Gene structure and expression profile of Manduca sexta prophenoloxidase-activating proteinase-3 (PAP-3), an immune protein containing two clip domains

Abstract

Prophenoloxidase-activating proteinase-3 (PAP-3) is a component of the defence system in Manduca sexta. We have isolated genomic clones and elucidated the organization of this gene. The 3′ end of exon 2, the entire exon 3 and the 5′ end of exon 4 encode the two amino-terminal clip domains. Southern blot analysis suggested a single copy of the PAP-3 gene in the genome. We identified several putative immune-responsive elements in the upstream region. The PAP-3 gene is not highly expressed in the fat body during larval development until the wandering stage begins. The mRNA level is high in the epithelium, fat body and haemocytes. Tissue-specific alternative splicing occurs in the fat body and trachea. A bacterial injection markedly induced the gene expression in the fat body and haemocytes.

Introduction

To survive and prosper in pathogen-rich environments, insects must possess an efficient defence system in order to prevent and control microbial infection (Gillespie et al., 1997). Sclerotized cuticle is effective in fending off infectious agents in the environment. Plasma proteins, haemocytes and fat body initiate a series of humoral and cellular responses to immobilize and kill invading microorganisms. Quinone production plays an important role in many of these immune mechanisms, including cuticle sclerotization, melanin synthesis, wound healing, encapsulation and free radical-mediated pathogen killing (Nappi & Vass, 2001). Phenoloxidase (PO), which catalyses quinone formation, is synthesized as an inactive zymogen prophenoloxidase (proPO) and activated by a cascade of serine proteinases (SPs), known as the proPO activation system (Ashida & Brey, 1998).

Analogous to the blood clotting and complement systems in vertebrates, the proPO activation cascade is triggered upon recognition of physiological or pathological signals. In the presence of microbial cell wall components (e.g. peptidoglycan, lipopolysaccharide and β-1,3-glucan), circulating or cell surface receptors specifically bind to these pathogen-associated molecular patterns and trigger autoactivation of the first proteinase precursor. This, in turn, leads to sequential cleavage activation of other cascade members. To date, our understanding of this pathway is largely limited to the final step, namely proPO activation by its activating enzyme prophenoloxidase-activating proteinase (PAP). PAP has been isolated and cloned from Manduca sexta, Bombyx mori, Holotrichia diomphalia and Pacifastacus leniusculus (Jiang et al., 1998, 2003a,b; Lee et al., 1998; Satoh et al., 1999; Wang et al., 2001). In M. sexta and H. diomphalia, PAP converts proPO to active PO only when clip-domain serine proteinase homologues (SPHs) are present.

The clip domain is a structural unit stabilized by three disulphide bonds. It is thought to have regulatory functions (Jiang & Kanost, 2000). While other PAPs contain one clip domain at the amino terminus, the B. mori proPO activating enzyme, M. sexta PAP-2 and PAP-3 have two – the role of the extra clip domain is not understood. Many arthropod SPs and SPHs contain clip domains. However, little is known about their structure, function or evolution. There is no information on the exon-intron organization and expression regulation of PAP genes.

With three terminal proteinases, the proPO activation system in M. sexta appears to be quite complex. A bacterial challenge significantly increased the mRNA and protein levels of PAP-1, PAP-2 and PAP-3 (Jiang et al., 1998, 2003a,b). Their tissue localization may also change at different developmental stages. For instance, PAP-1 is present in the prepupal cuticle whereas PAP-2 and PAP-3 accumulate in the haemolymph of the same stage insect. Profiling these PAP mRNAs in various tissues and cells should allow us to better understand the transcriptional regulation of these genes in a developmental context. In this paper, we report the exon-intron organization and expression profile of the M. sexta PAP-3 gene.

