CLIPB8 is part of the prophenoloxidase activation system in Anopheles gambiae mosquitoes

Abstract

In insects and other arthropods the formation of eumelanin (melanization) is a broad spectrum and potent immune response that is used to encapsulate and kill invading pathogens. This immune response is regulated by the activation of prophenoxidase (proPO), which is controlled by proteinase cascades and its serpin inhibitors, together forming the proPO activation system. While the molecular composition of these protease cascades are well understood in insect model systems, major knowledge gaps remain in mosquitoes. Recently, a regulatory unit of melanization in Anopheles gambiae was documented, comprised of the inhibitory serpin-clip-serine proteinase, CLIPB9 and its inhibitor serpin-2 (SRPN2). Partial reversion of SRPN2 phenotypes in melanotic tumor formation and adult survival by SRPN2/CLIPB9 double knockdown suggested other target proteinases of SRPN2 in regulating melanization. Here we report that CLIPB8 supplements the SRPN2/CLIPB9 regulatory unit in controlling melanization in An. gambiae. As with CLIPB9, knockdown of CLIPB8 partially reversed the pleiotropic phenotype induced by SRPN2 silencing with regards to adult survival and melanotic tumor formation. Recombinant SRPN2 protein formed an SDS-stable protein complex with activated recombinant CLIPB8, however did not efficiently inhibit CLIPB8 activity in vitro. CLIPB8 did not directly activate proPO in vitro nor was it able to cleave and activate proCLIPB9. Nevertheless, epistasis analysis using RNAi placed CLIPB8 and CLIPB9 in the same pathway leading to melanization, suggesting that CLIPB8 either acts further upstream of CLIPB9 or is required for activation of a yet to be identified serine proteinase homolog. Taken together, this study identifies CLIPB8 as an additional player in proPO activation cascade and highlights the complexity of the proteinase network that regulates melanization in An. gambiae.

Graphical Abstract

1. Introduction

Melanization, an arthropod-specific immune response, is used against a wide variety of pathogens (Cerenius and Soderhall, 2004; Cerenius et al., 2008). Indeed, insects deposit melanin on virtually all foreign surfaces naturally or experimentally introduced into their hemocoel, including nylon, epoxy, glass, and living tissues from other insects (Grimstone et al., 1967; Lackie, 1983; Salt, 1965, 1963). As all insects, the African malaria mosquito, Anopheles gambiae utilizes melanization as an anti-bacterial (Binggeli et al., 2014; Hillyer et al., 2003) and anti-fungal immune response (Yassine et al., 2012) that can also kill malaria parasites (Collins et al., 1986; Michel et al., 2005). Killing of pathogens through melanization is thought to occur through nutrient starvation (Chen and Chen, 1995) as well as direct toxic effects of reaction intermediates and byproducts (Nappi et al., 2009), although experimental support in An. gambiae is still lacking. Successful malaria transmission is therefore only ensured if Plasmodium parasites can escape the melanization response. Indeed, in transmission-permitting Plasmodium sp./mosquito species combinations, melanized parasites are rarely observed in the field (Cohuet et al., 2006; Lambrechts et al., 2007; Riehle et al., 2006; Schwartz and Koella, 2002), as Plasmodium sp. have evolved either active suppression (Boete et al., 2002) or avoidance mechanisms (Michel et al., 2006; Molina-Cruz et al., 2012).

Insect eumelanin is the product of oxidative polymerization of 5,6-dihydroxyindoles (DHIs). The substrate pathway that leads from tyrosine to DHI includes a number of enzymatic and nonenzymatic reactions, causing the initial production of the o-diphenols dopa and dopamine and their conversion to dopachrome and dopaminechrome, which form the precursors of DHI (Nappi and Christensen, 2005; Sugumaran, 2002). The key enzyme initiating melanization is phenoloxidase (PO), a monophenoloxygenase that catalyzes the initial hydroxylation of tyrosine to dopa as well as the oxidation of dopa and dopamine to their respective quinones (Nappi et al., 2009). PO are enzymes of about 80 kDa molecular weight that contain two binuclear copper binding sites (Hu et al., 2016). They are expressed as zymogens, called proPOs that require proteolytic cleavage at the N-terminus. While lepidopteran insect genomes encode only two, and D. melanogaster encodes three proPOs, mosquito genomes encode between nine to ten proPOs, all with high sequence similarity (Bartholomay et al., 2010; Neafsey et al., 2015; Waterhouse et al., 2007), but it remains unclear if individual mosquito POs differ in their biological function. Several, but not all An. gambiae proPOs are upregulated by blood feeding, and expression of at least five proPOs can be altered by 20-hydroxy ecdysone (Ahmed et al., 1999; Muller et al., 1999), suggesting that temporal and physiological specification of individual PO function.

The production of eumelanin generates a number of toxic byproducts including semiquinones and reactive oxygen species (Nappi and Christensen, 2005), which ultimately can be deleterious to the insect. Overstimulation of PO activity and ultimately melanization in An. gambiae causes a number of deleterious effects cumulating in shortened life span (An et al., 2011; Michel et al., 2005). It is therefore not surprising that melanization is tightly controlled, most prominently through regulation of the extracellular proteolytic cleavage of proPO to PO by the proPO activation cascade. Key enzymes in this cascade are clip-domain containing serine proteinases (CLIPs) that contain one or more amino-terminal clip domain that is separated by a linker region from a carboxyl-terminal S1A family serine proteinase domain (Smith and DeLotto, 1992). CLIPs are secreted into the hemolymph as zymogens, which require proteolytic activation by cleavage within the linker region. CLIPs are found in insects and other arthropods and can be subdivided into four evolutionarily distinct clades (A–D, Waterhouse et al., 2007; Kanost and Jiang, 2015), of which members of clade A have lost their proteolytic function. These proteinases are inhibited by serine proteinase inhibitors of the serpin family, the largest family of proteinase inhibitors in metazoans. Serpins act as suicide substrate inhibitors, forming covalent, stable inhibitory complexes with their target proteinases (Gettins, 2002), including CLIPs that are required for melanization in insects (Jiang and Kanost, 2000).

