A Hemocyte-like Cell Line Established from the Malaria VectorAnopheles gambiae Expresses Six Prophenoloxidase Genes*

Cell lines from the malaria vectorAnopheles gambiae have been established as a tool for the study of the mosquito innate immune system in vitro. Here, we describe the first continuous insect cell line that produces prophenoloxidase (PPO). This cell line (4a-3B) expresses constitutively six PPO genes, three of which are novel (PPO4, PPO5, andPPO6). The PPO genes show distinct temporal expression profiles in the intact mosquito, spanning stages from the embryo to the adult in an overlapping manner. Transient induction of larva-specific PPO genes in blood-fed adult females suggests that the developmental hormone 20-hydroxyecdysone may be involved in PPO gene regulation. Indeed, exposure of 4a-3B cells to 20-hydroxyecdysone in culture results in induction of those PPO genes that are mainly expressed in early developmental stages, and repression of PPO5, which is preferentially expressed at the adult stage. The cell line shows bacteria-induced immune transcripts that encode defensin and Gram-negative bacteria-binding protein, but no induction of PPO transcripts. This cell line most likely derives from a hemocyte lineage, and represents an appropriate in vitro model for the study of the humoral and cellular immune defenses of A. gambiae.

Anopheles gambiae mosquitoes are the principle vectors of Plasmodium falciparum, the parasite causing the most severe form of human malaria. However, within the A. gambiae complex mosquitoes may differ in their efficiency of transmitting malaria (1). Transmission requires completion of a complex sporogonic cycle, which takes place over 2 weeks within the mosquito and can be aborted by innate immune responses, such as the encapsulation of early oocysts soon after invasion of the midgut (2). Genes responsible for such refractory phenotypes would be candidates for inclusion in a vector control strategy based on genetically manipulated mosquitoes (3,4).
Vertebrate immune defense has been attributed to two general systems, innate and adaptive immunity, which are interconnected (5). In contrast, insect immune defense lacks the adaptive component and therefore an antibody-mediated immune response. Instead, it fully relies on innate immune mechanisms, such as the inducible synthesis of antimicrobial peptides, and the coagulation and melanization cascades (6,7). The ongoing exploration of immune mechanisms in A. gambiae is hampered by the small size of this insect, limited knowledge of its genetics, and the absence of an efficient method for germline transformation. Insect cell lines were previously shown to exhibit immune properties, including the inducible synthesis of antibacterial peptides (8 -10). Therefore, we have taken the approach of establishing A. gambiae cell lines as an in vitro system to analyze the immune mechanisms of the mosquito. We have reported that one such line expresses in an inducible manner a panel of immune marker genes (11).
In the course of establishing a larger variety of immuneresponsive A. gambiae cell lines, we identified some that secrete prophenoloxidase (PPO) 1 in the culture medium and adopted one such line, 4a-3B, as a standard. No continuous cell lines of this type have been reported previously in any insect. Phenoloxidases (POs) are enzymes that serve multiple tasks in insects, including cuticle pigmentation and sclerotization, wound healing and the melanotic encapsulation of protozoan (plasmodia) and metazoan (filaria) pathogens (12,13). The PPO zymogens are known to be synthesized in hemocytes, the cellular component of the insect immune defense, and are activated by trypsin-like serine protease components of the PPO activation system, also called the phenoloxidase cascade (see under "Discussion"). The PO cascade is inducible in vitro by microbial cell wall constituents and is considered part of the insect non-self recognition system (14,15). PO catalyzes the hydroxylation of tyrosine to dihydroxyphenylalanine and the oxidation of dihydroxyphenylalanine and dopamine to their respective quinones, mediators of protein cross-linking, and precursors of the melanin polymer that is a central component of the melanotic capsule (16,17). In addition, melanin synthesis leads to local increase of toxic quinones and free radicals (16). Molecular cloning has yielded 12 PPO sequences to date, 1 crustacean, 6 lepidopteran, and 5 from diptera. All the sequences are closely related and show homology to hemocyanin, the arthropod oxygen transporter, which is functionally replaced in insects by the tracheal system. Using degenerate oligonucleotides and cDNA from the PPO producing A. gambiae cell lines as template, we amplified a PCR product consisting of a mixture of six distinct PO sequences. Three of these corresponded to PPO genes found in studies then in press (18,19); the other three were novel and were designated PPO4, PPO5, and PPO6. Corresponding full-length cDNA clones were isolated and their sequences were determined. Reverse transcrip-tion PCR (RT-PCR) expression analysis using specific primer pairs confirmed that all six PPO genes are expressed constitutively in the same cell line but show distinct expression profiles during mosquito development. Transcription of the genes that are larva-specific is transiently up-regulated in the adult female mosquito following a blood meal. The same genes are up-regulated when our standard cell line is exposed to 20-HOE. Interestingly, the only adult-specific PPO gene, PPO5, shows a converse behavior: it is repressed upon a blood meal in the mosquito and upon 20-HOE treatment of the cell line. Thus, this cell line is a useful in vitro model for the study of hormoneregulated as well as immune-regulated gene expression, phenomena that are frequently connected in insects.