Results

PAP-3 genomic clones

A M. sexta genomic library was screened with the full-length PAP-3 cDNA and two positive clones, λ2 and λ3, were obtained (Fig. 1A). Sequence analysis indicated that a 5′ cDNA fragment (115 bp) was absent in the genomic clones. Using the corresponding fragment derived from a PCR, we screened the genomic library again and isolated another positive clone, λ7. From these three bacteriophages, we isolated DNA, mapped the restriction sites and subcloned ten genomic fragments into a vector. The inserts in the resulting recombinant plasmids, 24.4 kb long in total, were completely sequenced.

Figure 1.

Structure of the Manduca sexta PAP-3 gene. (A) Restriction map of PAP-3 genomic inserts in the positive clones λ2, λ3 and λ7. Arrow heads and vertical bars indicate the start and end points of the clones. B, BamHI; E, EcoRI; H, HindIII; S, SacI; X, XbaI. The underlined HindIII site denotes its presence in the cDNA. The 4.51 kb fragment at the 5′ end of λ3 contains two HindIII sites (in parentheses) at positions 1110 and 1500. Horizontal bars mark the PAP-3 genomic fragments used for subcloning. The segment ‘3–3′ represents the 3′ end of λ3 whose sequence does not match λ2. Dotted lines, regions not sequenced; broken lines, 3′ end of λ7 and 5′ end of λ3, sequenced. (B) Exon-intron organization of the PAP-3 gene. Numbered vertical bars denote the exons, with the non-coding regions shown as open boxes. S, signal peptide; C1 and C2, clip domains 1 and 2; L, linker region; P, proteinase domain. The same scale and starting site are used for the restriction map and exon-intron structure. (C) Sequence of the BamHI-Bg/II fragment at the 3′ end of λ3. The mismatching region (clone 3–3, panel A), starting two nucleotides before the EcoRI site, is shown in italics. The five nucleotides AGATC (underlined) resemble the recognition sequence (AGATCT) of Bg/II, a restriction enzyme used for the library construction.

Clone λ7, containing the 5′ end of the PAP-3 gene, did not overlap with clone λ3 (Fig. 1A). Long-distance PCR using primers located at the 3′ end of λ7 and 5′ end of λ3 did not yield any PCR product (data not shown), indicating that the gap between the genomic clones is long. After the BamHI site near the 3′ end of λ3 (Fig. 1A), there was a 167 bp sequence identical to the 5′ end of λ2 (Fig. 1C). While this result strongly suggested that these two genomic fragments overlap, we identified an extra 633 bp sequence at the 3′ end of the λ3 insert (Fig. 1C, clone 3–3). This fragment did not match the λ2 sequence. To test whether or not it represents a cloning artefact, we amplified the genomic DNA using primers flanking the 167 bp overlap (Fig. 1A) and obtained a single 954 bp product. Cloning and sequence analysis indicated that the PCR product was identical to the overlap and flanking genomic sequences in λ2 and λ3, but does not contain the 633 bp fragment.

Exon–intron organization

A comparison between the genomic and cDNA sequences revealed that the M. sexta PAP-3 gene is composed of eight exons and seven introns (Fig. 1B). Exon 1 includes a 5′ untranslated region and a coding region for the nineteen-residue signal peptide of PAP-3 (Fig. 2). Exon 2 and the 5′ end of exon 3 code for the first clip domain, whereas the rest of exon 3 and the 5′ end of exon 4 encode the second clip domain. The remaining part of exon 4 covers the linker sequence and the first two residues of the SP domain. Most of the catalytic domain is encoded by exon 5, exon 6 and the 5′ end of exon 7. The other part of exon 7 (1040 bp) corresponds to the 3′ untranslated region in the PAP-3 cDNA, which contains the polyadenylation signal (AATAAA) ten nucleotides before the poly(A) tail.

Figure 2.