Our current understanding of the molecular make-up of the proPO activation system is derived from studies in a few model organisms such as Drosophila melanogaster (Veillard et al., 2015), Manduca sexta (Kanost and Jiang, 2015), and Tenebrio molitor (Kan et al., 2008). A generalized view of the insect proPO activation system can be described as follows (Kanost and Jiang, 2015): The system consists of protease cascades that are triggered by the recognition of molecular patterns associated with pathogens or aberrant cells by soluble receptor molecules leading to the activation of a modular serine proteinase (MSP). MSP in turn activates a CLIPC that then activates the terminal CLIPB proteinase in this cascade, also called proPO activating proteinase (PAP) or proPO activating enzyme (PPAE). Active PAP then cleaves proPO to PO. In addition, the formation of the final active phenoloxidase complex on the foreign surface is mediated by one or more proteolytically inactive CLIPAs, which themselves require proteolytic activation in order to function. Furthermore, several protease cascades, characterized by specific PAPs and distinct upstream CLIPCs can act in parallel to regulate the melanization response in insects (An and Kanost, 2010). Under normal physiological conditions, proPO activation cascades are turned off, most prominently by a single highly conserved serpin (Park et al., 2000), referred to as Spn27A in Drosophila melanogaster (De Gregorio et al., 2002; Ligoxygakis et al., 2002), Serpin-3 in Manduca sexta (Zhu et al., 2003), and SRPN2 in mosquitoes (Bartholomay et al., 2010; Christophides et al., 2002; Waterhouse et al., 2007; Zou et al., 2010). The only experimentally confirmed inhibitory target of these serpins in proPO activation cascades is the terminal PAP within each cascade (An et al., 2013, 2011; Zhu et al., 2003). The depletion of this serpin results in continuous activation of PO in the hemolymph, the formation of large melanotic pseudotumors, and increased mortality rates (Ligoxygakis et al., 2002; Michel et al., 2005; Zou et al., 2010).

In An. gambiae, CLIPB9 is currently the only known bona fide member of the proPO activation system. This proteinase functions as a PAP, directly cleaving and activating proPO in vitro. CLIPB9 and SRPN2 form a functional regulatory unit of a proPO activation system, as CLIPB9 is required for melanotic tumor formation in adult female An. gambiae that are depleted of SRPN2 and is inhibited directly by SRPN2 in vitro and in vivo (An et al., 2011). However, targeted reverse genetic screens have identified a number of additional CLIPBs required for melanization of foreign surfaces (Paskewitz et al., 2006; Volz et al., 2006). In addition, reversion of the SRPN2-depleted phenotype is incomplete after knockdown of CLIPB9 suggesting additional proteinase cascades are required for complete activation of proPO in An. gambiae.

The current study was designed to test the specific hypothesis that CLIPB8 is part of the proPO activation system in An. gambiae. Previous studies indicate CLIPB8 is required for the ookinete melanization of the rodent malaria parasite, Plasmodium berghei, in CTL4kd An. gambiae mosquitoes (Volz et al., 2006) and also for melanization of Sephadex beads in An. gambiae hemolymph (Paskewitz et al., 2006). Phylogenetic analysis revealed that CLIPB8 is closely related to M. sexta PAP2 and 3 (An et al., 2010), suggesting a conserved role in mosquitoes. This is further substantiated by observations that the ortholog of An. gambiae CLIPB8 in the Yellow fever mosquito, Aedes aegypti, binds SRPN2 in vitro and is required for melanotic tumor formation in this species (Zou et al., 2010). Using genetic and biochemical methodologies our laboratory established previously, we initially determined whether CLIPB8 is part of the proPO activation system that is controlled by SRPN2 in An. gambiae, and subsequently explored its position within this cascade and in relation to its paralog, CLIPB9.

2. Results

2.1 CLIPB8 partially reverts the SRPN2-depletion phenotype

To test whether CLIPB8 is involved in SRPN2kd-mediated melanotic tumor formation, we performed double knockdown (dkd) analysis of CLIPB8 and SRPN2 in An. gambiae adult females using dsRNA injection. Kd efficiencies of CLIPB8 and SRPN2 were assessed at the transcript levels by RT-quantitative (q)PCR in each of the treatment groups (Fig. S1). CLIPB8 transcript levels were reduced by 90–94 % after treatment with dsCLIPB8 or dsCLIPB8/dsSRPN2 as compared to mosquitoes treated with only dsGFP or dsSRPN2, respectively. Transcript reduction of SRPN2 was slightly lower at 71–73%, which is comparable to previously reported knockdown levels for this gene (An et al., 2011; Michel et al., 2005).

Kd of CLIPB8 in a SRPN2-depleted genetic background decreased melanotic tumor formation caused by SRPN2 depletion (Fig. 1A). Melanized area per abdomen of dsSRPN2/dsCLIPB8-treated female mosquitoes were significantly smaller than in SRPN2-depleted mosquitoes (Mann-Whitney U test, P=0.002) (Fig. 1B). As observed previously, dsGFP or dsCLIPB8 treatments alone did not result in melanotic tumors in adult mosquitoes.

Figure 1. CLIPB8 knockdown partially reverts the SRPN2 depletion phenotype.

(A) Abdominal images 12d post dsRNA injection. (B) Total melanotic area per abdomen decreased significantly in CLIPB7SRPN2dkd mosquitoes as compared to SRPN2kd animals (Median with interquartile range, U-Test, P = 0.0002). (C) CLIPB8/SRPN2dkd also partially decreased the mortality rate observed in SRPN2-depleted mosquitoes. Survival curves with different letters were statistically significantly different (Log-rank Test, P<0.0001).

