The Salivary Adenosine Deaminase Activity of the Mosquitoes Culex quinquefasciatus and Aedes aegypti

Adenosine deaminase (ADA) hydrolyzes adenosine to inosine and ammonia. Its presence in the salivary glands of a New World sand fly and tsetse flies was recently reported (Charlab et al., 2000; Li and Aksoy, 2000) following the finding of salivary cDNA sequences with homology to mammalian and bacterial ADAs. Salivary homogenates of the sand fly Lutzomyia longipalpis had relatively high enzyme activities compared with other enzyme sources. Activity was significantly reduced following a blood meal, suggesting that the enzyme was secreted, in accordance with the identification of a secretory signal peptide deduced from the cDNA sequence (Charlab et al., 2000). The finding of salivary ADA in hematophagous flies is puzzling because adenosine causes vasodilation and inhibits platelet aggregation (Collis, 1989; Dionisotti et al., 1992; Chinellato et al., 1994), functions that should be of help in blood feeding (Ribeiro, 1995). Conversion of adenosine to inosine abrogates these vasodilatory and antiplatelet functions. Furthermore, the Old World sand fly Phlebotomus papatasi actually secretes large amounts of adenosine in its saliva (Ribeiro et al., 1999) as the main salivary vasodilator and does not have salivary ADA activity (J. M. C. Ribeiro, unpublished observations). The presence of a salivary ADA in some hematophagous insects and not in others is puzzling. A possible adaptive role for salivary ADA is to reduce local pain in the host while the insect is feeding, since adenosine A2 and A3 receptors stimulate peripheral pain receptors (Burnstock and Wood, 1996; Poon and Sawynok, 1999; Charlab et al., 2000). In addition, while adenosine induces mast cell degranulation and the release at the feeding site of histamine, a substance that causes itching and burning, inosine is more than 100 times less effective in producing this effect (Tilley et al., 2000). It has also been shown that inosine formed by ADA inhibits the production of inflammatory cytokines (Hasko et al., 2000).

Whether salivary ADA is present in other hematophagous arthropods is not known. In this report, we show that the full-length clone of a cDNA from the salivary glands of the mosquito Culex quinquefasciatus has high homology to vertebrate and invertebrate ADA and identify large amounts of this activity in the salivary glands of adult females of both C. quinquefasciatus and Aedes aegypti but not Anopheles gambiae. Only Aedes aegypti appear to secrete the enzyme.

Organic reagents were obtained from Sigma Chemical Corporation (St Louis, MO, USA). All water used was of 18MΩ quality and was produced using a MilliQ apparatus from Millipore (Bedford, MA, USA).

Insects were reared in the Medical Entomology Section of the Laboratory of Parasitic Diseases of the National Institute of Allergy and Infectious Diseases under the expert supervision of Mr Andre Laughinghouse. Insectary rooms were kept at 26±0.5°C, with a relative humidity of 70% and a 16h:8h light:dark photoperiod. The strains of mosquito used were the Liverpool/black eye strain of Aedes aegypti, the G3 strain of Anopheles gambiae and the Vero Beach strain of Culex quinquefasciatus. The most recent strain (C. quinquefasciatus) has been in continuous culture for 8 years. Adult mosquitoes used in the experiments were given no blood meals and were maintained on a diet of 10% Karo syrup solution. Salivary glands were dissected from adult female mosquitoes aged 3–10 days that had either never been fed blood or immediately after a blood meal, as indicated in the experiments. The glands were transferred to 10 or 20μl of Hepes saline (HS; 0.15moll⁻¹ NaCl, 10mmoll⁻¹ Hepes, pH7.0) in 1.5ml polypropylene vials, usually as groups of 20 pairs of glands in 20μl of HS or as individual glands in 10μl of HS. In some experiments, salivary glands from male mosquitoes were dissected. Salivary glands were kept at −75°C until needed, when they were disrupted by sonication using a Branson Sonifier 450 homogenizer (Danbury, CT, USA). Salivary homogenates were centrifuged at 10000g for 2min, and the supernatants were used for experiments. To obtain dilute saliva samples from mosquitoes, a small artificial feeder containing 100μl of HS was offered to a cage containing approximately 100 adult female mosquitoes. The feeder membrane consisted of stretched Parafilm (American National Can, Menasha, WI, USA). The feeder was warmed by continuous circulation of water at 38°C.

