophic membrane incubated with increasing amounts of HaCDA5a. The results are the means ± standard deviations for three independent infection experiments.

Effect of HaCDA5a overexpression on infectivity and virulence of baculovirus.

Overexpression of HaCDA5a by the recombinant virus did not influence virus growth in cell culture compared to that of virus lacking HaCDA5a (Fig. 6).

FIG. 6.

One-step growth curve analysis of Δccph-Hacda5a and Δccph viruses in Sf21 cells infected at an MOI of 1. The results are the means ± standard deviations (error bars) for independent infection and titration experiments. BV accumulation is shown as the titer, expressed in TCID₅₀ units/ml, calculated for each time point.

Due to the effect of HaCDA5a on PM permeability, we tested whether expression of this protein by baculovirus would change its performance in vivo. We compared the virulence of Δccph and Δccph-Hacda5a in two insect species, S. frugiperda and S. exigua. S. frugiperda was selected for bioassays because this species was also used to measure PM permeability after HaCDA5a treatment. S. frugiperda, however, is known for its low level of susceptibility to AcMNPV, and this virus is the basis of the bacmid-derived virus employed here (Δcc). For this reason, we also included S. exigua larvae, which are more susceptible to AcMNPV, in the bioassays.

Mean time to death was measured for both insect species and both viruses. Only larvae that died were included in the calculations. Calculation of the mean times to death was possible due to equal mortalities that both viruses caused in each species (26 and 28% for Δccph and Δccph-Hacda5a, respectively, in S. frugiperda larvae and 95 and 100% for Δccph and Δccph-Hacda5a, respectively, in S. exigua larvae). For both species, we found significant differences in the mean times to death between both viruses, but only when the viruses were administered orally (Fig. 7). S. frugiperda larvae infected per os with Δccph-Hacda5a died 32 h earlier (18% faster), on average, than larvae infected with the virus lacking HaCDA5a. In the case of S. exigua, the difference was smaller. The Δccph-Hacda5a virus killed infected larvae 11 h earlier (9% faster), on average, than larvae infected with the virus not expressing HaCDA5a. In contrast, intrahemocoelic infections using BVs showed a similar speed of kill for both viruses in the two insect species.

FIG. 7.

Mean times to death for fourth-instar S. frugiperda (A) and S. exigua (B) larvae infected orally or by intrahemocoelic injection with Δccph-Hacda5a in comparison to those for larvae infected with Δccph. The results are the means ± standard deviations for independent infection experiments. Asterisks denote a significant difference (two-way ANOVA; P < 0.05) from infection with Δccph.

The doses used in the first experiment gave very low mortality for S. frugiperda larvae when the virus was administered orally (28 and 26% for Δccph and Δccph-Hacda5a, respectively) and almost 100% mortality in the case of S. exigua L4 larvae. Therefore, the infectivities of Δccph and Δccph-Hacda5a were compared in third-, fourth-, and fifth-instar S. exigua larvae to assess the influence of larval instar on virus infectivity. In all tested instars, we found significant differences in mortality caused by both viruses (Fig. 8). The mortalities caused by Δccph and Δccph-Hacda5a were 50 and 63.4% for third-instar larvae, 33.8 and 54.1% for fourth-instar larvae, and 21.9 and 35.4% for fifth-instar larvae, respectively.

FIG. 8.

Mortality of S. exigua third-, fourth-, and fifth-instar larvae orally infected with Δccph-Hacda5a in comparison to Δccph. The results are the means ± standard deviations (error bars) for independent infection experiments. Asterisks denote a significant difference (two-way ANOVA; P < 0.05) compared to infection with Δccph.

DISCUSSION

In this study, a protein with homology to chitin deacetylases was found to be downregulated due to baculovirus infection in the midguts of H. armigera and S. exigua larvae. Quick downregulation of CDA as a consequence of baculovirus infection suggests an active role of this protein in the insect response to baculovirus. Our experiments on HaCDA5a influence on PM permeability have demonstrated, for the first time, an activity of insect CDA-like proteins on the PM. The decrease of HaCDA5a expression due to baculovirus infection would lead to less HaCDA5a protein present in the PM and thus increase the PM stiffness, and this may constitute an early mechanism of protection from oral infection by baculovirus.

In agreement with this hypothesis, several studies have revealed the effect of PM-disrupting agents or enzymes in enhancing baculovirus infectivity (6, 33, 45, 46). According to our expression results at different h p.i., the downregulation occurs during the first 6 h of infection, while the viral occluded derived viruses (ODVs) are still present in the gut lumen, and expression returns to normal levels once the gut lumen has been cleared of viral particles. Whether downregulation occurs as a response to virus penetration into the midgut cells or as a result of the detection of certain viral elements that could activate the response to baculovirus by the midgut epithelium is unknown.

The character of HaCDA5a activity on the PM remains unknown. Whether this activity is due to the physical interaction of HaCDA5a disrupting the PM, to its hypothetical deacetylation activity, or to any other enzymatic activity is something yet to be determined. We were unable to demonstrate deacetylase activity of HaCDA5a by measuring chitin deacetylation by native gel electrophoresis (41; data not shown), confirming previous results with TnPM-P42 CDA from T. ni (13). In contrast, Toprak et al. (40), using a similar methodology, have shown deacetylase activity of a 26-kDa form of the HaCDA5a ortholog from M. configurata (MacoCDA) recombinantly expressed in E. coli.

