Mosquito Phenoloxidase and Defensin Colocalize in Melanization Innate Immune Responses

Abstract

Mosquitoes mount strong humoral and cellular immune responses against foreign organisms. Two components of the mosquito immune response that have received much attention are the phenoloxidase cascade that leads to melanization and antimicrobial peptides. The purpose of the current study was to use immunocytochemistry and transmission electron microscopy to identify the location of the melanization rate-limiting enzyme phenoloxidase and the antimicrobial peptide defensin in innate immune reactions against Escherichia coli and Micrococcus luteus by the mosquito Aedes aegypti. Our results show that both phenoloxidase and defensin are present at the sites of melanin biosynthesis in immune reactions against bacteria. Furthermore, both proteins are often present inside the same melanotic capsules. When hemocytes were analyzed, phenoloxidase was present in the cytosol of oenocytoids, but no significant amounts of defensin were detected inside any hemocytes. In summary, these data show that phenoloxidase and defensin colocalize in melanization reactions against bacteria and argue for further studies into the potential role of defensin in phenoloxidase-based melanization innate immune responses in mosquitoes.

Mosquitoes vector viral, protozoan, and metazoan parasites that cause great morbidity and mortality worldwide (World Health Organization 1996; Roberts 2002). Currently, the control of mosquito-borne diseases has consisted of treating infected individuals, killing the mosquito vector, and using physical barriers to limit vector-human contact. Although these approaches have reduced disease prevalence in certain regions, their efficacy is rapidly diminishing. A primary example is the emergence of drug resistance by Plasmodium parasites; virtually all Plasmodium falciparum are chloroquine resistant (Wellems 2002) and resistance to mefloquine is on the rise (Brasseur et al. 1992). It is thought that resistance to artemisinin will likely occur because it has already been observed in laboratory models (Meshnick 2002). Hence, because of the diminishing efficacy of current control methods and the failure to discover new drugs, insecticide replacements, and effective vaccines, it has been necessary to develop new control strategies. It has been suggested that understanding the immune mechanisms used by mosquitoes against pathogens may lead to novel control methods (Beaty 2000).

The mosquito immune response elicited by invading organisms is robust and involves humoral and cellular components. The humoral component includes the phenoloxidase cascade system of parasite melanization and wound healing (Zhao et al. 1995; Lai et al. 2002), inducible antimicrobial peptides (Lowenberger 2001; Vizioli et al. 2001a), and reactive oxygen and nitrogen intermediates (Luckhart et al. 1998; Lanz-Mendoza et al. 2002; Kumar et al. 2003). The cellular component includes phagocytosis (Hernández-Martínez et al. 2002; Hillyer et al. 2003a,b,2004,2005) and encapsulation (Forton et al. 1985; Christensen and Forton 1986) by hemocytes. Of these, two areas of intense study are phenoloxidase-based melanization and the antimicrobial peptides. Melanization involves a complex series of reactions that include the hydroxylation of tyrosine by phenoloxidase to form 3,4-dihydroxyphenylalanine (dopa) and its sequential conversion to melanin by phenoloxidase and other enzymes (Christensen et al. 2005). This process is used by certain mosquito species and strains as a resistance mechanism against filarial nematodes and malaria (Plasmodium) by sequestering parasites inside a hardened proteinaceous capsule (Collins et al. 1986; Beerntsen et al. 1989). Antimicrobial peptides have been studied mostly for their antibacterial activity. They have also been shown or suggested to have anti-parasitic activity and have been hypothesized to function by triggering lysis (Lowenberger et al. 1996,1999a; Vizioli et al. 2001a,b), although this notion has been recently challenged (Bartholomay et al. 2004).

The yellow fever mosquito, Aedes aegypti, is the natural vector of dengue virus and yellow fever virus (World Health Organization 1996; Roberts 2002). Aside from diseases naturally transmitted by Ae. aegypti, this mosquito is often used as a laboratory model because it is easy to rear, its biology and genetics are well understood, and it serves as a model for studies relating to malaria and filarial nematode transmission (Beerntsen et al. 2000). Furthermore, we have previously shown that Ae. aegypti mount strong immune responses against bacteria and that different bacterial species elicit different immune responses (Hillyer et al. 2003b,2004,2005). For example, Gram(-) Escherichia coli are primarily phagocytosed and Gram(+) Micrococcus luteus are primarily melanized (Hillyer et al. 2003b), but the factors triggering phagocytic vs melanization immune responses against bacteria are independent of Gram type (Hillyer et al. 2004). In the current study, we used immunocytochemistry and transmission electron microscopy (TEM) to localize the melanization rate-limiting enzyme phenoloxidase and the antimicrobial peptide defensin in immune reactions against bacteria.