To obtain saliva, female mosquitoes were lightly anesthetized by exposure to 0°C for 1min. After their wings and legs had been removed, 0.1μl of a 1mmoll⁻¹ serotonin solution in HS was injected into the coelomic cavity using a heat-pulled glass capillary micropipette operated by mouth pressure (Ribeiro et al., 1984a). After the injection, the feeding fascicle was exposed from within the sheath and inserted into a 2cm piece of polyethylene tubing (PE-10, inner diameter 0.28mm, outer diameter 0.61mm; Clay Adams, Parsippany, NJ, USA) containing a column of mineral oil approximately 0.5cm long. This operation was carried out under a stereomicroscope using fine-tipped tweezers. After 10min, to allow salivation into the oil at room temperature, the mouthparts were removed from the tubing. A no. 30 needle was inserted into the other end of the tubing, and this was used to blow the oil+saliva into 20μl of HS, pH7.2, with the help of a 20μl Hamilton (Reno, NV, USA) syringe with a Luer-lock fitting. Ten samples were accumulated, each in 20μl of HS. The samples were then centrifuged at 10000g for 5min; the aqueous phase was removed for analysis.

Protein concentration was assessed by the Bradford method (Bradford, 1976) using a kit from Bio-Rad (Hercules, CA, USA).

mRNA was isolated from 80 pairs of salivary glands dissected from 0- to 2-day-old female C. quinquefasciatus using the Micro-FastTrack mRNA isolation kit (Invitrogen, San Diego, CA, USA) as described previously (Charlab et al., 2000). A polymerase chain reaction (PCR)-based cDNA library was subsequently prepared according to the instructions for the SMART PCR cDNA Library Construction Kit (Protocol PT3000-1, Version PR92334, Clontech, Palo Alto, CA, USA). Briefly, the C. quinquefasciatus salivary gland poly(A)⁺ RNA was reverse-transcribed to cDNA for 1h at 42°C using Superscript RNase H reverse transcriptase (Life Technologies, Inc., Gaithersburg, MD, USA), the SMART III oligonucleotide and a modified oligo(dT) primer (CDS III/3′ PCR primer) (Clontech). The combined use of the SMART III oligonucleotide and the CDS III/3′ primer allows the incorporation of asymmetrical SfiI restriction enzyme sites at both the 5′ and 3′ ends of the cDNA.

Second-strand synthesis was performed through a PCR-based protocol using the 5′ PCR primer and the CDSIII/3′ PCR primer as sense and antisense primers, respectively. Conditions for PCR were: 95°C for 1min, 20 cycles of 95°C for 15s and 68°C for 6min. After proteinase K treatment for 20min at 45°C, double-stranded cDNA was digested with SfiI for 2h at 50°C and size-fractionated using CHROMA SPIN-400 columns supplied by the manufacturer. The fraction profile was checked by electrophoresis on a 1.1% agarose/ethidium bromide gel. Peak fractions were visualized under ultraviolet illumination, and the first two fractions containing cDNA were pooled, precipitated and resuspended in 7μl of deionized water. The SfiIB-digested, size-fractionated and concentrated cDNA was then ligated into the SfiI-digested Lambda TriplEx2 vector (Clontech). The ligation reaction was packed using Gigapack Gold III packaging extract from Stratagene/Biocrest (Cedar Creek, TN, USA) following the supplier’s recommended protocol. The library was plated by infecting log-phase Escherichia coli XL1-Blue (Clontech). The percentage of recombinants was determined by PCR with vector primers flanking the inserted cDNA and visualization on a 1.1% agarose gel with ethidium bromide.