In addition, CLIPB8kd partially reverted the accelerated daily mortality rates observed in SRPN2-depleted mosquitoes (Fig. 1C). Dkd of CLIPB8 and SRPN2 increased significantly mosquito life span as compared to SRPN2kd (Log Rank Test, χ² = 14.42, df = 1, Bonferroni-corrected threshold of P < 0.01). However, mortality rates of CLIPB8/SRPN2 dkd mosquitoes remained accelerated as compared to CLIPB8kd mosquitoes (Log Rank Test, χ² = 10.83, df = 1, Bonferroni-corrected threshold of P < 0.01). As expected, CLIPB8kd did not alter mortality rates as compared to dsGFP-injected mosquitoes (Log Rank Test, χ² = 1.167, df = 1, Bonferroni-corrected threshold of P > 0.05). These results identify CLIPB8 as part of the melanization cascade in An. gambiae that is regulated by SRPN2 and, if misregulated, leads to melanotic tumor formation and increased mortality.

2.2 CLIPB8 is a functional serine proteinase with trypsin-like specificity

Based on sequence analysis, CLIPB8 is a typical Clip proteinase of the subfamily B. Its single clip domain between residues 17–71 contains 25 residues between Cys-3 and Cys-4, and activation cleavage is likely to occur after lysine, both hallmarks of subfamily B clip proteinases (Jiang and Kanost, 2000). The catalytic triad required for proteinase function is conserved and is predicted to consist of residues H158, A225, and S329. Proteolytic cleavage of the proCLIPB8 zymogen required for activation of clip-serine proteinases is likely to occur between CLIP and proteinase domain with the P1 site at K112 (Paskewitz et al., 2006). To determine the mode of action of CLIPB8, we expressed and purified the full-length recombinant proCLIPB8 zymogen that includes a C-terminal His tag using a baculovirus expression system. Endogenous activating proteinases for any mosquito CLIPs still await identification, and it is unknown which proteinase activates the proCLIPB8 zymogen in vivo. We therefore replaced the endogenous putative activation cleavage site of proCLIPB8 with that of bovine Factor Xa, so that recombinant proCLIPB8 (proCLIPB8_Xa) could be cleaved and activated in vitro.

Based on Western blot analysis, the full length purified recombinant proCLIPB8_Xa has a molecular weight of around 50 kD, which is significantly larger than its calculated molecular weight of 43.4 kD, indicating that the protein is glycosylated (Fig. S2A). Incubation of the recombinant protein with bovine Factor Xa resulted in cleavage of the protein and appearance of a second band of 36-kDa on the Western blot, the predicted size of the CLIPB8_Xa proteinase domain (Fig. 2A). Next, we screened a number of small peptides for their ability to function as a substrate for CLIPB8. CLIPB8_Xa was able to cleave a range of peptide substrates with arginine at the P1 site, including IEGR, IEAR, VGR and FVR (Figs. S2B and S3), suggesting a trypsin-like substrate preference. Importantly, FVRase activity was not detected when incubated with the proCLIPB8_Xa zymogen, confirming that proteolytic cleavage is required to activate proteinase activity of CLIPB8 (Fig. S2B).

Figure 2. CLIPB8 forms SDS-stable higher molecular weight complexes with SRPN2.

Incubation of active rCLIPB8Xa (His tag at C-terminus) with rSRPN2 (His tag at N-terminus) in vitro leads to the appearance of higher molecular bands as determined by Western blot analysis by using anti-His (A) and anti-SRPN2 (B) antibodies. Circle, CLIPB8_Xa zymogen; square, CLIPB8_Xa proteinase domain; triangle, SRPN2; star, complex containing both CLIPB8_Xa proteinase domain and SRPN2.

2.3 CLIPB8 is not a molecular target of SRPN2 inhibition

To determine whether CLIPB8 can by inhibited directly by SRPN2, we first examined whether SRPN2 can form a covalent protein complex with CLIPB8 in vitro by Western blot analysis. A higher molecular weight band appeared after incubation of the activated CLIPB8_Xa with SRPN2, indicating the formation of an inhibitory complex (Fig. 2). In the absence of active CLIPB8_Xa, anti-His antibodies recognized the 52-kDa CLIPB8_Xa zymogen and 43-kDa recombinant SRPN2 (Fig. 2A). When SRPN2 was mixed with active CLIPB8_Xa, a 36-kDa band corresponding to the CLIPB8 catalytic domain was detected. In addition, a new immunoreactive band at ~72-kDa, corresponding to the expected mass of the SRPN2/CLIPB8_Xa complex was detected. This band was also recognized by anti-SRPN2 antibodies (Fig. 2B). Analysis of tryptic peptides from the ~72-kDa band by MS/MS identified several peptides of CLIPB8 and SRPN2 (Fig. S4), confirming that SRPN2 forms a covalent complex with active CLIPB8_Xa. The peptides obtained for CLIPB8 covered both Clip and proteinase domain, indicating that the Clip domain remained attached to the proteinase domain upon activation by Factor Xa.

While complex formation was observed, the ratio of intensities of the complex band compared to CLIPB8_Xa proteinase domain and SRPN2 was low (Fig. 2), suggesting that SRPN2 may not be an efficient inhibitor of CLIPB8. We therefore analyzed the ability of SRPN2 to inhibit the FVRase activity of CLIPB8_Xa in vitro. While increasing concentrations of SRPN2 slightly dampened CLIPB8_Xa activity, the stoichiometry of inhibition (SI) was as high as 55 (Fig. 3). Therefore, SRPN2 is not an effective inhibitor of CLIPB8 in vitro, suggesting that, in contrast to CLIPB9, CLIPB8 is not a physiological target of SRPN2 in vivo.

