The Spl Serine Proteases Modulate Staphylococcus aureus Protein Production and Virulence in a Rabbit Model of Pneumonia

Staphylococcus aureus is a versatile human pathogen that produces an array of virulence factors, including several proteases. Of these, six proteases called the Spls are the least characterized. Previous evidence suggests that the Spls are expressed during human infection; however, their function is unknown. Our study shows that the Spls are required for S. aureus to cause disseminated lung damage during pneumonia. Further, we present the first example of a human protein cut by an Spl protease. Although the Spls were predicted not to cut staphylococcal proteins, we also show that an spl mutant has altered abundance of both secreted and surface-associated proteins. This work provides novel insight into the function of Spls during infection and their potential ability to degrade both staphylococcal and human proteins.

Histopathology scoring of the lungs of all of the infected rabbits in both groups was also performed, regardless of the time of death. The lungs were scored on the presence of pleuritis, edema, bronchopneumonia, necrosis, lymphoid cell/heterophil infiltration, and Gram-positive bacteria. Each of the six categories was given a score of 0 (within normal limits) to 3 (severe and extensive). The average total histopathology scores were nearly identical in WT strain-infected rabbits and Δspl::erm mutant-infected rabbits (Fig. 1C), and rabbits in both groups exhibited bronchopneumonia (Fig. 1D). Interestingly, the Δspl::erm mutant-infected rabbits displayed a pattern of one high-scoring lung and one low-scoring lung, while lung scores appeared more evenly distributed in the WT strain-infected rabbits ( Fig. 1C and E). A D'Agostino-Pearson normality test determined that the WT strain-infected lung scores follow a normal distribution (P ϭ 0.27), while mutant-infected lung scores do not (P ϭ 0.01), suggesting that the mutant-infected lungs represent two distinct populations. When the difference in score between the two lungs was calculated (Δ lung score) for each rabbit, the spl mutantinfected rabbits had a significantly greater difference between their lungs than WT strain-infected rabbits did (Fig. 1E, P ϭ 0.012). On the whole, these findings indicate that although no significant difference in the lethality of the Δspl::erm mutant was . Two experiments were performed, with three rabbits per group in each experiment. The P value (0.11) was calculated with a log rank (Mantel-Cox) test. (B) Gross pathology (top) and histopathology (bottom) of USA300 WT and spl mutant strain-infected rabbit lungs euthanized on day 6. Histopathology of H&E-stained infected lung tissue is shown. (C) Total histopathology score of each lung, combined from six scoring categories. n ‫؍‬ 12 lungs per group. (D) Bronchopneumonia score of each lung. n ‫؍‬ 12 lungs per group. P ‫؍‬ 0.09 calculated with an unpaired, two-tailed t test. (E) Difference between the left and right lung scores of each rabbit. For each rabbit, the value of the lung with the lower total clinical score was subtracted from the value of the lung with the higher clinical score. n ‫؍‬ 6 rabbits per group. The P value was calculated with an unpaired, two-tailed t test.
observed, this mutant also produced a clear phenotype of lung damage that has a constrained localization within the lungs.

SplA induces mucin 16 release from CalU-3 cells. Mucin 16 is an~25-MDa,
heavily glycosylated cell surface protein that is found on multiple epithelial tissues in the body, including the ocular surface and airway epithelia (19)(20)(21). Like other mucins, mucin 16 provides lubrication of epithelia, as well as a protective barrier against pathogens. Two studies have reported that RNA interference-mediated knockdown of mucin 16 in human corneal-limbal epithelial cells renders these cells more susceptible to S. aureus adherence (22,23). Additionally, the Streptococcus pneumoniae metalloprotease ZmpC has been shown to cleave mucin 16 from a variety of human epithelial cell types, allowing the bacteria to increasingly invade these cells in vitro (24). An adenovirus capable of causing keratoconjunctivitis also was recently shown to induce mucin 16 release from ocular epithelial cells, facilitating infection (25). These findings demonstrate that S. aureus and other pathogens encounter mucin 16 as a barrier to the host epithelium and are able to disrupt this barrier in order to colonize or infect the host.
