Using Every Trick in the Book: The Pla Surface Protease of Yersinia pestis Marjo Suomalainen, Johanna Haiko, Päivi Ramu, Leandro Lobo, Maini Kukkonen,Benita Westerlund-Wikström, Ritva Virkola, Kaarina Lähteenmäki, and Timo K. Korhonen Faculty of Biosciences, General Microbiology, University of Helsinki, timo.korhonen@helsinki.fi Abstract. The Pla surface protease of Yersinia pestis, encoded by the Y. pestis-specific plas-mid pPCP1, is a versatile virulence factor. In vivo studies have shown that Pla is essential in the establishment of bubonic plague, and in vitro studies have demonstrated various putative virulence functions for the Pla molecule. Pla is a surface protease of the omptin family, and its proteolytic targets include the abundant, circulating human zymogen plasminogen, which is activated by Pla to the serine protease plasmin. Plasmin is important in cell migration, and Pla also proteolytically inactivates the main circulating inhibitor of plasmin, α2-antiplasmin. Pla also is an adhesin with affinity for laminin, a major glycoprotein of mammalian basement membranes, which is degraded by plasmin but not by Pla. Together, these functions create uncontrolled plasmin proteolysis targeted at tissue barriers. Other proteolytic targets for Pla include complement proteins. Pla also mediates bacterial invasion into human endothelial cell lines; the adhesive and invasive charateristics of Pla can be genetically dissected from its proteolytic activity. Pla is a 10-stranded antiparallel β-barrel with five surface-exposed short loops, where the catalytic residues are oriented inwards at the top of the β-barrel. The se-quence of Pla contains a three-dimensional motif for protein binding to lipid A of the lipopolysaccharide. Indeed, the proteolytic activity of Pla requires rough lipopolysaccharide but is sterically inhibited by the O antigen in smooth LPS, which may be the selective advan-tage of the loss of O antigen in Y. pestis. Members of the omptin family are highly similar in structure but differ in functions and virulence association. The catalytic residues of omptins are conserved, but the variable substrate specificities in proteolysis by Pla and other omptins are dictated by the amino acid sequences near or at the surface loops, and hence reflect differ-ences in substrate binding. The closest orthologs of Pla are PgtE of Salmonella and Epo of Erwinia, which functionally differ from Pla. Pla gives a model of how a horizontally trans-ferred protein fold can diverge into a powerful virulence factor through adaptive mutations. 24.1 Pla as a Member of the Omptin Family of Aspartic Proteases Surface-associated proteolysis has for long been recognized as an important feature in migration of eukaryotic cells across tissue barriers that are formed by extracellular matrices (ECM) and basement membranes (BM). This holds for migration of phago-cytes to infected tissue sites as well as migration of metastatic tumor cells into

the circulation and secondary tissue sites (Plow et al. 1999; Myöhänen and Vaheri

24

Using Every Trick in the Book: The Pla Surface Protease of Yersinia pestis 269

Fig. 1. A cladogram presentation of omptin protein sequences. The omptin sequences were from the MEROPS database (http//:merops.sanger.ac.uk), except for PgtE which was from Guina et al. 2000. The protein sequences were aligned by the Clustal W programme (http:// www.ebi.ac.uk/clustalw/index.html) and the tree was drawn with the Phylodraw software (http://pearl.cs.pusan.ac.kr/phylodraw).

2004). Much less attention has been focused on the mechanisms of how invasive bacteria spread across tissue barriers of the host. Secreted hydrolytic enzymes, e.g. proteases and lipases, that damage tissue structures and barriers are obvious viru-lence factors of a number of bacterial pathogens. However, several pathogens, such as most species in the Enterobacteriaceae do not express active surface-bound or secreted proteolytic enzymes. Recent evidence has indicated that these bacteria turn themselves into proteolytic organisms by using host derived proteolytic systems. For this aspect, Yersinia pestis and its Pla surface protease make a prime example of how bacteria engage the host-derived plasminogen (Plg) system to cause tissue damage and to enhance bacterial spread.