Figure 3. SRPN2 does not inhibit CLIPB8 in vitro.

Purified rSRPN2 and activated rCLIPB8_Xa were incubated at increasing molar ratios. The residual FVRase activities were plotted as mean ± S.D. (n=3) against the corresponding molar ratios of serpin and active proteinase. CLIPB8_Xa activity was marginally affected by the presence of SRPN2, with a theoretical Stoichiometry of inhibition (SI) of 55.

2.4 CLIPB8 neither cleaves nor activates proPO

Biochemical analysis of the melanization cascade in mosquitoes is greatly enabled by the use of heterologous model systems such as M. sexta (Kanost and Gorman, 2008; Michel et al., 2006; Zou et al., 2010). We previously determined that CLIPB9_Xa functions as a PAP in An. gambiae (An et al., 2011), and used established assays to determine if CLIPB8 functions similarly as a terminal proteinase in the proPO activation cascade. ProPO circulates in M. sexta hemolymph as a heterodimer of 79-kDa PPO-p1 and 80-kDa PPO-p2 (Jiang et al., 1997). Western blot analysis of purified M. sexta PO with either Factor Xa or proCLIPB8_Xa using anti-M. sexta PO antibody identified two bands corresponding to the full-length proPO heterodimer at around 80-kDa (Fig. 4A), as well as a single smaller band of 56-kDa, suggesting partial proteolytic cleavage of M. sexta proPO. However, incubation with active CLIPB8_Xa did not cause an increase in PO activity above background levels. (Fig. 4B). The same bands were detected after incubation of purified M. sexta proPO with active CLIPB9_Xa. In addition, and as shown previously (An et al., 2011), a 72-kDa doublet band corresponding to mature M. sexta PO was detected, and PO activity increased significantly (Fig. 4A, B). As the activation cleavage sites between An. gambiae and M. sexta proPOs are conserved (Christophides et al., 2002), it is unlikely that CLIPB8 activates An. gambiae proPOs. Therefore, while CLIPB8 is part of the proPO activation cascade, CLIPB8 does not have proPO as its proteolytic target and consequently must act upstream of the An. gambiae PAP.

Figure 4. CLIPB8 does not function as a PAP.

rCLIPB8_Xa did neither cleave (A) nor activate (B) purified M. sexta proPO in vitro. When purified M. sexta proPO was incubated with CLIPB9_Xa as a positive control, a smaller PO band appeared (arrow), and increased PO activity was observed. Data are presented as mean ± one S.D. (n=3); different letters indicate significant statistical differences between treatment groups (one-way ANOVA followed by Newman-Keuls test, P < 0.05).

2.5 CLIPB8 and B9 are not proteolytic substrates of each other

Our results indicate that CLIPB8 acts upstream of a PAP in the proPO activation cascade. The only confirmed PAP in the PO activation cascade in An. gambiae is CLIPB9 (An et al., 2011). We therefore tested next whether CLIPB8 can function as the activating proteinase of proCLIPB9, and vice versa. To this end, we expressed and purified the wild-type forms of proCLIPB8 and proCLIPB9 in Sf9 cells, which contained the unmutated activation cleavage sites. However, we found no evidence that these two proteinases can utilize each other as substrates (Fig. 5). When activated CLIPB8_Xa was incubated with proCLIPB9 in equal molar ratios, the 56 kD zymogen of proCLIPB9 remained, and no additional band corresponding to the proteinase domain of CLIPB9 appeared (Fig. 5A right panel). Likewise, incubation of activated CLIPB9_Xa with proCLIPB8 had no effect on the 50 kD CLIPB8 zymogen (Figure 5B, right panel). These results are consistent with the conclusion that CLIPB8 is not the activating proteinase of proCLIPB9 or vice versa, and suggest that CLIPB8 either functions further upstream of CLIPB9 or may be part of a distinct proPO activation cascade that functions in parallel to CLIPB9.

Figure 5. CLIPB8 and B9 do not interact directly.

rCLIPB8_Xa does not cleave recombinant proCLIPB9 (A) or vice versa (B) as determined by Western blot analysis using anti-CLIPB8 and anti-CLIPB9-specific antibodies. (A) Recombinant proCLIPB8_Xa was activated by Factor Xa in vitro and incubated with recombinant wild-type proCLIPB9. The resulting Western blot was probed sequentially with anti-CLIPB8 and anti-CLIPB9 antibody. While anti-CLIPB8 antibody was able to detect a band corresponding to the proteinase domain of activated CLIPB8, no cleaved proteinase form was detectable by anti-CLIPB9 antibody. (B) The reciprocal experiment did not detect any cleavage product for wild-type CLIPB8 after incubation with activated CLIPB9_Xa. Both polyclonal antibodies were only able to detect the antigen they were raised against, and no cross reactivity to other proteins, including Factor Xa was observed. Circles, proteinase zymogen; squares, proteinase domain.

2.6 CLIPB8 and B9 act within the same proPO activation cascade

Knockdown of either CLIPB8 or CLIPB9 partially reverted the SRPN2 depletion phenotypes. However, based on the biochemical data presented above, it remained unclear if these two proteinases function in the same proPO activation cascade or if they function in parallel. To test whether CLIPB8 and B9 indeed belong to the same proteolytic cascade, we set out to perform epistasis experiments using triple knockdown (tkd) of SRPN2, CLIPB8 and CLIPB9. All dsRNA injections induced specific transcript reduction for their target transcripts, regardless if injected with one or more additional dsRNAs (Fig. S1). Furthermore, kd efficiencies remained at the same level between dkd and tkd experiments. Together, these results not only confirm efficacy of the tkd experimental design, but also reveal that neither SRPN2, CLIPB8, nor CLIPB9 influence the expression of the other two genes on a transcriptional level.