EXPERIMENTAL PROCEDURES
Mosquito Rearing-The A. gambiae strains 4a r/r, L3-5, and Suakoko 2La were maintained at 27°C and 75% relative humidity on a 12-h light-dark cycle. Larvae were kept in deionized water supplemented with 0.1% marine salt and fed with dry cat food. Adult mosquitoes had access to 20% sucrose solution ad libidum and were blood-fed on a volunteer's arm (H.-M. M.).
Mosquito Perfusions-Schneider medium (Sigma) supplemented with 10% FCS, 20 mM EGTA, and 10 mM thiourea was used to perfuse 1-day-old female 4a r/r mosquitoes. No further protease inhibitors were added, as FCS alone was found to be an efficient inhibitor of the PO cascade otherwise triggered by exposure of the hemolymph to air (blackening). Mosquitoes were chilled on ice and transferred on a 8-well glass slide containing 5 l of perfusion medium per well, and a cut was made at the second-to-last abdominal segment. A fine glass needle containing 5 l was inserted in the neck, and the perfusion medium, injected by air pressure, was collected in the well. Cells were allowed to settle for 2 min, and then 10 l of 4% formaldehyde in PBS (130 mM NaCl, 7 mM Na 2 HPO4, 3 mM NaH 2 PO4, pH 7.2) were added. After 20 min of fixation at ambient temperature, samples were processed as described below for immunofluorescence.
Hemolymph Collection-Pupae or decapitated female mosquitoes of the 4a r/r strain were placed in a drop of 5-l PBS containing 2 mg/ml of the serine protease inhibitor Pefabloc (Boehringer Mannheim) and 20 mM EGTA. After cutting apart thorax and abdomen, the body parts and the fluid were transferred in a vial containing 100 l of the same buffer, prechilled on ice. After the dissection of 10 animals, thoraces and abdomens were gently vortexed. The fluid phase was recovered and mixed with an equal volume of 2-fold concentrated reducing SDSpolyacrylamide gel electrophoresis sample buffer. Samples were boiled for 5 min and stored at Ϫ20°C.
Establishment and Maintenance of A. gambiae Cell Lines-Cell lines were established according to the method of Pudney et al. (20) using neonate A. gambiae larvae as starting material. Repeated agitation starting 36 h after egg deposition led to synchronous hatching 6 -7 h later; only larvae hatched within 1 h were used for primary cultures. Large scale primary cultures were set up in 50-ml tissue culture flasks (Falcon 3014) in a volume of 5 ml using about 1000 -3000 larvae, each cut, on average, in 3-4 pieces, followed by a trypsin treatment (20). Two alternative media were used, Schneider medium (Sigma) or MK/VPmedium (20), supplemented with 20% heat-inactivated FCS, penicillinstreptomycin (300 units/ml-300 g/ml), gentamycin (100 g/ml), and Fungizone (2.5 g/ml); the cultures were kept at 27°C. Once per week, half of the culture medium was replaced by fresh medium containing 20% FCS and penicillin-streptomycin (100 units/ml-100 g/ml). After subculture 10, the FCS content in the medium was gradually decreased to 10%. All the lines are kept continuously in Schneider medium.
Preparation of Serum-free Cell Growth-conditioned Medium-Cells were seeded in 75 cm 2 tissue culture flasks (Greiner) containing 25 ml of Schneider medium/10% FCS. After having reached half confluency, cell layers were rinsed several times with protein-free medium (Insect-Xpress, BioWhittaker), 15 ml of protein-free medium was added, and cultures were grown for 2 more weeks at 27°C. Cell growth was retarded in Insect-Xpress medium, although without any sign of deterioration of the cells. The conditioned medium was transferred in a dialysis bag and 5-fold concentrated on polyethylene glycol-35,000 powder. After dialysis against 10 mM phosphate buffer, pH 6.0, preparations were stored at Ϫ20°C.
