Cell and Tissue Research

, Volume 313, Issue 1, pp 117–127

Hemocyte-mediated phagocytosis and melanization in the mosquito Armigeres subalbatus following immune challenge by bacteria


  • Julián F. Hillyer
    • Department of Animal Health and Biomedical SciencesUniversity of Wisconsin-Madison
  • Shelley L. Schmidt
    • Department of Animal Health and Biomedical SciencesUniversity of Wisconsin-Madison
    • Department of Animal Health and Biomedical SciencesUniversity of Wisconsin-Madison
Regular Article

DOI: 10.1007/s00441-003-0744-y

Cite this article as:
Hillyer, J.F., Schmidt, S.L. & Christensen, B.M. Cell Tissue Res (2003) 313: 117. doi:10.1007/s00441-003-0744-y


Mosquitoes are important vectors of disease. These insects respond to invading organisms with strong cellular and humoral immune responses that share many similarities with vertebrate immune systems. The strength and specificity of these responses are directly correlated to a mosquito's ability to transmit disease. In the current study, we characterized the hemocytes (blood cells) of Armigeres subalbatus by morphology (ultrastructure), lectin binding, enzyme activity, immunocytochemistry, and function. We found four hemocyte types: granulocytes, oenocytoids, adipohemocytes, and thrombocytoids. Granulocytes contained acid phosphatase activity and bound the exogenous lectins Helix pomatia agglutinin, Galanthus nivalis lectin, and wheat germ agglutinin. Following bacteria inoculation, granulocytes mounted a strong phagocytic response as early as 5 min postexposure. Bacteria also elicited a hemocyte-mediated melanization response. Phenoloxidase, the rate-limiting enzyme in the melanization pathway, was present exclusively in oenocytoids and in many of the melanotic capsules enveloping bacteria. The immune responses mounted against different bacteria were not identical; gram(−) Escherichia coli were predominantly phagocytosed and gram(+) Micrococcus luteus were melanized. These studies implicate hemocytes as the primary line of defense against bacteria.


MosquitoHemocytePhenoloxidasePhagocytosisArmigeres subalbatus (Insecta)


Mosquitoes kill and incapacitate millions of people each year by transmitting malaria, lymphatic filariasis, dengue fever, Japanese encephalitis, and other diseases (Roberts 2002). The human and economic toll resulting from mosquito-borne diseases led the World Health Organization in 1996 to label mosquitoes "mankind's most indomitable foe" and "public health enemy number one" (WHO 1996). Many efforts directed at curtailing disease transmission have focused on vector control. Initially, these programs emphasized the use of pesticides; however, with the widespread emergence of insecticide resistance it has been necessary to develop alternative control methods.

Mosquitoes and other insects are able to mount powerful immune responses against invading pathogens, and the strength and specificity of these responses are directly correlated to vector competence (Beerntsen et al. 2000). Interactions between mosquitoes and pathogens are complex and vary greatly even with closely related organisms. For example, strains of Aedes aegypti have been selected to be effective vectors of the filarial worms Brugia malayi and B. pahangi, whereas Armigeres subalbatus is naturally resistant to B. malayi but susceptible to B. pahangi (Beerntsen et al. 1989). Even within a species, mating and genetic engineering strategies have produced resistant and susceptible strains of mosquitoes (Collins et al. 1986; Ito et al. 2002), but the factors conferring these traits remain poorly understood. By understanding the immune responses mounted by different species and strains, it may be possible to genetically manipulate these organisms to make parasite-vector biologic interplay incompatible (Alphey et al. 2002; Curtis 1968).

