We identified 4047 genes expressed in hemocytes, using An. gambiae genome-wide microarrays. While 279 transcripts were significantly enriched in hemocytes relative to whole adult female mosquitoes, 959 transcripts exhibited immune challenge-related regulation. The global transcriptomic responses of hemocytes to challenge with different species of bacteria and/or different stages of malaria parasite infection revealed discrete, minimally overlapping, pathogen-specific signatures of infection-responsive gene expression; 105 of these represented putative immunity-related genes including anti-Plasmodium factors. Of particular interest was the specific co-regulation of various members of the Imd and JNK immune signaling pathways during malaria parasite invasion of the mosquito midgut epithelium.

Previous studies with the mosquitoes Anopheles albimanus, Aedes aegypti and Armigeres subalbatus suggest that E. coli is primarily phagocytosed by hemocytes, while Micrococcus spp. are melanized extracellularly within the hemolymph [20-22]. Our own results (see above) noted a significant increase in the abundance of oenocytoids, which constitutively express phenoloxidase activity, following infection by M. luteus. Prior studies with An. gambiae in contrast indicate that granulocytes are the only hemocytes that phagocytize foreign targets and also inducibly express phenoloxidase activity following immune challenge by bacteria [31]. We therefore characterized phagocytosis and melanization toward E. coli or M. luteus in An. gambiae to assess whether: 1) oenocytoids and granulocytes differentially respond to these two bacterial species and 2) whether this response is qualitatively consistent with transcriptomic profiles. Phagocytosis assays revealed that significantly more granulocytes phagocitized E. coli than M. luteus (t-test; P < 0.01) (Figure ​(Figure2c).2c). However, we also noted that a higher proportion of granulocytes with internalized M. luteus contained bacteria that were melanized compared to granulocytes with internalized E. coli (t-test; P < 0.01) (Figure ​(Figure2d).2d). Although oenocytoids cannot phagocytize bacteria, challenge with M. luteus also induced a significantly greater proportion of these hemocytes to melanize than E. coli (t-test; P < 0.01) (Figure ​(Figure2e).2e). These results indicate that similar to other mosquitoes, M. luteus induces a much stronger melanization response than E. coli in An. gambiae, even though granulocytes phagocytize both species of bacteria. Although broadly consistent with the up-regulation of phagocytosis and melanization-related genes following M. luteus challenge, these results also do not explain why so few genes with phagocytic or immune functions are up-regulated by E. coli.

During the two major spatial transition stages of infection by Plasmodium sp., ookinete invasion of the midgut and sporozoite migration through the hemolymph, the parasite experiences considerable loss of abundance in An. gambiae [60,61] that is in part attributed to hemocyte-mediated immune responses [28,32]. Overall, transcripts of 431 genes were differentially expressed in hemocytes during malaria parasite infection (either at 24 hours and/or 19 days after P. berghei infection) (Figure ​(Figure1b1b and ​and1d;1d; Additional file 1). When considered relative to the total number of expressed genes, the magnitude of gene regulation was similar for the two different stages of malaria parasite infection (12.1 versus 11.0% of transcribed genes were differentially expressed, respectively, at 24 hours and19 days after P. berghei infection). Strikingly, only 23 (5.3%) of the differentially regulated transcripts had similar expression profiles during the two infection stages, while 16 (3.7%) transcripts were expressed in opposite directions. The remaining 392 (91.0%) transcripts were differentially expressed exclusively during one of the two stages of parasite infection. Only 5 (6.8%) of the 74 putative immune genes were regulated in the same direction while 7 (9.5%) transcripts were regulated in opposite directions at 24 hours and 19 days after P. berghei infection.

Phagocytosis assays were performed as previously described [31]. Briefly, 2 × 10³ heat-killed. fluoresceing isothiocyanate (FITC)-conugated E. coli or M. luteus were injected intrathoracically into cold anethetized mosquitoes. After 1 h at room temperature, hemocytes were collected, placed into primary culture, and identified as outlined by Castillo et al (2006) [31]. Briefly, granulocytes were identified by their spread morphology, oenocytoids were identified by morphology, differential labeling with monochlorobimane (MCB) and an anti-phenoloxidase antibody (PP06, generously donated by K. Michel and F. Kafatos), and prohemocytes were identified by morphology and an absence of labeling by MCB and the anti-PO antibody. The proportion of granulocytes that had ingested particles was determined by counting 100 cells in a randomly selected field of view using the fluorescent quenching method [31]. Melanization was quantified by counting hemocyte-internalized bacteria. The proportion of oenocytoids and granulocytes that had melanized without ingesting any bacteria was also determined by visual inspection. We infected mosquitoes with malaria parasites by allowing 4 day-old adult females to feed on a BALB C mouse infected with the PbGFP_CON strain of P. berghei, which constitutively expresses green fluorescent protein (GFP) under control of the P. berghei elongation factor 1α promoter [83]. A Plasmodium berghei reference line that constitutively expresses GFP at a high level throughout the complete life cycle. Molecular and Biochemical Parasitology [137, [23-33]]. Only mice exhibiting 3–5% parasitemia were used for infection of mosquitoes. Hemocytes were collected either 24 hours or 19 days after blood-feeding. All cohorts of infected mosquitoes were also monitored by dissection to determine levels of midgut infection by oocysts at 24 h and infection of salivary glands by sporozoites on day 19. Only cohorts in which ≥ 80% of individuals were infected were used for analysis. Controls included day 4 mosquitoes with no bleed feeding and mosquitoes blood-fed on uninfected mice. All data were analyzed by ANOVA followed by Dunnetts comparison procedure or t-test using the JMP 7.0 statistical platform (SAS, Gary, NC). Proportional data were arcsin transformed before analysis.

Hemocyte samples from bacteria-infected, P. berghei-infected and control mosquitoes were prepared and collected as described above. A minimum of 30 individuals were bled and their hemocytes pooled to obtain at least 500 ng of total RNA per replicate. However, for the day 19 post-blood feeding samples, more than 100 mosquitoes were bled per sample due to the reduced number of hemocytes present per individual (see Results). Total RNAs were then isolated from hemocytes using the RNAEasy kit (Qiagen) as outlined by the manufacturer. For each treatment and time point, three independent samples were prepared for use in subsequent microarray and QRT-PCR analyses.