The Salivary Catechol Oxidase/Peroxidase Activities of the Mosquito Anopheles Albimanus

When attempting to feed, blood-sucking arthropods injure their hosts and thus face a complex vertebrate defense system against blood loss, the hemostatic response. Platelet aggregation, vasoconstriction and blood coagulation are three interconnected components of the hemostatic system. Tissue destruction releases or exposes to the blood molecules such as ADP (usually millimolar concentration inside cells, but submicromolar concentrations in the extracellular compartment) and collagen that trigger platelet aggregation and platelet granule secretion. Platelets aggregate in seconds, forming a plug at the site of injury. In this process, platelets also secrete more ADP, serotonin and thromboxane A₂. These two last components induce constriction of smooth muscle around venules or arterioles, further restricting blood loss, and tightening the vessel against the platelet plug. Activated platelets and damaged tissue also trigger blood coagulation, leading to the formation of thrombin. Thrombin serves as an activator of platelets and catalyzes the formation of fibrin, which further reinforces the platelet plug structure at the damaged site. Blood-sucking animals thus have to counteract their host’s blood coagulation, platelet aggregation and vasoconstriction to obtain a blood meal (Ribeiro, 1987; Law et al. 1992).

Anticoagulants have been recognized in the saliva or salivary gland homogenates of blood-sucking animals since early in this century (Cornwall and Patton, 1914), although the characterization at the molecular level of such substances has only recently begun (Law et al. 1992). Similarly, only within the last 10 years has antiplatelet activity been found in saliva or salivary glands of blood-sucking arthropods, leeches and bats (Ribeiro, 1987; Law et al. 1992). Vasodilators have also recently been found in blood-sucking arthropods, with a surprising molecular variety (Law et al. 1992).

From the very few arthropods studied, a large diversity of salivary vasodilatory compounds is emerging: salivary glands of ixodid ticks contain vasodilatory prostaglandins (Ribeiro, 1987; Ribeiro et al. 1991), those from the bug Rhodnius prolixus contain a nitrovasodilator (Ribeiro et al. 1990), the sand fly Lutzomyia longipalpis contains the most potent peptidic vasodilator presently known (Ribeiro et al. 1989; Lerner et al. 1991) and the mosquito Aedes aegypti has a peptidic vasodilator of the tachykinin family (Pappas et al. 1986; Ribeiro, 1992; D. Champagne and J. M. C. Ribeiro, unpublished observations). This diversity of pharmacologically active compounds appears to derive from the convergent evolution to blood-feeding behavior found in many invertebrate and vertebrate animals.

Anopheles albimanus is a mosquito vector of malaria, and laboratory colonies of this insect are readily available. Salivary vasodilator compounds from anopheline mosquitoes have never been described, although antiplatelet (Ribeiro et al. 1985) and anticoagulant (Yorkee and Macfie, 1924; Metcalf, 1945; Ribeiro et al. 1985) activities have been reported. In this paper, we describe a salivary enzyme activity that destroys catecholamines and serotonin, substances that are important vertebrate vasoconstriction agents.

The Santa Tecla strain of Anopheles albimanus was maintained at the insectary of the University of Arizona in Tucson, at 27°C and 80% relative humidity. Adults were offered cotton swabs containing 20% sucrose in water. Salivary glands from female mosquitoes at least 3 days old were dissected and kept (20 pairs in 20 μl of phosphate-buffered saline: 10mmol l⁻¹sodium phosphate at pH7.0 with 150mmol l⁻¹NaCl) at −75°C until needed. They were then disrupted by ultrasound with a Branson sonifier (model 450) with the probe immersed for 2cm in a 100ml beaker containing 80ml of water at room temperature. The 1.5ml conical tube containing the glands was held with clamping forceps so that its conical tip was just under the tip of the probe. Power was set at 6 and a 50% cycle was run for 1min. Disruption of the glands was assessed under a stereoscope, and a new cycle was repeated if some glands appeared intact. The homogenate was centrifuged for 2min at 10000 g and the supernatant was recovered for use in all assays.

Mosquito blood feeding was accomplished by anesthetizing a rat with a mixture of ketamine and xylazine and exposing mosquitoes to its shaved abdomen. To increase salivation, feeding was disturbed at 1min intervals by lifting the rat from the cage. Fully engorged mosquitoes were used to test for the loss of enzyme activity and protein with salivation.