The mosquito immune system has both humoral and cellular components and shares many similarities with vertebrate innate immune systems (Barillas-Mury et al. 2000; Hernández-Martínez et al. 2002; Hillyer and Christensen 2002; Hillyer et al. 2003; Levashina et al. 2001; Lowenberger 2001). To date, most studies have focused on the humoral response and have virtually ignored hemocytes (blood cells). Melanotic encapsulation is a humoral process that has received much attention (Beerntsen et al. 2000; Zhao et al. 1995). It is a complex enzymatic process in which melanin polymers cross-link with proteins, a reaction that is believed to facilitate parasite killing either through starvation or oxidative damage. Ar. subalbatus is the only mosquito species known to use melanotic encapsulation as a natural defense mechanism against ingested filarial parasites and it is this process that is responsible for its resistance to B. malayi. Whereas other mosquito species manifest a limited melanization potential, Ar. subalbatus has a virtually unlimited melanization capacity because in the adult stage hemocytes can synthesize new phenoloxidase, a requisite enzyme in the pathway (Cho et al. 1998; Huang et al. 2001). Aside from melanotic encapsulation, the cellular immune response of Ar. subalbatus has gone essentially unstudied. In the current investigation we morphologically and cytochemically characterized the hemocytes from Ar. subalbatus and explored their reaction to foreign invaders.

Materials and methods

We previously described in detail the methodologies for studying mosquito hemocytes at the light- and electron-microscopic levels (Hillyer and Christensen 2002; Hillyer et al. 2003). The term hemocyte used in this study refers to the cells initially collected by saline perfusion of the hemocoels of adult Ar. subalbatus, and subsequently differentiated by both light and electron microscopy and immunocytochemical methods. Hemocyte classifications were determined using the methods of Hillyer and Christensen (2002).

Mosquito maintenance and hemolymph collection

Ar. subalbatus were reared as previously described (Beerntsen et al. 1989). Hemolymph (blood) from 1- to 4-day-old adult female mosquitoes was collected by volume displacement (perfusion) (Beerntsen and Christensen 1990).

Bacterial inoculations and latex bead injections

Escherichia coli K12, green fluorescent protein (GFP)-expressing E. coli DH5 alpha, and Micrococcus luteus cultures were grown at 37°C overnight in Luria-Bertani's rich nutrient medium. A 0.15-mm-diameter stainless steel probe was dipped in a bacterial pellet and briefly inserted through the mosquito's neck membrane.

Latex beads were purchased from Sigma (St. Louis, MO). A volume of 0.75 µl (approximately 130,000 beads) 0.88-µm-diameter carboxylate-modified polystyrene latex beads was injected through the neck membrane using a microinjection needle.

Light microscopy

Hemolymph from individual naïve, E. coli inoculated, GFP-expressing E. coli inoculated, or M. luteus inoculated mosquitoes was perfused onto uncoated slides or poly-l-lysine-coated coverslips. For hemocyte morphology and phagocytosis studies, cells were examined live using differential-interference-contrast (DIC) optics and/or GFP epifluorescence on an Olympus Provis AX70 light microscope connected to a Diagnostic Instruments Spot RT Slider digital camera. Digital images were taken, levels adjusted using Spot software, and DIC images passed through a HiGauss filter (kernel size = 7×7, pass = 1, strength = 10) using Image-Pro Plus. Enzyme activity assays were done as described by Kiernan (1990) and modified by Hillyer and Christensen (2002). Briefly, hemocytes were allowed to adhere to coverslips for 20 min, fixed in 4% formaldehyde, enzyme activity assays carried out, and coverslips mounted onto slides using Aqua Poly/Mount. The activity of phenoloxidase (monophenol monooxygenase, tyrosinase), the rate-limiting enzyme of the melanization pathway, was detected by the oxidation of l-dihydrophenylalanine (l-DOPA) to melanin and melanin precursors. Acid phosphatase activity was detected by the precipitation of lead phosphate and subsequent color conversion using ammonium sulfide. Enzyme activity slides were examined using bright-field illumination and DIC, and images collected and processed as described above. Of the four hemocyte types found, enzyme activities could only be studied in granulocytes and oenocytoids because the other two cell types, adipohemocytes and thrombocytoids, were washed away during the procedure.