Nucleotide sequence and structural features of the Manduca sexta PAP-3 gene. Nucleotides in the 5′ flanking region are assigned negative numbers. Nucleotide 1 is assigned based on the primer extension results (Fig. 3). Exon sequences are underlined with the encoded amino acid sequences listed below translated exons, using a one-letter code under the second nucleotide of each codon. While some regions of the intron sequences (marked ‘–’) are not shown, their sizes and positions are indicated. GATA boxes (6-nucleotide) and ISRE sites (13-nucleotide), bold and double underlined; NF-κB motifs (10-nucleotide) and EcRE (15-nucleotide), bold and single underlined; TATA box (6-nucleotide), bold italic. The mismatches in these motifs are in lower case. Sequences of the 3′ end of λ7 (4.32 kb), gap, and the 5′ end of λ3 (4.51 kb) are not shown. The SacI and EcoRI recognition sites (in bold) are numbered consecutively as 2336–2341 and 2342–2347, respectively. Single nucleotide polymorphic sites are in bold italic on the DNA sequence. Among them, nonsynonymous substitutions are further indicated on the affected amino acid residues (bold and underlined). The Cys residues in the clip domain and the catalytic residues in the SP domain are indicated with ‘+’ and ‘#’, respectively.

Introns 1, 2 and 3 are > 10, 2.6 and 0.8 kb in length, respectively, whereas the average size of introns 4–7 is only 0.4 kb. We compared the 5′ and 3′ ends of these introns and identified the consensus sequence: 5′-GTR(/G)W(/A)D(/G)K(/T) and B(/T)TY(/T)BCAG-3′, where the nucleotide in each set of parentheses appears in five to six out of the seven sequences at that position.

Transcription initiation, sequence variations and copy number

We determined the transcriptional initiation site of the PAP-3 gene by primer extension. After annealing with RNA from the fat bodies of bacteria-induced larvae, the primer (derived from nucleotides 28–56 of the PAP-3 coding region) was extended by ninety-six nucleotides by reverse transcriptase (Fig. 3). Therefore, the RNA synthesis started at an A (nucleotide +1), and there was a TCAGT sequence at nucleotides −2 to +3. This motif is typically present within ten nucleotides either side of the transcription initiation site in arthropod genes (Cherbas & Cherbas, 1993). We did not identify a TATA or Goldberg-Hogness box around the ‘-30 region’. Nevertheless, a perfect TATA sequence (TATAAA) was present at nucleotides −94 to −89 (Fig. 2).

Figure 3.

Determination of the transcription initiation site in the Manduca sexta PAP-3 gene. A primer, complementary to nucleotides 28–56 of the PAP-3 coding region near the 3′ end of exon 1, was terminally labelled with γ-³²P-dATP and annealed to the total RNA from fat bodies from bacteria-injected larvae (15 µg). After annealing to RNA, the primer was extended with MMLV reverse transcriptase. The set of sequencing reactions (ACGT) on the left of the primer extension lane for use as a sizing ladder was from dideoxynucleotide sequencing of single-stranded M13 mp18 DNA using −40 primer. The arrow indicates the 125 bp extension product from fat body RNA isolated from the induced larvae.

The exons were nearly identical in sequence to the PAP-3 cDNA clone isolated from the bacteria-induced fat body library (Jiang et al., 2003b). There are nine nucleotide differences in the coding region, most likely caused by allelic variations. Two of these give rise to amino acid residue changes (T¹³⁷ to S¹³⁷, Y³⁸³ to F³⁸³).