The C. quinquefasciatus cDNA library was plated at a density of approximately 150 plaques per plate (150mm Petri dish). The plaques were picked at random and transferred to 96-well polypropylene plates containing 50–75μl of deionized water per well. The plates were covered and placed on a gyratory shaker for 1h at room temperature (25°C). Of the phage sample, 5μl was used as a template to amplify and sequence the cDNA according to Valenzuela et al. (Valenzuela et al., 2000). Briefly, the primers for this reaction were sequences from the triplEX2 vector, PT2F1 (5′-AAGTACTCTAGCAATTGTGAGC-3′), which is positioned upstream of the cDNA of interest (5′ end), and PT2R1 (5′-CTCTTCGCTATTACGCCAGCTG-3′), which is positioned downstream of the cDNA of interest (3′ end). High-fidelity platinum Taq polymerase (Gibco-BRL, Gaithersburg, MD, USA) was used for these reactions. Amplification conditions were as follows: one hold at 75°C for 3min, one hold at 94°C for 3min and 34 cycles at 94°C for 30s, 49°C for 30s and 72°C for 80s. Amplified products were visualized on a 1.1% agarose gel with ethidium bromide. The concentration of double-stranded cDNA was measured using Hoechst dye 33258 on a Flurolite 1000 plate fluorimeter (Dynatech Laboratories, Chantilly, VA, USA).

To remove the primers from the PCR reaction products, PCR samples (3–4μl) containing between 100 and 200ng of DNA were treated with Exonuclease I (0.5unitsμl⁻¹) and shrimp alkaline phosphatase (0.1unitsμl⁻¹) for 15min at 37°C and 15min at 80°C on a 96-well PCR plate.

This mixture was used as a template for a cycle-sequencing reaction using the DTCS labeling kit (Beckman Coulter Inc., Fullerton, CA, USA). The primer used for sequencing (PT2F3) is upstream of the inserted cDNA and downstream of the primer PT2F1. The sequencing reaction was performed in a Gene Amp PCR system 9700 from PE Applied Biosystems (Foster City, CA, USA). Conditions were 75°C for 2min, 94°C for 4min and 30 cycles at 96°C for 20s, 50°C for 20s and 60°C for 4min.

To clean up the samples after cycle sequencing, they were subjected to a step using the multiscreen 96-well plate-cleaning system (Millipore); this plate was prepared by adding a fixed amount (manufacturer’s specification) of Sephadex-50 (Amersham Pharmacia Biotech, Piscataway, NJ, USA) and 300μl of deionized water. After 1h of incubation at room temperature, the water was removed from the multiscreen plate by centrifugation at 750g for 5min. After partially drying the Sephadex in the multiscreen plate, the whole cycle-sequencing reaction was added to the center of each well and centrifuged at 750g for 5min. The clean sample was collected on a sequencing microtiter plate (Beckman Coulter, Inc.). The plate was dried using a SpeedVac SC 110 with a microtiter plate holder (Savant Instruments Inc., Holbrook, NY, USA). The dried samples were immediately resuspended in 25μl of deionized ultraPure formamide (J. T. Baker; Phillipsburg, NJ, USA), and one drop of mineral oil was added to the top of each sample. Samples were either sequenced immediately on a CEQ 2000 DNA sequencing instrument (Beckman Coulter Inc.) or stored at −30°C for several days.

Sequenced cDNA inserts were analysed using the BLAST programs with BLOSUM-62 matrix (http://www.ncbi.nlm.gov/BLAST/; Altschul et al., 1997). During this process, we identified a full-length cDNA with significant sequence similarity to Lutzomyia longipalpis ADA (GenBank AF 234182) (Charlab et al., 2000). This cDNA was then reamplified using the PT2F1 and PT2R1 primers, and the entire cDNA was fully sequenced using custom-designed primers as described above.