A study describing the crystal structure and cleavage site preference of SplA found that this enzyme has highly specific substrate requirements. In an experiment with a cellular library of peptide substrates, SplA was found to cleave exclusively substrates containing the Trp/Tyr-Leu-Tyr-Thr/Ser (W/Y-L-Y-T/S) motif (26). That study also noted that mucin 16 contains a Y-L-Y-S sequence and could be a likely target for SplA cleavage ( Fig. 2A) (26). On the basis of this evidence, as well as our observation that the Δspl::erm mutant exhibited more localized disease in the rabbit pneumonia model, we hypothesized that SplA cleaves mucin 16 from lung cells. To test this, confluent monolayers of the lung epithelial cell line CalU-3 were treated with purified, His6tagged SplA. Because of the large size of mucin 16 ( Fig. 2A), CalU-3 medium was tested for mucin 16 with a slot blot apparatus rather than by gel electrophoresis. Probing with a mucin 16 antibody revealed that mucin 16 is endogenously released from CalU-3 cells (Fig. 2B). This is in accordance with previous reports that mucin 16 is naturally shed from epithelia (19,27,28). Incubation with SplA increased this shedding in a dosedependent manner. When SplA was treated with the serine protease inhibitor 3,4dichloroisocoumarin (3,4-DCI) before it was added to the cells, mucin 16 shedding was decreased to a level similar to that in untreated cells (Fig. 2B). This suggests that SplA activity is required for mucin 16 removal, by either cleaving it at the Y-L-Y-S site or inducing its shedding indirectly by degradation of another substrate. This activity may allow S. aureus to perturb the airway epithelial barrier and cause more disseminated disease.
The ⌬spl::erm mutant retains viability in whole blood, hemolysis activity, and protease production. S. aureus produces a multitude of virulence factors that function by thwarting the innate immune defenses of the infected host (29). Based on the in vivo phenotype of the Δspl::erm mutant, we hypothesized that it may have an altered capacity to survive these immune assaults. To test this, we compared the survival of USA300 WT and Δspl::erm mutant cells in whole human blood. Whole human blood contains cellular and complement-mediated innate immunity components and thus is used as a general assay of bacterial survival under these conditions. When log-phase bacteria were inoculated into whole human blood and incubated for 3 h, 60% of USA300 WT bacteria remained viable. However, 40% of Δspl::erm mutant bacteria remained viable (Fig. 3A), although this difference was not statistically significant.
Alternatively, we hypothesized that the Δspl::erm mutant had altered production of one or more secreted virulence factors whose levels could be modulated by proteolytic degradation by the Spl proteases in S. aureus. To investigate this, the activity of the secreted alpha toxin (Hla) and the secreted proteases staphopain A (ScpA) and aureolysin (Aur) were compared in USA300 WT and Δspl::erm mutant bacteria. Hla is a potent cytolysin and immunomodulatory virulence factor that has been shown to contribute to pathogenesis in a number of S. aureus infection models, including a murine model of pneumonia (30,31). In an Hla activity assay (32), the Δspl::erm mutant did not exhibit a change in Hla activity from the USA300 WT strain (Fig. 3B), indicating that hemolysis is unaffected by the Spls. Next, we tested the activity levels of the S. aureus secreted proteases staphopain A (ScpA) and aureolysin (Aur). Both of these proteases inhibit the classical and alternative pathways of complement activation (33). ScpA also cleaves the neutrophil chemotaxis receptor CXCR2, blocking neutrophil recruitment (34), and exhibits activity against the extracellular matrix protein collagen, whose degradation could promote host damage during infection (35). Aur is able to inhibit complement activation by cleaving the complement protein C3 (36). ScpA activity in cell-free spent medium was tested with a FRET substrate containing the ScpA cleavage site from human CXCR2 (34,37). Over the course of growth, the Δspl::erm mutant displayed the same ScpA activity as the USA300 WT strain, indicating that the Spls do not affect ScpA activity (Fig. 3C). A milk plate assay to detect Aur activity showed no decrease in the Δspl::erm mutant (Fig. 3D). On the whole, these results suggest that the Spls do not modulate ScpA or Aur activity.