Materials and Methods

Colloidal Gold Suspension Preparation

Colloidal gold was prepared by the reduction of gold chloride with sodium citrate. Briefly, 200 ml double distilled H₂O was placed in an Erlenmeyer flask and heated while stirring with a magnetic bar. When the water was at a near boil, 2 ml of 1% gold(III) chloride trihydrate (HAuCl₄ • 3H₂O; Sigma, St Louis, MO) was added and mixed for 30 sec. A volume of 2.5 ml of 2.5% sodium citrate (Sigma) was then added and the solution soft boiled for 30 min. After the solution cooled to room temperature, the volume was readjusted to 200 ml and stored at 4C. To determine colloidal gold bead size, a nickel 200-mesh pioloform-coated TEM grid was floated on a 30-μl drop of colloidal gold suspension for 10 min. The grid was then removed, blotted to remove excess liquid, dried, and colloidal gold particles visualized by transmission electron microscopy. Micrographs were taken and bead size determined using Image Pro-Plus software (Media Cybernetics; Silver Spring, MD). Calibration was done using the scale bar of the micrograph. Suspensions were made twice: the first colloidal gold suspension averaged 12.7 nm in diameter (±1.02 nm SD) and was conjugated to mouse anti-rabbit IgG, and the second colloidal gold suspension averaged 14.36 nm in diameter (±1.17 nm SD) and was conjugated to protein A (see below).

Conjugation of Proteins

Protein A (Sigma) and mouse anti-rabbit IgG (Pierce Biotechnology; Rockford, IL) were conjugated to colloidal gold particles by hydrophobic bonding. Initially, concentration and pH adsorption isotherms were done to determine the optimal conditions for conjugation (Albrecht et al. 1993). Briefly, 200 μl of colloidal gold at various pHs (ranging from 5 to 9) was added to 20 μl of Protein A or mouse anti-rabbit IgG at varying concentrations (ranging from 200 μg/ml to 25 μg/ml, in water) and mixed. After 20 min, 100 μl of a saturated NaCl solution was added and mixed. After another 20 min the color of the conjugate was assessed: red/pink was indicative of a stable conjugate and blue/purple was indicative of an unstable conjugate. The colloidal gold pH and protein concentration combination that required the minimum amount of protein to stabilize the colloidal gold, plus 10% excess protein, was used to make the final conjugates. For the final conjugation of mouse anti-rabbit IgG to 12.7 nm colloidal gold, 18 ml of colloidal gold (pH 7.3) was added to 1.8 ml of 75 μg/ml mouse anti-rabbit IgG (in water). For the final conjugation of Protein A to 14.4 nm colloidal gold, 10 ml of colloidal gold (pH 5.9) was added to 1 ml of 50 μg/ml Protein A (in water). Both solutions were individually mixed and allowed to stand for 30 min. The conjugates were then spun at 10,600 × g for 8 min, the supernatant removed, and the loose pellet resuspended in storage buffer (20 mM Tris base, 20 mM NaN₃, 150 mM NaCl, 0.1% BSA, pH 7.2) to one tenth the original volume.

Mosquito Rearing, Immune Challenges, and Sample Collections

Because only female mosquitoes take blood meals and, hence, are the only mosquitoes capable of transmitting diseases through infective bites, the current study focused exclusively on the immune response elicited in females. Female Ae. aegypti, Liverpool strain, were reared and maintained inside an environmental chamber at 26.5 ± 0.3C and 80 ± 5% relative humidity and adults fed ad libitum on 0.3 M sucrose-soaked cotton pads as described (Hillyer et al. 2004). E. coli K12 and M. luteus bacteria were grown separately and overnight in Luria-Bertani's rich nutrient medium (10 g tryptone, 5 g yeast extract, 10 g NaCl in 1 liter H₂O; Fisher Scientific, Pittsburgh, PA) inside a 37C shaking incubator until stationary phase had been reached.