We next examined the formation of melanotic tumors in the SRPN2/CLIPB9/CLIPB8tkd mosquitoes compared to dkd and single kd controls. As reported previously, CLIPB8kd or CLIPB9kd in SRPN2-depleted mosquitoes significantly reduced melanotic tumor formation in adult female mosquitoes. The average area of melanization observed in SRPN2kd mosquitoes was reduced by nearly 50 % in either CLIPB8kd or CLIPB9kd mosquitoes in a SRPN2-depleted genetic background (Fig. 6A). However, we did not observe any significant differences in total melanized area per abdomen in the tkd mosquitoes as compared to those from SRPN2/CLIPB9dkd or SRPN2/CLIPB8dkd mosquitoes (Fig. 6A). These results therefore provide no evidence for additive phenotypic effects of kd of these two proteinases on SRPN2 depletion-mediated melanization, and demonstrate that CLIPB8 and CLIPB9 function in the same proteolytic cascade that activates proPO in An. gambiae

Figure 6. Analysis of the PO activation cascade by tkd of SRPN2, CLIPB8 and B9.

(A) Total melanotic area per abdomen in dkd and tkd mosquitoes was significantly reduced as compared to SRPN2kd animals (Median ± interquartile range, Kruskal-Wallis test, P<0.0001, Dunns post test, P,0.05). (B) Survival curves of dkd and tkd mosquitoes as compared to SRPN2kd. Data are were obtained from three independent biological replicates. Survival curves with different letters were statistically significantly different (Log-rank Test, P<0.0001)

Similarly, the accelerated death rates of SRPN2-depleted mosquitoes were reduced significantly in SRPN2/CLIPB9dkd and SRPN2/CLIPB8dkd mosquitoes. Survival curves between these dkd populations were statistically significantly different from the dsGFP-injected controls, but identical between the two dkd treatment groups. As observed for the melanization phenotype, no further increase in survival in SRPN2kd mosquitoes was observed when transcripts of both proteinases were depleted (Fig. 6B). These results provide further evidence that kd of both CLIPB8 and CLIPB9 does not have an accumulative effect of the pleiotropic SRPN2-depletion phenotype and confirm that they function in the same pathway.

3. Discussion

Proteinase cascades, in which proteinases activate each other in a sequential order, play a central role in the mosquito immune system, regulating melanization, the Toll pathway, and most likely the complement system. To investigate the complexity anti-parasite immunity regulation in An. gambiae, we examined further the molecular make-up of the system that controls melanization via proteolytic proPO activation. We previously identified the first functional regulatory unit of melanization in this mosquito species, composed of CLIPB9 and its inhibitor, SRPN2. This unit directly regulates cleavage of proPO and thus the production of eumelanin. In this study, we provide functional evidence that a second CLIPB proteinase, CLIPB8 is required for melanization as part of the proPO activation system. In addition, we demonstrate that CLIPB8 and B9 function in the same proPO activation cascade, albeit hold distinct positions within this cascade.

3.1. CLIPB8 and B9 have distinct roles within melanization

CLIPB8 and B9 are closely related paralogs, separated by only 2.1 kb on chromosome 2R 14A (An et al., 2011), suggesting that these genes may be co-regulated (Li et al., 2013; Michalak, 2008). Indeed, their expression pattern is similar across wide variety of experimental conditions. In the An. gambiae expression map, which clusters the expression patterns of all annotated An. gambiae genes across 93 experimental conditions (Maccallum et al., 2011), CLIPB8 and B9, are located in the same map region within nodes 5,16 and 3,15, respectively. Both genes have similar tissue expression profiles, with enrichment in circulating hemocytes (Pinto et al., 2009), and little to no expression in midgut or ovaries (Baker et al., 2011). Furthermore, both genes are upregulated after blood feeding (Marinotti et al., 2006), further supporting co-regulation of these two proteinases. In addition, CLIPB8 and B9 exhibit the highest sequence similarity among An. gambiae CLIPBs: the amino acid sequences of their proteinase domains are 40% identical (An et al., 2011). We therefore expected and experimentally confirmed that the recombinant proteins of these two proteinases have similar peptide substrate specificities in vitro. Finally, our RNAi data confirm that both CLIPB8 and B9 reduce the melanotic phenotype conferred by SRPN2kd, supporting our hypothesis that, just as CLIPB9, CLIPB8 functions with the proPO activation system.

However, while CLIPB8 and B9 are similar in their expression profiles and biochemical function, we provide several independent lines of evidence that these two proteins have distinct roles in vivo. First, depletion of either one by RNAi in a SRPN2kd background causes a phenotype without affecting the other proteinase’s expression level, demonstrating that these two proteinases cannot complement each other’s function. Second, CLIPB8 and B9 have distinct physiological cleavage targets, as only CLIPB9, not CLIPB8 can cleave proPO. Third, only CLIPB9 provides an intervention point for melanization by SRPN2. As we have shown previously, CLIPB9 and SRPN2 form inhibitory complexes at rates that are physiologically relevant (An et al., 2011). In contrast, inhibition of CLIPB8 amidase activity requires extremely high levels of SRPN2 that are not observed in vivo. Therefore, although these two proteinases function within the same proPO activation cascade, their roles are not redundant. CLIPB8 and B9 are thus an excellent example of paralogous proteins in An. gambiae have undergone functional diversification. This functional diversification was not driven by divergent gene expression (Hughes and Friedman, 2005), nor different cellular localization (Marques et al., 2008) or biochemical mode of action. Rather, it is likely the consequence of altered substrate recognition through mutations in the recognition domains, reminiscent of alterations recently described in transcription factors (Singh and Hannenhalli, 2008; Voordeckers et al., 2015).