Rabbit aorta smooth muscle ring bioassays were performed isometrically using Krebs solution at 37°C bubbled continuously with 95% O₂ and 5% CO₂ (Perry, 1968). Dexamethasone (0.1 μmol l⁻¹) was added to the bath to prevent induction of NO synthase (Moncada et al. 1991), while EDTA (30 μmol l⁻¹) was added to chelate heavy metal contaminants. Rabbit aortic strips were also mounted, in some cases, according to Ferreira and Souza-Costa (1976) in an oil-immersed, laminar-flow superfusion mode. Aortic endothelium was removed by gently rubbing the inner portion of the rings with a wooden stick. Success of endothelium removal was confirmed by the inability of acetylcholine (1 μmol l⁻¹) to relax the rings preconstricted with 1 μmoll⁻¹ noradrenaline. The estrogenated rat uterus bioassay was performed isotonically using De Jalon solution at 30°C (Gaddum et al. 1949) in a laminar-flow superfusion mode. The guinea pig ileum bioassay was performed isotonically using Tyrode’s solution at 37°C and bubbled with O₂ and CO₂ as for the aortic assays (Webster and Prado, 1970).

Catechol oxidase and peroxidase reaction media contained 10 mmol l⁻¹sodium phosphate at pH7.0 (for catechol oxidase) or 10 mmol l⁻¹Hepes at pH7.0 (for peroxidase activity, because in the presence of phosphate the peroxidase substrate tended to crystallize), 150 mmol l⁻¹NaCl, the indicated amounts of salivary gland homogenates and either 1 mmol l⁻¹ noradrenaline for the catechol oxidase reaction or 0. 2mgml⁻¹o-dianisidine and 0.1 mmol l⁻¹H₂O₂ for peroxidase activities, in a total volume of 100 μl. Indicated amounts of homogenized and diluted salivary glands, usually 0.1 pairs of glands per well, were added. Progression of the reaction, at 37 °C, was followed colorimetrically at 450nm in a 96-well microplate read intermittently on a Molecular Devices Thermomax microplate reader equipped with a computer-based kinetics measurement module. For peroxidase assays, measurements were made at 15s intervals for the first 3min. For catechol oxidase measurements, a lag phase of 30–40 min often occurred, followed by a linear increase in the optical density. The reaction was followed for 3h, with measurements at 15min intervals, and the rate of the reaction was measured after the lag phase. Reaction rates were determined from the slopes of the linear regression of the optical density on reaction time. Correlation coefficients used were always larger than 0.95. We define 1 unit of activity as that resulting from a change in optical density of 1 absorbance unit per minute under the conditions described.

Serotonin destruction in a solution of 1ml of phosphate-buffered saline (PBS) at pH7.0, with 1 μmol l⁻¹serotonin and 10 μmoll⁻¹H₂O₂ was followed fluorometrically on a Perkin Elmer model 650-10S spectrofluorimeter, with excitation set at 295nm and emission at 330nm. Salivary gland homogenates were added, and the change in the fluorescent intensity was followed in an Intel80486 processor based computer using a Dataq analog/digital 12-bit converter. Reaction rates were calculated by linear regression of the linear portion of the fluorescent intensity change, sampled at 5s intervals.

Salivary gland histochemistry for peroxidase activity was carried out by fixing freshly dissected glands in PBS containing 2.5% glutaraldehyde and then transferring the glands to a small Petri dish containing 0.25mgml⁻¹4-chloronaphthol and 3.6mmol l⁻¹H₂O₂ in PBS. The colour developed within 2–5min.

Salivary secretion of peroxidase activity was monitored by exposing female mosquitoes to a 10cm diameter nitrocellulose membrane soaked (but not wet) with PBS containing 10mmol l⁻¹sodium bicarbonate to stimulate probing (Galun et al. 1985) and warmed to 37°C. After 5min of exposure to mosquitoes, the membrane was transferred to a Petri dish containing 0.25mgml⁻¹4-chloronaphthol and 3.6mmol l⁻¹H₂O₂ in PBS. The membrane was carefully placed on the surface of this solution, with the side probed by the mosquitoes facing up, in order to allow diffusion of the substrates from under the membrane while preventing diffusion of the peroxidase out of the membrane. After 15min, the coloured precipitate became fixed on the nitrocellulose mesh and the membrane could be washed with water, dried and stored.