To immune challenge mosquitoes, cultured bacteria at stationary phase were centrifuged at 2655 × g for 5 min and the supernatant withdrawn. A 0.15-mm steel probe was dipped in the bacterial pellet and inserted through the neck membrane of 1–5-day-old adult female mosquitoes (Hillyer et al. 2003b). At 1 hr and 24 hr postchallenge, hemolymph was collected by volume displacement (perfusion) directly into formaldehyde (Electron Microscopy Sciences; Hatfield, PA) and glutaraldehyde (Electron Microscopy Sciences) fixative as described (Hillyer et al. 2003a). One min after completing mosquito hemolymph collection, the supernatant was transferred to a new tube, allowed to fix for an additional 1 hr, and centrifuged for 20 min at 210 × g. The supernatant was removed and the cellular pellet was embedded in a plug of low-melting-temperature agarose. The agarose plug was then fixed for 20 min, rinsed with buffer, cut into 1.5-mm cubes, dehydrated through 70% ethanol, and embedded in LR-White resin (Polysciences; Warrington, PA). Each sample was comprised of the pooled hemolymph of 50–100 mosquitoes of the same age. A total of six sets of perfusions were done, each yielding at least two samples after cutting the agarose plugs. Furthermore, samples were prepared for 1-, 4-, and 5-day-old mosquitoes at 24-hr postchallenge and for 5-day-old mosquitoes at 1-hr postchallenge.

Labeling and Transmission Electron Microscopy

All labeling experiments were done at room temperature, and all solutions and rinses were made in storage buffer (20 mM Tris base, 20 mM NaN₃, 150 mM NaCl, 0.1% BSA, pH 7.2) unless otherwise noted. One hundred-nm-thick sections were cut using a diamond knife on a Reichert Om U3 ultramicrotome (Reichert; Vienna, Austria) and transferred to 200-mesh pioloform-coated nickel grids. Sections were allowed to dry for 10–30 min and rehydrated for 5 min. Sections were then blocked (5% fetal calf serum when using IgG conjugate or 1% BSA when using Protein A conjugate) for 30 min, rinsed three times for 3 min each, incubated in primary antibody [rabbit anti-defensin (Bartholomay et al. 2004) or rabbit anti-phenoloxidase (Lai et al. 2002)] for 2 hr, rinsed three times for 3 min each, incubated in Protein A-colloidal gold or mouse anti-rabbit IgG-colloidal gold for 2 hr, rinsed three times for 3 min each, rinsed three times for 3 min each in water, and allowed to dry. Sections were then counterstained for 5 min in 1% uranyl acetate and for 3 min in Reynold's lead, and viewed with a Philips CM120 TEM (Philips Electron Optics; Eindhoven, The Netherlands) as described (Hillyer and Albrecht 1998).

A total of 10 experiments were done, each including multiple samples, and four of the samples were independently labeled in separate trials at least twice. For all labeling experiments, serial sections for all samples were labeled for both defensin and phenoloxidase. Out of the 10 labeling experiments, 6 were done using mouse anti-rabbit IgG-colloidal gold and 4 with Protein A-colloidal gold. These experiments yielded similar results, with the exception that experiments using IgG-colloidal gold resulted in higher labeling intensities. Controls included labeling cultured bacteria (bacteria never injected into mosquitoes), omission of the primary antibody, and the use of an irrelevant antibody. Control samples were run on three of the labeling experiments with identical results. Finally, labeling experiments for samples collected at 1 hr and 24 hr postchallenge were done within days of each other and with the same reagents. Because they were not labeled on the same day, colloidal gold particles on bacterial melanotic capsules and on other regions of the section were counted, and statistical analysis (Student's t-test) was done to compare the background labeling of both samples and the specific labeling of both samples. Student's t-test was also used to compare labeling of bacteria in experimental samples vs background labeling of bacteria in control samples. Differences were deemed significant at p<0.05.

When observing hemolymph samples from bacteria-challenged mosquitoes collected by perfusion, the biological structures present are hemocytes, fat body, cellular debris from lysed cells (predominantly nuclei, mitochondria, and membranes), and bacteria (melanized and unmelanized) (Hillyer and Christensen 2002; Hillyer et al. 2003a). For all labeling experiments, all structures were examined for the presence of colloidal gold particles.

