Portal de Periódicos da CAPES

Autoprocessing of the Escherichia coli AIDA-I Autotransporter

2009; Elsevier BV; Volume: 284; Issue: 25 Linguagem: Inglês

10.1074/jbc.m109.010108

1083-351X

Marie‐Ève Charbonneau, Julie Janvore, Michaël Mourez,

Escherichia coli research studies

The cleavage of the autotransporter adhesin involved in diffuse adherence (AIDA-I) of Escherichia coli yields a membrane-embedded fragment, AIDAc, and an extracellular fragment, the mature AIDA-I adhesin. The latter remains noncovalently associated with AIDAc but can be released by heat treatment. In this study we determined the mechanism of AIDA-I cleavage. We showed that AIDA-I processing is an autocatalytic event by monitoring the in vitro cleavage of an uncleaved mutant protein isolated from inclusion bodies. Furthermore, by following changes in circular dichroism spectra and protease resistance of the renaturated protein, we showed that the cleavage of the protein is correlated with folding. With site-directed deletions, we showed that the catalytic activity of the protein lies in a region encompassing amino acids between Ala-667 and Thr-953, which includes the conserved junction domain of some autotransporters. With site-directed point mutations, we also found that Asp-878 and Glu-897 are involved in the processing of AIDA-I and that a mutation preserving the acidic side chain of Asp-878 was tolerated, giving evidence that this carboxylic acid group is directly involved in catalysis. Last, we confirmed that cleavage of AIDA-I is intramolecular. Our results unveil a new mechanism of auto-processing in the autotransporter family. The cleavage of the autotransporter adhesin involved in diffuse adherence (AIDA-I) of Escherichia coli yields a membrane-embedded fragment, AIDAc, and an extracellular fragment, the mature AIDA-I adhesin. The latter remains noncovalently associated with AIDAc but can be released by heat treatment. In this study we determined the mechanism of AIDA-I cleavage. We showed that AIDA-I processing is an autocatalytic event by monitoring the in vitro cleavage of an uncleaved mutant protein isolated from inclusion bodies. Furthermore, by following changes in circular dichroism spectra and protease resistance of the renaturated protein, we showed that the cleavage of the protein is correlated with folding. With site-directed deletions, we showed that the catalytic activity of the protein lies in a region encompassing amino acids between Ala-667 and Thr-953, which includes the conserved junction domain of some autotransporters. With site-directed point mutations, we also found that Asp-878 and Glu-897 are involved in the processing of AIDA-I and that a mutation preserving the acidic side chain of Asp-878 was tolerated, giving evidence that this carboxylic acid group is directly involved in catalysis. Last, we confirmed that cleavage of AIDA-I is intramolecular. Our results unveil a new mechanism of auto-processing in the autotransporter family. Monomeric autotransporters, secreted by the type Va secretion pathway, constitute one of the largest family of secreted proteins in Gram-negative bacteria (1Henderson I.R. Navarro-Garcia F. Desvaux M. Fernandez R.C. Ala'Aldeen D. Microbiol. Mol. Biol. Rev. 2004; 68: 692-744Crossref PubMed Scopus (637) Google Scholar). Various virulence attributes have been associated with most autotransporters, such as adhesion or invasion, self-association, biofilm formation, serum resistance, and cytotoxic activity to name just a few (2Henderson I.R. Nataro J.P. Infect. Immun. 2001; 69: 1231-1243Crossref PubMed Scopus (346) Google Scholar, 3Girard V. Mourez M. Res. Microbiol. 2006; 157: 407-416Crossref PubMed Scopus (39) Google Scholar). Autotransporters are synthesized as pre-proproteins with modular organizations. An N-terminal sec-dependent signal sequence permits the secretion of the protein across the inner membrane (4Peterson J.H. Szabady R.L. Bernstein H.D. J. Biol. Chem. 2006; 281: 9038-9048Abstract Full Text Full Text PDF PubMed Scopus (55) Google Scholar). A C-terminal membrane-embedded domain promotes secretion of parts of the protein across the outer membrane and is composed of a β-barrel and a α-helix spanning its lumen (5Oomen C.J. van Ulsen P. van Gelder P. Feijen M. Tommassen J. Gros P. EMBO J. 2004; 23: 1257-1266Crossref PubMed Scopus (303) Google Scholar, 6Barnard T.J. Dautin N. Lukacik P. Bernstein H.D. Buchanan S.K. Nat. Struct. Mol. Biol. 2007; 14: 1214-1220Crossref PubMed Scopus (137) Google Scholar). The central domain of the protein is, thus, extracellular and bears its functional part. In some cases the extracellular part of the protein is cleaved and remains associated with the outer membrane or is secreted in the extracellular milieu. Directly preceding the membrane-embedded domain, a functional subdomain of ∼100 amino acids is sometimes present in the extracellular domain and has been called the junction region (also named autochaperone domain or stable core) (7Oliver D.C. Huang G. Nodel E. Pleasance S. Fernandez R.C. Mol. Microbiol. 