The Pla surface protease is a member of the omptin family of transmembrane β-barrels in the outer membrane of Gram-negative bacteria. These proteins share a high identity in their predicted amino acid sequence, a similar size of 290-301 amino acids, and the same predicted protein fold (discussed in more detail in section 24.3 below). At present, the omptin family contains 13 members from 9 Gram-negative bacterial genera (Fig. 1). The individual omptins share ca. 40-50% sequence identity to any member of the family, and three subfamilies are obvious. Pla forms a subfam-ily with PgtE of Salmonella and Epo (also called PlaA) of Erwinia; these proteins share 74% sequence identity. The pla gene is located in the Y. pestis-specific plasmid pPCP1 (Sodeinde and Goguen 1989), and epo (plaA) is also encoded on a plasmid (McGhee et al. 2002), whereas pgtE has a chromosomal location (Yu and Hong 1986). The ompT gene – encoding the prototype omptin OmpT - of Escherichia coli is located in a cryptic prophage (Grodberg et al. 1988), ompP in F-plasmid (Matsuo et al. 1999), and sopA in the large virulence plasmid of Shigella (Egile et al. 1997).

270 Suomalainen et al. The ycoA and ycoB genes are chromosomal, and YcoA and YcoB (Yersinia chromo-sal omptin) form a subfamily with 99% sequence identity (Fig. 1). Thus the Y. pestis genome contains two omptin genes, ycoA shared with Yersinia pseudotuberculosis and the Y. pestis-specific pla. The individual omptins differ in their in vitro functions (target specificity in proteolysis, adhesiveness, invasiveness) as well as in regulation of their expression. The omptin fold appears to have spread horizontally amongst Gram-negative bacteria infecting humans, animals, or plants, and the functional differences probably result from mutations and adaptation to the life style of the host bacterium.

24.2 Virulence Functions of Pla Pla is an essential virulence factor in bubonic plague, which is an extracellular zoonotic infection where the bacteria invade from the subcutaneous infection site to lymphatic tissue and multiply in the lymph nodes (Perry and Fetherston, 1997). Of the individual omptins, Pla stands out as a dramatic virulence factor. Early studies indicated that the presence of the pPCP1 plasmid and the pla gene are associated with the invasive character of plague (Beesley et al. 1967; Ferber and Brubaker 1981). Deletion of pla in pPCP1 attenuates Y. pestis by a millionfold in a subcutane-ous infection model (Sodeinde et al. 1992), and recent in vivo evidence shows that the pla gene is needed for the establishment of bubonic plague by Y. pestis (Sebbane et al. 2006a) and is one the most highly expressed genes in the bubo (Sebbane et al. 2006b).

We have expressed in E. coli most of the omptins shown in Fig. 1 and compared their in vitro functions in an attempt to determine what makes Pla such a potent viru-lence factor. We have not identified a single function (substrate specificity in prote-olysis, adhesion, or invasion) that would be shared by all omptins, it rather seems that the omptins have diverged functionally and adapted to support the life style of their host bacterium. In short, the results show that the virulence role of Pla is de-pendent on: 1) its high capacity to engage the human proteolytic Plg system to dam-age tissue barriers, 2) its efficient adhesive and invasive characteristics, 3) surface architecture of Y. pestis that allows high Pla activity in extracellular conditions, i.e. outside host cells, and 4) high constitutive transcription of the pla gene in extracellu-lar conditions.

A main virulence function of Pla is the activation of Plg, a circulating, abundant mammalian precursor of plasmin, which is a potent serine protease with numerous physiological and pathological functions (Plow et al. 1999; Myöhänen and Vaheri, 2004). A major function of plasmin is to dissolve fibrin clots, it also is important in cell migration through its capacity to digest laminin of basement membranes and to activate precursors of mammalian collagenases and gelatinases (pro-MMPs; matrix metalloproteinases) to their active forms that further damage the tissue barriers. A central role of the Plg/plasmin system in plague is supported by the finding that Plg-deficient mice show increased resistance to the disease (Goguen et al. 2000). Activa-tion of Plg results from a single cleavage of the molecule between residues Arg561 and Val562 to give the two-chain plasmin, it is noteworthy that Pla and PgtE cleave