Results

Localization of Phenoloxidase

We have previously reported in naïve adult Ae. aegypti that phenoloxidase is present exclusively in oenocytoid hemocytes (Hillyer and Christensen 2002), and that in Armigeres subalbatus phenoloxidase is involved in the melanization of bacteria (Hillyer et al. 2003a). In the current study, we initially aimed to determine whether phenoloxidase was also involved in the melanization of bacteria by Ae. aegypti. We found this to be the case. At 24 hr postchallenge, phenoloxidase was present in the melanotic capsules of E. coli and M. luteus (Figure 1). Phenoloxidase was detected on bacteria that had been fully melanized (Figure 1C), partially melanized (Figure 1A), and occasionally on bacteria that showed little or no evidence of melanization (Figure 1B). Within melanized bacteria, phenoloxidase was most often present in the melanotic component of the bacteria (Figures 1A–1C). However, on rare occasions, colloidal gold particles were observed in the cytosol (not shown).

When hemocytes were examined, oenocytoids were the only cell type that contained phenoloxidase independent of bacteria and, similar to naïve mosquitoes (Hillyer and Christensen 2002), phenoloxidase was scattered throughout the cytoplasm and not inside membrane-delimited vesicles (Figure 1D). Phenoloxidase was also detected bound to bacteria that had been phagocytosed by granulocytes (Figure 1A). However, because in naïve and infected mosquitoes phenoloxidase is only present in the cytoplasm of oenocytoids (Hillyer and Christensen 2002; this study), phenoloxidase likely binds bacteria and initiates melanization in the hemolymph prior to phagocytosis. In addition to hemocytes and melanized bacteria, phenoloxidase was also occasionally observed on ruptured membranes (Figure 1B) in a manner similar to that seen in Ar. subalbatus (Hillyer et al. 2003a).

Localization of Defensin

Defensin is one of the most studied antimicrobial peptides in mosquitoes. Because it has been widely speculated that defensin functions by lysing bacteria, we originally expected to see defensin bound to unmelanized bacteria. This was not the case. Similar to phenoloxidase, at 24 hr postchallenge, defensin localized to E. coli and M. luteus melanotic capsules (Figure 2). Defensin was commonly present in fully melanized bacteria (Figures 2A–2C), but in a small percentage of cases defensin localized to bacteria that appeared to be in the early stages of melanization (Figure 2D). Within melanized bacteria, defensin was most often found on the melanized capsule (Figure 2A), but it was also observed in the melanized capsule and the cytosol (Figure 2B) or mostly in the cytosol (Figure 2C).

Figure 1

Colloidal gold labeling of mosquito hemolymph tissues for phenoloxidase. (A) Phenoloxidase surrounding partially melanized E. coli that have been phagocytosed by a granulocyte. (B) Phenoloxidase surrounding M. luteus (open arrow), E. coli at what appears to be the onset of melanization (double open arrow), and ruptured membranes (open arrowheads) in the hemolymph. (C) Phenoloxidase surrounding fully melanized M. luteus in the hemolymph. (D) Phenoloxidase in the cytosol of an oenocytoid. N, hemocyte nucleus; Solid arrows, hemocyte plasma membrane; Solid arrowheads, phagosome membrane. Bars = 400 nm.

When hemocytes were examined, defensin was occasionally found inside membrane delimited vesicles of granulocytes (Figures 2B and 2E). In most instances, the defensin molecules in these vesicles were attached to phagocytosed bacteria (Figure 2B), but in rare occasions defensin was also observed in these vesicles in the absence of bacteria (Figure 2E). However, given the rarity of this event and that these membrane-delimited vesicles are resorptive in nature (Hillyer et al. 2003b,2004), it is unlikely that granulocytes produce defensin. Also, defensin was not present in oenocytoids (Figure 2F) or adipohemocytes (not shown).

Figure 2

Colloidal gold labeling of mosquito hemolymph tissues for defensin. (A–C) Defensin surrounding melanized M. luteus. Three labeling patterns were observed: in the melanin material surrounding the bacteria (A), in the melanin material and the cytosol (B), and mostly in the cytosol (C). Furthermore, defensin was present in melanized bacteria present in the hemolymph (A,C), and in melanized bacteria that had been phagocytosed and sequestered inside membrane-delimited vesicles (arrowheads in B denote vesicle membrane). (D) Defensin surrounding an E. coli in the hemolymph. In most instances the bacterium had not been fully melanized. (E) Defensin in a granulocyte membrane-delimited vesicle, an uncommon observation. (F) No defensin was detected inside oenocytoids. Bar = 400 nm.

