α2-Macroglobulin-Proteinase Complexes Protect Streptococcus pyogenes from Killing by the Antimicrobial Peptide LL-37*

The significant human bacterial pathogen Streptococcus pyogenes expresses GRAB, a surface protein that binds α2-macroglobulin (α2M), a major proteinase inhibitor of human plasma. α2M inhibits proteolysis by trapping the proteinase, which, however, still remains proteolytically active against smaller peptides that can penetrate the α2M-proteinase complex. Here we report that SpeB, a cysteine proteinase secreted by S. pyogenes, is trapped by α2M bound to protein GRAB. As a consequence, SpeB is retained at the bacterial surface and protects S. pyogenes against killing by the antibacterial peptide LL-37.

(␣ 2 M) 1 with high specificity and affinity (3). ␣ 2 M bound to the bacterial surface via GRAB entraps and inhibits host and S. pyogenes proteinases and thereby protects bacterial surface proteins and virulence determinants from degradation (3). Among these proteinases is SpeB, a classical enzyme of S. pyogenes (4).
SpeB is a cysteine proteinase that is secreted in large quantities by S. pyogenes. It has broad proteolytic activity and cleaves a large number of human proteins as well as surface proteins of S. pyogenes (for review, see 2). Like several other secreted bacterial proteinases, SpeB degrades and inactivates LL-37 (5), one of the major human antibacterial peptides (for review, see Ref. 6). LL-37 is small enough (4.5 kDa) to penetrate into the cage that ␣ 2 M forms around the trapped proteinase, which commonly remains proteolytically active in the cage. The hypothesis that active SpeB could be retained at the bacterial surface in complex with ␣ 2 M and protect S. pyogenes against killing by LL-37 initiated the present investigation.

EXPERIMENTAL PROCEDURES
Proteins and Peptides-The streptococcal cysteine proteinase was purified as described previously (7). The activity of the purified protein was examined by active site titration as described (8). The SpeB preparation used in this paper was 46% active. Throughout the paper the amount of active SpeB is given. To activate SpeB, the proteinase was incubated in activation buffer (1 mM EDTA, and 1 mM DTT in 0.1 M NaAc-HAc, pH 5.0) for 30 min at 37°C, and activated SpeB was used throughout this investigation. Since high concentrations of DTT were found to reduce the inhibitory capacity of ␣ 2 M, the amount of DTT used in the assays was reduced compared with previous investigations (7,8). The final concentration of DTT in all assays was 0.1 mM, which did not affect SpeB activity toward azocasein. To block SpeB activity, E64 (Sigma) was added to the proteinase to a final concentration of 10 M. ␣ 2 M was purified from human plasma as described (9), or purchased from Sigma. For some experiments ␣ 2 M was pretreated with 0.5 M methyl amine for 20 min at 37°C to convert ␣ 2 M to the fast-migrating form (10). Plasmin from human placenta and bovine trypsin were from Sigma. The SpeB was radiolabeled using the Bolton and Hunter reagent (Amersham Biosciences, Amersham, UK) according to the manufacturer's instructions. Labeled proteins were separated from free 125 I on a PD10 column (Amersham Biosciences, Uppsala, Sweden). The cleavage site of SpeB in ␣ 2 M was determined by Edman degradation of fragments obtained after incubation of ␣ 2 M with activated SpeB for 15 min. LL-37 and Texas Red-labeled LL-37 were synthesized by Innovagen AB, Lund, Sweden.
SpeB Trap Assay-Radiolabeled SpeB was mixed with 2.5 g of ␣ 2 M or 1 l of heparinized human plasma in 10 l of PBS and followed by incubation for 15 min at 37°C. The mixture was subjected to SDS-PAGE using non-reducing conditions, followed by autoradiography using a BAS-III imaging plate. The plate was scanned with a Bio-Imaging Analyzer BAS-2000 (Fuji Photo Films Co. Ltd.), and intensity was calculated using the Image Gauche program. For competition experiments, radiolabeled SpeB and different amounts of proteinase were added to ␣ 2 M simultaneously. The reactions were terminated by the addition of SDS-PAGE sample buffer and the samples were put on ice.
Analysis of Proteolysis-SpeB was mixed with 1% (w/v) azocasein (Sigma) in PBS and incubated for 1 h at 37°C. Non-degraded azocasein was precipitated with 5% trichloroacetic acid, centrifuged at 10 000 ϫ g for 5 min, and A 366 of the resulting supernatant was determined. Alternatively SpeB was incubated with 40 nM Z-Arg-AFC (Enzyme Systems Products) (11) (3)) were grown to mid-log phase (A 620 ϭ 0.5), washed in PBS, and resuspended in PBS to a final concentration of 5 ϫ 10 9 cfu/ml. The bacteria were preincubated for 30 min with PBS or ␣ 2 M, washed three times in PBS, and resuspended in PBS. Radiolabeled SpeB was activated and preincubated for 15 min with PBS or E64 and added to the bacterial suspensions. After 15 min, the bacteria were washed three times, and the radioactivity of bacterial pellets was determined. Alternatively, incubation proceeded for various time points after the washing steps, and the radioactivity in the pellets was determined as described above.