Transmission electron microscopy

Transmission electron microscopy (TEM) sample blocks were prepared from the pooled hemolymph of 100–200 mosquitoes. Phosphate buffer (pH 7.0; PB) was used to perfuse hemolymph directly into tubes containing fixative. For ultrastructural studies the initial fixative used was 3% glutaraldehyde and 2% formaldehyde in PB. For immunocytochemical studies the initial fixative used was 0.5% glutaraldehyde and 2% formaldehyde in PB. However, the final fixative concentrations were lower because during the course of hemolymph collection the fixative was diluted 1:3. Samples were centrifuged to form a cellular pellet, rinsed, postfixed in 2% osmium tetroxide (omitted for immunocytochemical studies), rinsed, dehydrated through a graded ethanol series, and embedded in epoxy or acrylic resins. The resins used were a 1:1 mixture of EMBED 812 and Spurr's low viscosity embedding media for ultrastructural studies, and LR-white for immunocytochemical studies. Ultrathin sections were cut, placed on pioloform-treated 200-mesh grids and dried. Sections were then subjected to antibody or lectin labeling (see below), or stained with uranyl acetate and Reynold's lead, and viewed using a Philips CM120 TEM operating at 80 kV accelerating voltage under conditions previously described (Hillyer and Albrecht 1998). Samples collected for ultrastructural studies were naïve, E. coli and M. luteus inoculated (24 h), E. coli inoculated (1 h, 2 h, 24 h), GFP-expressing E. coli inoculated (2 h), M. luteus inoculated (1 h, 2 h, 24 h), and latex bead injected (24 h). For labeling experiments, hemocyte samples were taken from naïve, E. coli and M. luteus inoculated (24 h), E. coli inoculated (0.5 h, 24 h), and M. luteus inoculated mosquitoes (0.5 h, 24 h).


Direct and indirect colloidal gold labeling using exogenous lectins was done to characterize the different hemocyte types found in Ar. subalbatus. The lectins Helix pomatia agglutinin (HPA), Galanthus nivalis lectin (GNL, snowdrop), concanavalin A (ConA), soybean agglutinin (SBA), pokeweed (PWM), wheat germ agglutinin (WGA), and peanut agglutinin (PNA) were chosen because they bind hemocytes from other arthropod and mosquito species (Hillyer and Christensen 2002; McKenzie and Preston 1992; Nappi and Christensen 1986; Pipe et al. 1997; Wilson et al. 1999). Labeling for phenoloxidase was done using a rabbit antibody directed to the copper-binding region of Ar. subalbatus prophenoloxidase I (Lai et al. 2002).

Colloidal gold suspensions were made as described by Albrecht et al. (1993), and particle size measured in a manner similar to that described by Hillyer and Albrecht (1998). Optimal conditions for the direct conjugation of colloidal gold nanoparticles to HPA, GNL, ConA, SBA, PWM, PNA, ovomucoid (OVO), protein A, and mouse anti-rabbit IgG were determined by pH and adsorption isotherms (Horisberger and Rosset 1977; Roth 1983) and conjugation done by hydrophobic bonding (Albrecht et al. 1993). The indirect labeling of WGA was done using OVO-colloidal gold because WGA does not form a stable conjugate (Geoghegan and Ackerman 1977). The indirect detection of phenoloxidase was done using protein A-colloidal gold (for naïve samples) or mouse anti-rabbit IgG-colloidal gold (for bacteria inoculated samples). Labeling procedures were those described by Hillyer and Christensen (2002). To assess cross-reactivity between the phenoloxidase antibody and the bacteria used during challenges, TEM samples of E. coli and M. luteus were prepared and labeled. E. coli showed slight, but minimal cross-reactivity (some bacterium had one to three beads) and M. luteus exhibited no cross-reactivity (data not shown).


Hemocyte types and morphology

The hemocytes from Ar. subalbatus were grouped into four subclasses: granulocytes, oenocytoids, adipohemocytes, and thrombocytoids.


