The serine protease homolog spheroide is involved in sensing of pathogenic Gram-positive bacteria

In Drosophila, recognition of pathogens such as Gram-positive bacteria and fungi triggers the activation of proteolytic cascades and the subsequent activation of the Toll pathway. This response can be achieved by either detection of pathogen associated molecular patterns or by sensing microbial proteolytic activities (“danger signals”). Previous data suggested that certain serine protease homologs (serine protease folds that lack an active catalytic triad) could be involved in the pathway. We generated a null mutant of the serine protease homolog spheroide (sphe). These mutant flies are susceptible to Enterococcus faecalis infection and unable to fully activate the Toll pathway. Sphe is required to activate the Toll pathway after challenge with pathogenic Gram-Positive bacteria. Sphe functions in the danger signal pathway, downstream or at the level of Persephone.


Introduction
The fruit fly, Drosophila melanogaster, spends its life among decaying matter and rotten fruit, where it coexists with different microorganisms. One of the main characteristics of the immune response of Drosophila melanogaster is the challenge-induced synthesis and secretion of antimicrobial peptides (AMPs). This response involves the activation of two signal transduction cascades-the Toll and IMD pathways [1]. Gram-positive bacteria and fungi activate the Toll pathway, whereas Gramnegative bacteria and Gram-positive bacteria of the genus Bacillus activate the IMD pathway [2]. In both cases, signaling leads to the activation of NF-κB transcription factors and expression of target genes including AMPs.
In the late 1980's, Charles Janeway proposed that the innate immune mechanisms are essential for the early detection and defense against infection. These mechanisms discriminate between self and microbial non-self. Janeway proposed the existence of germ-line encoded pathogen recognition receptors (PRRs) that recognize conserved signature molecules expressed by pathogens, referred to as Pathogen Associated Molecular Patterns (PAMPs) [3]. A few years later, Polly Matzinger proposed the danger signal hypothesis. This hypothesis proposed that the activation of immune mechanisms is not due to discrimination between self and non-self, but a1111111111 a1111111111 a1111111111 a1111111111 a1111111111 rather to sensing of danger signals: either recognition of pathogens, or alarm signals produced by microbial activities or by the host's own damaged cells or tissues [4].
The Toll pathway can be activated in two ways: recognition of PAMPs by circulating Pathogen Recognition Receptors (PRRs) in the hemolymph or by virulence factors, mostly proteases, secreted by the pathogens. This activation triggers proteolytic cascades in the hemolymph. The terminal protease in the cascade cleaves Spaetzle to its activated ligand form, which is able to bind the Toll receptor and activate the intracellular pathway. Depending on the triggering signal, two proteolytic cascades can be distinguished. First, the recognition cascade activated by PAMPs which includes 3 serine proteases, ModSP [5], Grass [6,7] and SPE [8]. Second, the danger signal cascade can be activated by pathogen-encoded, secreted proteases. Such abnormal protease activity indicates that potentially dangerous changes are happening. Danger signaling involves the serine protease Persephone (Psh) [6].
There are over two hundred genes coding for serine proteases (SPs) and serine protease homologs (SPHs) in the Drosophila genome [9]. SPHs maintain the serine protease fold but lack amidase activity since at least one of the catalytic triad residues is missing [9]. The physiological functions of SPHs are poorly understood, although they have been implicated in different arthropod immune responses, in the horseshoe crab [10], Manduca sexta [11] and Anopheles gambiae [12,13].
We identified the protease Grass as being required for Toll pathway activation downstream of PRRs [6]. Grass was initially identified during an RNAi based screen of serine proteases and serine protease homologs [7], but its function was incorrectly assigned, probably due to the incomplete knockdown of the gene mediated by RNAi. We decided to verify the function of the other candidates identified in this work and focused on the serine protease homolog Spheroide (Sphe). It has been reported that Sphe is involved in the activation of Toll pathway. The knockdown of sphe by RNAi induced the same phenotype upon immune challenge as that of SPE, implying that Sphe might function as an adaptor or regulator of SPE [7].
Here, we use a null mutant of sphe to demonstrate that Sphe is involved in the activation of Toll pathway. By using protease-deficient bacteria we conclude that Sphe is sensing the virulence factors (proteases) produced by pathogenic Gram-positive bacteria. Furthermore, using flies that are double mutants for both sphe and grass, we show that Sphe is involved in the danger signal cascade.