Adenosine deaminase activity was measured in quartz microcuvettes holding 60–100μl samples (Starna Cells, Atascadero, CA, USA). The indicated concentrations of adenosine, AMP or adenine in 10mmoll⁻¹ Hepes plus 150mmoll⁻¹ NaCl (HS) (or buffer as otherwise indicated) were added to the cuvette, followed by saliva or salivary gland homogenate (usually equivalent to 5–10% of one pair of glands, or approximately 50–100ng of protein). After mixing the solution by several cycles of aspiration and deposition of the sample, the absorbance (A) either at 265nm or at both 241 and 265nm was read at 5s intervals for the single-wavelength protocol or at 15s intervals for the dual-wavelength protocol using a Lambda 18 spectrophotometer from Perkin Elmer (Norwalk, CT, USA). The cuvettes were thermostatted to 30°C. One unit of enzyme activity is defined as the amount of enzyme that catalyzes the deamination of 1μmol of adenosine per minute (ΔA=8.6min⁻¹ml⁻¹ at 265nm or ΔA=13.76min⁻¹ml⁻¹ when the difference in optical densities between 249 and 265nm is used; Agarwal and Parks, 1978, and our own measurements). In some experiments, the cuvette was scanned at the indicated time intervals from 230 to 320nm.

Gel permeation chromatography experiments were performed with a Superdex 75 column (3.2mm×300mm) isocratically perfused with 10mmoll⁻¹ Hepes, pH7.4, 0.15moll⁻¹NaCl at a flow rate of 50μlmin⁻¹. A CM4100 pump and SM 4100 UV detector (ThermoSeparation Products, Riviera Beach, FL, USA) were used. Fractions were collected at 30s intervals. Of each fraction, 5μl was used to determine ADA activity in the assay described above, following the rate of absorbance change at 265nm for 2min. The column was calibrated with cytochrome c, myoglobin, carbonic anhydrase, ovalbumin and the monomer and dimer of bovine serum albumin.

Statistical tests were carried out using SigmaStat software (Jandel Scientific, San Raphael, CA, USA). Mann–Whitney tests were conducted in comparisons where the variances of two groups were not homogeneous; otherwise, Student’s t-tests were performed.

We randomly sequenced 600 clones from a salivary gland cDNA library obtained from female C. quinquefasciatus. Among the sequenced clones, we identified a cDNA with a high degree of similarity to ADA and invertebrate growth factors when the translated protein sequence was compared with the non-redundant protein database of the National Center for Biotechnology Information of the National Library of Medicine using the BlastP program (Altschul et al., 1997). This cDNA (GeneBank accession number AF298886) has the adenosine/AMP deaminase (A_deaminase) Pfam signature (Bateman et al., 2000) and a signal peptide indicative of secretion (Nielsen et al., 1997) (Fig.1). The mature protein is predicted to have an amino acid sequence starting at the 18th amino acid residue (sequence AKLI...), an isoelectric pH of 5.78 and a relative molecular mass of 55575.

To verify whether salivary gland homogenates of C. quinquefasciatus contain ADA activity, salivary homogenate equivalent to 0.2 of a pair of glands was incubated with 100μmoll⁻¹ adenosine in 100μl of HS buffer, and the mixture was scanned at 10min intervals from 230 to 320nm. The spectrum shifted in a time-dependent manner from a maximum at 258nm to one at 249nm, consistent with the conversion of adenosine to inosine. The differential spectrum shows a maximum decrease at 265nm and a maximum increase at 241nm, as found for the conversion of adenosine to inosine (Agarwal and Parks, 1978) (Fig.2).

Many ADAs can also utilise AMP as a substrate, and so we measured the ability of C. quinquefasciatus salivary homogenates to convert AMP to IMP. The rate of conversion of AMP by the enzyme in the homogenate is faster than that of adenosine (Fig.3). Although part of the activity could be ascribed to the conversion of AMP to adenosine by 5′-nucleotidase, this mechanism cannot explain the increased enzyme activity with AMP. In other words, an infinite amount of 5′-nucleotidase could instantly convert all AMP to adenosine and, thus, the ADA activity would be the same for AMP and adenosine. It could not, however, convert AMP to inosine faster than adenosine is converted to inosine. We conclude, therefore, that there is AMP deaminase activity in C. quinquefasciatus salivary gland homogenates. In three different pools of C. quinquefasciatus salivary homogenates, the ratio of AMP deaminase activity to ADA activity was 3.3±0.07 (mean ± s.e.m., N=3). When adenine was used as a substrate, no reaction was observed (results not shown).