Using Every Trick in the Book: The Pla Surface Protease of Yersinia pestis 271

Fig. 2. Cleavage (top panels) and activation or inactivation (lower panels) of the human Plg (left) and α2AP (right) by recombinant E. coli XL1 expressing pla, ompT, or pgtE cloned into the vector plasmid pSE380. In the cleavage assays, the bacteria were incubated with the target protein, which was then analyzed by Western blotting. In assays for Plg activation (bottom left), bacteria and Plg were incubated with a chromogenic plasmin substrate, whose degrada-tion was measured by spectrophotometry. In assays for α2AP inactivation (bottom right), bacteria and α2AP were first incubated for 2 h (black bars) or 5 h (grey bars), plasmin was then added, and its activity was measured after a 90-min incubation. PgtE has inactivated nearly all α2AP in 2 h, whereas inactivation by Pla remains partial after 5 h, the inactivation of the antiprotease by OmpT remains close to the control assay with pSE380. The original data has been published in Kukkonen et al. 2001 and in Lähteenmäki et al. 2005a.

the plasmin chains further to smaller, angiostatin-like fragments and that OmpT of E. coli is very poor in the cleavage as well as activation of Plg (Fig. 2). Cleavage of Plg by Pla leads to formation of plasmin activity, which is detectable with Pla more rapidly than with PgtE; however, PgtE cleaves the Plg chain slightly more rapidly (Fig 2). Our hypothesis is that PgtE has more than one initial cleavage site within the Plg molecule, the other one(s) leading to enzymatically inactive Plg fragments. This

272 Suomalainen et al. is supported by our ongoing research showing that many of the other omptins cleave the Plg molecule without forming active plasmin. It follows that one of the important virulence properties of Pla is the rapid cleavage of Plg, which precisely mimics that of mammalian Plg activators. Plasmin is a potent protease, and its formation as well as activity are tightly con-trolled in the mammalian body at various levels. Circulating plasmin is rapidly inac-tivated by circulating antiproteases, of which α2-antiplasmin (α2AP) is the main inhibitor of soluble plasmin (Lijnen and Collen 1995). α2AP binds to the plasmin molecule and reduces the half-life of soluble plasmin to 0.1s. Pla overcomes this control mechanism simply by cleaving and inactivating α2AP (Kukkonen et al. 2001), and thus creates uncontrolled plasmin proteolysis. Figure 2 shows the cleav-age and inactivation of α2AP by recombinant E. coli expressing Pla, PgtE or OmpT. Both Pla and PgtE inactivate the antiprotease, with PgtE being slightly more effi-cient. Our conclusion is that destroying the control system is more important for the intracellular pathogen Salmonella as it can rely – at least partially - on Plg activation performed by activated and migrating phagocytes (Lähteenmäki et al. 2005b). Prote-olytic targets for Pla also include complement proteins (Sodeinde et al. 1992). Whether this leads to inactivation of the complement cascade has not been shown, and the biological significance of this function remains open as cells of Y. pestis are resistant to serum killing, due to e.g. binding of the inhibitory C4bp protein onto the bacterial cell surface (Ngampasutadol et al. 2005). PgtE and OmpT degrade α-helical antimicrobial peptides (Stumpe et al. 1998; Guina et al. 2000), which could increase bacterial infectivity in the host; however, this function has not been ob-served with Y. pestis.

Pla also has nonproteolytic functions in adhesion and invasion of Y. pestis. It is an adhesin with affinity to BMs, where the main target is the major glycoprotein of BM, laminin (Lähteenmäki et al. 1998; Lobo 2006). Conflicting results have been reported for Pla binding to collagens, which could indicate that Pla actually binds to gelatins, i.e. denaturated collagens, which are present in variable amounts in the commercial collagen preparations that were used (Kienle et al. 1992; Lähteenmäki et al. 1998; Lobo 2006). Laminin is not a target for Pla proteolysis but is efficiently degraded by plasmin. In concert, Plg activation, cleavage of α2AP, and bacterial adherence to laminin creates uncontrolled proteolysis which is directed onto a sus-ceptible target in BMs (Lähteenmäki et al. 2005a). Pla is a more efficient adhesin to BM than the other omptins (Kukkonen et al. 2004). Cowan et al. (2000) reported that pPCP1-positive Y. pestis invades HeLa cells from a cervical cancer, and we observed that recombinant E. coli with Pla invaded human umbical vein endothelial cells (HUVECs) as well as the endothelial-like cell line ECV304 (Lähteenmäki et al. 2001; Kukkonen et al. 2004). The invasion is inhibited by small amounts of fetal calf serum and by normal human serum (unpublished), which indicates that it is not func-tional from the luminal side of blood vessels. At present, the biological mechanisms and significance of the Pla-mediated invasion remain open.