Immune and Hormonal Stimulation of A. gambiae Cell Lines-Cell cultures were immune-stimulated by adding to a confluent culture for 4 h a mixture of heat-inactivated Escherichia coli 1106 and Micrococcus luteus A270, 1000 bacteria per mosquito cell. 20-Hydroxyecdysone (Sig-ma) in ethanol was added yielding a final concentration of 0.5 g/ml (1 M), with the ethanol concentration below 0.1%. For the wash-out experiment, 20-HOE-containing medium was aspirated, and cell layers were rinsed several times with Schneider medium before conditioned 20-HOE-free medium, taken from a culture of identical age, was added.
Reverse Transcription-PCR Analysis-Unless otherwise indicated, all molecular techniques were performed as described (21). Total RNA from cell lines and the various mosquito developmental stages was purified using the RNaid PLUS kit (bio101) according to the manufacturer's instructions. First strand cDNA was synthesized using Moloney murine leukemia virus-reverse transcriptase and supplied solutions (Life Technologies, Inc.) with 2 g total RNA in a reaction volume of 15 l. cDNA synthesis was primed using the equivalent of 3 l of oligo(dT) 25  PCR were subjected to electrophoresis on 6% nondenaturing polyacrylamide gels. Dried gels were exposed on Kodak X-Omat AR films with Ilford Fast Tungstate screens. The ribosomal gene S7 (22) served as internal control and was used for normalization as follows: the S7 signals resulting from the initial PCRs of a given set of cDNA samples were quantified with a PhosphorImager (Molecular Dynamics), the S7 values were equalized, and the template amounts were adjusted accordingly. The S7 signals obtained from the second round of PCRs were again quantified by phosphorimaging and normalized against S7 in order to fine-tune the amount of sample loaded on the final gel, resulting in deviations between any S7 signal in a given experiment of no more than Ϯ5%. The resulting lengths of the corresponding PCR products were as follows: S7, 460 bp; defensin, 404 bp; GN BP, 511 bp; PPO1, 358 bp; PPO2, 344 bp; PPO3, 343 bp; PPO4, 246 bp; PPO5, 346 bp; and PPO6, 349 bp. Except for PPO2, the primer combinations denoted above do not work on genomic DNA, e.g. on bacterial artificial chromosomes (BACs), as the polymorphic sequence block chosen for the design of the antisense primers turned out to span an intron/exon border that occurs in PPO1, 2 which is obviously conserved in all PPO genes except PPO2.
Screening of A. gambiae cDNA Libraries-Two lambda ZAP cDNA libraries were screened, one was constructed using mRNA isolated from fourth instar larvae of A. gambiae G3 (23) and one using mRNA isolated from abdomens of adult female A. gambiae Suakoko 2La mosquitoes (24). Both libraries were constructed using the ZAP Express system (Stratagene). As probe, the gel-purified PCR product amplified from cDNA of cell line 4a-3B using the primers PO-CuA and PO-CuB was enriched with gel-purified PPO4, PPO5, and PPO6 fragments and 32 Plabeled via PCR. From a total of 700,000 plaques screened, 200,000 larval and 500,000 abdominal, approximately 60 positive signals were obtained in each library. Before starting the plaque purification procedure, the supernatants of the positive phage plaque zones of the primary screen were PCR-typed using the set of six specific PPO primer pairs. The identified PPO4, PPO5, and PPO6 samples were plaque purified via hybridization, followed by PCR typing.
Cloning and Sequencing-The 0.6-kb PCR product amplified from cell line cDNA using the degenerate primers PO-CuA (5Ј-CAICA(TC)T-GGCA(TC)TGGCA(TC)CTIGT(GATC)TA(TC)CC-3Ј) and PO-CuB (5Ј-CITGCCAICG(AG)TA(AG)AAIA(ACT)(CG)GG(AG)TCIC(TG)CAT-3Ј) (1 min at 94°C, 1 min at 50°C, and 1 min at 72°C, 25 cycles) was gel-purified and cloned using the TA Cloning Kit (Invitrogen). Plasmid clones containing PPO cDNA inserts were obtained from the purified lambda ZAP phage clones using the in vivo excision method according to the manufacturer's instructions (Stratagene). The sequences of the cloned insertions were determined by the EMBL sequencing facility. The RGD sequence in PPO6 was confirmed by sequencing the corresponding region of four independent cDNA clones; in PPO4 and PPO5, the corresponding region of a second cDNA clone was sequenced.
In Situ Hybridization to Polytene Chromosomes-Hybridizations were essentially performed as described (25), using semi-gravid female mosquitoes of the Suakoko 2La strain, during the second gonotrophic cycle. DNA of whole BACs that were shown by PCR typing to contain PPO3 and PPO6 (BAC clone 27E19) or PPO4 and PPO5 (BAC clone 28C11) were used for in situ hybridization. In addition, in situ hybridizations were carried out using gel-purified insertions of cDNA clones corresponding to PPO4, PPO5, and PPO6.