Results
Sphe is required in the activation of the immune response after a challenge with Enterococcus faecalis As previously shown for Grass, RNAi mediated knock-down could potentially give different results compared to null mutants [6]. The reasons for this are not always clear and could be attributed to incomplete knock-down and OFF-target effects that are not detected by bioinformatics tools. To circumvent limitations of RNAi-mediated knockdown, we looked for another way of inactivating sphe. In the fly line Mi{ET1}spheroide MB11555 a Minos transposon element is inserted 412 bp downstream of the start codon, in an intronic sequence. This insertion reduced the expression of the sphe transcript compared to wild type (S1A Fig). Flies in which sphe expression was reduced were more susceptible than wild type flies to infection with the Gram-positive bacterium Enterococcus faecalis (S1C Fig). Sphe mutants showed reduced activation of the Toll pathway after immune challenge compared to wild type flies as measured by the levels (30%) of drosomycin (drs) antimicrobial peptide gene expression (S1B and S1D Fig).
We excised the Minos insertion element to confirm that the susceptibility phenotype was due to its insertion in sphe. We obtained a line with precise excision of the element (sphe Δ11 ) that expresses wild type sphe mRNA levels. When sphe Δ11 flies were challenged with Enterococcus faecalis they showed normal expression of drs (S1A and S1B Fig). This excision line is used as a wild type control in subsequent experiments (ctrl) unless otherwise stated. This demonstrates that the Minos insertion was indeed responsible for the susceptibility phenotype. We also obtained two imprecise excisions, sphe Δ49 and sphe Δ104 , in which no sphe expression was detected. The sphe Δ49 deletion includes the entire transcript as well as 974 bp upstream that include 46 bp of the 3'UTR of CG9673, and 307 bp downstream that include the 5'UTR of CG9676 (S1I Fig). The sphe Δ104 deletion starts at the Minos insertion site and includes 835 bp of upstream sequence. At the protein level, the first 74 amino acids residues are missing, which include the signal peptide and 50 amino acids residues of the catalytic domain, including the His residue from the catalytic triad. Both deletions are therefore null alleles of sphe.
When sphe null mutant flies were challenged with pathogenic Gram-positive bacterium Enterococcus faecalis we observed a significant decrease of drs levels 24 hours after infection compared to that of wild type flies (drs reaches 45% of wild type level) ( Fig 1A). Furthermore sphe flies are more susceptible to this immune challenge than wild type flies ( Fig 1B). Since both null alleles sphe Δ49 and sphe Δ104 show the same phenotype, we will describe only the results obtained with sphe Δ104 .
When sphe null mutant flies, were challenged with the non-pathogenic Gram-positive bacterium, Micrococcus luteus, or by natural infection with the entomopathogenic fungus Beauveria bassiana, drs expression was comparable to that in wild type flies ( Fig 1C and 1E and S1E-S1G Fig). Accordingly, sphe null mutant flies showed the same susceptibility to Beauveria bassiana infection as wild type flies ( Fig 1D).
To confirm that the phenotype we observed is due to sphe inactivation, we overexpressed Sphe with the UAS-Gal4 system using the ubiquitous Actin5C>Gal4 driver. Sphe overexpressing flies are viable and show no obvious phenotype. Sphe overexpression does not induce the Toll pathway as measured by levels of drs mRNA. We therefore expressed Sphe in sphe mutant background, and observed the rescue of the phenotype as assayed by the induction of drs expression in response to Enterococcus faecalis infection, as well as an enhanced survival to the infection (Fig 2A and 2B).
Taken together, these data show that Sphe is involved in Toll pathway activation and is required to activate a full and efficient response to the pathogenic Gram-positive bacterium Enterococcus faecalis.

Sphe is involved in the "danger" signal Toll activation cascade
Enterococcus faecalis is a pathogenic Gram-positive bacterium that activates the Toll pathway via both proteolytic branches, through recognition of Lys-type peptidoglycan [14], as well as through production of virulence factors that activate the "danger" signal cascade [6]. To assess in which of these branches Sphe is functioning, we generated double mutants sphe Δ104 ;grass hrd in which the recognition cascade is blocked and we challenged these flies with Enterococcus faecalis. The levels of drs expression 24 hours after immune challenge are significantly decreased (drs reaches 20% of wild type level) compared to both sphe and grass hrd single mutants, to a level comparable to that of spz mutants flies (Fig 3A). This additive effect indicates that Sphe is acting in a parallel pathway to Grass, in the "danger" signal Toll activation cascade.