The most abundant cell type found in Ar. subalbatus is the granulocyte, which represents approximately 95% of the total cell population that adheres to glass (Figs. 1A–D, 2A). In freshly collected samples, these spherical cells average 9 μm in diameter. On settling, granulocytes readily adhere to glass microscope slides and spread to slightly greater than four times their original size (Fig. 2A). Spreading occurs in a fan-like manner with star-like filopodia extending from the cell body (Fig. 2D). Often, granulocytes clump and spread while in contact with one another (Fig. 2A, B, E, I). The main characteristic of granulocytes is the presence of numerous membrane-delimited vesicles that vary from the more common electron-lucent to electron-dense (Fig. 1A–D). At the light-microscopic level different vesicles can be observed, including clear (most common), metallic red (second most common), pale green, pale yellow, and pale blue. Although their function and composition remain to be determined, many vesicles exhibit acid phosphatase activity indicative of lysosomes (Fig. 2A). Another organelle commonly found in granulocytes was rough endoplasmic reticulum (RER), which varies from very high levels (Fig. 1A) to almost none (Fig. 1B), with most cells having intermediate levels. Additionally, granulocytes contain numerous polyribosomes and few mitochondria. When naïve hemocytes were labeled with the panel of seven lectins, three lectins strongly bound granulocytes: WGA, GNL, and HPA (Fig. 1B–D). Binding for all three lectins exhibited a similar pattern; most labeling was concentrated in the membrane-delimited vesicles with some labeling of the plasma membrane. Granulocytes labeled most intensely with HPA, followed by GNL, and then WGA. The lectins ConA, SBA, PWM, and PNA did not bind granulocytes.
Fig. 1A–F.

Armigeres subalbatus granulocytes and adipohemocytes. A Granulocyte with electron-lucent vesicles, electron-dense vesicles, and rough endoplasmic reticulum. Nucleus contains both heterochromatin and euchromatin. B Granulocyte labeled with the lectin Helix pomatia agglutinin (HPA). C Granulocyte labeled with Galanthus nivalis lectin. D Granulocyte labeled with the lectin wheat germ agglutinin. Binding for all three lectins followed a similar pattern; most labeling was localized in the membrane-delimited vesicles. E Adipohemocyte with large lipid droplets and mitochondria. Most of the cytoplasm is filled with glycogen particles arranged in the alpha configuration (E, inset). F Adipohemocyte labeled with the lectin HPA. Labeling is concentrated in the cytoplasm. Scale bars 2 µm (A), 0.5 µm (B, C, D, F), 5 µm (E)

Fig. 2A–J.

Light microscopy of Armigeres subalbatus hemocytes. A Granulocytes possess acid phosphatase activity (brown staining). B Oenocytoids possess phenoloxidase activity (brown staining, arrow). Note that granulocytes do not stain for the activity of this enzyme (e.g., arrowheads). Most oenocytoids are round and do not attach strongly to glass (top inset) while a small proportion spread, and extend filopodia (bottom inset). C Adipohemocyte with large lipid droplets (e.g., arrow) and metallic red vesicles (e.g., arrowhead). D, E Differential interference contrast and epifluorescence image overlays showing that granulocytes phagocytose large numbers of E. coli as early as 5 min postinoculation. F Melanization of E. coli 5 min postinoculation (arrow). Capsules are brown. G Melanization of E. coli at 24 h postinoculation (arrow). Note a phagocytosed unmelanized E. coli (arrowhead). H Clump of melanized (e.g., arrow) and unmelanized (e.g., arrowhead) E. coli at 24 h postinoculation. I Melanization of M. luteus at 5 min postinoculation (e.g., arrows). Capsules are black. Notice the presence of unmelanized or not significantly melanized M. luteus (arrowhead). Melanized bacteria were also seen loose in the hemocoel (insets). J Melanization of M. luteus at 24 h postinoculation. Some capsules are black and others are brown. Scale bar 10 µm


Oenocytoids average 9 μm in diameter and manifest a homogeneous cytoplasm rich in polyribosomes (Fig. 3A). They contain few mitochondria and few to no membrane-delimited vesicles. Phenoloxidase-positive reactivity was detected by both light and electron microscopy, with this enzyme and its activity scattered throughout the cytoplasm, and not localized within membrane-delimited vesicles (Figs. 2B, 3B, C). Phenoloxidase labeling of the nucleus was also observed. The relative abundance of oenocytoids was hard to determine because of the limited capacity of these cells to adhere to glass microscope slides. However, based on phenoloxidase-positive staining of the cells adhering to glass slides, oenocytoids represent 5.2% of the total hemocyte population. When naïve hemocytes were labeled with exogenous lectins, WGA, GNL, and HPA very weakly bound some oenocytoids (data not shown). Whenever binding occurred, it was localized to the few membrane-delimited vesicles present. The lectins ConA, SBA, PWM, and PNA did not bind oenocytoids.
Fig. 3A–D.