Using Every Trick in the Book: The Pla Surface Protease of Yersinia pestis 273

Fig. 3. The β-barrel structure of Pla modeled using the coordinates of the OmpT crystal struc-ture (Vandeputte-Rutten et al. 2001) as a template, the Swiss-Model homology modeling server (http://www.expasy.org/swissmod/SWISS-MODEL.html), and the Swiss-Pdb viewer 3.7 (SP4) programme. On the left is a side view with the outer membrane borders and the five surface loops L1-L5 indicated. On the right, a top view with the four catalytic residues indicated.

24.3 Structure-Function Relationships in Pla Resolution of the crystal structure of OmpT led to reclassification of omptins as apartate proteases (Vandeputte-Rutten et al. 2001). The structure of OmpT serves as a template to model other omptins, including Pla (Kukkonen et al. 2004). Pla is a β-barrel with 10 antiparallel β-strands connected by four short periplasmic turns and five extracellular loops L1-L5 (Fig. 3). The barrel is long, ca. (70 Å) and vase-shaped and protrudes ca. 40 Å from the lipid bilayer, with the outermost loops lo-cated just above the lipopolysaccharide (LPS) core region. The catalytic residues Asp84, Asp86, Asp206, and His208 are located in an acidic groove at the top of the barrel (Fig. 3). They are oriented inwards in the barrel and bordered by the mobile, short loops L1-L5. The catalytic residues are conserved in the omptin sequences, excepting YcoA and YcoB which have Asn84 instead of Asp84 and are classified in the MEROPS database as nonprotease members of the family, due to this sequence difference. However, there is no experimental evidence showing that YcoA/B really are nonproteolytic, neither have their adhesive or invasive properties been assessed.

Omptins cleave polypeptide targets after basic residues but differ substantially in recognition of protein substrates. Correct substrate binding is dictated by the loop structures. In general, the β-barrel is a stable membrane-embedded structure, which

274 Suomalainen et al.

Fig. 4. Isoforms of the Pla molecule. Western blotting of recombinant E. coli cell wall prepa-rations obtained by sonication reveals four forms of the Pla molecule. The pre-Pla is the un-processed form and α-Pla is the mature form of the protein. β-Pla is formed by autoprocessing at Lys262 in L5, and formation of γ-Pla is dependent on the presence of the lipid A–binding motif formed by Arg138 and R171. The original data was published in Kukkonen et al. 2001, 2004.

tolerates large deletions or insertions in surface loops (Schulz 2000), and this seems to provide the basis for the variable proteolysis specificities of the omptins. We have been able to change the substrate specificity of OmpT and to turn it into an efficient Plg activator and α2AP inactivator by cumulative substitutions of OmpT surface residues with the corresponding residues of Pla (Kukkonen et al. 2001). Reverse substitutions rendered Pla incapable in both functions. Another important feature in Pla functions is the dependency on LPS. Omptin sequences contain a consensus motif for protein binding to lipid A (Vandeputte-Rutten 2001; Kukkonen et al. 2004). This motif occurs in several prokaryotic and eukaryotic proteins that bind LPS and is formed by basic amino acids – Arg138 and Arg171 in Pla (Fig. 4) - that bind to phosphates in lipid A (Ferguson et al. 2000). Purified, detergent-solubilized Pla requires addition of LPS to be active (Kukkonen et al. 2004). On the other hand, Pla and PgtE are sterically inhibited by smooth LPS with a long O chain; this was seen both with recombinant bacteria expressing Pla or PgtE as well as by successful reactivation of purified His6-Pla with rough but not by smooth LPS. Y. pestis is genetically rough and lacks smooth LPS (Skurnik et al. 2000), and our conclusion is that full activity of Pla is the selective advantage for loss of the O antigen in Y. pestis (Kukkonen et al. 2004). Isolates of Salmonella enterica invariably express smooth LPS, and S. enterica overcomes this problem by shortening the LPS to a rough type inside macrophages, where PgtE is fully active (Lähteenmäki et al. 2005b). Thus these two bacterial species obviously are able to