Expression of PO Fragments in E. coli, Purification of Recombinant Proteins, and Antibody Production-The TA-plasmid vectors (Invitrogen) containing the PCR fragments amplified by the degenerate oligonucleotides PO-CuA and PO-CuB were EcoRI-digested, and the PO inserts were subcloned in the EcoRI site of the polylinker of the modified expression vector pDS56/RBSII,6xHis/E Ϫ (-2) (26), resulting in a fusion of six histidine residues at the amino termini. Expression of recombinant proteins in E. coli and purification by nickel-chelate affinity chromatography was performed as described previously (27). The purified PO proteins were insoluble and were stored at Ϫ20°C in 8 M urea, 0.1 M sodium phosphate, 0.01 M Tris-HCl, pH 8.0, at a concentration of 5 mg/ml. Two male rats were immunized with each construct using 100 g of purified protein and the Ribi Adjuvant System following the manufacturer's instructions (RIBI ImmunoChem Research). Rats were bled 1 week after the fourth booster injection.
Immunofluorescence-4a-3A and 4a-3B cells were seeded on round glass coverslips placed in 24-well tissue culture plates (Nunc) yielding half confluent cell layers. After 3 days of growth at 27°C the Schneider medium was removed, and the coverslips were washed once with PBS and then fixed in PBS/4% formaldehyde for 20 min. After fixative removal followed by one PBS wash, cells were exposed for 2 min to 0.2% Triton X-100 in PBS and then washed twice with PBS. Coverslips were blocked for 2 h with 1% bovine serum albumin in PBS and then incubated overnight at 4°C with anti-PPO4 serum or preimmune serum diluted 1:500 in 1% bovine serum albumin/PBS. After three PBS washes, samples were incubated for 1 h with a fluorescein isothiocyanate-conjugated anti-rat IgG antibody (The Jackson Laboratory) diluted 1:300 in 1% bovine serum albumin/PBS. After the first PBS wash, nuclei were counterstained with DAPI (Boehringer Mannheim) diluted 1:15,000 in PBS for 5 min, and three further PBS washes followed before mounting.

Isolation and Preliminary Characterization of Two A. gambiae Cell
Lines-In a previous report, we described the cell line Sua1B, which has immune-responsive properties, such as inducible expression of the antibacterial peptide defensin (11). In a further effort to establish A. gambiae cell lines with novel properties, primary cultures were set up starting from neonate larvae of three A. gambiae strains, Suakoko 2La, 4a r/r, and L3-5. Further variation was brought in by differing culture conditions, e.g. initial seeding density, medium, and splitting strategy. Some primary cultures gave rise to cell lines that developed relatively slowly and were finally identified serendipitously as PPO producers, when we observed spontaneous blackening of stored culture supernatants. Such lines were most frequently derived from the 4a r/r strain. The presence of PPO in the culture supernatant could be monitored directly by artificial activation on nitrocellulose (28). This rather simple test enabled us to follow PPO expression in all the lines during increasing subculture number. Whereas some primary cultures gradually lost the ability to produce PPO during establishment of the line, probably due to overgrowth by non-PPO producing fast growing cell types, others manifested the PPO phenotype stable, over 2 years (subculture Ͼ 60), until submission of this study for publication. We have chosen cell line 4a-3B as an example for our stable PPO producing cell lines and cell line 4a-3A as a PPO-negative control. These lines were derived from two primary cultures initiated from a single batch of minced neonate larvae of the A. gambiae strain 4a r/r. Both lines show inducible expression of certain immune markers (see below).
