Uptake of the Necrotic Serpin in Drosophila melanogaster via the Lipophorin Receptor-1

The humoral response to fungal and Gram-positive infections is regulated by the serpin-family inhibitor, Necrotic. Following immune-challenge, a proteolytic cascade is activated which signals through the Toll receptor. Toll activation results in a range of antibiotic peptides being synthesised in the fat-body and exported to the haemolymph. As with mammalian serpins, Necrotic turnover in Drosophila is rapid. This serpin is synthesised in the fat-body, but its site of degradation has been unclear. By “freezing” endocytosis with a temperature sensitive Dynamin mutation, we demonstrate that Necrotic is removed from the haemolymph in two groups of giant cells: the garland and pericardial athrocytes. Necrotic uptake responds rapidly to infection, being visibly increased after 30 mins and peaking at 6–8 hours. Co-localisation of anti-Nec with anti-AP50, Rab5, and Rab7 antibodies establishes that the serpin is processed through multi-vesicular bodies and delivered to the lysosome, where it co-localises with the ubiquitin-binding protein, HRS. Nec does not co-localise with Rab11, indicating that the serpin is not re-exported from athrocytes. Instead, mutations which block late endosome/lysosome fusion (dor, hk, and car) cause accumulation of Necrotic-positive endosomes, even in the absence of infection. Knockdown of the 6 Drosophila orthologues of the mammalian LDL receptor family with dsRNA identifies LpR1 as an enhancer of the immune response. Uptake of Necrotic from the haemolymph is blocked by a chromosomal deletion of LpR1. In conclusion, we identify the cells and the receptor molecule responsible for the uptake and degradation of the Necrotic serpin in Drosophila melanogaster. The scavenging of serpin/proteinase complexes may be a critical step in the regulation of proteolytic cascades.