Enterococcus faecalis produces several virulence factors, including cytolysin, aggregation substance, the zinc metalloprotease gelatinase GelE, and the serine protease SprE [14,15]. We focused on the secreted extracellular proteases GelE and SprE as potential virulence factors that might be sensed by the danger signal cascade. To confirm the involvement of sphe in this danger signal sensing, we used protease-deficient strains of Enterococcus faecalis that were mutant for either gelE (TX5264), sprE (TX5243), or both gelE and sprE (TX5128) [16,17]. We observe slight, but reproducible, reductions in drs levels when wild type flies are challenged with protease-deficient bacteria compared to those challenged with wild type bacteria. The observed decrease in drs levels is similar to the one observed in psh mutant flies challenged with wild type bacteria suggesting that these proteases are required for activating the danger signal pathway. This is confirmed with the fact that there is no additive effect when psh mutant flies are challenged with protease-deficient bacteria compared to the same infection in wild type flies (Fig 3B).
After protease-deficient bacteria immune challenge, sphe mutants behave as wild type flies and show no susceptibility to the protease-deficient bacteria ( Fig 3C) and the levels of drs 24 hours after infection are as in wild type controls indicating normal activation of Toll pathway ( Fig 3D). The same result was found using either of the single mutants for gelE or sprE (S2A and S2B Fig) indicating that both of these virulence factors contribute to the activation of Toll pathway. We confirmed this observation by using the non-pathogenic bacterium Enterococcus faecium that is closely related to E. faecalis but lacks these virulence factors [18]. After immune challenge with E. faecium, sphe mutant flies show no susceptibility ( Fig 3E) and drs levels 24 hours after infection are as in wild type controls indicating normal activation of Toll pathway (Fig 3F). Taken together these data demonstrate that Sphe is involved in sensing proteases produced by Enterococcus faecalis for Toll pathway activation.  Sphe is involved in sensing Gram-positive pathogenic bacteria We tested another pathogenic Gram-positive bacterium, Staphylococcus aureus. We observed that sphe null mutant flies showed the same susceptibility as psh mutant flies to infection compared to wild type flies (Fig 4A), but drs levels 24 hours upon immune challenge are as in wild type controls (Fig 4B). The wild type activation of Toll pathway could however be due to the PRR pathway. In order to confirm the involvement in the "danger" signal cascade, we used double mutant sphe Δ104 ;grass hrd flies in which the recognition cascade is also blocked. drs levels are significantly decreased in sphe Δ104 ;grass hrd double mutants compared to the levels in wild type flies and to both sphe or grass hrd single mutant flies (it reaches 30% of wild type flies) ( Fig  4B). These data demonstrate that Sphe and Grass act in parallel in the sensing of virulence factors produced by Staphylococcus aureus.

Sphe is acting in the Persephone pathway
Over expression of serine proteases that lead to proteolytic cleavage of Spz constitutively activates the Toll pathway and induces drs expression in the absence of immune challenge. By over expressing the Toll pathway serine proteases in an sphe mutant background we assessed the position of Sphe in the cascades. As expected from the phenotype of sphe flies, Toll pathway activation after over expression of both PGPR-SA and GNBP1, or of SPE, is not blocked in sphe mutant background (Fig 5A and 5B). However, Toll pathway activation after Psh over expression is strongly reduced in sphe mutant background (Fig 5C). These observations demonstrate that Sphe is acting downstream of (or at the same level as) Psh in the danger signal cascade.

Discussion
A previous RNAi screen suggested that Sphe was required for Toll pathway activation [7]. By analyzing the null mutant phenotype of this SPH, we show that Sphe is not required for different kind of infections as previously reported but only after the immune challenge with the pathogenic Gram-positive bacteria Enterococcus faecalis and Staphylococcus aureus. Furthermore, we demonstrate that Sphe is a component of the danger signal cascade acting downstream (or at the same level) of Psh, and is involved in the sensing of virulence factors (proteases) secreted by these pathogenic bacteria.