To investigate whether ADA and AMP deaminase activities co-elute following gel-permeation chromatographic procedures and to obtain an estimate of their molecular masses, we submitted 20 pairs of homogenized salivary glands from C. quinquefasciatus to chromatography in a Superdex 75 column. The activities of both substrates co-elute with an apparent molecular mass of 62kDa (Fig.4). Although this estimate is larger than the 55kDa predicted from the translated cDNA, it is still within the margin of error of the technique. The ranges of retention times of the fraction containing the maximum activity correspond to molecular masses of 55.7–69kDa. In addition, the mature molecule, which has several glycosylation sites, could have been post-translationally modified. In any case, these data support the hypothesis that ADA and AMP deaminase activities reside in the same molecule.

To determine whether other mosquitoes could have salivary ADA activity, we incubated salivary homogenates from Aedes aegypti and Anopheles gambiae with adenosine and scanned the cuvette for changes in ultraviolet absorption. Even when incubating two pairs of homogenized gland equivalents from the anopheline mosquito, no activity was found; however, incubation of salivary homogenate equivalent to 0.02 of a pair of glands from Aedes aegypti resulted in measurable activity (Fig.5). Aedes aegypti salivary homogenates had little AMP deaminase activity. In three different pools of Aedes aegypti salivary homogenates, the ratio of AMP deaminase activity to adenosine activity was 0.07±0.02 (mean ± s.e.m., N=3).

To investigate whether mosquito salivary ADA could be secreted during a blood meal, we measured the enzyme activity in salivary homogenates obtained from unfed mosquitoes and from mosquitoes immediately after a blood meal. While more than 95% of Aedes aegypti offered a blood meal would feed readily, only 10–20% of C. quinquefasciatus would do so. For this reason and to obtain salivary glands from mosquitoes of the same physiological status, control (pre-blood-fed) salivary glands from C. quinquefasciatus were removed from mosquitoes attracted to a human hand. Because virtually all (>95%) Aedes aegypti were attracted to the human hand, this selection was not performed in these mosquitoes. In both mosquito species, there was a significant decrease in salivary protein content following a blood meal. Levels fell from 2.6±0.5 to 1.2±0.2μgproteinglandpair⁻¹ for Aedes aegypti (mean ± s.e.m., N=12, P=0.016, Mann–Whitney rank sum test) and from 1.00±0.36 to 0.32±0.08μgproteinglandpair⁻¹ (mean ± s.e.m., N=12, P=0.027, Mann–Whitney test) for C. quinquefasciatus. While there was a significant decrease in the amount of ADA activity in Aedes aegypti salivary gland homogenates before and after a blood meal (from 17.0±1.6 to 12.0 1.3munitsglandpair⁻¹, mean ± s.e.m., N=12, P=0.028, t-test), only a small non-significant decrease was observed in C. quinquefasciatus (from 2.64±0.37 to 2.53±0.34munitsglandpair⁻¹, mean ± s.e.m., N=12, P>0.1, t-test) (Fig.6). Although these data suggest that the Aedes aegypti enzyme is secreted during a blood meal, we cannot conclude the same for the C. quinquefasciatus enzyme.

Unlike females, which feed on both blood and sugar, adult male mosquitoes take only sugar meals. Salivary glands of male and female mosquitoes have similar amounts of the enzymes associated with sugar feeding (Marinotti et al., 1990), but males have much less, or none, of the enzyme activities associated with blood feeding, e.g. antiplatelet apyrase enzyme or anticlotting activities (Rossignol et al., 1984; Stark and James, 1995). Accordingly, to investigate whether mosquito salivary ADA activity is associated with blood feeding, we measured ADA activity in salivary homogenates of male Aedes aegypti and C. quinquefasciatus. Male mosquitoes had very little ADA activity, amounting to less than 2% of that of their female counterparts (Fig.6), suggesting that female salivary ADA activity is associated with a role in blood feeding.