The proposed model of CDA-like protein expression regulation due to baculovirus infection as a protective mechanism was supported by the increase in virulence and pathogenicity of HaCDA5a-expressing virus in comparison to the parental virus. The virus expressing HaCDA5a killed both tested species, S. frugiperda and S. exigua, faster than the parent virus, but only when administered orally. The same recombinant virus injected intrahemocoelically killed the larvae with the same speed as the virus lacking HaCDA5a, also revealing that this protein does not have any apparent detrimental effect on virus multiplication in non-midgut cells. Additional experiments with S. exigua larvae revealed that the virus expressing Hacda5a causes significantly higher mortality than the parental virus. These data suggest that expression of CDA changes the performance of the virus in vivo, and most likely it is active at the entrance of the virus to midgut epithelial cells while bypassing the PM barrier. We have found CDA in the polyhedra of recombinant virus by Western analysis (data not shown). Possibly, CDA-associated polyhedra release this protein while being dissolved in the alkaline insect midgut lumen, and CDA may act on the PM, changing its permeability and thus enabling easier virus entrance into epithelial cells. Another explanation for the better performance of recombinant virus expressing HaCDA5a could be the secretion of active protein by epithelial cells back to the ectoperitrophic compartment. Evidence against this hypothesis is the fact that the protein is most likely expressed after all the remaining ODVs disappear from the midgut lumen. It is also possible that HaCDA5a protein synthesized after infection facilitates virus spread from the epithelial cells to the trachea. Insect tracheas are lined with a protein-chitin layer (5), and tracheal cells serve as a conduit for the systemic spread of baculovirus infection Midgut epithelial cells are primary target cells, but the underlying basal lamina prevents the progeny virus from directly entering the hemocoel. Passage through the basal lamina is facilitated by immediate infection of tracheoblasts, the cells that cross the basal lamina and can deliver the virus to the trachea and to the hemocoel (8).

CDA-like proteins have recently been described by proteomic approaches for the midguts of H. armigera (4, 30) and Mamestra configurata (40). Also, for T. ni, an HaCDA5a ortholog was identified in the larval midgut (13). The availability of the whole B. mori genome enabled the identification of eight chitin deacetylase homologs in this species. In general, it appears that CDAs constitute a common protein composition in Lepidoptera and apparently belong to one of two major protein classes of the PM.

So far, many CDA-like proteins have been identified from insects, but it is likely that genomic and proteomic data from new species will add many more to this list. The largest number (nine CDAs) has been identified in the red flour beetle, Tribolium castaneum. Expression profiles and double-stranded-RNA-mediated downregulation of all nine CDA transcripts revealed spatial and functional specialization among T. castaneum CDAs (2). Clear division into midgut-specific CDAs and non-gut CDAs was observed. Similar to lepidopteran CDAs, T. castaneum midgut-specific CDAs lack ChBD, while all non-gut CDAs carry this domain. Differences in the distribution of T. castaneum midgut-specific CDAs along the gut suggest that each of the CDAs may function specifically in different parts of the gut, locally changing the properties of PM, including permeability.

Bombyx mori CDA tissue specificity analysis using the Bombyx mori Microarray Database (BmMDB) (http://silkworm.swu.edu.cn/microarray/) revealed that group V CDAs are mostly expressed in the midgut, while CDAs from the other groups (I to IV) are expressed in other tissues, such as silk glands, Malpighian tubules, or hemocytes. HaCDA expression analysis by RT-PCR confirmed that the CDA grouping reflects not only domain content but also tissue distribution. HaCDA5a, expressed mainly in the midgut, and HaCDA5b, expressed in the midgut and other tissues, fall into group V, together with BmCDAs that are mostly expressed in the midgut and together with other CDAs from Lepidoptera that have been described as midgut proteins.

The lack of a typical chitin binding domain in HaCDA5a and other lepidopteran CDAs implies the existence of alternative modes of interaction with chitin. Apart from CDAs recently identified from PMs of a few insect species, to date most of the known PM proteins, including mucins, contain multiple C6 motifs of six cysteine residues that interact with chitin. Of the three H. armigera CDAs described here, only HaCDA1 contains a C6 motif, but interestingly, this CDA is hardly expressed in the midgut and its level is not regulated due to virus infection. HaCDA5a, which was demonstrated to bind chitin, contains 15 cysteines along its sequence, but their distribution does not resemble the C6 motif. It was proposed that the CDA domain may serve as a chitin binding domain (13). Alternative non-cysteine interactions with chitin have been described for arthropods (32). Many cuticular proteins have a single conserved domain, known as the R&R consensus, first reported by Rebers and Riddford (31). However, this consensus is not found in HaCDA5a or other CDA-like proteins from Lepidoptera. It would be interesting to investigate which CDA epitopes are able to bind chitin and if those as yet undescribed motifs are multiple and equally distributed, enabling strong binding, as in the case of the C6 motif.

In summary, we have found the downregulation of a midgut-specific CDA-like protein as a possible mechanism used by H. armigera to reduce the susceptibility to baculovirus by decreasing PM permeability. Recombinant baculoviruses expressing this protein have shown significant increases in the speed of kill and in pathogenicity against larvae from different Spodoptera spp., two parameters highly appreciated in baculovirus-based biopesticides. We believe that the study of the larval response will point out candidate host genes to be used, by either silencing or overexpression, for improvement of the insecticidal characteristics of baculovirus.