Proteomic studies reveal differences between USA300 WT and ⌬spl::erm mutant strain secreted and surface proteins. Finally, the surface and secreted proteomes of the USA300 WT and Δspl::erm mutant strains were analyzed. The objective of these experiments was to identify changes in protein levels that might explain the altered virulence phenotype of the Δspl::erm mutant and suggest potential S. aureus proteins that are degraded by the Spls. The proteomic data were analyzed with the Scaffold program using Fisher's exact test, and all hits with a P value of Յ0.010 were considered significant. The surface proteomics identified 63 proteins that were detected with increasing abundance in the Δspl::erm mutant relative to USA300 WT (Table 1) and 21 proteins detected with decreasing abundance in the Δspl::erm mutant ( Table 2). A few genes involved in virulence and immune evasion were increased in the Δspl::erm mutant, including those encoding IsdA and SpA. IsdA promotes S. aureus adherence to desquamated nasal epithelial cells (38), fibrinogen, and fibronectin (39) and enhances S. aureus survival of bactericidal fatty acids and peptides in human skin (40). SpA binds to IgG, which protects S. aureus from opsonophagocytosis and dysregulates the host adaptive immune response (41,42). The cell wall-associated protein EbpS was also increased on the surface of spl mutant bacteria. Although EbpS was once thought to act as an adhesin by binding to elastin, recent studies have found that its contribution to bacterial adhesion is minimal (43). However, EbpS does mediate Zn 2ϩ -dependent growth and biofilm formation (44). Further, an investigation of the differential expression of S. aureus genes in murine models of colonization and pathogenesis found that ebpS was significantly upregulated in the blood relative to the nares (45). Therefore, this protein may play a role during infection. Increased levels of these proteins could enhance S. aureus immunomodulation and virulence.
Proteins with decreased levels on the surface included the cell wall-anchored adhesins SdrC, SdrD, and SdrE. These proteins are members of the Clf-Sdr MSCRAMM family, and all have been demonstrated to promote S. aureus adhesion or immune evasion (46)(47)(48). SdrC self-associates to promote biofilm formation (49), while both SdrC and SdrD promote S. aureus adherence to desquamated nasal epithelial cells (38). SdrE inhibits classical (50) and alternative (51) complement activation. The decrease of these proteins in the spl mutant suggests that one or more Spl proteases indirectly affect their abundance, rather than directly degrading them.
Secreted proteomics identified 48 proteins increased (Table 3) and 28 proteins decreased (Table 4) in the spl mutant relative to the USA300 WT strain. As in the surface proteomics, IsdA was increased in the Δspl::erm mutant secretome, suggesting that it is more abundant both on the cell surface and in shedding from the cell due to cleavage or cell lysis. Multiple virulence factors were also increased, including LukD, Sbi, Nuc, and Efb. LukD is part of the LukED bicomponent pore-forming toxin, which is cytolytic to several cell types, including monocytes and polymorphonuclear leukocytes (52,53). LukED has also been reported to promote S. aureus growth in human blood by facilitating iron scavenging (54). Like SpA, Sbi binds IgG, which prevents opsonophagocytosis of S. aureus (55). Nuc is a potent secreted DNase that is able to degrade neutrophil extracellular traps (NETs), promoting S. aureus survival in the lung and host killing in a murine model of respiratory infection (56). NET degradation by Nuc also induces macrophage death (57). Efb is a secreted protein that binds fibrinogen and complement C3 to form a protective shield around S. aureus and inhibit its phagocytic clearance (58,59). The increased production of these proteins in the Δspl::erm mutant suggests that it may have enhanced immunomodulatory properties that contribute to its increased lethality in the rabbit model of pneumonia. In contrast, multiple secreted virulence factors are decreased in abundance in the spl mutant, including the ⌼ hemolysin component HlgB, the staphylococcal complement inhibitor (SCIN), and the ␦ hemolysin Hld (Table 4). HlgB is the F subunit of the ⌼ hemolysin pore-forming toxins (60-62). Hlg toxins have been shown to contribute to virulence in a murine model of septic arthritis (63) and to induce eye injury when directly injected intraocularly into rabbits (64), as well as promote S. aureus virulence in rabbit endophthalmitis infection models (65)(66)(67). SCIN is a secreted protein that inhibits complement activation by binding to C3 convertases (68,69). The Hld cytolysin is translated from RNAIII, the regulatory RNA that is the effector of the staphylococcal agr system (70). Decreased RNAIII would indicate lower agr activation in the spl mutant; however, our finding that Hla and ScpA activities are not decreased in the spl mutant suggests that this is not the case (Fig. 3B and C). Altered Hld levels therefore may be due to a posttranslational modification.