To obtain further evidence for secretion of salivary ADA during probing, we exposed approximately 100 female Aedes aegypti for 20min to an artificial feeder containing a solution of 50μmoll⁻¹ adenosine in HS. During this period, there were always 3–5 mosquitoes probing the membrane but, perhaps because of the absence of phagostimulation, individual mosquitoes probed the membrane for only 10–30s and then left. Nevertheless, when the solution was removed from the feeder, its spectrum had changed to one characteristic of inosine (Fig.7A). When 5μl of the feeder solution was added to 100μmoll⁻¹ adenosine in 95μl of HS and the cuvette scanned at 12min intervals (Fig.7B,C), the resulting spectral change (Fig.7D) demonstrated an ADA activity corresponding to 2.5munits contained in the feeder after Aedes aegypti probing, or nearly 15% of the mean amount of ADA contained in one pair of female salivary glands. Thus, if each of the approximately 100 mosquitoes probed the membrane during the 20min interval, the equivalent of 0.15% of the ADA content of one pair of glands was delivered to the feeder solution per mosquito, assuming an average of 16munits of ADA activity per pair of female Aedes aegypti salivary glands (from Fig.6).

When the same experiment was attempted with C. quinquefasciatus, they did not attempt to probe the membrane of the feeder. To entice these mosquitoes to probe the membrane, a human finger was placed on the surface of the net covering the mosquito cage. While the mosquito was attempting to probe the finger, the finger was gently withdrawn from the cage surface and replaced by the feeder. In this manner, three mosquitoes were enticed to probe the membrane for periods of 15–30s each. The ADA activity measured in the feeder amounted to less than 0.1% of the ADA content of one pair of glands (results not shown), indicating that insignificant amounts of ADA were lost from C. quinquefasciatus while probing the membrane feeder.

As an independent test for possible ADA secretion in C. quinquefasciatus and Aedes aegypti saliva, we measured ADA and AMP deaminase activity in individual samples of female mosquito saliva. While both activities were readily detected in Aedes aegypti saliva, with higher values for ADA than for AMP deaminase, ADA and AMP deaminase activities for C. quinquefasciatus were at the detection limit of the assay, and were not significantly different from zero (Fig.8). The ADA values for Aedes aegypti corresponded to the equivalent of 2.3% of the content of one pair of glands, on average, per collected saliva sample. These results indicated that Aedes aegypti salivary ADA and AMP deaminase are secreted, and failed to reveal secretion from C. quinquefasciatus salivary glands.

Finally, to exclude the possibility that the presence of salivary ADA in C. quinquefasciatus was due to contamination of the salivary homogenates by fat body cells, fat body from C. quinquefasciatus was removed from adult female mosquitoes and homogenized as for the salivary glands. Using the same protein content as that of one whole salivary gland in our ADA assay, no ADA activity was found in three independent fat body homogenates (results not shown).