2003; 47: 1367-1383Crossref PubMed Scopus (126) Google Scholar, 8Renn J.P. Clark P.L. Biopolymers. 2008; 89: 420-427Crossref PubMed Scopus (49) Google Scholar, 9Mogensen J.E. Tapadar D. Schmidt M.A. Otzen D.E. Biochemistry. 2005; 44: 4533-4545Crossref PubMed Scopus (27) Google Scholar). This region is essential for stabilizing the β-barrel and/or to promote folding of the extracellular domain (7Oliver D.C. Huang G. Nodel E. Pleasance S. Fernandez R.C. Mol. Microbiol. 2003; 47: 1367-1383Crossref PubMed Scopus (126) Google Scholar, 8Renn J.P. Clark P.L. Biopolymers. 2008; 89: 420-427Crossref PubMed Scopus (49) Google Scholar, 9Mogensen J.E. Tapadar D. Schmidt M.A. Otzen D.E. Biochemistry. 2005; 44: 4533-4545Crossref PubMed Scopus (27) Google Scholar).The adhesin involved in diffuse adherence (AIDA-I) 3The abbreviations used are: AIDA-Iadhesin involved in diffuse adherenceAahautotransporter adhesin heptosyltransferaseIPTGisopropyl-β-thiogalactopyranosideSPATEserine protease autotransporters of EnterobacteriaceaeTBSTris-buffered salineTricineN-[2-hydroxy-1,1-bis(hydroxymethyl)ethyl]glycineWTwild type. 3The abbreviations used are: AIDA-Iadhesin involved in diffuse adherenceAahautotransporter adhesin heptosyltransferaseIPTGisopropyl-β-thiogalactopyranosideSPATEserine protease autotransporters of EnterobacteriaceaeTBSTris-buffered salineTricineN-[2-hydroxy-1,1-bis(hydroxymethyl)ethyl]glycineWTwild type. is a monomeric autotransporter that has been extensively studied. AIDA-I was originally identified as a plasmid-encoded protein from a diffusely adhering Escherichia coli strain isolated in a case of infantile diarrhea (10Benz I. Schmidt M.A. Infect. Immun. 1989; 57: 1506-1511Crossref PubMed Google Scholar). This adhesin was then shown to play a role in neonatal and postweaning diarrheal diseases in piglets, both of which cause major economic losses in farms worldwide (11Niewerth U. Frey A. Voss T. Le Bouguénec C. Baljer G. Franke S. Schmidt M.A. Clin. Diagn. Lab. Immunol. 2001; 8: 143-149Crossref PubMed Scopus (78) Google Scholar, 12Chapman T.A. Wu X.Y. Barchia I. Bettelheim K.A. Driesen S. Trott D. Wilson M. Chin J.J. Appl. Environ. Microbiol. 2006; 72: 4782-4795Crossref PubMed Scopus (191) Google Scholar, 13Zhang W. Zhao M. Ruesch L. Omot A. Francis D. Vet. Microbiol. 2007; 123: 145-152Crossref PubMed Scopus (184) Google Scholar). Besides its role as an adhesin, AIDA-I has been shown to mediate self-association and biofilm formation (14Sherlock O. Schembri M.A. Reisner A. Klemm P. J. Bacteriol. 2004; 186: 8058-8065Crossref PubMed Scopus (139) Google Scholar) as well as invasion of epithelial cells (15Charbonneau M.E. Berthiaume F. Mourez M. J. Bacteriol. 2006; 188: 8504-8512Crossref PubMed Scopus (34) Google Scholar). Additionally, this protein undergoes a modification that is rare in bacteria, as it is O-glycosylated by the specific cytoplasmic protein autotransporter adhesin heptosyltransferase (Aah) (16Benz I. Schmidt M.A. Mol. Microbiol. 2001; 40: 1403-1413Crossref PubMed Scopus (146) Google Scholar, 17Charbonneau M.E. Girard V. Nikolakakis A. Campos M. Berthiaume F. Dumas F. Lépine F. Mourez M. J. Bacteriol. 2007; 189: 8880-8889Crossref PubMed Scopus (58) Google Scholar). AIDA-I has been suggested to be a member of a new group of autotransporter called self-associating autotransporters, which includes the Ag43 aggregation factor and the TibA invasin (18Klemm P. Vejborg R.M. Sherlock O. Int. J. Med. Microbiol. 2006; 296: 187-195Crossref PubMed Scopus (78) Google Scholar).After secretion at the cell surface, the extracellular domain of the protein, called mature AIDA-I, is cleaved from the C-terminal domain, called AIDAc, which encompasses the β-barrel, the α-helix, and the junction region (19Suhr M. Benz I. Schmidt M.A. Mol. Microbiol. 1996; 22: 31-42Crossref PubMed Scopus (72) Google Scholar, 20Benz I. Schmidt M.A. Mol. Microbiol. 1992; 6: 1539-1546Crossref PubMed Scopus (160) Google Scholar). Mature AIDA-I, however, remains strongly associated with AIDAc (15Charbonneau M.E. Berthiaume F. Mourez M. J. Bacteriol. 2006; 188: 8504-8512Crossref PubMed Scopus (34) Google Scholar). The processing of AIDA-I is neither essential for the multiple functions of the protein nor for its stability (15Charbonneau M.E. Berthiaume F. Mourez M. J. Bacteriol. 2006; 188: 8504-8512Crossref PubMed Scopus (34) Google Scholar). Questions remain about the mechanism involved in this cleavage. In a few autotransporters, the cleavage reaction involves an exogenous protease. IcsA for instance, an autotransporter of Shigella flexneri, is cleaved by a dedicated outer membrane protease called IcsP, which is related to the OmpT protease found in E. coli (21Shere K.D. Sallustio S. Manessis A. D'Aversa T.G. Goldberg M.B. Mol. Microbiol. 