LL-37 Killing Assays-KTL3 or MR4 bacteria were grown to mid-log phase (A 620 ϭ 0.5), washed, and resuspended in 0.1 M Tris, pH 7.5, with 10 mM glucose (Tris-glucose). LL-37 (100 pmol) was pretreated for 1 h with different concentrations of SpeB, or SpeB complexed with ␣ 2 M, and added to the bacterial suspension. After 2 h, the bacteria were washed in Tris-glucose, and viable counts were performed. Alternatively, for determining the protective effects of the ␣ 2 M-SpeB complex bound to the bacterial surface via GRAB, KTL3 or MR4 bacteria were pretreated with ␣ 2 M for 30 min. The bacteria were washed three times in Tris-glucose and incubated with activated SpeB for 15 min. After three additional washes, different concentrations of LL-37 were added, and viable counts were performed after 2 h.

RESULTS AND DISCUSSION
Initial experiments were performed to expand previous studies (3) on the relationship between SpeB and ␣ 2 M. Proteinase entrapment by ␣ 2 M results in inhibition of proteinase activity against larger protein substrates, and when SpeB was incubated with different concentrations of ␣ 2 M, a dose-dependent inhibition of SpeB activity against azocasein was observed (Fig.  1A). The SpeB-␣ 2 M interaction was further investigated by mixing radiolabeled SpeB with ␣ 2 M or plasma, followed by SDS-PAGE and autoradiography. A fraction of the radiolabeled SpeB co-migrated with ␣ 2 M, demonstrating the formation of ␣ 2 M-proteinase complexes (Fig. 1B). Pretreatment of ␣ 2 M or plasma with methylamine, converting ␣ 2 M to an inactive state, or pretreatment of radiolabeled SpeB with the cysteine proteinase inhibitor E64, prevented complex formation, as did the addition of a molar excess of unlabeled SpeB (Fig. 1B).
Proteinases are captured by ␣ 2 M after cleaving a bait region within the ␣ 2 M molecule. To determine the cleavage site of SpeB in ␣ 2 M, SpeB was incubated with ␣ 2 M and the resulting ␣ 2 M-SpeB complexes were subjected to SDS-PAGE. The cleavage pattern indicated that SpeB cleaved ␣ 2 M in the bait region (data not shown), which was confirmed by NH 2 -terminal sequencing of excised protein bands. A sequence corresponding to the ␣ 2 M bait region was obtained, identifying a cleavage site between His 699 and Val 700 . Papain has the same cleavage site (12), further emphasizing the similarity in substrate specificity between SpeB and papain (13,14).
The SpeB trap assay was used to investigate the time dependence of the interaction between radiolabeled SpeB and a molar excess of ␣ 2 M. The association between SpeB and ␣ 2 M followed a hyperbolic curve reaching 50% association after ϳ1 min (Fig. 1C). Both trypsin and plasmin are efficiently inhibited by ␣ 2 M, and unlabeled trypsin, plasmin, and SpeB were used to compete with the binding of radiolabeled SpeB to ␣ 2 M (Fig. 1D). SpeB and trypsin had similar competition profiles, whereas plasmin was less efficient in inhibiting ␣ 2 M-SpeB complex formation under the conditions used.
These data show that SpeB is inactivated by ␣ 2 M when tested against azocasein in solution, and previous work has demonstrated that SpeB cleaves and inactivates the antimicrobial peptide LL-37 (5). Since proteinases in complex with ␣ 2 M generally retain their activity against small substrates, we investigated whether SpeB in complex with ␣ 2 M was still active against LL-37. SpeB was pretreated with a molar excess of ␣ 2 M, followed by incubation with Texas Red-labeled LL-37. Samples were taken at various time points and separated by SDS-PAGE ( Fig. 2A). The degradation pattern of LL-37 by SpeB is very similar to that of Pseudomonas aeruginosa elastase (5), which cleaves LL-37 in the region responsible for its antibacterial activity (15). Unexpectedly, LL-37 was cleaved more efficiently when incubated with SpeB-␣ 2 M complexes than with only SpeB. ␣ 2 M alone did not induce degradation of LL-37 (data not shown). When the activity of SpeB against the small fluorescent substrate Z-Arg-AFC (11) was tested, SpeB in complex with ␣ 2 M was found to be three to five times more active than SpeB alone (data not shown). The molecular basis for the enhanced peptide degradation by SpeB in complex with ␣ 2 M is unclear, but the observation was further substantiated in experiments with S. pyogenes bacteria. Thus, ␣ 2 M-SpeB complexes inhibited LL-37-mediated killing more efficiently than SpeB alone (Fig. 2B). Again, ␣ 2 M alone had no effect (Fig.  2B, 0 pmol of SpeB, filled bar).