Using Every Trick in the Book: The Pla Surface Protease of Yersinia pestis 275

utilize their omptins in different environments, Y. pestis in extracellular conditions but S. enterica inside or immediately after being released from macrophages. The substitution of Arg138 and Arg171 renders PgtE and Pla enzymatically inactive and abolishes formation of the β- and the γ-forms, but at present it is not known exactly how this is associated with enzymatic activity of Pla/PgtE and whether these protein forms are formed within the outer membrane. Formation of β-Pla is prevented by catalytic-residue substitutions as well as by substitution of the residue Lys262 at L5 (Fig. 4), indicating that it is formed by auto-processing (Kukkonen et al. 2001). Change of other lysines at L5 does not prevent the autoprocessing, which indicates that Lys262 is the cleavage site. Autoprocessing is also prevented by certain substitutions near the catalytic site, these residues proba-bly function in self recognition by Pla. Different substrate specificities of the omptins are also reflected in the fact that the autoprocessing site in Pla is different from those in PgtE and OmpT (Kramer et al. 2000; Kukkonen et al. 2004). It was earlier as-sumed that formation of β-Pla and γ-Pla represent activation processes similar to those in mammalian Plg activators; however, prevention of β-Pla formation does not affect Plg activation by Pla (Kukkonen et al. 2001), and the biological significance of the autoprocessing remains open. 24.4 Conclusions The Pla surface protease/adhesin is an essential and multifunctional virulence factor in the plague disease process. Its high virulence potential results in part from its versatile character in generating uncontrolled and targeted proteolysis as well as being an invasin. Important also is the inactivation of the O antigen genes in Y. pes-tis, which exemplifies how a loss of one virulence function (the synthesis of the O antigen) is compensated by the subsequent high activity of another virulence func-tion (Pla). Y. pestis and S. enterica also demonstrate how a coordinated control and modification of the cell wall and surface proteolysis is obtained by two pathogens having very different life styles and pathogenic mechanisms.

The omptin family has most likely resulted from horizontal gene transfer and subsequent genetic adaptation to gain new or modified functions. Our on-going work has shown that the gain of novel functions in omptins can result from a few amino acid substitutions at critical protein regions. The omptin barrel with its five mobile loops seems an adjustable template to gain novel functions. The nearest ortholog of Pla is Epo of the plant pathogen Erwinia. The gene encoding epo (plaA) is located in a 36-kb mosaic plasmid, which contains homologs to transposases and integrases as well as to genes encoding surface proteins of several bacterial species (McGhee et al. 2002). In functions and regulation, Pla, PgtE and Epo show similarities as well as differences, and our hypothesis is that they have diverged from a common omptin ancestor to serve the different lifestyles of the host bacteria. Pla has clearly been demonstrated to be an essential virulence factor in bubonic plague. So far these studies have used deletion mutants only, and a challenging task will be to create Pla derivatives impaired in one function only so that the relative virulence roles of specific Pla functions can be addressed.

276 Suomalainen et al. 24.5 Acknowledgements We thank the Academy of Finland (grant numbers 105824, 211300, 80666, 201967, the Microbes and Man Research Programme, Network of Excellence in Europatho-genomics, project number 110716), and the University of Helsinki for financial support. 24.6 References Beesley, E.D., Brubaker, R.R., Janssen, W.A. and Surgalla, M.J. (1967) Pesticins. III. Expres-

sion of coagulase and mechanisms of fibrinolysis. J. Bacteriol. 94, 19-26. Cowan, C., Jones, H.A., Kaya, Y.H., Perry, R.D. and Straley, S.C. (2000) Invasion of epithe-

lial cells by Yersinia pestis: evidence for a Y. pestis-specific invasion. Infect. Immun. 68, 4523-4530.