1997; 25: 451-462Crossref PubMed Scopus (101) Google Scholar). Similarly, the Neisseria meningitidis serine protease NalP is specifically responsible for the partial processing of various autotransporters, including the IgA protease (22van Ulsen P. van Alphen L. ten Hove J. Fransen F. van der Ley P. Tommassen J. Mol. Microbiol. 2003; 50: 1017-1030Crossref PubMed Scopus (111) Google Scholar), App (22van Ulsen P. van Alphen L. ten Hove J. Fransen F. van der Ley P. Tommassen J. Mol. Microbiol. 2003; 50: 1017-1030Crossref PubMed Scopus (111) Google Scholar), AusI (23van Ulsen P. Adler B. Fassler P. Gilbert M. van Schilfgaarde M. van der Ley P. van Alphen L. Tommassen J. Microbes Infect. 2006; 8: 2088-2097Crossref PubMed Scopus (32) Google Scholar), and MspA (24Turner D.P. Marietou A.G. Johnston L. Ho K.K. Rogers A.J. Wooldridge K.G. Ala'Aldeen D.A. Infect. Immun. 2006; 74: 2957-2964Crossref PubMed Scopus (66) Google Scholar). In contrast, the cleavage of AIDA-I is not mediated by the known E. coli periplasmic or outer membrane proteases (DegP, OmpT, or OmpP), and it occurs in E. coli as well as in Shigella or Salmonella strains (19Suhr M. Benz I. Schmidt M.A. Mol. Microbiol. 1996; 22: 31-42Crossref PubMed Scopus (72) Google Scholar). Based on those observations, this processing has been suggested to be an autocatalytic event, even though no protease catalytic site can be identified in the protein.Two types of autocatalytic processing mechanisms have been described in other autotransporters. The Hap autotransporter of Haemophilus influenzae (25Fink D.L. Cope L.D. Hansen E.J. Geme 3rd, J.W. J. Biol. Chem. 2001; 276: 39492-39500Abstract Full Text Full Text PDF PubMed Scopus (55) Google Scholar), the SphB1 protein of Bordetella pertussis (26Coutte L. Willery E. Antoine R. Drobecq H. Locht C. Jacob-Dubuisson F. Mol. Microbiol. 2003; 49: 529-539Crossref PubMed Scopus (58) Google Scholar), and the NalP (22van Ulsen P. van Alphen L. ten Hove J. Fransen F. van der Ley P. Tommassen J. Mol. Microbiol. 2003; 50: 1017-1030Crossref PubMed Scopus (111) Google Scholar), App (27Serruto D. Adu-Bobie J. Scarselli M. Veggi D. Pizza M. Rappuoli R. Aricó B. Mol. Microbiol. 2003; 48: 323-334Crossref PubMed Scopus (84) Google Scholar), and IgA proteases of N. meningitidis (28Pohlner J. Halter R. Beyreuther K. Meyer T.F. Nature. 1987; 325: 458-462Crossref PubMed Scopus (476) Google Scholar) cleave themselves by an endogenous serine protease domain. More recently, it was shown that the serine protease autotransporters of Enterobacteriaceae (SPATE) family cleave themselves in the α-helix located in the lumen of the membrane-embedded β-barrel by a catalytic reaction that involves an aspartate and a conserved asparagine (6Barnard T.J. Dautin N. Lukacik P. Bernstein H.D. Buchanan S.K. Nat. Struct. Mol. Biol. 2007; 14: 1214-1220Crossref PubMed Scopus (137) Google Scholar, 29Dautin N. Barnard T.J. Anderson D.E. Bernstein H.D. EMBO. J. 2007; 26: 1942-1952Crossref PubMed Scopus (97) Google Scholar). The same autocatalytic reaction has been shown for two distantly related B. pertussis autotransporters, pertactin and BrkA (29Dautin N. Barnard T.J. Anderson D.E. Bernstein H.D. EMBO. J. 2007; 26: 1942-1952Crossref PubMed Scopus (97) Google Scholar). However, none of these mechanisms can be hypothesized for AIDA-I cleavage; AIDA-I possesses neither a serine protease domain like Hap nor the asparagine residue required for the SPATE cleavage. In addition, the AIDA-I cleavage site is not located in the α-helix. Thus, the mechanism of AIDA-I processing must be different and remains unknown.The mechanisms by which a diversity of other autotransporters are cleaved remain unknown. This is for instance the case for the E. coli Ag43 self-associating autotransporters (30Henderson I.R. Owen P. J. Bacteriol. 1999; 181: 2132-2141Crossref PubMed Google Scholar), the Helicobacter pylori VacA cytotoxin (31Nguyen V.Q. Caprioli R.M. Cover T.L. Infect. Immun. 2001; 69: 543-546Crossref PubMed Scopus (68) Google Scholar), the Chlamydia pneumoniae polymorphic membrane proteins (32Vandahl B.B. Pedersen A.S. Gevaert K. Holm A. Vandekerckhove J. Christiansen G. Birkelund S. BMC Microbiol. 2002; 2: 36Crossref PubMed Scopus (36) Google Scholar), and the Ssp proteins of Serratia marcescens (33Ohnishi Y. Beppu T. Horinouchi S. J. Biochem. 1997; 121: 902-913Crossref PubMed Scopus (21) Google Scholar) as well as the cohemolysin Cfa (34Litwin C.M. Johnson J.M. Infect. Immun. 2005; 73: 4205-4213Crossref PubMed Scopus (19) Google Scholar) and the Arp immunogenic protein of Bartonella henselae (35Litwin C.M. Rawlins M.L. Swenson E.M. Infect. Immun. 