Southern blot analysis was used to estimate the copy number of the PAP-3 gene in the M. sexta genome (Fig. 4). The BamHI digest (Fig. 4, lane 1) contained a 6 kb fragment that hybridized with the cDNA probe before exon 4 (Fig. 1A). The > 12 kb band with higher radioactivity probably included exons 4–8. A 2.3 kb HindIII fragment (Fig. 4, lane 2) matched the sequence between exon 4 and the 5′ end of exon 7 (Fig. 1) and produced a strong hybridization signal. The other three bands (> 10 kb) probably included exons 1, 2–3 and 7–8. With two recognition sites in the PAP-3 gene (Fig. 1A), SacI yielded three high molecular weight bands that did not separate well (Fig. 4, lane 3). In contrast, XbaI generated a much smaller band at 3.1 kb (Fig. 4, lane 4), probably corresponding to exons 7 and 8. The other part of the PAP-3 gene migrated to the ~11 kb position and hybridized with the probe much more strongly. EcoRI digestion generated two strong bands at 2.4 and > 10 kb and two weak bands at 2.7 and ~8 kb (Fig. 4, lane 5). Finally, SacI and XbaI double digestion generated at least two bands (Fig. 4, lane 6). The lower one has the same mobility as the 3.1 kb XbaI fragment. The upper one (3.6 kb) strongly reacted with the cDNA probe in the region corresponding to exons 3–6. Longer exposure revealed three other bands at 5.5, 7 and 12 kb (data not shown). In summary, the number, size and intensity of these radioactive restriction fragments agree well with those predicted from the restriction enzyme map of the cloned gene (Fig. 1A). There is only one copy of the PAP-3 gene in the M. sexta genome and the sequence variations (Fig. 2) are likely due to allelic differences in the insects used for library constructions.

Figure 4.

Southern blot analysis of Manduca sexta genomic DNA using ³²P-labelled PAP-3 cDNA. Samples of the genomic DNA (15 µg) were digested with BamHI (lane 1), HindIII (lane 2), SacI (lane 3); XbaI (lane 4), EcoRI (lane 5), SacI and XbaI (lane 6). After separation by agarose gel electrophoresis and transfer to a nitrocellulose membrane, the DNA fragments were hybridized with the full-length cDNA probe. The positions and sizes of the molecular markers are indicated on the right.

Putative regulatory elements

In order to identify putative regulatory elements, we searched the 1.8 kb upstream sequence of the PAP-3 gene by computer analysis (Table 1). Nine NF-κB motifs are present in this region, four on the plus strand and five on the minus strand. The motifs starting at nucleotides −1018 and −427 match nine out of ten positions in the NF-κB consensus sequence, originally derived from Drosophila melanogaster (Hultmark, 1993). A perfect match is located on the minus strand from −1756 to −1765. There are five interferon-stimulated response elements (ISREs) beginning at nucleotides −1724, −944, −292, −213 and −193. In comparison with the thirteen-base mammalian ISRE consensus, all these sequences contain two mismatches. An ISRE was identified in the Anopheles gambiae prophenoloxidase gene-1 (Ahmed et al., 1999).

Table 1.

Sequence analysis of the 5′ flanking region of the Manduca sexta PAP-3 gene

Motif name and consensus*	Sequence found**	Strand	Location	Matched
NF-κB: GGGRAYYYYY	GGGGACTTCT	−	−1765	10/10
	GGcAATTCCC	+	−1018	9/10
	GcGGAgTCTC	+	−922	8/10
	aGGAATTCaT	−	−760	8/10
	GGtAACTTgT	−	−694	8/10
	GaGGAaTCTT	+	−683	8/10
	GGaGATTCTC	+	−427	9/10
	tGaAACTTTT	−	−196	8/10
	GGtcACTCTT	−	−35	8/10
GATA: WGATAA	TGATAA	+	−1713	6/6
	AGATAA	+	−1137	6/6
	TGATAA	+	−536	6/6
	AGATAA	+	−145	6/6
ISRE: GGAAANNGAAANN	GGtAAAAaAAATG	+	−1724	11/13
	GtAAACTGAtACA	−	−944	11/13
	tGAAtTTGAAATT	−	−292	11/13
	GGgcACGGAAAAG	+	−213	11/13
	aGAAgGAGAAACT	+	−193	11/13
EcRE: RRGKTCANTGAMCYY	AGtTTgAATGcACTT	+	−804	12/15
	AAGaTtATTGcACTC	+	−291	12/15
	AGGTTCAGTGcgCgT	+	−6	12/15

^*R: A,G; Y: C,T; W: A,T; K: G,T; M: A,C. W: A,T; N: A,C,G,T.