Because defensin is not detectable in naïve mosquitoes and becomes measurable for the first time several hours following challenge (Lowenberger et al. 1999b; Bartholomay et al. 2004), we labeled, in parallel, samples collected at 1 hr and 24 hr following bacterial challenge. Defensin was present in melanotic capsules at both 1 hr and 24 hr postchallenge (Figure 3). The overall labeling pattern was similar for both timepoints. However, the intensity of labeling at 24 hr postchallenge was 3.6 times higher than at 1 hr postchallenge (Figure 3). When comparing samples labeled at 1 hr postchallenge vs 24 hr postchallenge, background for both was negligible and statistically similar (p=0.1), and specific labeling was statistically different (p=0.01).

Co-localization of Phenoloxidase and Defensin

Prior to the initiation of this work, our working hypothesis was that phenoloxidase would localize to melanized bacteria and defensin would localize to unmelanized bacteria. Surprisingly, both localized to melanized bacteria. Because of this finding, serial sections of hemolymph samples collected at 24-hr postchallenge were cut and one labeled with defensin and the other with phenoloxidase. On many occasions, defensin and phenoloxidase colocalized to the same E. coli and M. luteus melanotic capsules (Figure 4). Not all melanotic capsules labeled with phenoloxidase or defensin, and not all capsules that labeled with one labeled with the other, but this was expected given the protein cross-linking and the reactive intermediates produced during this enzymatic melanization cascade (Christensen et al. 2005), and our previous observations in Ar. subalbatus (Hillyer et al. 2003a). Finally, during these experiments most labeled E. coli were at the earlier stages of melanization, and most labeled M. luteus were at the later stages of melanization. This observation was also expected because the primary immune response against E. coli is phagocytosis, but the primary immune response against M. luteus is melanization (Hillyer et al. 2003b).

Figure 3

Comparison of defensin labeling at 1 hr and 24 hr postchallenge. Samples collected at 24 hr postchallenge (A,B) labeled several times more intensely than samples collected at 1 hr postchallenge (C,D). N, hemocyte nucleus; V, membrane-delimited vesicle; arrows, hemocyte plasma membrane; arrowheads, phagosome membrane. Bar = 400 nm.

Evaluation of Control Samples

Labeling of defensin and phenoloxidase in melanization reactions is specific. When identical labeling experiments were carried out on embedded cultured bacteria (bacteria never injected into mosquitoes), no significant labeling was observed (Figure 5; p>0.001 for all comparisons). This indicated that neither the primary antibodies nor the colloidal gold conjugates bind cultured bacteria. When labeling experiments were repeated on hemolymph samples from challenged mosquitoes but omitting the primary antibodies, no significant labeling was observed. This indicated that the colloidal gold conjugates do not bind mosquito tissues or mosquito immune reactions (Figure 6). When an irrelevant primary antibody was used to label experimental samples, no significant labeling was observed either, indicating that IgGs do not nonspecifically bind mosquito tissues or mosquito immune reactions (not shown). Finally, results for all experimental samples were similar, independent of the colloidal gold conjugate used (protein A-colloidal gold vs mouse anti-rabbit IgG-colloidal gold).

Figure 4

Colocalization of phenoloxidase and defensin in melanization reactions. (A,B) Serial sections were labeled with phenoloxidase (A) and defensin (B). The areas inside the squares are magnified in (C–F). (C–F) The same bacterial melanotic capsules label with both phenoloxidase (C,E) and defensin (D,F). Bars: A,B = 2 μm; C–F = 400 nm.