2007; 75: 5255-5263Crossref PubMed Scopus (13) Google Scholar).In this study we determined the mechanism of AIDA-I cleavage. We first showed that a truncated mutant of AIDA-I can refold in vitro and undergo self-cleavage, providing definite proof for the autocatalytic mechanism. By constructing site-directed deletions and point mutations, we then showed that the catalytic domain lies in the region encompassing the amino acids between Ala-667 and Thr-953, and we identified two residues in the junction region, Asp-878 and Glu-897, that are essential for processing. Last, we confirmed that the mechanism of proteolysis for AIDA-I is intramolecular. This is the first report of this type of autocatalytic activity.DISCUSSIONIn this study we proved the autocatalytic nature of the cleavage of the AIDA-I autotransporter by following the in vitro refolding of a mutant lacking its C-terminal α-helix and β-barrel. We noted that in vitro cleavage was less efficient that processing in vivo. This could be because of incomplete folding. Indeed, we observed that folding and processing occur at similar rates, which strongly suggests that the two events are closely related. This observation is also evidence against the cleavage, resulting from a contaminant protease, as the folding of AIDA-I is strikingly slow, and an exogenous protease would most likely cleave AIDA-I faster. Furthermore, we observed that the in vitro refolding of AIDA-I is dramatically pH-dependent and without folding (pH 10), or with an aberrant folding (pH 4) the protein remains uncleaved, showing more evidence of a strong association between folding and cleavage.We identified by deletion analysis a 286-amino acid region directly involved in the processing of AIDA-I. This region includes the junction domain, which ends just before the cleavage site and the C-terminal part of the AIDA-I functional domain. We chose to mutate acidic residues of this region because of the observation that in vitro refolding and processing of AIDA-I occurred even in the presence of serine and cysteine protease inhibitors. Some acidic proteases, like the omptin family, rely on catalytic aspartate residues that cannot be identified by classical consensus motifs (42Hritonenko V. Stathopoulos C. Mol. Memb. Biol. 2007; 24: 395-406Crossref PubMed Scopus (61) Google Scholar). We found that Asp-878 and Glu-897 in the junction region are involved in the processing of AIDA-I. The complete elimination of cleavage when these residues are mutated strongly suggests that they play a catalytic role. Mutations that preserve the acidic side chain were tolerated, giving more evidence that the carboxylic acid groups are directly involved in catalysis. The requirement of Asp-878 for catalytic activity was confirmed by in vitro refolding experiments with the D878N mutant, showing that it can refold but remains uncleaved. We observed that the uncleaved mutant presents a normal global conformation, as this mutant is as functional and as stable as the wild-type AIDA-I. Last, we confirmed that the cleavage of AIDA-I relies on an intramolecular mechanism of processing.As shown in Fig. 9, our results are consistent with a three-dimensional structural model of the junction region and the AIDA-I cleavage site (the amino acids between Ile-807 and Leu-956), which we generated using the known crystal structure of B. pertussis pertactin (39Emsley P. Charles I.G. Fairweather N.F. Isaacs N.W. Nature. 1996; 381: 90-92Crossref PubMed Scopus (260) Google Scholar). This model predicts that Asp-878 and Glu-897 are located close to the cleaved peptidic bond, which is found in a loop. This gives an explanation as to why folding and cleavage occur at similar rates; the junction domain of AIDA-I and the region encompassing the cleavage site have to be properly folded for the correct positioning of the catalytic residues in relation to the cleaved peptidic bond. The model also explains why mutations such as E845A or N849A lead to a decrease in the processing efficiency, as these changes would likely destabilize the loop bearing the bond that has to be cleaved.The mechanism of autocatalytic cleavage we describe here for AIDA-I is unique among the monomeric autotransporter family. Alignment of the junction region sequences of several autotransporters highlights that Asp-878 of AIDA-I is conserved among autotransporters whether they are processed or not (Fig. 9). However, Glu-897 is unique to AIDA-I among autocatalytic-cleaved autotransporters, which could explain why this mechanism of proteolysis is unique to AIDA-I. Our model also differs from the two previously described intramolecular mechanisms of cleavage for autotransporters; that is, the processing mediated either by endogenous serine protease domains or SPATE and pertactin-like autotransporters.The cleavage mechanism for several other autotransporters, including the Ag43 aggregation factor, remains unknown. As Ag43 and AIDA-I are closely related in sequence and possess similar functional features (18Klemm P. Vejborg R.M. Sherlock O. Int. J. Med. Microbiol. 