The results described above show that SpeB in complex with ␣ 2 M effectively degrades LL-37 and that the presence of the complexes reduces the ability of LL-37 to kill S. pyogenes bacteria. S. pyogenes induces potent inflammatory responses leading to increased vascular permeability and leakage of plasma proteins, including ␣ 2 M, into the site of infection. As mentioned, SpeB has broad substrate specificity, and when secreted the enzyme will encounter a complex mixture of proteins. It seems unlikely that a non-selective cleavage of a large number of potential substrates would be advantageous for the pathogen. By forming complexes with ␣ 2 M a much more restricted activity, focused on small peptides like LL-37, is achieved. However, to be fully effective against antibacterial peptides attacking the bacterial cell membrane, the proteolytic activity should probably also be located at the bacterial surface.
Most strains of S. pyogenes express GRAB, an ␣ 2 M-binding cell wall-attached protein (3,16). To determine whether GRAB could mediate binding of SpeB to the bacterial surface via ␣ 2 M, radiolabeled SpeB was added to the S. pyogenes strain KTL3 or to MR4, a mutant of KTL3 lacking GRAB (3), that had been preincubated with PBS or ␣ 2 M (Fig. 3A). The results demonstrate that SpeB is bound to the surface of GRAB-expressing KTL3 bacteria preincubated with ␣ 2 M, whereas only background levels of SpeB were associated with KTL3 bacteria preincubated with PBS, or with MR4 bacteria preincubated with ␣ 2 M. Background binding was also obtained when radiolabeled SpeB was inactivated with the cysteine proteinase in-hibitor E64 and then added to GRAB-expressing KTL3 bacteria preincubated with ␣ 2 M (Fig. 3A, bar c). Following binding to ␣ 2 M at the bacterial surface, the amount of surface-associated SpeB remained approximately the same for at least 6 h (Fig.  3B).
To investigate whether ␣ 2 M-SpeB complexes at the bacterial surface protect S. pyogenes against killing by LL-37, mid-log cultures of KTL3 or MR4 bacteria were incubated with ␣ 2 M, followed by SpeB, and then washed. Subsequently, different amounts of LL-37 were added to the bacteria. After 2 h, the number of colony-forming units was determined (Fig. 3C). Compared with MR4 bacteria, KTL3 bacteria preincubated with ␣ 2 M survived significantly higher concentrations of LL-37. No difference in sensitivity to LL-37 between the bacterial strains was seen without preincubation with ␣ 2 M and SpeB (data not shown). Fig. 4 summarizes the observations of this study, demonstrating that ␣ 2 M-SpeB complexes can be formed at the surface of S. pyogenes through protein GRAB and that these complexes protect the bacteria from killing by LL-37. Previously, other mechanisms for S. pyogenes defense against antimicrobial peptides have been described. SIC, another secreted protein of S. pyogenes (17), inactivates both LL-37 and the neutrophilderived antimicrobial peptide ␣-defensin (18). Moreover, apart from cleaving LL-37, SpeB degrades proteoglycans, thereby releasing dermatan sulfate that binds to and inactivates ␣-defensin (19). Thus, S. pyogenes has developed different defense strategies against antimicrobial peptides. This apparent redundancy could be explained by the temporal aspects of an S. pyogenes infection. It has been proposed that the early phase of S. pyogenes infection is characterized by inhibition of proteolytic activity, including inhibition of complement activation, at the bacterial surface, while later phases are characterized by massive proteolysis due to the release of SpeB and host proteinases (2). In this context, SIC would be an important early defense against antimicrobial peptides, being produced in the early growth phase (17). As the infection proceeds, SpeB production will start and soluble and surface bound ␣ 2 M-SpeB complexes are formed. Thus, at this stage bacterial surface proteins are protected, and SpeBs activity is directed against smaller substrates, like antimicrobial peptides. When SpeB production further increases, the protective mechanism is overridden and bacterial surface proteins responsible for bacterial attachment to host structures will be degraded, probably facilitating bacterial spread (2,8). When the local ␣ 2 M pool is depleted, SpeB will degrade proteoglycans, thus generating free dermatan sulfate that inactivates ␣-defensin. At this late phase of infection, activation of potent host proinflammatory systems by SpeB (20,21) will induce clinical symptoms of S. pyogenes disease.
Since the discovery of antibacterial peptides (22), the significance of these peptides in the initial clearance of bacteria at epithelial surfaces has been firmly established (for review, see Refs. [23][24][25][26]. It has also become increasingly evident that proteolysis and inhibition of proteolysis contribute to bacterial pathogenicity and virulence. The finding that ␣ 2 M-SpeB complexes protect S. pyogenes from killing by LL-37 represents a novel principle for the evasion of innate immunity and unites these lines of research. In this context, it is interesting that ␣ 2 M-like proteins were recently identified in a large number of bacterial species (27), further emphasizing the need for regulation of proteolysis also in bacteria. Notably, an ␣ 2 M homologue was not found in S. pyogenes (27), which, on the other hand, can bind ␣ 2 M via protein GRAB. These observations support the notion that entrapment of proteinases through ␣ 2 M and ␣ 2 M-like proteins could be a widespread mechanism to direct bacterial proteolysis against smaller substrates, such as antibacterial peptides.