Armigeres subalbatus oenocytoids and thrombocytoids. A Oenocytoid with a homogeneous cytoplasm. B Oenocytoid labeled with anti-phenoloxidase antibody. The area inside the square is magnified in C. C Phenoloxidase is present scattered throughout the cytoplasm of oenocytoids. D Thrombocytoid with a homogeneous cytoplasm. Thrombocytoids do not spread on glass (inset) or contain phenoloxidase (not shown). Scale bars 2 µm (A), 1 µm (B), 0.4 µm (C), 5 µm (D)


Adipohemocytes are the second most common cell perfused from naïve mosquitoes. These cells are large (approximately 30 µm in diameter) and several times the size of granulocytes and oenocytoids (Figs. 1E, F, 2C). They do not attach to glass and are fragile, with many lysing within 30 min postperfusion. Adipohemocytes have a large nucleus and large lipid droplets. Most of the remaining cytoplasm is composed of glycogen particles arranged in the alpha configuration (Fig. 1E, inset). Additional structures present are mitochondria and a few electron-dense vesicles, some of which resemble the metallic red vesicles seen in granulocytes (Fig. 2C). When adipohemocytes were labeled with lectins, HPA strongly bound the cytoplasm (Fig. 1F). WGA and GNL very weakly bound the few lysosomes present (data not shown). The lectins ConA, SBA, PWM, and PNA did not bind adipohemocytes.


Thrombocytoids are similar in size to adipohemocytes (approximately 35 µm in diameter), few in number, do not attach to glass, and lyse shortly after perfusion. Thrombocytoids are similar to oenocytoids in possessing a homogeneous cytoplasm with many polyribosomes and few mitochondria (Fig. 3D), but they differ from oenocytoids in lacking phenoloxidase. When naïve hemocytes were labeled with lectins, WGA, GNL, and HPA very weakly bound the few lysosomes inside thrombocytoids (data not shown). The lectins ConA, SBA, PWM, and PNA did not bind thrombocytoids.

Bacteria inoculations and latex bead injections

Ar. subalbatus hemocytes mount strong and rapid phagocytic and melanization responses towards E. coli and M. luteus. The responses against both species start as early as 5 min postinoculation, but the mechanisms differ markedly. The primary response towards the gram(−) bacterium E. coli was phagocytosis. When mosquitoes were inoculated with GFP-expressing E. coli and hemolymph collected 5 min later, granulocytes were observed actively phagocytosing multiple bacteria (Fig. 2D, E). In some instances, in excess of 40 bacteria could be seen within or on the surface of a single granulocyte. At 1 h postinoculation, 100-nm-thick sections of individual granulocytes often contained upwards of 30 E. coli, suggesting that a single granulocyte can phagocytose over 100 bacteria (Fig. 4A). Phagocytosis of E. coli, as well as degradation (darkening and breaking up of bacterial membranes), was also observed at 2 h (Fig. 4C) and 24 h (Fig. 4D) postinoculation. In all instances, phagosome membranes wrapped tightly around phagocytosed E. coli. Melanization of E. coli was also noted as early as 5 min postinoculation, but this process was not as common and seemed to be a secondary response (Fig. 2F). At the ultrastructural level, well-developed melanized capsules were formed as early as 1 h postinoculation, and some of these capsules had been phagocytosed (Fig. 4B, C). Because oenocytoids are the only hemocytes that contain phenoloxidase, bacterial melanization must have occurred in the hemocoel prior to them being phagocytosed by granulocytes. Clumping of melanized bacteria was also observed (Fig. 2H).
Fig. 4A–I.