The Sphe serine protease homolog has a signal peptide and a trypsin-like protease fold. Within the catalytic triad, the active serine residue is mutated to a glycine residue, blocking the proteolytic activity. Since it has no amidase activity, Sphe cannot directly activate a downstream zymogen.
Recent studies on serine proteases involved in the activation of Toll pathway during embryonic development reported that Gastrulation Defective (GD) forms a complex with Snake (Snk) and Easter (Ea) and that this association is required for the activation of Ea by Snk. When GD is itself activated, its NH 2 -terminal region interacts with sulfated proteins located in the ventral region of the perivitelline membrane. This localization acts to bring Ea and Snk together and promote the ventrally restricted processing of Ea. Surprisingly, this mediation of Snk activity is not dependent on the proteolytic activity of GD but still occurs in GD mutants that lack one of the active catalytic residues [19]. This result establishes that a proteolytically inactive SP can function as a mediator to promote zymogen activation via another SP component of a proteolytic cascade.  Serine protease homologs have been implicated in various physiological processes. In 1991, Hogg et al. reported that a mammalian serine protease homolog, protein Z (PZ), a vitamin Kdependent glycoprotein, binds to thrombin causing its conformational change and its association with phospholipid membrane vesicles. This membrane localization is important during coagulation and clotting as it partitions thrombin to the site of an injury [20]. Later studies of protein crystal structure have demonstrated that PZ functions as a cofactor regulating proteolytic activity of Factor Xa (FXa) on phospholipid vesicles [21]. This is achieved through interaction between the N-terminal domain of PZ and FXa and PZ-dependent Protease Inhibitor (ZPI), which forms a serpin/protease complex with FXa. This interaction is required for the assembly of a protein complex on the phospholipid vesicle surface, which leads to the formation of an effective inhibitory complex containing PZ/FXa/ZPI.
One reported serine protease homolog in Drosophila, Masquerade, is necessary during embryonic development to promote and/or stabilize cell-matrix interactions [22]. Homologs of Masquerade in Tenebrio molitor and Holotrichia diomphalia larvae are required for the proteolytic activation of prophenoloxydase, suggesting a function as a cofactor of the active protease [23]. The Manduca sexta serine protease homolog SPH3 is required in the immune response of the moth to infection with Gram-negative bacterium Photorabdus luminescens. SPH3 was initially identified as a target for the Repeats-in-toxin (RTX)-metalloprotease, protease A (PrtA), which is secreted by the bacterium. Upon infection, SPH3 is upregulated in both the fat body and hemocytes. RNAi-mediated knockdown of SPH3 increased the susceptibility of moths to infection by Photorabdus luminescens [11].
Sphe is involved in the sensing of proteases produced by Gram-positive pathogenic bacteria. We can hypothesize that Sphe is recruited in a complex that mediates the activation of a target SP that could be Persephone on infection with a virulent strain. However, Sphe it is not sensing infection by the entomopathogenic fungus Beauveria bassiana nor commercial Aspergillus oryzea proteases directly injected in the body cavity even both events induce activation of the Toll pathway through Psh [6] (S1H Fig). Two further SPHs, sphynx1 and sphynx2, were identified in the RNAi screen of Kambris et al., together with sphe, as putative components of the Toll pathway proteolytic cascade. Since Sphe functions upon infection with specific pathogens, it is likely that Sphynx1 and Sphynx2 might also be involved in the activation of the immune response against other pathogens.
Further investigation is necessary to elucidate the mechanisms by which Sphe is functioning and how it contributes to the sensing of these virulence factors. In addition, characterization of other Drosophila SPHs would give insight into their mechanisms of action in other proteolytic cascades.
Natural infection with Beauveria bassiana was performed as described [2]. Injection of sublethal doses of commercially available proteases from Aspergillus oryzae (over 500 U/g; P6110; Sigma-Aldrich) was previously described [6].
At least three independent survival experiments were performed. In each experiment and for each genotype, a mix of 20-30 (both males and females) six to eight days-old flies were infected with E. faecalis, protease-deficient E. faecalis or S. aureus, or by natural infection with B. bassiana at 29˚C [27]. The survival data was plotted using GraphPad Prism Software and for statistical analysis we used Log-rank (Mantel-Cox) test.