Identification of Three Novel PPO Genes-To verify the expression of PPO on the molecular level, we PCR-amplified RNA sequences from several PPO-positive lines using degenerate oligonucleotides designed according to two conserved sequence blocks within the PO copper binding sites CuA (HHWH-WHLVYP) and CuB (MRDP(F/V/I)FYRWH), respectively ( Fig.  1) (29). A PCR product of the expected size (approximately 600 bp) was obtained and cloned. Among 13 sequenced clones, we identified six distinct PPO sequences, which are different enough to represent distinct genes rather than polymorphism (see below). Subsequently, we designed primer pairs specific for each of the six PPO sequences (see under "Experimental Procedures"); analysis of 20 additional PCR-derived PPO clones with these specific primer pairs did not yield any evidence of an additional PPO gene: for every clone tested, one (and only one) of the six PPO primer pairs amplified a fragment of the expected size, thus confirming that the gene-specific primer pairs are diagnostic. All six PPOs were expressed in each of the six independent PPO-positive cell lines analyzed. 3 Three of the sequences matched the PPO1 gene published by Lee et al. (19) and the PPO2 (formerly PPO-p1) and PPO3 (formerly PPO-p2) sequences published by Jiang et al. (18). However, the three other PPO genes were novel and were named PPO4, PPO5, and PPO6. The mixed PCR products generated by the degenerate primers were enriched with these novel sequences, and were used as probe to screen two A. gambiae cDNA phage libraries: one constructed with mRNA of whole fourth instar larvae (23), and one with mRNA purified from abdomens of female mosquitoes (24). The screening yielded approximately 60 positive signals from each of the two libraries, and rescreening of a subset of the positives, using the six PPO gene-specific primer pairs, identified PPO clones that were evidently differentially represented: from the adult library 10 PPO5, 5 PPO6, and 1 PPO2 clone were identified, whereas from the larval library, 5 PPO2 and 4 PPO3 clones were identified. The original 120 positives were rescreened with the PPO4-specific primers, leading to the identification of four PPO4 clones from the larval and one from the adult library. At least three full-length cDNA clones were characterized for each of the novel PPO genes PPO4, PPO5, and PPO6. Fig. 1 shows the amino acid sequences of PPO4, PPO5, and PPO6 as deduced from their cDNA sequences, aligned with those of the previously identified A. gambiae PPO genes PPO1, PPO2, and PPO3. Like all the arthropod PPO sequences published so far, the mosquito PPOs do not encode a potential signal peptide sequence. An arginine-phenylalanine zymogen cleavage site (Fig. 1, vertical arrow) that has been experimentally determined in Bombyx mori (30) and Holotrichia diomphalia (31) is conserved in both sequence and position relative to the initiator methionine. A second cleavage site (Fig. 1,  arrowhead) was found in Musca domestica PPO (32), and the sequence encompassing that site (REE) is present in all six A. gambiae PPO sequences; furthermore, the amino-terminal sequence of the resulting activated PO (EEATVVPDG) is conserved with only one substitution in A. gambiae PPO1. Cleavage sites at similar positions were determined experimentally in H. diomphalia PPO (31) and Pacifastacus leniusculus PPO (33), located three amino acids before and five amino acids following the M. domestica cleavage site, respectively. There is evidence that the sequential cleavage of both sites in H. diomphalia PPO is necessary for PPO activation (31). The two copper binding sites including the 6 histidine residues that interact with the copper atoms (34) are conserved. PPO2 is the only A. gambiae PPO that has both of the crucial residues in a conserved motif (gCgwQhm) corresponding to a potential thiol ester bridge; this motif was noted in the Manduca sexta PPO (now designated PPO-p2) 3A). Both lines are immune-inducible, because bacterial challenge significantly increases the signals corresponding to A. gambiae defensin (35) and Gram-negative bacteria-binding protein (11), which are also induced under these conditions in the intact animal (11,35). However, cell line 4a-3A does not express any PPO gene, either before or after bacterial challenge. In contrast, all the PPO genes are expressed in line 4a-3B constitutively and are not induced by bacterial challenge. Indeed, further experiments have established that five of the genes, but not PPO5, show modest repression in the presence of bacteria. 3 The various PPO sequences correspond to unprocessed polypeptides with calculated molecular masses ranging between 78.1 and 79.2 kDa and isoelectric points ranging between pH 6.1 and pH 6.9. We examined PPO expression in these cell lines at the protein level (Fig. 3B), using antisera generated against recombinant PPO4 and PPO6 fragments expressed in E. coli. Our antisera are specific for PPO, as indicated by the total absence of signal in immunoblot of cell line 4a-3A, but detect three major bands in line 4a-3B, centered at approximately 80 kDa, in the size range expected for prophenoloxidases. The same bands, but in different proportions, were found endogenously in hemolymph collected from pupae and female mosquitoes, respectively. A fourth band migrating around 75 kDa, visible in the cell line and pupal lanes, may be derived from proteolytic degradation or processing. Antibodies raised against PPO purified from M. sexta (18) and B. mori (36) cross-reacted, but detected fewer bands than were seen with the anti-PPO6 serum. 3 Both the anti-PPO4 and the anti-PPO6 sera were equally suitable for immunofluorescence experiments, and gave results comparable with those obtained with antibodies against B. mori PPO. 3 Fig. 4 shows an immunofluorescence experiment using anti-PPO4. The antiserum (but not preimmune serum) stained specifically and intensely a subset of presumably differentiated cells (approximately 5%) in the PPO producing line 4a-3B. These cells showed variable morphology, permitting the detection of cell pairs, presumably reflecting relatively recent cell divisions. No staining was detected with the 4a-3A cell line.