Molecular sieving HPLC was performed on salivary gland homogenates using a TSK 250 column (300mm×7.5mm) and pre-column (75mm×7.5mm, obtained through Bio Rad, USA) with 10mmol l⁻¹Hepes at pH7.0 and 150mmol l⁻¹NaCl at a flow rate of 0.8mlmin⁻¹. The absorbances at 280 and 413nm were recorded, and fractions were collected every 0.4min. To assay for vasodilatory activity, gland homogenates were run with 0.2mol l⁻¹ ammonium formate at pH6.0. Fractions were collected as above, then 10 μl of 10mgml⁻¹bovine serum albumin was added before drying under vacuum in a centrifuge. This allowed concentration of the samples and removal of the volatile buffer. The relative molecular mass markers used were cyanocobalamin, myoglobin, ovalbumin and immunoglobulin G. Retention times of these markers did not differ in the two buffer systems used. Dried fractions were resuspended in 0.1ml of Krebs medium and assayed in the rabbit aortic ring preparation.

Light absorption spectra were measured with a Hewlett Packard model 8452A diode array spectrophotometer. Spectra were stored to a computer disk, and difference spectra were determined with a spreadsheet program.

All animals used in these experiments were treated according to approved protocols reviewed by the University of Arizona Institutional Animal Care and Use Committee.

When salivary gland homogenates of Anopheles albimanus were added to an aortic ring preconstricted with 1 μmol l⁻¹noradrenaline, a slow vasodilation followed (Fig. 1). In six experiments, with different rabbit aortic rings and different salivary gland homogenates, full relaxation of the aorta was obtained when four pairs of salivary glands were added per milliliter of the bioassay media. This vasodilation was independent of the presence of an intact endothelium, and was not seen if the aorta was contracted with 60 mmol l⁻¹KCl instead of noradrenaline. It was also destroyed by exposure to 100 °C for 5min, partially destroyed by exposure to 56°C for 5min and destroyed by exposure to 90% ethanol, or 50% acetonitrile and 0.1% trifluoroacetic acid (data not shown).

The possibility that the salivary homogenate exhibited its vasodilatory activity either by slowly acting on the tissue to release some vasodilatory substance or by changing its susceptibility to noradrenaline was investigated. Aortic rings were preconstricted with noradrenaline for various periods and then 0.5ml of the Krebs medium was added to an aortic strip that was being superfused without NA. As time progressed, less vasoconstrictory activity was retained in the preparation incubated with noradrenaline and salivary homogenates (Fig. 2A). This previous finding could be explained by destruction of noradrenaline by the salivary gland homogenate. To test this hypothesis, noradrenaline in Krebs–Ringer solution was incubated with salivary gland homogenates (without aortic rings) and, at predetermined time intervals, a sample was added to the superfused aortic strip. The results indicate that the salivary homogenate was destroying noradrenaline (Fig. 2B).

It is known that noradrenaline is susceptible to oxidation, producing the corresponding adrenochrome, which has typical absorbance maxima at 300 and 490nm (Klebanoff, 1959). After incubation of 0.1mmol l⁻¹noradrenaline with salivary homogenates for 90min, the difference spectrum obtained was consistent with the oxidation of noradrenaline (Fig. 3).

Incubation of 1mmol l⁻¹noradrenaline with 1 pair of homogenized salivary glands per milliliter in phosphate-buffered saline showed a lag phase of 20–40min (Fig. 4) before the reaction started. This lag phase could be eliminated and the rate of reaction increased by the addition of both xanthine oxidase and xanthine (but not by either alone) to give a reaction producing superoxide. The oxidation of adrenaline could be totally blocked by the combined addition of superoxide dismutase and catalase. Addition of superoxide dismutase alone increased the lag phase, which was followed by a burst of activity (Fig. 4). Salivary gland homogenates also oxidized dihydroxyphenylalanine (DOPA), which could be followed by increased absorbance at 450nm (not shown).