Introduction
The immune response to pathogen challenge in humans consists of an immediate ''innate'' response (mediated via cellular responses and the action of antimicrobial peptides), followed by a delayed ''acquired'' response (mediated by antibodies). In insects, the antibody response is absent, but these organisms synthesise antibiotic peptides, activate macrophage-like cells and mount a melanization response [1]. The necrotic (nec) gene encodes a proteinase inhibitor of the serpin family (serine proteinase inhibitor). Nec controls a proteolytic cascade which activates the innate immune response to fungal and Gram + bacterial infections [2]. The Nec serpin carries an N-terminal extension which modifies its substrate specificity and is cleaved following immunechallenge [3]. In nec null mutants, the Toll-mediated immune response is constitutively activated, even in the absence of infection, implying that Nec continually restrains this immune response. The serpins have been extensively studied in mammals, where they regulate many extracellular proteolytic cascades. The coagulation, inflammatory and complement pathways are controlled by a 1 -Antithrombin, a 1 -Antitrypsin and C1-Inhibitor, respectively [4][5][6]; while Plasminogen Activator Inhibitor-1 modulates angiogenesis, affecting both wound-healing and tumour growth [7]. Disorders in serpin metabolism underlie a major group of human genetic diseases, the serpinopathies, which are associated with failure to clear inert serpin polymers [5,8,9] and homologous mutations in Necrotic similarly form inactive polymers [10,11].
Serpins interact with their target proteinase via a ''suicideinhibition'' mechanism, which destroys both serpin and proteinase and generates a covalently-linked complex [12]. Inert serpin/ proteinase complexes are removed from circulation by endocytosis in the liver [13], via receptors of the low-density lipoprotein (LDLR) family [14,15]. The LDLR family consists of a diverse group of cell surface receptors [16] that is evolutionarily conserved [17]. LDLR/ligand binding is pH-dependent, so that the complex dissociates in the low pH environment of the endosomal compartment, allowing LDLR to be recycled to the cell surface [18][19][20].
During endocytosis, the internalization of receptor-bound proteins requires Dynamin function for the pinching-off of clathrin-coated vesicles [21]. Endocytosed proteins are transported to various intracellular compartments, with the Rab family of Rasrelated GTPases being critical for coordinating vesicle formation, transport and fusion with the target membrane [22]. In particular, maturation of the early endosome coincides with the replacement of Rab5 by Rab7 and the accumulation of lumenal vesicles to form multivesicular bodies (MVB) [23]. MVB correspond to a class of late endosome, which requires Hook and Fab1 for maturation [24]. Following LDLR/ligand dissociation, transport of the free LDLR back to the plasma membrane is mediated via Rab11positive, recycling endosomes. The contents of MVBs are delivered by direct fusion, either to lysosomes, or to the plasma membrane [25]. A key component in protein sorting from late endosomes to lysosomes is ubiquitination. In this process, HRS (hepatocyte growth factor-regulated kinase substrate) binds ubiquitin and interacts with ubiquitinated cargos in the early endosome [26], while Fab1 encodes a phosphatidylinositol(3)phosphate 5-kinase which acts downstream of HRS [27,28]. In Drosophila, MVB fusion with the lysosome requires the function of the deep orange (dor) and carnation (car) subunits of the HOPS (homeotypic vacuole fusion and protein sorting) trafficking complex [29].
The garland cells represent an attractive model system to study endocytotic processes, since they are giant cells that are highly active in clathrin-mediated endocytosis. They form two rows of loosely-connected cells surrounding the oesophagus at the junction with the proventriculus, but little is known about their biological function [30][31][32][33]. Garland cells are bi-nucleated with a highly vacuolated structure featuring deep invaginations of the plasma membrane [34]. The pericardial cells, on the other hand, are arranged in two rows of approximately 20 cells, on either side of the heart. Both cell types have a similar morphology and express the WD40 domain protein, Rudhira [34,35] and the dot-Gal4 enhancer trap [36], while their positioning around the proventriculus and dorsal vessel ensures good contact with flowing haemolymph. In contrast to the garland cells, pericardial cells are mono-nucleate; although both cell types are bi-nucleate in Hyalophora cecropia and Calliphora erythrocephala [37,38]. These two cell types are homologous to athrocytes in other insects [39] and share a nephrin/podocin based filtration mechanism with the vertebrate kidney glomerulus [40]. Instead of being passed to a nephric tubule, however, the ultrafiltrate is endocytosed from lacunae in the garland and pericardial cells [34,41,42]. There is no connection between these athrocytes and the Malpighian tubules, which regulate the concentration of plasma ions and metabolites in Insects. In this respect, the garland and pericardial cell clusters resemble more the vertebrate reticulo-endothelial cells [41,43] than the kidney glomerulus. The reticulo-endothelial system consists of groups of phagocytic cells that take up denatured proteins, bacterial and viral components [44][45][46].
In this study, we investigate the role of LDL-family receptors in the Toll-mediated immune response of Drosophila by dsRNAknockout. We have tested the role of the Lipophorin Receptors, LpR1 and LpR2, in Nec processing. We show that Nec is taken up in the garland and pericardial cells by LpR1, probably as a serpin/ proteinase complex. Antibody staining against different proteins involved in clathrin-mediated endocytosis and intracellular trafficking establishes that Nec is sorted and degraded in the garland and pericardial athrocytes, but not recycled to the haemolymph. In general, the paucity of Rab11-positive vesicles indicates that these cells do not recycle endocytotic components, but are the site of protein and peptide degradation.

Necrotic expression pattern and activation by infection
Northern-blot analysis showed that nec is expressed at low levels in larvae and moderate/high levels in pupae and adults, in unchallenged flies [10]. Here we use quantitative RT-PCR to quantify nec transcript levels with and without infection. We found no nec expression in embryos, moderate levels in larvae and adults, and higher levels in pupal stages. Six hours after infection with M. luteus, nec transcript was up-regulated 10 fold in larvae and 7 fold in adults, to reach similar levels ( Figure S1). Nec expression was induced 4-6 fold after septic injury with Micrococcus luteus and 3.4 fold after fungal infection with Beauvaria bassiana in adult flies [47]. Given these results, we analysed Nec uptake after infection in both larvae and adults.
In situ hybridization showed strong nec expression in the fat-body of infected wild-type larvae, which is lost in nec transcript null larvae ( Figure S2). In addition, nec transcript was detected in midand hind-gut. We did not detect nec transcript in Malpighian tubules, imaginal discs, brain tissue, pericardial or garland cells ( Figure S2) of wild-type larvae. These results are in broad agreement with expression array data from the FlyAtlas (http:// flyatlas.org/atlas). The anti-Nec antibody detects protein staining in the fat-body of wild-type larvae, which is lost in protein null larvae (data not shown).