Egile, C., d´Hauteville, H., Parsot, C., and Sansonetti, P.J. (1997) SopA, the outer membrane protease responsible for polar localization of IcsA in Shigella flexneri. Mol. Microbiol. 23, 1063-1073.

Ferber, D.M. and Brubaker, R.R. (1981) Plasmids in Yersinia pestis. Infect. Immun. 31, 839-841.

Ferguson, A.D., Welte, W., Hofmann, E., Lindner, B., Holst, O., Coulton, J.W. and Diedrerichs, K. (2000) A conserved structural motif for lipopolysaccharide recognition by prokaryotic and eukaryotic proteins. Struct. Fold. Des. 8, 585-592.

Goguen, J.D., Bugge, T. and Degen, J.L. (2000) Role of the pleiotropic effects of plasminogen deficiency in infection experiments with plasminogen-deficient mice. Methods 21, 179-183.

Grodberg, J., Lundrigan, M.D., Toledo, D.L., Mangel, W.F. and Dunn, J.J. (1988) Complete nucleotide sequence and deduced amino acid sequence of the ompT gene of E. coli K-12. Nucleic Acids Res. 16, 1209.

Guina, T., Yi, E.C., Wang, H., Hackett, M. and Miller, S.I. (2000) A PhoP-regulated outer membrane protease of Salmonella enterica serovar Typhimurium promotes resistance to alpha-helical antimicrobial peptides. J. Bacteriol. 182, 4077-4086.

Kienle, Z., Emödy, L., Svanborg, C. and O´Toole, P.W. (1992) Adhesive properties conferred by the plasminogen activator of Yersinia pestis. J. Gen. Microbiol. 138, 1679-1687.

Kramer, R.A., Zandwijken, D., Egmond, M.R. and Dekker, N. (2001) In vitro folding, purifi-cation and characterization of E. coli outer membrane protease OmpT. Eur. J. Biochem. 267, 885-893.

Kukkonen, M., Lähteenmäki, K., Suomalainen, M., Kalkkinen, N., Emödy, L., Lång, H. and Korhonen T.K. (2001) Protein regions important for plasminogen activation and inactiva-tion of α2-antiplasmin in the surface protease Pla of Yersinia pestis. Mol. Microbiol. 40, 1097-1111.

Kukkonen, M., Suomalainen, M., Kyllönen, P., Lähteenmäki, K., Lång, H., Virkola, R., Helander, I.M., Holst, O. and Korhonen, T.K. (2004) Lack of O-antigen is essential for plasminogen activation by Yersinia pestis and Salmonella enterica. Mol. Microbiol. 51, 215-225.

Lähteenmäki, K., Kukkonen, M. and Korhonen, T.K. (2001) The Pla surface protease/adhesin of Yersinia pestis mediates bacterial invasion into human endothelial cells. FEBS Lett. 504, 69-72.

Using Every Trick in the Book: The Pla Surface Protease of Yersinia pestis 277

Lähteenmäki, K., Virkola, R., Sarén, A., Emödy, L. and Korhonen, T.K. (1998) Expression of plasminogen activator Pla of Yersinia pestis enhances bacterial attachment to the mammal-ian extracellular matrix. Infect. Immun. 66, 5755-5762.

Lähteenmäki, K., Edelman, S. and Korhonen, T.K. (2005a) Bacterial metastasis: the host plasminogen system in bacterial invasion. Trends Microbiol. 13, 79-85.

Lähteenmäki, K., Kyllönen, P., Partanen, L. and Korhonen, T.K. (2005b) Antiprotease inacti-vation by Salmonella enterica released from infected macrophages. Cell. Microbiol. 7, 529-538.

Lobo, L.A. (2006) Adhesive properties of the purified plasminogen activator Pla of Yersinia pestis. FEMS Microbiol. Lett. 262, 158-162.