DISCUSSION
The Spl proteases are an intriguing topic because, despite evidence that they contribute to S. aureus-host interaction, their specific functions have not been identified. Structural and activity analyses of SplA, SplB, and SplD have revealed that these enzymes are likely highly specific, given that they have cleavage preferences at subsites beyond the P1 residue of cleavage (26,(71)(72)(73). Further, the Spls are not produced as zymogens but, in some cases, appear to require the binding of a preferred substrate in order to take on an enzymatically active conformation (72), as has been demonstrated by characterization of SplB activity (71). The goals of this study were to investigate the function of the spl operon in virulence and to identify possible substrates of the Spl proteases.
The virulence phenotype of the spl mutant is complex, since the mutant-infected animals had a statistically insignificant change in lethality but more localized, less disseminated lung damage. We tested SplA cleavage of predicted substrate mucin 16 (26) and found that it induces the shedding of this protein from CalU-3 cells. SplA may therefore promote S. aureus invasion and spreading by removing mucin 16 from epithelial cells. Since mucin 16 is found on ocular, airway, and female reproductive tract epithelial cells (19), this function could facilitate infection at multiple body sites. This finding is also the first report of an Spl enzyme potentially cleaving a host target. Proteomic studies of the secreted and surface proteins present on the USA300 WT and spl mutant strains revealed many proteins altered in abundance. The fact that intracellular and housekeeping genes were identified is one caveat of this experiment, suggesting that cell lysis may have occurred during sample preparation. However, none of the proteins identified contain the consensus motifs that have been identified for substrates of the Spls. These are W/Y-L-Y-S/T for SplA (26), W-E-L-Q for SplB (71), and R-Y/W-P/L-T/L/I/V-S for SplD (73). We therefore cannot predict any specific hits that are directly cleaved. However, it is possible that the Spls are able to cleave substrates with some deviation from these consensus motifs. The abundance of some proteins may also be indirectly affected by the Spls. Alternatively, SplC or SplE could be responsible   (3), their activities are likely identical. The proteomic results show virulence factors in both the increased and decreased protein groups, so it is difficult to predict specific virulence factors that are responsible for the virulence phenotype observed. A few of the proteomic hits also do not match our in vitro studies. For example, ScpA was found to be decreased in the spl mutant (Table 4) but unaltered when we tested its activity with a specific ScpA substrate   (Fig. 3C). The decrease in Hld also could correspond to lower RNAIII, indicating lower agr activation, but this was also not observed in our Hla and protease studies ( Fig. 3B and C). Hld translation could therefore be altered, or it may be degraded posttranslationally. Hld is a member of the phenol-soluble modulin (PSM) family of peptides, which are known to be protease labile (74,75). Thus, the lower Hld levels in the spl mutant could be due to degradation by an unknown protease. Our findings have some similarities to a previous study that tested the virulence and proteomics of a total protease knockout of S. aureus (74). In a murine model of sepsis, the protease null mutant had increased lethality in mice but reduced invasion of organs. Proteomics also identified an increase in the abundance of agr-regulated virulence factors such as Hla and other leukocidins. Although we did not observe an increase in Hla, some of our proteomic results mirrored those of the previously reported study. For example, the previous paper reported increased abundance of Sbi, LukE, Seq, Geh, Efb, IsdA, IsaA, Spa, and EbpS in the protease null mutant, all of which were increased in our spl mutant as well (Tables 2 and 4). This suggests that the Spl proteases contributed to the previously reported phenotypes and corroborates the model in which secreted proteases modulate virulence factor production.
A recent study found that the Spls are important mediators of the host allergic airway response to S. aureus (8). In the serum of five healthy S. aureus carriers, strong IgG4 antibody binding to the Spls was observed. Purified SplD induced a Th2-type response in a tissue explant model with tissue from human nasal polyps. In a mouse allergy model, inhalation of SplD also induced eosinophil infiltration and IgE antibodies to SplD (8). IgG4, IgE, and other Th2-mediated responses are involved in allergic airway disease (75). Interestingly, many allergens that induce allergic responses in the lungs are proteases, including proteases produced by Aspergillus fumigatus (76) and mites and cockroaches (77). These findings indicate that, in addition to SplD, the other Spls may cleave substrates in airway cells and contribute to allergic airway disease and other immune responses in the airway. Our finding that SplA induces mucin 16 shedding adds to this model, as mucin 16 removal could facilitate the ability of the other proteases to act on host cell targets.