Nec is taken up from the haemolymph by the garland cells
We were initially unable to detect Nec protein in any larval or adult tissues, apart from the fat-body, using a rabbit polyclonal antibody that works on Western blots. Similar experiments using an antibody to Serpin27A also failed to detect antibody staining outside the fat-body. These preliminary findings are consistent with the hypothesis that serpin turnover in the haemolymph is rapid and that tissue-fixation might be slower than serpin degradation. To test this hypothesis, we blocked the formation of clathrin-coated vesicles, using the temperature-sensitive Dynamin mutation, shi ts1 [30]. Third instar shi ts1 larvae were grown, infected and dissected at the permissive temperature of shi ts1 and shifted to the restrictive temperature during fixation. This procedure combines rapid fixation with a block in endocytosis.
Under these conditions, Nec antibody staining was absent, or barely detectable, in the garland cells of infected wild-type larvae, but strong staining was seen in the garland cells of infected shi ts1

Author Summary
Serpin inhibitors control a wide range of rapid physiological responses that are activated by proteolytic cascades, such as blood coagulation, inflammation, the complement pathway, and angiogenesis. They interact with their target proteinases by a ''suicide inhibition'' mechanism, which generates an inert, denatured, serpin/proteinase complex. In mammals, humoral serpins are secreted from the liver into the blood plasma. The denatured complex is later endocytosed back into the liver and degraded. In Drosophila, the Necrotic serpin is secreted from the fatbody into the haemolymph, where it controls the humoral immune response. We show here, however, that Necrotic is not endocytosed in the fat-body, but in the garland and pericardial athrocytes. These cells clear serpins from the haemolymph extremely rapidly. The Necrotic-binding receptor for this process is LpR1, a member of the LDLR family. The endocytosed serpin is targeted for lysosomal degradation, with none being recycled to the haemolymph. More importantly, we show that mutations in LpR1 cause a profound effect on the immune response. Thus, our results indicate that the scavenging of serpin/ proteinase complexes might be a critical step in the regulation of proteolytic cascades. larvae ( Figure 1). The basal level of Nec expression was detected, at reduced intensity, in uninfected shi ts1 larvae ( Figure 1D). Similar results were seen in pericardial cells and in adult stages. The specificity of the Nec antibody was confirmed by the loss of protein staining in a shi ts ; nec null background ( Figure S2D). Given that in situ hybridization failed to detect nec transcript ( Figure S2C), the Nec protein in garland cells is presumably taken-up from the haemolymph. RNAi knockdown of nec transcript in the garland cells (of shi ts ; UAS-nec dsRNAi ; dot-Gal4 UAS-GFP larvae) has no effect on Nec antibody staining, which remained strong ( Figure S3). Note that garland cells bud-off endocytotic vesicles from labyrinthine channels, within a cortical region of several microns depth, rather than from the outer cell surface [30,31]. As a result, blocking endocytosis leaves the walls of the labyrinthine channels coated with bound Nec protein.
To confirm that the Nec visualised in garland cells derives from protein synthesis in the fat-body we knocked down nec transcription in the fat-body, in shi ts ; UAS-nec dsRNAi /c564-Gal4 larvae. Garland cells from these larvae show no detectable Nec protein ( Figure S4).
Taken together, these results indicate that garland and pericardial cells take-up Nec from the haemolymph and confirm that this protein is synthesised in the fat-body.

Time-course of Nec uptake in response to infection
The intensity of Nec-antibody staining in garland cells showed a rapid increase after infection with M. luteus, which is visible by 30 mins post-infection, peaks between 6 and 8 h and returns to basal levels within 24 h ( Figure S5).