3.2. CLIPB8 functions upstream of CLIPB9 within the same proPO cascade

In the generalized model of the prototypical proPO activation cascade, the modular serine proteinase MSP autoactivates in response to binding to a pattern recognition receptor (PRR) bound to microbe- or pathogen-associated patterns (Takahashi et al., 2015). Active MSP then cleaves and activates a CLIPC, which in turn activates a CLIPB. Active CLIPB then executes the terminal step in the cascade by functioning as a PAP, cleaving and activating proPO to PO (Kanost and Jiang, 2015). Based on this model and given that CLIPB8 belongs to the subfamily B of clip-domain serine proteinases, we hypothesized that CLIPB8 also acts as PAP, functioning in parallel to PAP activation activity of CLIPB9 (An et al., 2011). However, data presented in this study do not provide evidence for this hypothesis. CLIPB8/B9dkd in a SRPN2-depleted background does not further reduce melanotic tumor formation as compared to the CLIPB8kd or CLIPB9kd alone, demonstrating that CLIPB8 and B9 do not control PO activity in parallel. Instead, at least in the context of SRPN2kd, CLIPB8 and B9 act within the same proPO activation cascade leading to melanotic tumor formation, likely regulating PO activity by cleavage of one or more proPOs through CLIPB9. Furthermore, CLIPB8 does not function as a PAP in vitro. Although we used proPO isolated from the hemolymph of M. sexta and not An. gambiae, insect proPO sequences are highly conserved and the vast majority of them share the same activation cleavage site, including those of M. sexta proPO1/2 and An. gambiae PPO1-8 (Christophides et al., 2002). Whether CLIPB8 can function as a PAP of e.g. An. gambiae ProPO9, which has an activation cleavage site distinct of M. sexta proPO1/2 and An. gambiae PPO1-8 (Christophides et al., 2002), warrants further investigation. However, the genetic and biochemical data detailed in this manuscript demonstrate that CLIPB8 does not function as a PAP in melanotic tumor formation in vivo, and must act upstream of CLIPB9.

Although belonging to the B subfamily, we nevertheless explored whether CLIPB8 functions directly upstream of CLIPB9. However, this is not the case, as CLIPB8 cannot cleave proCLIPB9 in vitro (or vice versa). It is nevertheless possible that CLIPB8 is further upstream in the pathway activating a yet to be identified proteinase, which is responsible for CLIPB9 activation, despite its designation in the B subfamily and the deviation from the cascade model elucidated in other insects. An alternative and more parsimonious explanation is that CLIPB8 is required for the activation of one or several proteolytically inactive CLIPAs that complement the proPO cascade. Published examples for proteolytic activation of CLIPAs by CLIPBs exist from other insect species (Kan et al., 2008; Wang et al., 2014). A candidate for such a proteinase homolog is CLIPA8, which is required for melanization of bacteria, fungi and P. berghei (Schnitger et al., 2007; Volz et al., 2006; Yassine et al., 2012). Thus far, it is the only CLIPA that has been shown to be required for melanization in An. gambiae. Future studies will have to determine whether CLIPA8 indeed requires proteolytic activation for its function, and if this activation is mediated by CLIPB8.

3.3. The complexity of proPO activation in An. gambiae

The proPO activation cascade, characterized by CLIPB8 and B9, is at least partially conserved among Culicidae, as the orthologs of CLIPB8 and B9 are also required for melanotic tumor formation in Ae. aegypti (Zou et al., 2010). However, the regulation of proPO in mosquitoes is probably more complex, as it is likely that several proPO activation cascades exist in mosquitoes. Besides CLIPB8 and CLIPB9, several other CLIPB proteinases, including CLIPB1, 3, 4, 14, and 17 also affect murine malaria parasite and/or bead melanization (Volz et al., 2005, 2006; Paskewitz et al., 2006). While their molecular function awaits identification, it is likely that these additional CLIPBs are active proteinases contributing directly to proPO activation. In addition, reversion of the SRPN2kd-induced phenotype by the CLIPB9 PAP is incomplete, indicate that other PAPs also mediate melanotic tumor formation (An et al., 2011). Indeed, preliminary data indicate that at least CLIPB4 functions as a PAP in vitro (Zhang and Michel, unpublished). Together, these findings strongly point to the existence of several proPO activation cascades in An. gambiae acting in parallel.

The notion of several proPO activation cascades in An. gambiae is perhaps not surprising, given their existence in other distantly related insect species, including M. sexta. However, the complexity of the proPO activation cascade(s) is further underscored by the observation that, dependent on the initiating trigger of proPO activation, the number and identity of CLIPBs required for melanization is partially overlapping but not congruent. Melanization of Sephadex beads implanted into adult female An. gambiae requires at least CLIPB4 and B8, and to a lesser extend the action of CLIPB1, but not CLIPB9 nor B10 (Paskewitz et al., 2006). Melanization of P. berghei in the laboratory selected L35 strain of An. gambiae (Collins et al., 1986) is mediated by CLIPB3, 4, and 17 (Volz et al., 2006), while melanization of the same parasite in CTL4-depleted mosquito host depends only CLIPB4, 8, 14, and 17 (Volz et al., 2006). While this current study demonstrates that CLIPB8 acts upstream of CLIPB9 in tissue melanization, published data show that CLIPB8 must act upstream of a different PAP, potentially CLIPB4, for melanization of glass beads and potentially also P. berghei parasites (Paskewitz et al., 2006; Volz et al., 2006).

In summary, this study identifies CLIPB8 as a non-redundant member of the proPO activation system in An. gambiae that is required for melanotic tumor formation. CLIPB8 supplements the SRPN2/CLIPB9 regulatory unit in controlling melanization, acting upstream of CLIPB9 potentially by proteolytic cleavage of CLIPA8. Our findings suggest that proPO activation cascades in An. gambiae are branched and act in parallel, highlighting the complexity of the regulatory network that controls melanization in this important vector species.

4. EXPERIMENTAL PROCEDURES

4.1 Mosquito strain and rearing

The An. gambiae G3 strain (MRA-112) was maintained as described previously (An et al., 2011).