^**the nucleotides in lower case do not match with the consensus.

ISRE, interferon-stimulated response elements; EcRE, ecdysone response element.

Four GATA boxes exist at positions −1713, −1137, −536 and −145 on the plus strand. We have also identified three ecdysone response elements (EcREs) on the same strand at positions −804, −291 and −6. While the GATA boxes perfectly match the consensus (WGATAA), three of the fifteen nucleotides in the putative EcREs are not consistent with the motif (RRGKTCANTGAMCY) (Cherbas & Cherbas, 1996).

A 151 bp segment in the 5′ upstream region is 85% identical in nucleotide sequence to an intron region in a Bombyx mori ABC transporter gene (Fig. 5A). It is unclear whether or not these sequences represent a lepidopteran repetitive sequence with a regulatory function. Using BlastN, we searched the GenBank database with the PAP-3 intron sequences and identified several high-score matches from M. sexta and other insects. For instance, a 68 bp fragment between exons 3 and 4 is 85–93% similar to intron sequences from the M. sexta juvenile hormone binding protein, eclosion hormone, serpin-1 and arylphorin genes (Fig. 5B). The significance of these similar sequences is not understood.

Figure 5.

Multiple sequence comparison. (A) A 151 bp sequence in the PAP-3 gene aligned with a Bombyx mori ABC transporter (ABC-T, AF245662) gene segment. Identical positions are indicated with *. (B) A highly similar intron sequence in Manduca sexta PAP-3, juvenile hormone-binding protein (JHBP), eclosion hormone (EH), serpin-1 and arylphorin genes. Based on a BlastN search, similar sequences in the PAP-3 (nucleotides 6150–6084), JHBP (nucleotides 6084–6150, AF527636), EH (nucleotides 3414–3480, M27808), serpin-1 (nucleotides 6052–6117, U58361) and arylphorin (nucleotides 2835–2900, M28394) genes were retrieved from GenBank and aligned using ClustalW program (Thompson et al., 1994). The JHBP, serpin-1 and arylphorin sequences are shown as reverse complements of the plus strand DNA sequence. *, Position identical in all five sequences; +, position identical in four of the five sequences.

Temporal and spatial regulation of transcription

To understand the regulation of PAP-3 transcription, we examined its mRNA levels in different tissues and development stages (Fig. 6). Semi-quantitative RT-PCR using PAP-3 specific primers resulted in PCR products of the expected sizes. Due to a 75 bp insert (Jiang et al., 2003b and data not shown), most of the PAP-3 transcripts in the midgut gave rise to a 438 bp PCR product that migrated slower than the 363 bp one from the fat body, muscle, integument, salivary gland, and nerve tissue (Fig. 6A). Both transcripts were detected in haemocytes, the trachea and Malpighian tubules. There was no major change in relative abundances of the mRNAs between feeding and wandering larvae. Relative intensities indicated that PAP-3 transcripts were present at a low level in the fat body and haemocytes of the naive larvae and, after a bacterial challenge, the mRNA levels increased considerably (data not shown). This result, consistent with our previous RT-PCR data (Jiang et al., 2003b), and the existence of immune-responsive elements in the PAP-3 gene, validated our methods for estimating the mRNA levels. It also agreed well with data from an immunoblot analysis of the haemolymph samples (Jiang et al., 2003b) and immunohistochemical analysis of haemocytes (Fig. 7).

Figure 6.

Profiling of PAP-3 gene transcription by RT-PCR. (A) Expression of the PAP-3 gene in different tissues from day 3 fifth instar larvae or day 3 wandering larvae. N, nerve tissue; S, salivary gland; Mt, Malpighian tubule; T, trachea; Mg, midgut; H, haemocytes; F, fat body; I, integument; Mu, muscle; l, 438 bp, long PCR product with the 75 bp insert; s, 363 bp, short PCR product; l/s, a mixture of l and s. Two forms of the PAP-3 cDNA clones (Jiang et al., 2003b) were used as templates to amplify l and s. (B) changes of PAP-3 mRNA level in fat bodies from Manduca sexta at different developmental stages. M. sexta ribosomal protein S3 (rpS3) transcripts were normalized in all these samples.