Discussion

To date, much work has been directed toward characterizing the enzymes involved in melanization reactions (Zhao et al. 1995; Beerntsen et al. 2000). In a similar manner, much work has been directed toward identifying mosquito proteins that have antimicrobial activity in vitro (Lowenberger 2001). Nevertheless, our understanding of these two humoral processes is far from complete. Melanization involves a complex series of reactions that involve the sequential conversion of dopa to melanin by the enzymes phenoloxidase, dopa decarboxylase, dopachrome conversion enzyme, and others (Zhao et al. 1995). This enzymatic cascade is well understood, and silencing genes that produce these enzymes result in decreased melanization efficiency (Shiao et al. 2001; Infanger et al. 2004). However, the molecules responsible for triggering these site-specific melanization reactions have not been well characterized, although recently a thioester-containing protein and a glucan-binding protein have been shown to be necessary for the successful melanization of Plasmodium and filarial worms, respectively (Blandin et al. 2004; Wang et al. 2005). The antimicrobial peptides, on the other hand, are small proteins that have been identified because they possess antimicrobial activity in vitro (Lowenberger 2001). Extensive work has been done to study their transcriptional and translational regulation. However, very little is known about their mechanism of action in vivo. The current study suggests a possible link between the phenoloxidase melanization cascade and the antimicrobial peptide defensin.

In Ae. aegypti, developmental Northern blots showed that the gene coding the antimicrobial peptide defensin is not transcribed in the larval stages or adult mosquitoes (Lowenberger et al. 1999b). However, high levels of transcript are detected in callow (white) pupae but not in black pupae, an interesting finding given that the development from callow pupae to black pupae usually takes 12 hr (Lowenberger et al. 1999b). Previously, it was suggested that defensin upregulation in this developmental stage is likely 1. to lyse bacteria entering through the delicate early pupal cuticle prior to sclerotization, 2. to lyse bacteria released during histolysis of specific tissues during metamorphosis, or 3. because defensin may have a dual role in defense and development (Lowenberger et al. 1999b). Upon further review, it is unlikely that in these instances defensin acts in defense because similar events occur between larval molts and between pupal-adult ecdysis, a time when transcription of the defensin gene is unchanged. However, the timing of defensin transcription correlates with the time of extensive melanization during cuticular sclerotization. These observations and our data showing the colocalization of defensin and phenoloxidase in the antibacterial immune melanization response in adults suggest that there may be a link between defensin and the melanization cascade. For example, it is possible that one function of defensin may be the initiation of phenoloxidase-based melanization reactions, explaining its necessity during pupal cuticular sclerotization. However, we cannot discount the possibility that hemolymph proteins are randomly incorporated into melanization reactions, and that the high concentration of defensin in the hemolymph at 24 hr postchallenge (45 μM; Lowenberger et al. 1995) explains its presence in melanization immune responses.

Figure 5

Phenoloxidase and defensin labeling of cultured bacteria (bacteria never injected into a mosquito). (A,B) Phenoloxidase labeling of E. coli (A) and M. luteus (B). (C,D) Defensin labeling of E. coli (C) and M. luteus (D). No significant labeling was observed. Bar = 400 nm.

Figure 6

Labeling of mosquito immune reactions in the absence of primary antibody. No significant labeling of E. coli (arrowheads in A,C,D), M. luteus (solid arrows in B,D), hemocytes (A,B), or ruptured membranes (open arrows in A) was observed. N, hemocyte nuclei. Bar = 400 nm.

Furthermore, in the current study, the cellular immunolocalization of phenoloxidase was as expected: in bacteria-challenged mosquitoes, phenoloxidase was found exclusively in oenocytoids. This is in accord with our previous studies in Ae. aegypti and Ar. subalbatus that showed that oenocytoids are the only hemocytes containing phenoloxidase in naïve, and naïve and bacteria-challenged mosquitoes, respectively (Hillyer and Christensen 2002; Hillyer et al. 2003a). However, we were unable to determine the source of defensin. Northern analyses have shown that defensin is produced in the fat body (Lowenberger et al. 1999a). Hence, we expected to find defensin in adipohemocytes: a population of cells also collected during mosquito perfusions that, because of morphological similarities, have been hypothesized to be cells sloughed off the fat body during hemolymph collection (Hillyer and Christensen 2002; Hillyer et al. 2003a). Possibilities for the lack of labeling in adipohemocytes include that they are different from fat body cells, or that defensin is produced in specific regions of the fat body, and that these regions are different from the ones adipohemocytes originate from.

In summary, these data show that defensin and phenoloxidase colocalize in melanization immune responses and suggest that the antimicrobial peptide defensin may be involved in the mosquito melanization response against bacteria. These data argue for further studies into the potential role of defensin in melanization innate immune responses in mosquitoes.