2006; 296: 187-195Crossref PubMed Scopus (78) Google Scholar), it is tempting to hypothesize that a similar mechanism of processing involving carboxylic amino acids exists for Ag43. However, Ag43 is cleaved at a different site compared with the one of AIDA-I (43Klemm P. Hjerrild L. Gjermansen M. Schembri M.A. Mol. Microbiol. 2004; 51: 283-296Crossref PubMed Scopus (115) Google Scholar) and in a region that has only poor homologies with known autotransporter structures.The question of the function of AIDA-I cleavage remains. In autotransporters that are cleaved the processing can play different roles, such as the release of cytotoxic domains in the extracellular milieu (44Navarro-García F. Canizalez-Roman A. Luna J. Sears C. Nataro J.P. Infect. Immun. 2001; 69: 1053-1060Crossref PubMed Scopus (58) Google Scholar), keeping a protein localized at a pole (45d'Hauteville H. Dufourcq Lagelouse R. Nato F. Sansonetti P.J. Infect. Immun. 1996; 64: 511-517Crossref PubMed Google Scholar), or controlling the expression level of an adhesin (46Fink D.L. St. Geme 3rd, J.W. J. Bacteriol. 2003; 185: 1608-1615Crossref PubMed Scopus (32) Google Scholar). The intramolecular nature of AIDA-I processing excludes the possibility that it serves to regulate the protein level at the surface of the bacteria. In addition, we did not observe that cleavage is important for function, as uncleaved mutants resulting from mutations of the catalytic residues or mutations of the residues of the cleaved peptidic bond are as functional as the wild-type protein. This puzzling observation is at least consistent with the fact that we could not find any conditions that lead to the release of mature AIDA-I in the extracellular milieu (15Charbonneau M.E. Berthiaume F. Mourez M. J. Bacteriol. 2006; 188: 8504-8512Crossref PubMed Scopus (34) Google Scholar). We can hypothesize that in the context of an animal infection, a specific condition may cause the release of mature AIDA-I, which would be beneficial for the bacteria, for example to escape the immune response. Alternatively, cleavage could generate a new unique surface representing an interaction site with other unknown proteins or host factors, which could play a role during infection. This mechanism was recently observed with the EscU protein, a structural component of the inner membrane ring of the type-three secretion system in E. coli (47Zarivach R. Deng W. Vuckovic M. Felise H.B. Nguyen H.V. Miller S.I. Finlay B.B. Strynadka N.C. Nature. 2008; 453: 124-127Crossref PubMed Scopus (111) Google Scholar). This last hypothesis would be consistent with the small differences we observed between CD spectra for the D878N mutant and the WT protein in their native conformations or after in vitro folding.Self-processing of autotransporters is a frequent feature and occurs by a variety of different mechanisms, with our study providing yet another one. This diversity is quite surprising and strongly suggests that this processing plays an important role. Further investigation is, therefore, required to establish what this role might be in the case of AIDA-I. Monomeric autotransporters, secreted by the type Va secretion pathway, constitute one of the largest family of secreted proteins in Gram-negative bacteria (1Henderson I.R. Navarro-Garcia F. Desvaux M. Fernandez R.C. Ala'Aldeen D. Microbiol. Mol. Biol. Rev. 2004; 68: 692-744Crossref PubMed Scopus (637) Google Scholar). Various virulence attributes have been associated with most autotransporters, such as adhesion or invasion, self-association, biofilm formation, serum resistance, and cytotoxic activity to name just a few (2Henderson I.R. Nataro J.P. Infect. Immun. 2001; 69: 1231-1243Crossref PubMed Scopus (346) Google Scholar, 3Girard V. Mourez M. Res. Microbiol. 2006; 157: 407-416Crossref PubMed Scopus (39) Google Scholar). Autotransporters are synthesized as pre-proproteins with modular organizations. An N-terminal sec-dependent signal sequence permits the secretion of the protein across the inner membrane (4Peterson J.H. Szabady R.L. Bernstein H.D. J. Biol. Chem. 2006; 281: 9038-9048Abstract Full Text Full Text PDF PubMed Scopus (55) Google Scholar). A C-terminal membrane-embedded domain promotes secretion of parts of the protein across the outer membrane and is composed of a β-barrel and a α-helix spanning its lumen (5Oomen C.J. van Ulsen P. van Gelder P. Feijen M. Tommassen J. Gros P. EMBO J. 2004; 23: 1257-1266Crossref PubMed Scopus (303) Google Scholar, 6Barnard T.J. Dautin N. Lukacik P. Bernstein H.D. Buchanan S.K. Nat. Struct. Mol. Biol. 2007; 14: 1214-1220Crossref PubMed Scopus (137) Google Scholar). The central domain of the protein is, thus, extracellular and bears its functional part. In some cases the extracellular part of the protein is cleaved and remains associated with the outer membrane or is secreted in the extracellular milieu. Directly preceding the membrane-embedded domain, a functional subdomain of ∼100 amino acids is sometimes present in the extracellular domain and has been called the junction region (also named autochaperone domain or stable core) (7Oliver D.C. Huang G. Nodel E. Pleasance S. Fernandez R.C. Mol. Microbiol. 