The Nec endocytotic pathway
Nec is endocytosed in clathrin-coated vesicles. Clathrincoated vesicles co-localised with Nec in garland and pericardial cells, in both larval and adult stages ( Figure 2). Many vesicles stained AP50-positive (clathrin-coated) and Nec-positive; but some AP50 2 Nec + and AP50 + Nec 2 vesicles were also seen. These results indicate that Nec is internalized in clathrin-coated vesicles and presumably re-sorted to other, clathrin-negative, compartments.
Nec is sorted through Rab-positive endosomes for degradation. Rab5 antibody staining partially co-localizes with Nec-positive vesicles in early endosomes ( Figure 3A). Driving mutant UAS-Rab5 S43N expression in garland cells, under the control of dot-Gal4, resulted in strong accumulation of Necpositive vesicles ( Figure 3J) even in the absence of infection. Similarly, Nec partially co-localized with Rab7-positive late endosomes ( Figure 3E) and expressing UAS-Rab7 Q67 mutant protein lead to accumulation of Nec-positive vesicles in the absence of infection ( Figure 3K). The UAS-Rab5 S43N and UAS-Rab7 Q67L transgenic constructs caused a similar block in Nec processing. Taken together, these results suggest that Rab5 and Rab7 are critical for Nec sorting and degradation in garland cells.
Nec requires sorting through multi-vesicular bodies for degradation. From the early endosome, the endocytotic cargo is sorted into multivesicular bodies and can be delivered to the lysosome for degradation. HRS antibody staining was used to identify ubiquitinated endosomes, in the sorting pathway from the early endosome [26] to multivesicular bodies. As shown in Figure 4A, Nec staining co-localizes with HRS staining in early endosomes.  To investigate lysosomal delivery, we used the dor, hook and car mutants of the HOPS complex to block late endosome/lysosome fusion [29,48,49]. In all three mutants, Nec-positive endosomes/ MVB accumulate even in the absence of infection ( Figure 5). We conclude therefore that Nec sorting through MVBs via the HOPS complex is required for lysosomal delivery.
Nec co-localises with Fab1. Fab1 acts downstream of HRS [27,28] and is required for vacuolar membrane trafficking and delivery to the lysosome. Antibody staining shows partial colocalization of Fab1 and Nec, indicating that Nec is destined for lysosomal degradation by Fab1 ( Figure 6).
Internalised Nec is targeted for lysosomal degradation and is not re-exported to the haemolymph. An antibody to a lysosomal membrane protein, Lamp1, was used to confirm that Nec is trafficked to the lysosomal compartment. Co-localisation of Nec and Lamp1 antibody staining confirms that Nec is sorted from MVB to lysosomes, where it is degraded ( Figure 6).
An alternative trafficking pathway for proteins within MVBs is that they can be re-exported to the haemolymph via Rab11positive recycling vesicles [50]. We found that garland cells contain few Rab11-positive vesicles in general, and that the Nec protein staining did not co-localize with those Rab11-positive vesicles that were present (Figure 7). Therefore, the Nec protein taken-up by the garland cells is targeted exclusively for lysosomal degradation.
Identification of the Nec trafficking receptor dsRNA knockdown assays. The 6 Drosophila orthologues of the mammalian LDLR family were tested for an immune response function by dsRNA knockdown. Adult flies were injected with LDLR dsRNAs and 4 days later infected with M. luteus [51]. Toll pathway activity was assayed by quantifying transcript levels of the   antibiotic peptide Drosomycin (Drs), ( Figure S6). Knockdown efficiencies for the D-LDLR orthologues were: LpR2 (92%), LpR1 (86%), CG33087 (87%), CG12139 (91%), arrow (56%) and CG8909 (88%), as estimated by qRT-PCR. Under these conditions, silencing of CG33087, CG12139, arrow and CG8909 does not affect Drs transcript levels, with a barely significant reduction with silencing of LpR2. Strikingly, LpR1-silencing doubles Drs transcript levels. This result implies that partial knockdown of LpR1 decreases Nec activity in the haemolymph. To confirm this result we measured nec and Drs transcript levels in Df(3R)lpr1, Df(3R)lpr2 and Df(3R)lpr1/2 in infected flies ( Figure S7). Deficiency of LpR1 reduces nec (58%) and increases Drs transcript levels (206%), while deficiency of LpR2 has a weaker effect in the opposite direction. In the absence of immune challenge, Drs transcript levels are (1030%) higher in Df(3R)lpr1 than in wild-type flies ( Figure S7).