Since heme peroxidases display catechol oxidase activity with kinetics similar to that shown above, including lag kinetics abbreviated by superoxide or hydrogen peroxide donors (Klebanoff, 1959; Paul, 1963), and since Anopheles albimanus salivary glands are a brownish-red color, suggesting the presence of a heme protein, we assayed the homogenates for peroxidase activity. The homogenates displayed a strong activity with o-dianisidine in the presence of hydrogen peroxide. This activity was inhibited by KCN and sodium azide, as expected of heme-containing peroxidases (Fig. 5). Additionally, the visible spectrum of 50 homogenized pairs of glands in 100 μl of PBS indicated a typical Soret band with a maximum at 413nm, which faded upon addition of dithionite and intensified with exposure to carbon monoxide, with a shift of the maximum to 425nm (Fig. 6).

To test whether substrates of pharmacological interest could be destroyed by Anopheles albimanus salivary homogenates, several substances were incubated at concentrations of 1 μmol l⁻¹with homogenates in the presence and absence of 10 μmol l⁻¹H₂O₂. Histamine, measured in the guinea pig ileum preparation, and prostaglandin E₂, measured in the rat uterus preparation, were not destroyed by the homogenates, either in the presence and in the absence of H₂O₂ (data not shown). However, serotonin activity, monitored on the rat uterus preparation, was inhibited.

Incubation of homogenates with serotonin and hydrogen peroxide caused a time-dependent destruction of the biological activity of serotonin (Fig. 7). The intrinsic fluorescence of serotonin was also diminished upon incubation with homogenates and H₂O₂ (Fig. 8A). The reaction was efficient in the use of H₂O₂, as indicated by the destruction of 1 μmol l⁻¹serotonin when submicromolar concentrations of H₂O₂ were added to the medium (Fig. 8B).

To test whether the catechol oxidase and peroxidase activities reside in the same molecule(s), homogenates were submitted to molecular sieving chromatography, and the eluate was monitored for the various activities. Activities for catechol oxidase, o-dianisidine peroxidation, serotonin peroxidation and vasodilation co-eluted with the same retention time as a peak absorbing at 413nm with an estimated relative molecular mass of 50000 (Fig. 9).

The peroxidase activity was localized within the glands by fixing them in glutaraldehyde and staining for peroxidase activity. Stain localized exclusively on the distal, female-specific lobes of the glands (Fig. 10A).

To verify whether the peroxidase activity was of a secretory or a constitutive nature, mosquitoes were allowed to probe a warm membrane of nitrocellulose, which was later incubated with peroxidase substrates. Results confirmed the secretion of the enzyme activity during probing (Fig. 10B). Additionally, the total protein content and peroxidase activity levels of control and blood-fed mosquitoes were measured. The results showed a reduction of 41.9% in salivary protein and 40.5% in salivary peroxidase content following the blood meal (P<0.01 for both comparisons on a Student’s t-test) (Fig. 11).

The results reported above indicate that salivary gland homogenates of the mosquito Anopheles albimanus oxidize noradrenaline. Additionally, homogenates also demonstrate a peroxidatic reaction towards o-dianisidine and serotonin. The enzyme resides in the posterior, female-specific regions of the female salivary glands, and enzymatic activity can be detected in sites probed by hungry mosquitoes. Protein and peroxidase activity are significantly reduced after a blood meal, indicating the secretory fate of this activity during hematophagy. In the following paragraphs we will advance the hypothesis that both oxidase and peroxidase activities reside on a heme protein, discuss the physiological significance of these activities for blood feeding, and speculate on the question of where the superoxide or hydrogen hydroperoxide comes from.

Peroxidases can, with certain substrates, promote oxidatic (as opposed to peroxidatic) reactions. For example, horseradish peroxidase and lactoperoxidase have been shown to oxidize adrenaline (Klebanoff, 1959; Paul, 1963). Additionally, substances such as dihydroxyfumaric acid (Buhler and Mason, 1961; Dordick et al. 1986) or thyroxine-related compounds (Klebanoff, 1959) can be used instead of hydrogen peroxide for the co-oxidation of typical peroxidase reactions (Paul, 1963). Enzymes displaying such behavior belong to the heme peroxidase family (Paul, 1963). Interestingly, salivary glands of Anopheles albimanus have a distinct brownish-red color, particularly in the posterior middle lobe, and homogenates display a typical Soret absorption band. This pigment eluted as a single symmetrical peak on a TSK molecular sieving column, co-eluting with vasodilatory, noradrenaline-oxidizing and peroxidase activities. The simplest explanation for our results is that Anopheles albimanus salivary glands contain a heme peroxidase displaying an oxidatic reaction towards catecholamines.