2003; 47: 1367-1383Crossref PubMed Scopus (126) Google Scholar, 8Renn J.P. Clark P.L. Biopolymers. 2008; 89: 420-427Crossref PubMed Scopus (49) Google Scholar, 9Mogensen J.E. Tapadar D. Schmidt M.A. Otzen D.E. Biochemistry. 2005; 44: 4533-4545Crossref PubMed Scopus (27) Google Scholar). This region is essential for stabilizing the β-barrel and/or to promote folding of the extracellular domain (7Oliver D.C. Huang G. Nodel E. Pleasance S. Fernandez R.C. Mol. Microbiol. 2003; 47: 1367-1383Crossref PubMed Scopus (126) Google Scholar, 8Renn J.P. Clark P.L. Biopolymers. 2008; 89: 420-427Crossref PubMed Scopus (49) Google Scholar, 9Mogensen J.E. Tapadar D. Schmidt M.A. Otzen D.E. Biochemistry. 2005; 44: 4533-4545Crossref PubMed Scopus (27) Google Scholar). The adhesin involved in diffuse adherence (AIDA-I) 3The abbreviations used are: AIDA-Iadhesin involved in diffuse adherenceAahautotransporter adhesin heptosyltransferaseIPTGisopropyl-β-thiogalactopyranosideSPATEserine protease autotransporters of EnterobacteriaceaeTBSTris-buffered salineTricineN-[2-hydroxy-1,1-bis(hydroxymethyl)ethyl]glycineWTwild type. 3The abbreviations used are: AIDA-Iadhesin involved in diffuse adherenceAahautotransporter adhesin heptosyltransferaseIPTGisopropyl-β-thiogalactopyranosideSPATEserine protease autotransporters of EnterobacteriaceaeTBSTris-buffered salineTricineN-[2-hydroxy-1,1-bis(hydroxymethyl)ethyl]glycineWTwild type. is a monomeric autotransporter that has been extensively studied. AIDA-I was originally identified as a plasmid-encoded protein from a diffusely adhering Escherichia coli strain isolated in a case of infantile diarrhea (10Benz I. Schmidt M.A. Infect. Immun. 1989; 57: 1506-1511Crossref PubMed Google Scholar). This adhesin was then shown to play a role in neonatal and postweaning diarrheal diseases in piglets, both of which cause major economic losses in farms worldwide (11Niewerth U. Frey A. Voss T. Le Bouguénec C. Baljer G. Franke S. Schmidt M.A. Clin. Diagn. Lab. Immunol. 2001; 8: 143-149Crossref PubMed Scopus (78) Google Scholar, 12Chapman T.A. Wu X.Y. Barchia I. Bettelheim K.A. Driesen S. Trott D. Wilson M. Chin J.J. Appl. Environ. Microbiol. 2006; 72: 4782-4795Crossref PubMed Scopus (191) Google Scholar, 13Zhang W. Zhao M. Ruesch L. Omot A. Francis D. Vet. Microbiol. 2007; 123: 145-152Crossref PubMed Scopus (184) Google Scholar). Besides its role as an adhesin, AIDA-I has been shown to mediate self-association and biofilm formation (14Sherlock O. Schembri M.A. Reisner A. Klemm P. J. Bacteriol. 2004; 186: 8058-8065Crossref PubMed Scopus (139) Google Scholar) as well as invasion of epithelial cells (15Charbonneau M.E. Berthiaume F. Mourez M. J. Bacteriol. 2006; 188: 8504-8512Crossref PubMed Scopus (34) Google Scholar). Additionally, this protein undergoes a modification that is rare in bacteria, as it is O-glycosylated by the specific cytoplasmic protein autotransporter adhesin heptosyltransferase (Aah) (16Benz I. Schmidt M.A. Mol. Microbiol. 2001; 40: 1403-1413Crossref PubMed Scopus (146) Google Scholar, 17Charbonneau M.E. Girard V. Nikolakakis A. Campos M. Berthiaume F. Dumas F. Lépine F. Mourez M. J. Bacteriol. 2007; 189: 8880-8889Crossref PubMed Scopus (58) Google Scholar). AIDA-I has been suggested to be a member of a new group of autotransporter called self-associating autotransporters, which includes the Ag43 aggregation factor and the TibA invasin (18Klemm P. Vejborg R.M. Sherlock O. Int. J. Med. Microbiol. 2006; 296: 187-195Crossref PubMed Scopus (78) Google Scholar). adhesin involved in diffuse adherence autotransporter adhesin heptosyltransferase isopropyl-β-thiogalactopyranoside serine protease autotransporters of Enterobacteriaceae Tris-buffered saline N-[2-hydroxy-1,1-bis(hydroxymethyl)ethyl]glycine wild type. adhesin involved in diffuse adherence autotransporter adhesin heptosyltransferase isopropyl-β-thiogalactopyranoside serine protease autotransporters of Enterobacteriaceae Tris-buffered saline N-[2-hydroxy-1,1-bis(hydroxymethyl)ethyl]glycine wild type. After secretion at the cell surface, the extracellular domain of the protein, called mature AIDA-I, is cleaved from the C-terminal domain, called AIDAc, which encompasses the β-barrel, the α-helix, and the junction region (19Suhr M. Benz I. Schmidt M.A. Mol. Microbiol. 1996; 22: 31-42Crossref PubMed Scopus (72) Google Scholar, 20Benz I. Schmidt M.A. Mol. Microbiol. 1992; 6: 1539-1546Crossref PubMed Scopus (160) Google Scholar). Mature AIDA-I, however, remains strongly associated with AIDAc (15Charbonneau M.E. Berthiaume F. Mourez M. J. Bacteriol. 