Lijnen, H.R. and Collen, D. (1995) Mechanisms of physiological fibrinolysis. Bailliere´s Clinical Haematology 8, 277-290.

Matsuo, E., Sampei, G., Mizobuchi, K. and Ito, K. (1999) The plasmid F OmpP protease, a homolue of OmpT, as a potential obstacle to E. coli-based protein production. FEBS Lett. 461, 6-8.

McGhee, G.C., Schnabel, E.L., Maxson-Stein, K., Jones, B., Stromberg, V.K., Lacy, G.H. and Jones A.L. (2002) Relatedness of chromosomal and plasmid DNAs of Erwinia pyrifoliae and Erwinia amylovora. Appl. Environ. Microbiol. 68, 6182-6192.

associated plasminogen. Cell. Mol. Life Sci. 61, 2840-2858. Ngampasutadol, J., Ram, S., Blom, A.M., Jarva, H., Jerse, A.E., Lien, E., Goguen, J., Gulati,

S. and Rice, P.A. (2005) Human C4b-binding protein selectively interacts with Neisseria gonorrhoeae and results in species-specific infection. Proc. Natl. Acad. Sci. USA 102, 17142-17147.

Perry, R.D. and Fetherston, J.D. (1997) Yersinia pestis – etiological agent of plague. Clin. Microbiol. Rev. 10, 35-66.

Plow, E.F., Ploplis, V.A. , Carmeliet, P. and Collen D. (1999) Plasminogen and cell migration in vivo. Fibrinol. Proteol. 13, 49-53.

Schulz, G.E. (2000) β-barrel membrane protens. Curr. Opin. Struct. Biol. 10, 443-462. Sebbane, F., Jarrett, C.O., Gardner, D., Long, D. and Hinnebusch, B.J. (2006a). Role of the

Yersinia pestis plasminogen activator in the incidence of distinct septicemic and bubonic forms of flea-borne plague. Proc. Natl. Acad. Sci. USA 103, 5526-5530.

Sebbane, F., Lemaite, N., Sturdevant, D.E., Rebeil, R., Virtaneva, K., Porcella, S.F. and Hinnebusch, B.J. (2006b) Adaptive response of Yersinia pestis to extracellular effectors of innate immunity during bubonic plague. Proc. Natl. Acad. Sci. USA 103, 11766-11771.

Skurnik, M., Peippo, A. and Ervelä, E. (2000) Characterization of the O-antigen gene clusters of Yersinia pseudotuberculosis and the cryptic O-antigen cluster of Yersinia pestis shows that the plague bacillus is most closely related to and has evolved from Y. pseudotubercu-losis serotype O1b. Mol. Microbiol. 37, 316-330.

Sodeinde, O.A. and Goguen J.D. (1989) Nucleotide sequence of the plasminogen activator gene of Yersinia pestis: relationship to ompT of E. coli and gene E of Salmonella typhi-

Sodeinde, O.A., Subrahmanyam, Y.V.B.K., Stark, K., Quan, T., Bao, Y. and Goguen, J.D. (1992) A surface protease and the invasive character of plague. Science 258, 1004-1007.

Stumpe, S., Schmid, R., Stephens, D.L., Georgiu, G. and Bakker, E.P. (1998) Identification of OmpT as the protease that hydrolyzes the antimicrobial peptide protamine before it enters growing cells of E. coli. J. Bacteriol. 180,4002-4006.

Vandeputte-Rutten, L., Kramer, R.A., Kroon, J., Dekker, N., Egmond, M.R. and Gros, P. (2001) Crystal structure of the outer membrane protease OmpT from E. coli suggests a novel catalytic site. EMBO J. 20, 5033-5039.

Myöhänen, H. and Vaheri, A. (2004) Regulation and interactions in the activation of cell-

murium. Infect. Immun. 57, 1517-1523.

278 Suomalainen et al. Yu, C.Q. and Hong, J.S. (1986) Identification and nucleotide sequence of the activator gene of

the externally induced phosphoglycerate transport system of Salmonella typhimurium. Gene 45, 51-57.

Picture 23. Attendees at the banquet. Photograph by R. Perry.