The secretory role of these activities is apparent from the decrease in activity in the salivary glands following a blood meal and by its presence in sites probed by mosquitoes. The physiological role in feeding of these activities appears to be associated with the inactivation of vasoconstrictor substances released during the hemostatic process. Interestingly, one of the sources of vertebrate peroxidase is the eosinophil, a leukocyte that is usually found as a latecomer to inflammatory reactions and is thought to be involved in down-regulating inflammation (Bruijnzeel, 1989; Kay, 1985; Venge et al. 1987). Its secreted peroxidase has been implicated in the destruction of leukotrienes (Henderson et al. 1982). Since heme peroxidases usually display a very large range of substrate specificity, it is possible that these enzymes also react with unknown plasma precursors, producing other biologically active substances with vasodilatory activity.

Although Anopheles albimanus salivary gland homogenates promote a slow vasorelaxation in noradrenaline-constricted aortic rings, this effect was seen with 1–3 pair of glands per milliliter. The volume of skin probed by mosquitoes could be conservatively estimated to be a 3mm×3 mm×3 mm cube, or around 30 μl. The long spectrophotometric assays were performed with 1 pair of glands per milliliter and 1 mmol l⁻¹noradrenaline, or 1000 times more noradrenaline than the pharmacological (1 μmol l⁻¹) concentrations used in the bioassay. These large amounts were needed for accumulation of enough adrenochrome to be detectable photometrically. It is probable that the concentrations of enzyme injected during feeding would be at least one or two orders of magnitude larger than those used in our assays, decreasing the time needed for its effect.

If the peroxidatic rather than the oxidatic reaction of the homogenate is the main physiological function of this activity, then superoxide, hydrogen peroxide or another suitable co-oxidizing substrate must be supplied to the feeding site while saliva is being secreted. Indeed, the effect of the homogenates on noradrenaline can be increased by orders of magnitude in the presence of superoxide. Additionally, significant destruction of serotonin was seen only in the presence of hydrogen peroxide, the decay product of the short-lived superoxide anion. Superoxide or hydrogen peroxide could be produced either by the mosquito or by the injured host tissue. Activated neutrophils, which respond to ATP and to other factors released during tissue injury by aggregation and superoxide production, could be one source of superoxide (Kuroki and Minakami, 1989; Moon et al. 1990; Stewart et al. 1990). Alternatively, superoxide or hydrogen peroxide could be supplied by the mosquito during salivation. These ideas remain to be tested.

Although this peroxidase activity is the first of its kind to be described in the saliva and salivary glands of a blood-sucking arthropod, it is interesting that both salivary peroxidase and catechol oxidase have been found in phytophagous Heteroptera (Miles and Peng, 1989). The peroxidase found in aphids is active against plant secondary compounds that are feeding deterrents to many other insects. Salivary peroxidase in phytophagous Heteroptera and in the mosquito Anopheles albimanus could be analogous in function, destroying host compounds that would otherwise disrupt feeding.

Many plant secondary compounds are amines, as are many vertebrate hemostatic and inflammatory mediators, and they are potential substrates for oxidases and peroxidases. Taking into consideration the enormous diversity and convergent evolution of insects towards plant-feeding and blood-feeding behavior, it is to be expected that many have found peroxidases to help the feeding process. It is possible that many may have amino-oxidases with a similar function. These enzymes are easily recognized by their FAD requirement and yellow color.

Plants and animals have sophisticated mechanisms to prevent blood, sap or tissue loss to insects. It has become apparent in the last few years that insect salivary glands contain the keys that unlock such defense responses. Since convergent evolution has resulted in the adaptation of different insects to the same mode of feeding on similar hosts, diverse solutions to the same problem may have been achieved. It is reasonable to expect that further knowledge of the biochemistry and pharmacology of arthropod salivary glands will lead to the discovery of novel compounds, to a better understanding of host/insect associations and to a new perspective in analyzing host hemostasis and injury responses.

We thank Drs Donald Champagne, Eddie Cupp, Mary Cupp, René Feyereisen, Mark Novak and Jan Veenstra for useful discussions and comments on the manuscript, Dr T. Meyer for use of the diode array spectrophotometer and Dr R. Feyereisen for the use of the spectrofluorimeter and ELISA reader. This work was supported by PHS grant AI-18694 from the National Institutes of Health.