Figure 7.

Immunolocalization of PAP-3 protein in granular cells from the bacteria-injected larvae. The same microscopic field of fixed haemocytes was photographed using fluorescence microscopy. The haemocytes were reacted with a polyclonal antiserum to PAP-3 and FITC-labelled secondary antibodies. (A) Haemocytes from uninjected control larvae. (B) Haemocytes collected at 24 h after injection of Micrococcus luteus. Scale bar = 50 µm. The bright spots in the granular haemocytes represent PAP-3 synthesized and stored after the immune challenge.

Considering its roles as a centre for intermediary metabolism and in haemolymph protein synthesis in insects, we closely monitored the PAP-3 mRNA level in the fat body at different developmental stages. The PAP-3 gene was transcribed at a low level in this tissue throughout the last two larval feeding stages. There was a small, daily increase in the first five days in the fourth instars and a decrease thereafter (Fig. 6B). The mRNA was almost undetectable after day 3 of the fifth instar. However, after the insects entered the wandering stage, PAP-3 mRNA level largely increased and then gradually decreased. The transcripts became abundant again in the early pupal stages and reduced subsequently. The mRNA was still detectable in the fat bodies from late wandering larvae, late pupae and adults.

Discussion

Although it is not well studied, transcriptional regulation of immune genes in lepidopteran insects is generally considered to be similar to that in Drosophila. In Drosophila, transcription factors of the Rel family (i.e. Dorsal, Dif and Relish) control the expression of many immune genes (Engstrom, 1998; Harshman & James, 1998). They form homodimers or heterodimers, bind to their recognition sequences (NF-κB motifs), and induce the synthesis of antimicrobial peptides (e.g. diptericin, cecropin and drosomycin) and other acute-phase proteins (Hoffmann, 2003). The nuclear translocation of these transcription factors are regulated by the Toll and Imd pathways. Pathogen recognition directly or indirectly stimulates these intracellular signalling pathways. For example, Drosophila PGRP-SA, Persephone (a clip-domain SP), and cleaved spätzle are components of an extracellular signalling pathway that activates the Toll receptor (Michel et al., 2001; Choe et al., 2002; Ligoxygakis et al., 2002; Ramet et al., 2002; Takehana et al., 2002). NF-κB motifs and their binding proteins exist in other insects. Cecropia immune-responsive factor was identified in nuclear extracts from induced Hyalophora cecropia pupae (Sun & Faye, 1992). Responsive elements similar to the Drosophila NF-κB motif (GGGRAYYYYY) are present in the 5′ flanking region of the M. sexta haemolin and lysozyme genes (Wang et al., 1995; Mulnix & Dunn, 1994).

In this work, we elucidated the structure of M. sexta PAP-3, an immune-responsive SP involved in melanization and other defence mechanisms. In particular, nine NF-κB motifs were identified, three of which closely resemble the Drosophila consensus (Table 1). Like other immune responsive elements (e.g. ISREs, GATA boxes), the role of these putative recognition sequences has to be further investigated.

Many arthropod SPs and SPHs involved in immunity and development contain clip domains (Jiang & Kanost, 2000). One of the five Drosophila SPs/SPHs has such a regulatory unit (Ross et al., 2003). So far, the structure and function of clip domains are poorly understood and their evolutionary history has not been closely examined. We found that the Drosophila SPs/SPHs with multiple clip domains are different in domain subgroups and linker sizes from B. mori proPO-activating enzyme (PPAE), and M. sexta PAP-2 and PAP-3 (Satoh et al., 1999; Jiang et al., 2003a,b). Perhaps PPAE-like SPs evolved after Lepidoptera and Diptera diverged from their common ancestor ~330 Ma (Gaunt & Miles, 2002), and there was a subsequent expansion of dual clip-domain SPs in lepidopteran insects. In this study, we discovered that each clip domain in PAP-3 did not evolve as a complete exon unit(s). While such an exon-domain relationship impedes domain shuffling/duplication, mutations at the splicing junctions may have contributed to the sequence diversity within the two domains. Exon duplication is responsible for the domain organization of Drosophila masquerade, whose clip domains 1, 2, 3–4 and 5 are separately encoded by four exons (http://www.flybase.org/).