The Spls are clearly highly specific proteases, and their natural targets have been difficult to identify. However, mounting evidence suggests that they are expressed in vivo and affect host immune responses. Future work will tease apart the roles of individual Spls in interaction with the host and unravel their host effects, hopefully by identification of their cleavage substrates in the host.

MATERIALS AND METHODS
Strains and plasmids. LAC is a USA300 CA-MRSA strain (78), and the erythromycin-sensitive version called USA300 WT (79) was used in this study. The LAC Δspl::erm mutant strain was generated by phage transduction from RN6390 Δspl::erm (3) into LAC with 80␣. Strains were cultured on tryptic soy agar (TSA) plates or in tryptic soy broth (TSB) at 37°C, with shaking at 200 rpm for broth culture.
The SplA purification plasmid pSK236/splA6xHis was kindly provided by Suzan Rooijakkers of the University Medical Center Utrecht. This plasmid contains C-terminally His6-tagged splA that is under control of the S. aureus scn (SCIN) promoter. The plasmid was moved by electroporation into S. aureus Newman to maximize SplA production, since scn is positively regulated by the sae system (80) and sae expression is enhanced in the Newman strain (81).
Rabbit pneumonia model. The rabbit pneumonia model was prepared in accordance with reference 18. The USA300 WT and Δspl::erm mutant strains were grown overnight in TSB, washed once in sterile TSB, and resuspended at 1 ϫ 10 10 CFU/ml. Adult Dutch belted rabbits (Bakkom) were anesthetized with ketamine (25 mg/kg) and xylazine (20 mg/kg) administered subcutaneously. A ventral midline incision was made through the skin and into the trachea; a catheter was then inserted and fed into the lungs. Approximately 2 ϫ 10 9 CFU of washed bacteria in 200 l of TSB was delivered through the catheter to each rabbit. Inocula were plated to confirm the dose given. The actual dose was 2 ϫ 10 9 to 3 ϫ 10 9 CFU. The incisions were closed, and the rabbits were monitored for 6 days. Over the course of the experiment, the rabbits were treated for pain with buprenorphine (0.05 mg/kg twice daily) through intramuscular injection. Rabbits were euthanized by intravenous injection of 1 ml of a mixture of the sodium salts of phenytoin and pentobarbital (Beuthanasia-D) either on day 6 or if they were found during the course of the experiment to be unable to right themselves or exhibit escape behavior. At the time of death, rabbit lungs were photographed; tissues were fixed (10% neutral buffered formalin), processed, and embedded; and tissue sections were stained with H&E and histopathologically scored in accordance with standard principles (82).
SplA purification. S. aureus Newman expressing pSK236/splA6xHis was grown overnight in 5 ml of TSB plus 10 g/ml chloramphenicol, and then the entire culture was added to 200 ml of TSB and grown overnight. The spent medium was harvested by centrifugation and filtration, and 121 g of ammonium sulfate was slowly dissolved in the spent medium to achieve approximately 85% saturation. The sample was centrifuged at 30,000 ϫ g at 4°C for 30 min to collect the pellet, which was dissolved in 10 ml of PBS and dialyzed into 50 mM Na 2 HPO 4 with 0.3 M NaCl at pH 8.0 and then into the same buffer with 10 mM imidazole for loading onto the His-Select nickel affinity gel (Sigma). Batch purification with the His-Select nickel affinity gel was performed according to the manufacturer's protocol. The final protein was dialyzed into PBS and stored at Ϫ80°C.
Mucin 16 slot blot assay. CalU-3 cells were seeded at 500,000/well into 24-well plates and cultured with 1 ml/well minimum essential medium (Gibco) containing 10% heat-inactivated fetal bovine serum (Gibco) and 1ϫ nonessential amino acids (Gibco) at 37°C in 5% CO 2 for 2 days. For the SplA-induced shedding experiment, purified SplA was added to fresh medium without serum at 500 l/well to avoid serum cross-reactivity with the mucin 16 antibody. For inhibition of SplA, 3,4-DCI was added at a final concentration of 10 M. The cells were incubated with SplA at 37°C in 5% CO 2 for 2 h, and the medium was removed for Western blotting. For the slot blot assay, samples were loaded onto polyvinylidene difluoride (PVDF) with a vacuum and a slot blot apparatus. The PVDF was first equilibrated in methanol for 15 s, in distilled H 2 O (dH 2 O) for 2 min, and in TBS plus 0.5% Tween 20 for 5 min. It was then placed into the slot blot apparatus on top of two pieces of filter paper wetted with TBS plus 0.5% Tween, and the vacuum was turned on low to drain excess buffer. A 100-l volume of each sample was loaded into the slots, and then the vacuum was turned on until the liquid was absorbed. TBS plus 0.5% Tween 20 at 100 l/well was then loaded to wash any unbound sample from the wells. The membrane was dipped into methanol and then dH 2 O and then blocked for 2 h in 5% BSA in TBS plus 0.5% Tween 20. A primary mouse anti-CA 125 antibody was diluted 1:200 in blocking buffer and incubated for 1 h, and a horseradish peroxidase-conjugated goat anti-mouse secondary antibody was diluted 1:5,000 in blocking buffer and incubated for 1 h. The membrane was washed with TBS plus 0.5% Tween 20 between steps. The membrane was incubated in chemiluminescent substrate and exposed to film.