4.2 Recombinant serine proteinase production, purification, and antibody generation

The recombinant serine proteinase zymogens proCLIPB8, proCLIPB9, proCLIPB8_Xa, and proCLIPB9_Xa were produced in the insect Sf9 cell line using the Bac-to-Bac expression system (Life Technologies), and purified as described previously (An et al., 2011). The complete coding regions of proCLIPB9 and proCLIPB8 were amplified by PCR from cDNA of adult female An. gambiae with forward primers including a NotI site and reverse primers containing six codons for histidine residues followed by a stop codon and a HindIII site (for primer sequences see Table S1). The expression construct was assembled by inserting the PCR product digested with NotI and HindIII into the vector pFastBac1 (Life Technologies) digested with the same restriction enzymes. The resulting expression plasmids were used as template to generate the two mutant proCLIPB9_Xa and proCLIPB8_Xa expression constructs, both encoding recombinant proteins with a His-tag at the C-terminus. Briefly, the activation cleavage sites, IGAK for proCLIPB8, and IGMR of proCLIPB9, were replaced by IEGR, the cleavage site for factor Xa, using QuickChange® Multi Site-Directed Mutagenesis Kit (Stratagene) following the manufactory’s instructions. The resulting plasmids were transfected into Sf9 cells to generate recombinant baculoviruses according to the manufacturer’s instructions (Life Technologies). Recombinant proteinase zymogens were expressed in Sf9 cells following the manufacturer’s protocol and purified by nickel-nitrilotriacetic acid (NTA) agarose chromatography followed by ion-exchange chromatography on a Q-Sepharose column, as described previously (An et al., 2011). Fractions containing recombinant protein were examined for purity by 10% SDS-PAGE and Coomassie stain, and pooled for further use.

For the generation of antibodies against full-length An. gambiae CLIPB8 and B9, cDNA fragments encoding the mature zymogen without signal peptide were amplified by PCR using cDNA as template with forward primers including a NcoI site and six histidine codons, and reverse primers containing a stop codon and a NotI site. The expression construct was assembled by inserting the PCR product digested with NcoI and NotI into vector pET28a (Novagen) cut with the same restriction enzymes. Recombinant N-terminally His-tagged rflCLIPB8 and rflCLIPB9 were expressed in a one liter culture of Escherichia coli strain BL21 with 0.1 mM of isopropyl β-D-thiogalactoside for 5 h at 20°C, 250 rpm. Both, rflCLIPB8 and rflCLIPB9 were expressed in an insoluble form and were purified under denaturing conditions by nickel-nitrilotriacetic acid agarose affinity chromatography (Qiagen, Valencia, CA, USA). 1.5 mg aliquots of the purified rflCLIPB8 and rflCLIPB9 were further resolved by preparative SDS-PAGE and used for polyclonal rabbit antiserum production (Cocalico Biologicals, Reamstown, PA, USA). Antibodies were purified as described previously (An et al., 2012).

All PCR primer sequences used for generation of the recombinant proteins described above are listed in Table S1.

4.3 Recombinant serpin production and purification

Recombinant SRPN2 was expressed in E. coli using the pET expression system (Novagen) and purified as described previously (An et al., 2011). Briefly, the cDNA fragment encoding the full-length mature SRPN2 was amplified with the forward primer including an NcoI site, a start codon, and six histidine residue codons, and the reverse primer containing a stop codon followed by a HindIII restriction site. The expression construct was assembled by inserting the results PCR products digested with NcoI and HindIII into the expression vector pET-28a (Novagen) digested with the enzymes. The resulting plasmids were transformed into E. coli BL21(DE3) strain followed by recombinant protein expression in 1 L of LB medium with 0.1 mM IPTG, at 20°C and 225 rpm for 16 h. Bacterial cells were lysed by sonication and subjected to NTA agarose chromatography followed by ion-exchange chromatography on a Q-Sepharose column. Fractions containing recombinant protein were examined for purity by 10% SDS-PAGE and Coomassie stain, and pooled for further use. All PCR primer sequences are listed in Table S1.

4.4 Activation of recombinant zymogens and amidase activity assays

The activation of rproCLIPB8_Xa and rproCLIPB9_Xa by factor Xa was performed as described previously (An et al., 2011) with some modification. Briefly, 5 μg of recombinant purified zymogen was incubated overnight with 2 μg of commercial bovine Factor Xa (New England Biolab) in a total reaction volume of 100 μl (20 mM Tris-HCl, pH 8.0, 100 mM NaCl, 2 mM CaCl₂) at 27°C. Two negative controls were set up in parallel, in which either Factor Xa or the zymogen was replaced with same volume of buffer. Cleavage of the recombinant zymogen was determined by Western blot. To examine the amidase activity following activation, 2 μl of the above reaction were transferred to 96 well flat bottomed black microplate (Corning) followed by the addition of 100 μl of 200 μM Benzoyl-FVR-AMC (Thrombin Substrate III, Fluorogenic, EMD Chemicals) in buffer (0.1 M Tris-HCl, pH 8.0, 0.1 M NaCl, 5 mM CaCl₂). Fluorescence changes were monitored immediately in a microplate fluorescence reader (Flx 800 microplate fluorescence reader, Bio-Tek Instruments, Inc.) at excitation wavelength 360 nm and emission wavelength 440 nm every 70 seconds for 20 min. One unit of amidase activity was defined as ΔRFU/min=1. Amidase activity of the serine proteinase was defined as the activity of enzyme minus the activity of Factor Xa alone.