Differential Expression of PPO Genes during Development-The expression of the six PPO genes at different developmental stages was assessed using RT-PCR, as shown in Fig. 5A. None of the genes were expressed in freshly laid eggs. Thereafter, different genes showed characteristic temporal profiles of expression. In broad terms and consistent with previous reports (18,19), PPO1, PPO2, and PPO3 were mostly expressed in embryos and larvae; their expression also persisted in young pupae and, in the case of PPO2, in older pupae and adults as well, albeit at a low level. These genes are not expressed coordinately. For example, in postembryonic life, PPO1 transcripts were most prominent in fourth instar larvae and early pupae, whereas PPO3 expression was maximal in the second instar and declines thereafter. Of the novel genes, PPO4 was expressed in late larvae, young pupae, and adults, somewhat reminiscent of the PPO2 pattern. PPO5 and PPO6 are the only genes that were not transcribed in embryos but mainly in the pupal and adult stages. Relative to the rest, PPO5 was the most adult-specific gene; PPO5 was found with highest frequency among the positive PPO cDNA clones identified in the adult library (see above).
Induced Transcription of Larval PPO Genes in Response to Blood Feeding-Blood feeding has a tremendous impact on mosquito physiology, for example, stimulating digestive enzyme production and resulting in a transient increase in ecdysteroid concentration. The hormone 20-HOE is produced by the ovaries following the blood meal and stimulates the production of vitellogenin and other serum proteins by the fat body cells (37). Fig. 5B shows that the four PPO genes that are mainly expressed in the embryonic and larval stages, PPO1 through PPO4, were induced after a blood meal. Their transcripts show maximal accumulation 24 h after feeding, a decline toward the end of blood digestion at 48 h, and approximate basal levels 2 days later. In clear contrast, PPO6 shows no significant change after blood feeding, and PPO5 is transiently repressed at 24 h at the mRNA level.

20-Hydroxyecdysone Effects on PPO Expression in Cell
Line 4a-3B-The transient induction of the larval PPOs following a blood meal suggested the possibility of hormonal regulation by the moulting hormone 20-HOE. The experiment shown in Fig.  6 confirmed this supposition and revealed that the effects of the hormone on the cell line are gene-specific, paralleling the effects of a blood meal in the intact mosquito (Fig. 5B). The larval specific PPO4 gene shows strong in vitro induction that began by 8 h, peaked at 16 h, and was still detectable at 40 h after hormone addition; PPO1, PPO2, and PPO3 yielded similar results. 3 PPO6 also showed induction that was unusually persistent. In contrast, the adult-specific PPO5 gene is repressed by hormone treatment as early as 2 h after hormone addition. The effects of the hormone were gradually reversed by removal of hormone from the medium ("washout experiment"). If exter-

FIG. 2. Evolutionary relationships and cytogenetic localization.
A, phylogenetic tree showing the relationship between the amino acid sequences of the A. gambiae PPOs and with the remaining arthropod PPOs identified in the diptera species A. subalbatus (45) and D. melanogaster (54); the lepidoptera species B. mori (30), M. sexta (29,46), and H. cunea (55); and the crayfish P. leniusculus (33). Numbers between the branches indicate the chromosomal divisions to which the A. gambiae PPO genes were mapped. B, in situ hybridization, to A. gambiae polytene chromosomes, of BAC clone 28C11 containing the PPO4 and PPO5 genes. BAC clone 27E19 containing the PPO3 and PPO6 genes yielded the same result; note the characteristic puff adjacent to division 21B (arrow) on the left arm of chromosome 2.
nal hormone was removed at 16 h, the PPO4 and PPO6 transcript levels at 24 and 40 h were lower than in the presence of the hormone; the effects of the washout are more prominent for PPO4 than for PPO6. In strong contrast, hormone washout results in strong superinduction of PPO5 by 40 h. Interestingly, full hormonal induction of PPO4 and maximal repression of PPO5 required relatively high but not unphysiological 20-HOE concentrations (1 M), whereas PPO6 was already fully induced by hormone treatment at the lowest 20-HOE concentration tested (50 nM). 3 High sensitivity of PPO6 to the hormone may explain its persistent induction, the modest effect of hormone washout, as well as the high expression level of PPO6 in unfed adults and its consequent unresponsiveness to the elevated ecdysteroid levels during blood digestion (Fig. 5B).