Blood survival assay. Blood survival of the USA300 WT and Δspl::erm mutant strains was measured as previously described (83). Strains were subcultured and grown to mid-log phase (optical density at 600 nm [OD 600 ] of 0.8) and centrifuged at 8,000 ϫ g for 5 min. Bacteria were washed twice in TSB and resuspended to 1 ml. A 100-l volume was added to 1 ml of heparinized human blood, and the culture was placed on a tumbler at 37°C for 3 h. Serial dilutions were plated on TSA and counted.
Hemolysis activity assay. An Hla activity assay was performed as previously described (84). Briefly, overnight cultures of the USA300 WT and Δspl::erm mutant strains were subcultured at 1:500 in TSB and grown for 24 h. Spent medium from these cultures was serially diluted 2-fold across a 96-well plate. Defibrinated rabbit blood (HemoStat Laboratories) was centrifuged to pellet rabbit erythrocytes, which were washed three times in 1.1ϫ PBS and resuspended to a final concentration of 3%. A 30-l volume of each spent medium dilution was mixed with 70 l of the erythrocyte solution in the 96-well plate and then incubated statically at room temperature for 1 h, after which the OD 630 was measured. The OD 630 was plotted versus the percent spent medium on a four-parameter logistic curve with the Prism program, and the midpoint of the curve (the percentage of medium needed to lyse 50% of the erythrocytes) was used as an indicator of activity.
ScpA activity assay. Overnight cultures of the USA300 WT and Δspl::erm mutant strains were diluted to an OD 600 of 0.1 in 25 ml of TSB and grown for 24 h. At selected time points, spent medium was collected and incubated with a fluorescent resonance energy transfer (FRET) substrate that is based on the CXCR2 ScpA cleavage site (34). The FRET assay was performed in accordance with reference 85. The substrate was resuspended to a concentration of 50 M in 20 mM Tris-HCl, pH 7.4. Spent medium was also buffered to 20 mM Tris-HCl, pH 7.4. The substrate (25 l) was mixed with 175 l of buffered spent medium, and fluorescence (excitation, 490 nm; emission, 520 nm) measurements were taken every 2 min for 30 min with a Tecan plate reader. The activity at each time point was expressed as the slope (fluorescence/time).
Aur and hemolysis plate activity assays. Overnight cultures of the USA300 WT and Δspl::erm mutant strains were spotted in 5-l amounts onto blood agar (5% [vol/vol] rabbit blood and 3% Bacto agar) or milk agar (5% nonfat dry milk and 3% Bacto agar) plates. The plates were incubated overnight at 37°C, and pictures of colonies with zones of clearing were taken.
Proteomics. Overnight cultures of the USA300 WT and Δspl::erm mutant strains were grown in TSB and harvested for secreted and surface proteomic profiling. Secreted proteome profiling was carried out in accordance with reference 86, and surface proteomic analysis was carried out in accordance with reference 87.

ACKNOWLEDGMENTS
We thank Les Shaw (University of South Florida, Tampa, FL) for assistance with proteomic studies and Suzan Rooijakkers (Universiteit Utrecht, Utrecht, the Netherlands) for kindly providing the SplA purification plasmid.
A.E.P. was funded by American Heart Association predoctoral fellowship award 14PRE19910005. Studies in the laboratory of A.R.H. are supported by Project 3 of NIH grant AI083211. W.S.-P. and P.M.S. were supported by startup funds from the University of Iowa Carver College of Medicine.