2006; 188: 8504-8512Crossref PubMed Scopus (34) Google Scholar). The processing of AIDA-I is neither essential for the multiple functions of the protein nor for its stability (15Charbonneau M.E. Berthiaume F. Mourez M. J. Bacteriol. 2006; 188: 8504-8512Crossref PubMed Scopus (34) Google Scholar). Questions remain about the mechanism involved in this cleavage. In a few autotransporters, the cleavage reaction involves an exogenous protease. IcsA for instance, an autotransporter of Shigella flexneri, is cleaved by a dedicated outer membrane protease called IcsP, which is related to the OmpT protease found in E. coli (21Shere K.D. Sallustio S. Manessis A. D'Aversa T.G. Goldberg M.B. Mol. Microbiol. 1997; 25: 451-462Crossref PubMed Scopus (101) Google Scholar). Similarly, the Neisseria meningitidis serine protease NalP is specifically responsible for the partial processing of various autotransporters, including the IgA protease (22van Ulsen P. van Alphen L. ten Hove J. Fransen F. van der Ley P. Tommassen J. Mol. Microbiol. 2003; 50: 1017-1030Crossref PubMed Scopus (111) Google Scholar), App (22van Ulsen P. van Alphen L. ten Hove J. Fransen F. van der Ley P. Tommassen J. Mol. Microbiol. 2003; 50: 1017-1030Crossref PubMed Scopus (111) Google Scholar), AusI (23van Ulsen P. Adler B. Fassler P. Gilbert M. van Schilfgaarde M. van der Ley P. van Alphen L. Tommassen J. Microbes Infect. 2006; 8: 2088-2097Crossref PubMed Scopus (32) Google Scholar), and MspA (24Turner D.P. Marietou A.G. Johnston L. Ho K.K. Rogers A.J. Wooldridge K.G. Ala'Aldeen D.A. Infect. Immun. 2006; 74: 2957-2964Crossref PubMed Scopus (66) Google Scholar). In contrast, the cleavage of AIDA-I is not mediated by the known E. coli periplasmic or outer membrane proteases (DegP, OmpT, or OmpP), and it occurs in E. coli as well as in Shigella or Salmonella strains (19Suhr M. Benz I. Schmidt M.A. Mol. Microbiol. 1996; 22: 31-42Crossref PubMed Scopus (72) Google Scholar). Based on those observations, this processing has been suggested to be an autocatalytic event, even though no protease catalytic site can be identified in the protein. Two types of autocatalytic processing mechanisms have been described in other autotransporters. The Hap autotransporter of Haemophilus influenzae (25Fink D.L. Cope L.D. Hansen E.J. Geme 3rd, J.W. J. Biol. Chem. 2001; 276: 39492-39500Abstract Full Text Full Text PDF PubMed Scopus (55) Google Scholar), the SphB1 protein of Bordetella pertussis (26Coutte L. Willery E. Antoine R. Drobecq H. Locht C. Jacob-Dubuisson F. Mol. Microbiol. 2003; 49: 529-539Crossref PubMed Scopus (58) Google Scholar), and the NalP (22van Ulsen P. van Alphen L. ten Hove J. Fransen F. van der Ley P. Tommassen J. Mol. Microbiol. 2003; 50: 1017-1030Crossref PubMed Scopus (111) Google Scholar), App (27Serruto D. Adu-Bobie J. Scarselli M. Veggi D. Pizza M. Rappuoli R. Aricó B. Mol. Microbiol. 2003; 48: 323-334Crossref PubMed Scopus (84) Google Scholar), and IgA proteases of N. meningitidis (28Pohlner J. Halter R. Beyreuther K. Meyer T.F. Nature. 1987; 325: 458-462Crossref PubMed Scopus (476) Google Scholar) cleave themselves by an endogenous serine protease domain. More recently, it was shown that the serine protease autotransporters of Enterobacteriaceae (SPATE) family cleave themselves in the α-helix located in the lumen of the membrane-embedded β-barrel by a catalytic reaction that involves an aspartate and a conserved asparagine (6Barnard T.J. Dautin N. Lukacik P. Bernstein H.D. Buchanan S.K. Nat. Struct. Mol. Biol. 2007; 14: 1214-1220Crossref PubMed Scopus (137) Google Scholar, 29Dautin N. Barnard T.J. Anderson D.E. Bernstein H.D. EMBO. J. 2007; 26: 1942-1952Crossref PubMed Scopus (97) Google Scholar). The same autocatalytic reaction has been shown for two distantly related B. pertussis autotransporters, pertactin and BrkA (29Dautin N. Barnard T.J. Anderson D.E. Bernstein H.D. EMBO. J. 2007; 26: 1942-1952Crossref PubMed Scopus (97) Google Scholar). However, none of these mechanisms can be hypothesized for AIDA-I cleavage; AIDA-I possesses neither a serine protease domain like Hap nor the asparagine residue required for the SPATE cleavage. In addition, the AIDA-I cleavage site is not located in the α-helix. Thus, the mechanism of AIDA-I processing must be different and remains unknown. The mechanisms by which a diversity of other autotransporters are cleaved remain unknown. This is for instance the case for the E. coli Ag43 self-associating autotransporters (30Henderson I.R. Owen P. J. Bacteriol. 1999; 181: 2132-2141Crossref PubMed Google Scholar), the Helicobacter pylori VacA cytotoxin (31Nguyen V.Q. Caprioli R.M. Cover T.L. Infect. Immun. 2001; 69: 543-546Crossref PubMed Scopus (68) Google Scholar), the Chlamydia pneumoniae polymorphic membrane proteins (32Vandahl B.B. Pedersen A.S. Gevaert K. Holm A. Vandekerckhove J. Christiansen G. Birkelund S. BMC Microbiol. 