4.5 Serpin-proteinase complex formation and detection by Western blot and Mass spectrophotometry (MS) analysis

The potential formation of CLIPB8-SRPN2 complexes in vitro was monitored by Western blot. Recombinant proCLIPB8_Xa was activated as described in 4.4 above, mixed with purified rSRPN2 at a molar ratio of 1:10, and incubated at RT for 10 min. Reaction mixtures were then separated by 10% SDS-PAGE followed by transferring the proteins to a PVDF membrane. After blocking in 5% milk in 1x Tris-buffered saline with 0.05% Tween 20 (TBST) for 1 h, the membrane was incubated with mouse anti-His antibody (1:2000, GenScript) or rabbit anti-SRPN2 antibody [(Michel et al., 2005), 1:2000] in 1x TBST containing 0.5% milk for 1h at RT. After three 10 min washes with 1xTBST, membranes were incubated with goat anti-mouse or anti-rabbit IgG HRP-conjugated secondary antibodies (Promega, 1:20,000 dilution) in 1xTBST containing 0.5% milk for 1h at RT. After three further 10 min washes in 1xTBST, primary antibody binding was visualized by Western Lightning Chemiluminescence Reagent Plus Kit (Perkin Elmer).

In addition, after separation by 10% SDS-PAGE, bands of interest were excised from the gel, and subjected to MS analysis. In gel trypsin digestion and MALDI-MS analysis (Bruker Daltonics Ultraflex III) were performed at Biotechnology/Proteomics Core Facility in Kansas State University. Mass spectra were searched against a SwissProt protein database and analyzed by mMass software (http://www.mmass.org).

4.6 Serpin inhibition assays

To examine the inhibitory effect of SRPN2 on proteinase activity, 100 ng of in vitro activated proteinase in a volume of 2 μl was incubated with 4 μl of purified rSRPN2 (at a 10:1 molar ratio serpin to active enzyme) in 20 mM Tris-HCl, pH 8.0, and 100 mM NaCl with the addition of 1 μl BSA (2 μg/ μl) at room temperature (RT) for 10 min. The complete reaction was the subjected to an amidase activity assay as described in 4.4 above. Substitution of rSRPN2 with 4 μl of buffer (20 mM Tris-HCl, pH 8.0, 100 mM NaCl) was used to determine 100% enzyme activity.

4.7 ProPO cleavage and in vitro PO activity assays

To examine the potential cleavage of proPO by CLIPB8 and/or CLIPB9, 40 ng of purified M. sexta proPO (kindly provided by M. Gorman, Kansas State University) were incubated with 250 ng of activated rCLIPB8_Xa or rCLIPB9_Xa at 37°C for 30 min. Samples were then subjected to SDS-PAGE and Western blot analysis using antibodies against M. sexta proPO (Jiang et al., 1997) as described previously (An et al., 2011). To determine PO activity, 1 μg purified M. sexta proPO was mixed with 250 ng activated rCLIPB8_Xa or rCLIPB9_Xa at 37°C for 30 min. PO activity was measured using dopamine as a substrate as described before (Jiang et al., 2003).

4.8 RNAi and phenotypic analysis in adults

DsRNA preparation and mosquito injections were performed as described previously (An et al., 2011). Primer pairs for template amplification of dsSRPN2, dsCLIPB8, dsCLIPB9, and dsGFP are listed in Table S1. In all experiments, 1–2 day old adult female G3 mosquitoes were used for dsRNA injection, and maintained on sugar water provided ad libitum throughout the entire experiment (see 4.1 above). To examine potential genetic interactions between CLIPB8, CLIPB9 and SRPN2 in vivo, double- or triple knockdown (dKD, tKD) was performed by injection of 138 nl of solution at 1.5 μg/μl for each dsRNA. For KD and dKD controls, dsGFP was added to keep the total dsRNA dose constant at 207 ng/mosquito between treatment and controls. Melanotic tumor formation and mortalities were recorded and analyzed as detailed in An et al., 2010. Mosquito abdomens were dissected and analyzed for melanotic tumors when mortality reached 70% in the dsSRPN2/dsGFP-injected control mosquitoes. All experiments were performed with three independent biological replicates using at least 50 mosquitoes per treatment.

4.9 Total RNA isolation, cDNA synthesis and RT-qPCR

To examine knockdown efficiency, total RNA was extracted from mosquitoes 4 days post injection using TRIzol reagent (Life Technologies). The cDNA was synthesized with the iScript cDNA synthesis kit (BioRad) using 100 ng of total RNA as template in a total reaction volume of 25 μl following the manufacturer’s instructions. All qPCRs was performed in a 25 μl reaction containing 10.5 μl of 10-fold diluted cDNAs, 0.4 μM of each primer, and 1× iQ SYBR Green Supermix (BioRad) using the ABI StepOnePlus system (Applied Biosystems). AgRPS7 was used as an internal reference gene, and relative gene expression was calculated by ΔΔCt method. All qPCR analyses were conducted using three biological replicates for each gene, each with three technical repeated measurements.

4.10 Statistical Analyses

Statistical analyses were executed using GraphPad Prism 5.1 Software (GrapPad Software Inc.). Survival data were analyzed using Kaplan-Meier, and potential differences between treatment groups were analyzed using Log-Rank test (Mantel Cox). Melanotic tumor formation data were evaluated for normality of distribution using Shapiro-Wilk normality test; data were analyzed using (i) Mann Whitney U-test, if comparing two treatment groups, or (ii) Kruskal Wallis test for multiple treatment groups, with Dunn’s Multiple Comparison post test. All enzymatic activity data were evaluated using One-Way ANOVA, with Newman-Keuls post test.

Highlights.

• The clip-domain serine protease CLIPB8 is required for melanotic tumor formation in Anopheles gambiae.

• CLIPB8 is a functional serine proteinase with trypsin-like specificity.

• CLIPB8 is a non-redundant member of the prophenoloxidase (proPO) activation cascade upstream of proPO-activating proteases.

• Epistasis analyses confirm mechanistic link between melanotic tumor formation and mosquito survival.