The PAP-3 mRNA levels varied in different tissues of M. sexta. In the fifth instar larvae, PAP-3 transcripts were more abundant in the haemocytes and trachea than in the fat body. After the larvae entered the wandering stage, however, the fat body became a major source of PAP-3 expression. Additionally, the size of the PCR product was 75 bp smaller in the fat body than that in some other tissues (Fig. 6A, Jiang et al., 2003b). At this moment, we do not understand the significance or mechanism of such tissue-specific regulation of PAP-3 transcription/splicing. In fact, the 75 bp sequence was not identified in the sequenced regions of λ7 and λ3.

The transcription profiles of PAP-3 and PAP-1 (Zou, unpubl. data) are somewhat similar in the fat body at different developmental stages. Their up-regulation during the wandering stage (days 1–4) and down-regulation in the middle-to-late pupae stage suggest that hormonal signals might play a role the transcriptional control of PAP-1 and PAP-3 expression. Indeed, upon treatment with 20-hydroxyecdysone, the PAP-1 mRNA level in the cultured larval fat body reduced in a concentration-dependent manner (Z. Zou, unpubl. data). We identified three ecdysone responsive elements in both genes. Nonetheless, we did not detect a significant change in PAP-3 mRNA level after the cultured fat body was incubated with the moulting hormone (Z. Zou & H. Jiang, data not shown). The hormonal regulation of immune proteins is worth exploring in the future.

Experimental procedures

Insect rearing and bacterial injection

M. sexta larvae were hatched from eggs (Carolina Biological Supply Co, Burlington, NC) and reared according to Dunn & Drake (1983). For bacterial induction, day 2 fifth instar larvae were injected with formalin-killed Escherichia coli XL1-blue (2 × 10⁸ cells/larva). Haemolymph and fat body samples were collected 24 h later for RNA preparation.

Library screening, subcloning, sequencing and computer analysis

To screen the genomic library, full-length or PCR-derived PAP-3 cDNA was labelled with α-³²P-dCTP using Multiprime DNA Labelling System (Amersham Pharmacia Biotech, Bucks, UK). The intact cDNA was retrieved from the PAP-3 clone by restriction enzyme digestion, whereas the 5′ PCR fragment (115 bp) was amplified using primers j662 (5′-CGA CGC TGA GGT AAC ACG T-3′, nucleotides 2–20) and j663 (5′-TTG TCC ACT AAC AAA GCA G-3′, reverse complement of nucleotides 98–116). A M. sexta genomic library, generously provided by Dr Yucheng Zhu at the Southern Insect Management Research Unit (USDA-ARS), was screened with the cDNA probes at 5 × 10⁵ cpm/ml by a standard protocol (Sambrook & Russell, 2001). Positive plaques were purified to homogeneity by repeated screening, and λ DNA was prepared from the amplified bacteriophages using Wizard Lambda Preps DNA Purification System (Promega, Madison, WI). Restriction maps were determined by single and double digestions of the DNA samples with EcoRI, SacI, and XbaI. Following agarose gel electrophoresis and capillary transfer, the separated digestion products on the nitrocellulose membrane were hybridized with the cDNA probes and visualized by autoradiography. PAP-3 gene fragments were subcloned into the same sites in pBluescript-(KS). The resulting recombinant plasmids were sequenced using a BigDye v2.0 Terminator Cycle Sequencing Ready Reaction Kit (PE Applied Biosystems, Foster City, CA). Sequence editing, assembly, and analysis were performed using MacVector Sequence Analysis Software (Version 6.5, Oxford Molecular Ltd, Oxford, UK).