Localization of PPO Transcripts in the Mosquito-We examined the localization of PPO gene expression in dissected body parts and organs of adult female mosquitoes at 24 h following a blood meal to permit detection of PPO1 and PPO3, which are not expressed in unfed females. The gene encoding the antibacterial peptide defensin is strongly expressed in the mosquito midgut (11,38) and was included as a positive control for gut-specific gene expression. As shown in Fig. 7A, transcripts of PPO2-PPO6 are abundant throughout the body of the mosquito, except for the midgut and ovaries, which show only marginal PPO expression. In particular, these transcripts are very abundant in the head and thorax, as well as in the abdomen after removal of midgut and ovaries. This very broad distribution is consistent with the hypothesis that much of PPO mRNA is associated with hemocytes, which are known to be widely distributed (39,40). The PPO1 transcript is very rare but of similarly wide distribution; in relative terms it is enhanced in the midgut, where it is comparable in abundance to other body parts.
We have perfused mosquitoes to collect hemolymph together with hemocytes. The perfusion also recovers loosely attached fat body cells, which can be recognized by large round lipid vesicles. As shown by immunofluorescence analysis of the per-fusate (Fig. 7B), the anti-PPO4 antiserum recognizes a subpopulation of hemocytes but not the fat body cells. DISCUSSION In the present report we have isolated three members of the PPO gene family in the most important malaria vector mosquito, A. gambiae; we have illuminated the evolutionary relationship and chromosomal arrangement of this gene family; we have documented the disparate developmental expression profiles of the six PPO genes and their differential regulation by the hormone 20-HOE; we have shown that they are expressed widely within the body, associated with expression in the hemocytes; and we have isolated stable PPO-expressing and nonexpressing cell lines, which provide an unprecedented tool for molecular fine analysis of gene regulation and enzyme activation in this important gene family.
Phenoloxidases are indeed of crucial importance for the physiology and defense reactions of insects and other arthropods. They are intimately related to the periodic production of an effective exoskeleton barrier during the life cycle, functioning in sclerotization and pigmentation of the new cuticle after each moult (41). They also function in wound healing by crosslinking and melanizing temporary barriers that are established at cut surfaces by hemolymph clotting (42). Phenoloxidaseassociated local melanization is an important defense response of insects, which is triggered by non-self recognition and frequently is associated with physical encapsulation of invading parasites (13). In A. gambiae, melanotic encapsulation of young Plasmodium oocysts is a prime example of a refractoriness mechanism that can prevent completion of the malaria life cycle in the mosquito (2,43). In general, the reactions set in motion by phenoloxidase activity result in chemical as well as physical protection, because the oxidations leading to melanin formation also generate free radicals and toxic quinone intermediate radicals (16). By the same token, the activity of these enzymes must be readily elicited but also finely regulated to avoid damage to the insect itself. Rapid local and fine regula- tion is thought to depend on a cascade of activation and inhibition reactions reminiscent of those that control blood clotting (15). In sum, phenoloxidases serve multiple and fundamental mechanisms of insect developmental physiology, non-self recognition and defense, which are frequently interconnected. Nevertheless, relatively little is known at the molecular level about the mechanisms that regulate prophenoloxidase production and activation.
Full-length sequence analysis of a small number of prophenoloxidases from insects and a crayfish has only been achieved recently. The present work, together with the recent studies of Brey and co-workers (19) and Kanost and co-workers (18), make A. gambiae a favorite system for the study of the phenoloxidase family. Based on the results of rescreening a total of 68 cDNA and PCR clones with PPO gene-specific primers, we suspect that the six genes that we now know in A. gambiae may well constitute the complete phenoloxidase family, the only one that has been fully elucidated in any insect. If additional PPO genes exist, they are either nonexpressed, highly divergent in sequence, or so similar to the known genes that they are amplified by our gene-specific primers.
The dendrogram of the sequence relationships ( Fig. 2A), which is rooted with the crayfish PPO sequence as outgroup, suggests that the anopheline PPO family has arisen by five gene reduplications. On the assumption of a constant molecular clock, and considering the present chromosomal locations of the genes, a plausible model of the evolution of this gene family is as follows. An ancient gene duplication and a subsequent gene dispersal gave rise to the ancestor of PPO1 (at chromosomal division 13B) and the common ancestor of the other five genes (at division 21B). In turn, the latter reduplicated without gene dispersal to give the ancestor of PPO4/PPO5, linked to the ancestor of PPO6/PPO3/PPO2. The latter underwent a third reduplication, and one of its products translocated to division 24B, ultimately becoming PPO2. Both genes remaining at 21B finally reduplicated in situ, with the fourth duplication giving rise to the PPO4 and PPO5 pair, and the fifth producing the PPO6 and PPO3 pair. As a result, division 21B currently consists of a four gene cluster of the composition (PPO4/PPO5)-(PPO6/PPO3). The precise order and spacing of the genes in the cluster is currently unknown, but we do know that the most closely related genes are proximate neighbors. Even the most recent gene duplications were followed by regulatory diversification: the paired PPO4 and PPO5 genes (72.2% identical at the protein level) differ substantially in terms of developmental specificity, as do the paired genes PPO3 and PPO6 (81.8% identical). Highly specific antibodies that discriminate individual members of the family will be required to study the tissue distribution of individual genes and to facilitate further analysis of their regulation.