The presence of salivary ADA activity in hematophagous insects is puzzling, because adenosine is a vasodilatory and anti-platelet substance (Collis, 1989; Dionisotti et al., 1992; Chinellato et al., 1994), which might be expected to increase blood availability at the feeding site. However, adenosine may be implicated in peripheral pain perception (Burnstock and Wood, 1996; Poon and Sawynok, 1999; Charlab et al., 2000), although its role in nociception is contradictory: while adenosine A2 and A3 receptors appear to be clearly nociceptive, adenosine A1 receptor stimulation appears to be anti-nociceptive (Poon and Sawynok, 1998; Poon and Sawynok, 1999). Another effect of salivary ADA may be related to a local accumulation of inosine at the feeding site. Inosine has been shown to be a potent inhibitor of the production of the inflammatory cytokines TNF-α, Il-1, Il-12, macrophage-inflammatory protein-1α and IFN-γ (Hasko et al., 2000). Il-1, in particular, is a potent peripheral nociceptive agent (Bianchi et al., 1998). However, it is doubtful whether these cytokines could be produced by the host and released during the few minutes that mosquitoes take to feed, although they may have an effect on the fate of the host immune response to salivary allergens and pathogens co-injected with the mosquito saliva. While the role of adenosine in nociception is unclear, adenosine is known to induce mast cell degranulation (Tilley et al., 2000), resulting in the release of histamine. From the point of view of the mosquito, the itching induced by histamine would have the undesirable effect of alerting the host to its presence. Although it is clear that adenosine could be beneficial to a hematophagous insect, to the point that phlebotomine sand flies produce copious amounts of the substance in their salivary glands (Ribeiro et al., 1999; Ribeiro and Modi, 2001), the presence of ADA in the saliva of other hematophagous insects indicates that, if it has any evolutionary significance, adenosine may not be a desirable substance at the feeding site.

In accordance with the ambivalent role of adenosine in blood feeding, the data presented in this paper indicate that the presence of salivary ADA in mosquitoes is not a consistent finding, unlike the ubiquitous occurrence of diverse anticoagulants or the enzyme apyrase (Ribeiro, 1995). No ADA activity was found in Anopheles gambiae, but relatively high activities were found in both Culex quinquefasciatus and Aedes aegypti. The specific activity of 16munitsglandpair⁻¹ for Aedes aegypti represents a specific activity of 5unitsmg⁻¹salivaryprotein, while the value of 2.6munitsglandpair⁻¹ for C. quinquefasciatus represents 2.5unitsmg⁻¹salivaryprotein. This amount is larger than the specific activity of erythrocyte ADA (0.2unitsmg⁻¹protein; Agarwal et al., 1975) and is similar to that in tissues specialized in digestion, such as the crude ADA sold by Sigma Chemical Co. obtained from calf intestinal mucosa (1–5unitsmg⁻¹protein) (Sigma Chemical catalog, 2000–2001 edition). In the case of Aedes aegypti, salivary ADA appears to be secreted during a blood meal, as demonstrated by the significant decrease in levels in the salivary glands following a blood meal, its presence in solutions probed by hungry female mosquitoes and its presence in serotonin-induced saliva. In C. quinquefasciatus, no activity could be detected in serotonin-induced saliva, although very low levels of the enzyme (less than 0.03% of the content of one pair of glands per probing mosquito) were left in feeders probed by this species. No significant decrease in enzyme levels was detected in C. quinquefasciatus salivary glands following a blood meal. This suggests that ADA serves a largely endogenous function in the salivary glands of C. quinquefasciatus.

We can only speculate about the evolutionary scenario that has led to the presence or absence of ADA in the saliva of hematophagous arthropods, but we may assume that adenosine has both positive and negative effects on blood feeding. The positive effects could be due to its actions on platelet aggregation and vasodilation. Perhaps for this reason, the sand flies Phlebotomus papatasi and P. argentipes have abundant adenosine in their salivary glands (Ribeiro et al., 1999; Ribeiro and Modi, 2001). It is possible that these phlebotomine hosts are less sensitive to the negative effects of adenosine, that they feed when their hosts are in profound sleep and unresponsive to relatively small disturbances or that sand flies have other mechanisms to counteract the negative effects of adenosine on feeding. However, both Aedes aegypti and C. quinquefasciatus have their own salivary vasodilatory molecules: the peptide sialokinin in Aedes aegypti (Champagne and Ribeiro, 1994) and an unknown substance in C. quinquefasciatus (Ribeiro, 2000). In addition, Aedes aegypti has large amounts of salivary apyrase, which inhibits platelet aggregation (Ribeiro et al., 1984b), while C. quinquefasciatus has a potent salivary phospholipase C that destroys platelet activating factor (J. M. C. Ribeiro and I. Francischetti, in preparation). Other molecules can therefore replace the positive functions of adenosine at the feeding site, and we are left only with its negative effects.