Determination of the PAP-3 transcription initiation site

Primer extension was carried out as described previously (Sambrook & Russell, 2001) in order to determine the transcription starting position. An oligonucleotide (j396, 5′-CCA CTA ACA AAG CAG AAA TAC GAA GCT AA-3′), corresponding to the reverse complement of nucleotides 28–56 of the PAP-3 coding region, was terminally labelled with γ-³²P-ATP using T4 polynucleotide kinase (Promega). The primer (1.0 × 10⁵ cpm/ml) was added to 15 µg total RNA, incubated at 60 °C for 15 min, and then slowly cooled to below 30 °C in 45 min. Annealed primer-RNA complexes were extended with MMLV reverse transcriptase (200 U/µl, 1 µl) for 1 h at 42 °C. The extension products as well as the ³⁵S-labelled DNA size standards were separated on 8% polyacrylamide gels containing 7 m urea and visualized by autoradiography.

Southern blot analysis

M. sexta genomic DNA was extracted from a single fifth instar larva. Aliquots of the DNA sample (15 µg each) were incubated with restriction enzymes (BamHI, EcoRI, HindIII, SacI, XbaI, SacI-XbaI) at 37 °C for 6 h. After electrophoretic separation on a 1% agarose gel and capillary transfer on to GeneScreen-Plus nitrocellulose membrane (NEN Life Science Products, Boston, MA), the digested genomic fragments on the membrane were hybridized with α-³²P-labelled full-length PAP-3 cDNA.

Immunocytochemistry

At 1 and 24 h after day 2 fifth instar larvae were injected with E. coli XL1-blue (1 × 10⁷/larva), haemocytes were collected, washed and fixed on multiwell slides as described by Willott et al. (1995). These cells were incubated with 1 : 1000 diluted PAP-3 antiserum overnight at 4 °C. After removing excess antibodies with Tris-buffered saline, the haemocytes were reacted with 1 : 500 diluted FITC-conjugated goat-anti-rabbit IgG (Bio-Rad, Hercules, CA) at room temperature for 1 h. Following a washing step, cells were observed and photographed with phase contrast or fluorescence optics under an Olympus BH-2 microscope.

RNA extraction and RT-PCR analysis

RNA samples were extracted from various tissues of M. sexta at different developmental stages (see Fig. 5 legend for details), using Micro-to-Midi Total RNA purification system (Invitrogen Life Technologies, Carlsbad, CA). Similarly, fat body and haemocyte RNA samples were isolated from naive and bacteria-challenged M. sexta larvae. First-strand cDNA synthesis was performed using 2–4 µg RNA, 10 pmol oligo(dT)₁₇, and 200 U MMLV reverse transcriptase (Invitrogen Life Technologies) at 37 °C for 1 h. M. sexta ribosomal protein S3 transcripts were used as an internal standard to control the template amount in a preliminary PCR experiment. Relative levels of PAP-3 cDNA in the normalized samples were determined by semiquantitative PCR using j699 (5′-CGT GTT GTT ATT AGC TTC GTA TTT CT-3′) and j700 (5′-CAT CCC CCC AGC CTC TAC-3′ for PAP-3′. The thermal cycling conditions were: 94 °C for 30 s; 50 °C for 30 s; 72 °C for 60 s. PCR cycle numbers were empirically chosen to show comparable band intensity and avoid reaction saturation. After electrophoretic separation on a 1.3% agarose gel, intensities of the PCR products were quantified and compared using Digital Science 1D Gel Analysis Software (Kodak, Rochester, NY).

GenBank accession numbers

The nucleotide sequences reported in this paper have been submitted to the GenBank™/EBI Data Bank with the accession numbers AY789466 and AY789467.