Phagocytosis and melanization of bacteria and latex particles by Armigeres subalbatus. A Granulocyte with approximately 30 phagocytosed E. coli at 1 h postinoculation. B Melanized E. coli at 1 h postinoculation. C Phagocytosis of unmelanized (e.g., arrows) and melanized (arrowhead) E. coli by a granulocyte at 2 h postinoculation. D Phagocytosis of E. coli by a granulocyte at 24 h postinoculation. E Melanized M. luteus at 1 h postinoculation. Note that the melanotic capsule is denser and thicker than that of E. coli at the same timepoint (B). F Phagocytosis of melanized and partially degraded M. luteus by a granulocyte at 1 h postinoculation. G Phagocytosis of melanized M. luteus by a granulocyte at 2 h postinoculation. Notice two melanotic capsules have been cut on different planes of section (arrowheads). H Phagocytosis of melanized M. luteus by a granulocyte at 24 h postinoculation (arrowhead). I Phagocytosis of latex particles by granulocytes at 24 h postinoculation (e.g., arrowheads). Scale bars 2 µm (A, C, F, G, H, I), 0.5 µm (B, E), 1 µm (D)

The primary response towards the gram(+) bacterium M. luteus was melanization, which was detected as early as 5 min postinoculation (Fig. 2I). By 24 h postinoculation, most M. luteus had been melanized (Fig. 2J). At the electron-microscopic level, completely melanized capsules measuring 2 μm in diameter were observed at 1 h postinoculation (Fig. 4E). This rapid host response may actually be a prerequisite for phagocytosis, as the only M. luteus internalized were those that were either completely melanized, or enveloped by a flocculent material that likely represents the early stages of melanin being deposited on the surface of the bacterium (Fig. 4F–H). This flocculent material often prevented the phagosome membranes from wrapping tightly around the bacteria. Melanization and phagocytosis of M. luteus was observed at all timepoints studied.

Interestingly, melanization of E. coli and M. luteus was different. At 5 min postinoculation, E. coli melanotic capsules were brown (Fig. 2F) and M. luteus capsules were black or very dark brown (Fig. 2I). At 1 h postinoculation, M. luteus melanotic capsules were more electron-dense than E. coli capsules (Fig. 4B, E). At this time, many M. luteus capsules had reached a density state where the embedding resin was unable to infiltrate them and by 24 h most had reached this impenetrable state. E. coli capsules with impeded resin infiltration were rarely observed at 1 h postinoculation and in modicum at 24 h postinoculation, but never at the levels seen for M. luteus.

Granulocytes also phagocytosed large numbers of latex particles (Fig. 4I). Although infrequent, these particles were also melanized (data not shown). In our previous study we reported that Ae. aegypti granulocytes phagocytosed but did not melanize latex particles (Hillyer et al. 2003). In this study we used latex particles from the same manufacturer, and although they were purportedly identical (same catalog number), there were obvious differences (e.g., color). Hence, we repeated the experiment in Ae. aegypti and found that this mosquito phagocytosed and also occasionally melanized these latex particles. We do not know the chemical and physical differences between the two bead preparations that led to different outcomes.

Granulocytes did the bulk of phagocytosis but other cells also participated. Oenocytoids occasionally phagocytosed small numbers of bacteria, bacterial melanotic capsules and latex beads, and, in an isolated event, one thrombocytoid had internalized E. coli. Finally, the amount of biological material recovered at 24 h postinoculation was 2–4 times greater than that obtained when hemolymph was collected from naïve mosquitoes. Whether this is the result of tissue damage during infection or the release of more hemocytes into circulation is currently under investigation.