If we assume that the negative role of adenosine is related to its ability to bring the attention of the host to the feeding site, then the intensity of this negative effect would be affected by two variables: the degree to which adenosine induces pain or itching at the feeding site and the behavioral response of the host to pain and itching. The first variable can vary from species to species, e.g. the density of mast cells in the skin of the host selected for feeding by the mosquito, the number and affinity of adenosine receptors inducing cell degranulation or the density of adenosine nociceptive receptors in the skin. The second variable depends, for example, on whether the vertebrate is asleep or alert and how it reacts to the same stimulus in these two conditions. Depending on these variables, it may be of more or less selective advantage to neutralize the negative effect of adenosine on blood feeding, and more than one strategy may be employed to counteract its effects. Thus, adenosine could be destroyed via ADA or its effects reduced by the use of physiological inhibitors, e.g. molecules that act in an opposite way on the target cell, such as mast cell degranulation inhibitors or anesthetics.

It is interesting to note that Aedes aegypti is a diurnal feeder and that its main hosts are humans. Salivary homogenates from Aedes aegypti have the ability to prevent the release of tumor necrosis factor (TNF) by mast cells (Bissonette et al., 1993), indicating that this insect may be under pressure to avoid causing degranulation of the mast cells of the host. In contrast, C. quinquefasciatus is a nocturnal feeder on humans and may avoid the negative behavioral effects of adenosine by feeding when the physiological state of the host is less responsive or unresponsive to an increase in the levels of adenosine at the feeding site. Alternatively, C. quinquefasciatus may have additional salivary components that antagonize the nociceptive and/or mast cell degranulating effects of adenosine. In this regard, it is interesting to note that both triatomine bugs (Ribeiro and Walker, 1994) and ticks (Paesen et al., 1999) have salivary proteins that neutralize histamine, and an anesthetic substance has recently been reported from the saliva of a triatomine bug (Dan et al., 1999). The presence of ADA in Aedes aegypti may represent one of several ways in which the negative effects of adenosine can be counteracted. While these considerations may help to explain why Aedes aegypti could benefit from secreting ADA, we have no clue as to the role of salivary ADA in C. quinquefasciatus, although we can speculate that it may be involved in some endogenous function, such as regulating transmitter signaling by deactivating adenosine or producing inosine or an inosine derivative as a secretory product. Such speculations suggest the hypothesis that anti-nociceptive and anti-mast cell salivary substances should be found abundantly in the salivary glands of hematophagous arthropods.

From a biochemical perspective, it is interesting that the salivary AMP deaminase activity of C. quinquefasciatus is approximately three times greater than that of ADA, while in Aedes aegypti it is one-tenth of the ADA activity. AMP deaminase and ADA are enzymes that probably evolved by gene duplications (Becerra and Lazcano, 1998). Assuming that the two activities reside in a single enzyme in mosquito salivary glands, cloning and expression of these two enzymes, one from each mosquito species, might indicate the amino acid residues that confer greater specificity towards either AMP or adenosine. Alternatively, each mosquito species may express the two enzymes (ADA and AMP deaminase) in their glands in different ratios.

The cDNA discussed in this paper may or may not be responsible for the majority of the enzyme activity seen in C. quinquefasciatus salivary glands. The work reported here confirms that sequencing cDNA libraries from salivary glands of hematophagous insects can yield information leading to the detection of many salivary enzymes not previously identified, such as amylases, hyaluronidases, ADA and 5′-nucleotidases (Charlab et al., 1999; Charlab et al., 2000). Coupled with appropriate biochemical and pharmacological assays, this approach lends impetus to our understanding of how insects have evolved to become one of the most successful and diverse groups of species in the animal kingdom.

We are grateful to Dr Robert Gwadz, Thomas Kindt and Louis Miller for encouragement and support, to Van My Pham for technical assistance and to Nancy Shulman for editorial assistance.