Chromosomal deletion of LpR1 blocks Nec uptake into garland cells
The LpR1 and LpR2 genes are adjacent transcripts in Drosophila. Flies carrying either a chromosomal deletion for LpR1, (Df(3R)lpr1), the pBac(LpR1) mutation, or a deletion of both LpR transcripts (Df(3R)lpr1/2), fail to take up Nec into garland cells. Conversely, deletion of the LpR2 transcript only (in Df(3R)lpr2 flies) leaves Nec uptake unaffected (Figure 8).

LpR1, LpR2, and Nec co-localize in garland and pericardial cells
Antibody staining of LpR1 and Nec in larval and adult garland and pericardial cells showed co-localisation ( Figure 9). The equivalent experiments with LpR2 antibody showed that this receptor also co-localises with Nec ( Figure S8). Taken together with the blocking of Nec uptake in Df(3R)lpr1 flies, these results confirm that LpR1 is the Nec trafficking receptor and that LpR2 is present, but does not traffic Nec, under these conditions. The LpR1 receptor is required for efficient uptake of a Nec-proteinase complex Digestion of the Nec DN core serpin (lacking the N-terminal peptide) with porcine pancreatic elastase (PPE) produces a covalently linked Nec DN -PPE complex with reduced levels of native Nec DN ( Figure 10A). This native serpin+serpin/protease complex mixture was taken up more readily by garland cells than the pure Nec DN sample, provided that the LpR1 transcript was present. Nec DN -PPE complex uptake was blocked by the Df(3R)lpr1 and Df(3R)lpr1/2 chromosomal deletions. In addition, uptake of the undigested, native Nec DN , serpin was decreased by the Df(3R)lpr2 and Df(3R)lpr1/2 deletions ( Figure 10G and 10J). These results are consistent with LpR1 being the receptor for the inert Nec-proteinase complex and LpR2 being the main receptor   for native Nec, although other receptors may well be responsible for some of the native Nec binding.