Phenoloxidase-based melanization of bacteria

To corroborate whether the electron-dense capsules formed around inoculated bacteria were the result of phenoloxidase-based melanization, we labeled samples with anti-Ar. subalbatus phenoloxidase I, the isoform of the enzyme responsible for immunologic melanization in this species (Shiao et al. 2001). Phenoloxidase labeling of E. coli and M. luteus melanotic capsules was observed at both timepoints studied (0.5 and 24 h; Fig. 5A, C). Labeling intensity was higher in capsules collected 0.5 h postinoculation, and was more frequently observed with capsules that were in their initial stages of melanization. Phenoloxidase was not detected in all capsules nor was labeling intensity particularly strong; however, labeling was consistently above background. We hypothesize that the low strength of labeling is the result of phenoloxidase molecules losing their antigenicity as other proteins cross-link around them during capsule hardening. At 0.5 and 24 h postinoculation, phenoloxidase also was detected in some unmelanized E. coli (Fig. 5B). This labeling was more common at 24 h postinoculation and is probably restricted to E. coli that were undergoing the initial stages of melanization. Furthermore, many of these labeled E. coli appeared damaged, as indicated by surrounding flocculent material and/or ruffled membranes. Similar labeling of unmelanized M. luteus was observed at 0.5 h, but unmelanized M. luteus were not observed at 24 h. When hemocytes themselves were examined for phenoloxidase, the cytoplasm and nucleus of oenocytoids labeled strongly, along with some loose membranes (Fig. 5B top inset, D). Granulocytes, adipohemocytes, and thrombocytoids did not label for phenoloxidase; thus, hemocyte production of new phenoloxidase I was limited to oenocytoids.
Fig. 5A–D.

Phenoloxidase labeling of bacteria inoculated into Armigeres subalbatus. A Colloidal gold labeling of melanized E. coli at 0.5 h postinoculation. B Strong labeling of unmelanized E. coli at 24 h postinoculation. Labeled bacteria often showed signs of degradation, such as membrane ruffling (bottom inset). In all samples, loose membranes that are likely ruptured oenocytoid plasma membranes also labeled (top inset). C Labeling of melanized M. luteus at 0.5 h postinoculation. Phenoloxidase was also detected surrounding some unmelanized M. luteus at 0.5 h postinoculation (insets). D Oenocytoid with phenoloxidase scattered throughout the cytoplasm. Oenocytoids were the only hemocyte type that contained this enzyme. Scale bar 0.5 µm


Ar. subalbatus is a natural vector of Japanese encephalitis and the filarial nematodes B. pahangi and Dirofilaria (Nochtiella) repens (Beerntsen et al. 1989; Chen et al. 2000; Dissanaike et al. 1997). Herein we characterized the hemocytes of Ar. subalbatus as granulocytes, oenocytoids, adipohemocytes, and thrombocytoids. Furthermore, we showed that mosquitoes rapidly mount strong phagocytic and melanization responses against inoculated bacteria and that this response is pathogen specific.

This investigation documents the existence of an efficient cellular innate immune system in adult Ar. subalbatus in response to intrahemocoelic inoculations of two species of bacteria, E. coli and M. luteus. Hemocyte-mediated phagocytosis and melanization responses, which result from the activities of granulocytes and oenocytoids, respectively, appear within 5 min following immune challenge. The identity of the factors involved in initiating phagocytosis and melanization are not known, but the speed at which these hemocyte-mediated responses are manifested suggests they constitute a first line of defense in Ar. subalbatus, and that the unidentified factors initiating these responses are either present before, or synthesized extremely rapidly after, immune challenge. This also appears to be the situation in bacteria inoculated Ae. aegypti (Hillyer et al. 2003). These data argue against the proposal that inducible immune peptides represent the primary mechanism for clearing bacteria and other microbial pathogens in these mosquito hosts as transcriptional data depict a later deployment of these molecules (Dimopoulos et al. 2001; Lowenberger 2001; Lowenberger et al. 1999a, 1999b; Richman et al. 1996).