Complete recovery, sequence determination and expression analysis of the PPO families in additional insects will be required to determine when the gene duplications, and the regulatory diversification, occurred relative to the phylogeny of insects. Currently, we know two widely diverged PPO genes in each of the three moths, B. mori, M. sexta, and Hyphantria cunea. In each case, one of the genes is located on the same branch of the dendrogram as PPO1 of A. gambiae, suggesting that at least the first of the postulated duplications, giving rise to PPO1, antedates the last common ancestor of diptera and lepidoptera. At the other extreme, the sequence similarity of PPO6 and PPO3 to the presently known PPO gene of the mosquito Armigeres subalbatus (a culicine mosquito belonging to the tribe Aedini, like Aedes) indicates that the third postulated reduplication occurred prior to the separation of anophelines and culicines. It is likely that the second model mosquito, Aedes aegypti, will prove to have several PPO genes as well.
In crustaceans and insects, PPO is synthesized in the hemocytes (33, 44 -47). In lepidoptera, it has been shown at the RNA and protein levels that insect PPO is synthesized in a specialized type of hemocytes, the oenocytes (44,46). In contrast to crustaceans, PPO is already present in the cell-free plasma of noninfected insects (47,48), and in the case of B. mori, it has been shown that PPO is transported to the cuticle (42). In the dipteran species D. melanogaster, the components responsible for melanin synthesis are located in the crystal cells (49). However, within the family Culicidae (mosquitoes), no general classification of hemocyte types has been achieved as yet (39,50,51). According to a recent analysis of the domino mutation in D. melanogaster (52), secondary sites of PPO synthesis cannot be ruled out. Indeed, in Anopheles PPO was localized in the midgut (36,53) and the salivary glands (36). Whether these epithelial organs are minor sites of PPO synthesis or take up PPO synthesized in hemocytes remains to be determined. The low level of PPO RNA detected in the A. gambiae midgut, relatively enriched in PPO1 RNA (Fig. 7A), is supportive of the first alternative.
In A. gambiae, our perfusion studies have shown the presence of PPO in hemocytes but not fat body cells. The wide distribution of transcripts of each PPO gene in head, and thoracic and abdominal body parts is consistent with the current view that these genes are largely expressed in the widely dispersed hemocytes. Future in situ hybridizations and immunofluorescence studies (with gene-specific antibodies that are not yet available), should reveal all the sites of PPO synthesis and protein localization, complementing in the spatial dimension our current knowledge of temporally specific expression of individual genes. It will be interesting to know whether PPO gene transcription or translation in the mosquito is immuneresponsive at any time during development or whether PPO regulation in response to immune challenge only engages the activators and inhibitors of the phenoloxidase cascade.
The two A. gambiae cell lines that we have described here, which were generated from the same mosquito strain, show intriguing properties. Both of these cell lines are immuneresponsive, in terms of induction of the defensin and Gramnegative bacteria-binding protein genes after challenge by bacterial constituents. Therefore, both lines presumably possess pattern recognition receptors and the capacity to respond to activation of these receptors. However, one of the lines (4a-3A) is incapable of expressing the prophenoloxidase genes, whereas the other (4a-3B) expresses all six of these genes in a constitutive manner. In the latter line, PPO transcript levels are unaffected or decrease slightly after bacterial challenge. In contrast, these levels are strongly regulated by 20-HOE in a reversible and gene-specific manner. Immunofluorescence analysis detects PPO protein in only a subset of approximately 5% of the cells in this line; the positive cells tend to be large and morphologically variable. Yet PPO expression is a stable property of this cell line, and the PPO-positive cells are frequently found in morphologically similar pairs, suggesting recent cell divisions. Our current interpretation is that the positive cells represent a differentiated state, in which infrequent cell divisions continue. This inference will need to be confirmed by cloning the cell line and subsequently investigating what exogenous signals might lead nonexpressing cell progenitors to turn on the PPO genes. It will be interesting to investigate whether all PPO genes can be turned on simultaneously or whether cell differentiation might recapitulate the developmental succession of PPO genes in the whole animal. To date, we only know that the various PPO transcripts in the cell line are subject to hormonal regulation, in a gene-specific manner that is related to their presumed hormonal regulation in the intact adult female mosquito, following a blood meal. Clearly, these unique cell lines are highly promising tools for the study of phenoloxidase regulation in insects.