Discussion
The Necrotic serpin controls activation of the Toll-mediated immune-response in Drosophila, which represents the best-studied example of serpin-regulated proteolytic cascade in insects [2,3,52]. The Drosophila melanogaster genome encodes 15 putative inhibitory serpin transcripts that carry secretion-signal peptides [53]. In addition to nec, the Spn27A serpin controls Toll-mediated morphogenesis in the embryo and the phenol-oxidase cascade in adults [54,55]. Spn28D (CG7219) also regulates the phenol-oxidase cascade [56], while Spn77Ba regulates tracheal melanization, which also can trigger systemic expression of Drosomycin via the Toll pathway [57]. In addition, the Spn42Da transcript inhibits furin, which is involved in the maturation of secreted proteins [58].
In mammals, serpins are removed from circulation by endocytosis and degraded in the liver as inactive serpin/proteinase complexes. This aspect of serpin metabolism, however, has not  been studied previously in Drosophila. In this study, we identify the mechanism of serpin clearance by endocytosis in the garland and pericardial cells. These cells are known to take up ferritin [59] and GFP [60] and have been suggested to be homologous to mammalian reticulo-endothelial cells [41,43] or nephrocytes [34,40,61]. As in mammals, serpin turnover in Drosophila is extremely rapid, so that we were unable to detect the immuneresponse serpins Nec and Spn27A under normal conditions. However, freezing the pinching-off of endocytotic vesicles, using the shi ts1 mutation, allows us to detect serpin uptake. Endocytosed Nec is sorted, first to Rab5-positive early-endosomes and then to Rab7-positive late-endosomes. Disrupting these steps by expression of dominant negative UAS-Rab 5S43 or UAS-Rab7 Q67L , leads to accumulation of Nec-positive vesicles. Similarly, co-localisation of anti-Nec and anti-HRS antibody staining indicates that the serpin is present in early endosomes in the ubiquitin-dependent sorting pathway; while anti-Nec and anti-Fab1 confirms that the serpin is destined for lysosomal degradation. Blocking late-endosome/ lysosome fusion using HOPS-complex mutants causes accumulation of Nec-positive endosomes/MVB, indicating that Nec sorting to MVB is required for lysosomal delivery. Co-localization of anti-Nec and anti-Lamp1 antibody staining confirms that Nec is delivered to lysosomes for degradation, while the absence of Nec staining in Rab11-positive vesicles indicates that none of the serpin is recycled from MVB to the haemolymph. In summary, Nec is cleared from the haemolymph and sorted through MVB, via the ubiquitin-dependent pathway, to lysosomes for degradation.
In vitro studies in mammals have shown that different members of LDLR family have different binding specificities to different native serpins and serpin/proteinase complexes [15,62]. In this study, we have shown that LpR1 is the Nec trafficking receptor in vivo, but that neither LpR1 nor LpR2 traffics Spn27A. By analogy to mammalian systems, Nec is probably taken up by LpR1 as a complex with its target proteinase. In addition, pre-digestion of Nec with PPE increases Nec uptake in garland cells that are not deficient for the LpR1 receptor. Our results establish that active trafficking of Nec from the haemolymph can modulate the immune response. Nec clearance is extremely rapid, but deletion of the LpR1 gene sensitises the immune response: nec transcript levels decrease and Drs transcript levels increase. These results imply a regulatory feedback loop at the transcriptional level. In this context, it is significant that LpR1 appears to bind the noninhibitory serpin/proteinase complex, in preference to the native Nec serpin. Clearance of the serpin/protease complex through the athrocytes appears to compete with a regulatory feedback loop affecting nec transcription.
In summary, we establish that the Nec serpin is taken-up via LpR1 from the haemolymph and degraded in the garland and pericardial athrocytes.

M. luteus infection
The Gram-positive bacterium Micrococcus luteus, strain CIP A270, was used to activate the Toll-mediated immune response. Adult flies were challenged by septic injury with a thin tungsten needle dipped into a concentrated M. luteus culture. Flies were collected after 6 h (for Nec assays) or 24 h (for Drs response). Larvae were infected by incubation on a Petri dish carrying an overnight culture of M. luteus.

Confocal microscopy
A Leica confocal microscope and software were used. Images were processed using Adobe Photoshop.

Complex formation
The Nec core serpin, DN-Nec was incubated with PPE in PBS as previously described [52]. Porcine pancreatic elastase (PPE) was purchased from Sigma. DN-Nec was purified with the IMPACT-NT protein purification system [52].

Endocytotic assay
Garland cells were dissected and transferred to PBS solution on ice. Garland cell serpin-uptake studies were performed by incubating cells in PBS solution containing 0.2 mg/ml purified DN-Nec solution or 0.2 mg/ml flN-Nec-PPE complex. After 20 minutes incubation at room temperature, cells were fixed and stained as described above.

Whole-mount in-situ hybridisation
The Nec antisense DIG-labelled RNA probe (cDNA nucleotides 586-1450) was prepared according to the manufacturer's instructions (Roche Biochemical).
cDNA preparation and quantitative Real Time-PCR Total RNA was extracted using RNAeasy Mini Kit (Quiagen) according to the manufacturer's instructions. For qRT-PCR, 1 mg of RNA was reverse-transcribed using ThermoScript RT-PCR (Invitrogen). cDNA was quantified by real-time PCR using iQ SYBERGreen Supermix (Biorad), with the iCycler (Bio-Rad) thermocycler. Probes were normalized against a ribosomal protein 49 (rp49) control. Samples were run three times and the relative levels of a given mRNA were normalized by cycling threshold analysis (DCT). Figure S1 qRT-PCR shows low levels of nec expression during embryonic and moderate levels during larval development. Expression levels are somewhat higher in unchallenged adults than third instar larvae, but reach very similar levels in both stages