Of considerable interest was our observation that phagocytosis was the predominant response made by Ar. subalbatus against E. coli, whereas against M. luteus melanization was more commonly employed. These data suggest the two species of bacteria elicit different recognition and/or immune effector mechanisms, as well as specific signal transduction cascades that regulate these processes. This differential response towards gram(+) and gram(−) bacteria has also been observed in Ae. aegypti and Anopheles albimanus, although the timings are not identical. Ar. subalbatus and Ae. aegypti both melanize and phagocytose bacteria as early as 5 min postinoculation but An. albimanus does not begin melanizing until 30 min and phagocytosing until 6 h (Hernández-Martínez et al. 2002; Hillyer et al. 2003). The differing responses towards gram(+) and gram(−) bacteria are in accord with studies in Drosophila melanogaster that have shown that two different pathways, Toll and IMD, are responsible for specific responses towards gram(+) and gram(−) bacteria (Hoffmann and Reichhart 2002).

Ar. subalbatus is the only mosquito known to utilize melanotic encapsulation as a natural form of resistance against filarial nematodes (Beerntsen et al. 2000). The underlying factors involved in recognition and melanization of filarial worms are not known, but they are likely associated with hemocytes as these cells are present at the encapsulation sites and possess enzymes that play critical roles in melanogenesis (Christensen and Forton 1986; Hillyer and Christensen 2002; Johnson et al. 2003). The melanization capacity of Ar. subalbatus versus Ae. aegypti is different. Following infection, Ar. subalbatus kills a much higher percentage of worms than Ae. aegypti (Beerntsen et al. 1989), a trait that can be attributed to the synthesis of new phenoloxidase exclusively by Ar. subalbatus hemocytes (Cho et al. 1998; Taft et al. 2001). Because of immune and genetic differences (Beerntsen et al. 1989; Ferdig et al. 1998), we carried out parallel experiments to address hemocyte structure and function in these two mosquito species in efforts to elucidate mechanisms underlying the differences in vector competence (Hillyer and Christensen 2002; Hillyer et al. 2003). We found that both species have four distinct hemocyte populations and that they possess the same enzyme activities and are cytochemically and ultrastructurally similar, with the exception of Ar. subalbatus granulocytes having considerably higher RER content. Similar to Ae. aegypti, we believe granulocytes and oenocytoids are true circulating hemocytes, but because of size, fragility, and lack of apparent immune involvement, we believe adipohemocytes and thrombocytoids are likely fixed-tissue cells that are artifacts of the perfusion process. Following immune challenge, both mosquito species rapidly respond to the gram(−) bacterium E. coli and the gram(+) bacterium M. luteus by phagocytosis and melanization, respectively. In Ae. aegypti, melanization can be observed at 5 min postinoculation as a slight yellowing of the bacteria (Hillyer et al. 2003), but melanization by Ar. subalbatus is much faster; at 5 min postinoculation complete melanotic capsules can be seen surrounding E. coli and M. luteus. Furthermore, melanization of E. coli and M. luteus by Ae. aegypti results in brown melanotic capsules but melanization by Ar. subalbatus results in brown E. coli capsules and black or dark-brown M. luteus capsules. The extent of melanization is also different; melanization of E. coli by Ar. subalbatus is more common and in a minority of samples studied even appeared to exceed the level of phagocytosis.

In recent years, research findings in the field of vertebrate immunology have demonstrated that the innate immune system is essential and of major importance in pathogen surveillance and clearance (Brown 2001). Furthermore, parallel studies in vertebrate and invertebrate immunology have shown many similarities between these two innate immune systems, i.e., Toll receptors, pattern recognition molecules, and complement (Barillas-Mury et al. 2000; Hoffmann et al. 1999; Hoffmann and Reichhart 2002; Levashina et al. 2001; Yu et al. 2002). Phagocytosis, a mechanism used by macrophages to kill microorganisms, was originally discovered in an invertebrate (Metchnikoff 1893). Taken altogether, these data and our current observations argue that continuing studies in mosquito innate immunity will yield answers explaining vector competence and at the same time shed light on potential vertebrate innate immune mechanisms.


We thank C.C. Chen for providing the anti-phenoloxidase I antibody and D. Schneider for providing GFP-expressing E. coli. Useful discussions with A.J. Nappi, C.C. Chen, L.C. Bartholomay, B.K. August, and R.J. Massey are greatly appreciated.

Copyright information

© Springer-Verlag 2003