2002; 2: 36Crossref PubMed Scopus (36) Google Scholar), and the Ssp proteins of Serratia marcescens (33Ohnishi Y. Beppu T. Horinouchi S. J. Biochem. 1997; 121: 902-913Crossref PubMed Scopus (21) Google Scholar) as well as the cohemolysin Cfa (34Litwin C.M. Johnson J.M. Infect. Immun. 2005; 73: 4205-4213Crossref PubMed Scopus (19) Google Scholar) and the Arp immunogenic protein of Bartonella henselae (35Litwin C.M. Rawlins M.L. Swenson E.M. Infect. Immun. 2007; 75: 5255-5263Crossref PubMed Scopus (13) Google Scholar). In this study we determined the mechanism of AIDA-I cleavage. We first showed that a truncated mutant of AIDA-I can refold in vitro and undergo self-cleavage, providing definite proof for the autocatalytic mechanism. By constructing site-directed deletions and point mutations, we then showed that the catalytic domain lies in the region encompassing the amino acids between Ala-667 and Thr-953, and we identified two residues in the junction region, Asp-878 and Glu-897, that are essential for processing. Last, we confirmed that the mechanism of proteolysis for AIDA-I is intramolecular. This is the first report of this type of autocatalytic activity. DISCUSSIONIn this study we proved the autocatalytic nature of the cleavage of the AIDA-I autotransporter by following the in vitro refolding of a mutant lacking its C-terminal α-helix and β-barrel. We noted that in vitro cleavage was less efficient that processing in vivo. This could be because of incomplete folding. Indeed, we observed that folding and processing occur at similar rates, which strongly suggests that the two events are closely related. This observation is also evidence against the cleavage, resulting from a contaminant protease, as the folding of AIDA-I is strikingly slow, and an exogenous protease would most likely cleave AIDA-I faster. Furthermore, we observed that the in vitro refolding of AIDA-I is dramatically pH-dependent and without folding (pH 10), or with an aberrant folding (pH 4) the protein remains uncleaved, showing more evidence of a strong association between folding and cleavage.We identified by deletion analysis a 286-amino acid region directly involved in the processing of AIDA-I. This region includes the junction domain, which ends just before the cleavage site and the C-terminal part of the AIDA-I functional domain. We chose to mutate acidic residues of this region because of the observation that in vitro refolding and processing of AIDA-I occurred even in the presence of serine and cysteine protease inhibitors. Some acidic proteases, like the omptin family, rely on catalytic aspartate residues that cannot be identified by classical consensus motifs (42Hritonenko V. Stathopoulos C. Mol. Memb. Biol. 2007; 24: 395-406Crossref PubMed Scopus (61) Google Scholar). We found that Asp-878 and Glu-897 in the junction region are involved in the processing of AIDA-I. The complete elimination of cleavage when these residues are mutated strongly suggests that they play a catalytic role. Mutations that preserve the acidic side chain were tolerated, giving more evidence that the carboxylic acid groups are directly involved in catalysis. The requirement of Asp-878 for catalytic activity was confirmed by in vitro refolding experiments with the D878N mutant, showing that it can refold but remains uncleaved. We observed that the uncleaved mutant presents a normal global conformation, as this mutant is as functional and as stable as the wild-type AIDA-I. Last, we confirmed that the cleavage of AIDA-I relies on an intramolecular mechanism of processing.As shown in Fig. 9, our results are consistent with a three-dimensional structural model of the junction region and the AIDA-I cleavage site (the amino acids between Ile-807 and Leu-956), which we generated using the known crystal structure of B. pertussis pertactin (39Emsley P. Charles I.G. Fairweather N.F. Isaacs N.W. Nature. 1996; 381: 90-92Crossref PubMed Scopus (260) Google Scholar). This model predicts that Asp-878 and Glu-897 are located close to the cleaved peptidic bond, which is found in a loop. This gives an explanation as to why folding and cleavage occur at similar rates; the junction domain of AIDA-I and the region encompassing the cleavage site have to be properly folded for the correct positioning of the catalytic residues in relation to the cleaved peptidic bond. The model also explains why mutations such as E845A or N849A lead to a decrease in the processing efficiency, as these changes would likely destabilize the loop bearing the bond that has to be cleaved.The mechanism of autocatalytic cleavage we describe here for AIDA-I is unique among the monomeric autotransporter family. Alignment of the junction region sequences of several autotransporters highlights that Asp-878 of AIDA-I is conserved among autotransporters whether they are processed or not (Fig. 9). However, Glu-897 is unique to AIDA-I among autocatalytic-cleaved autotransporters, which could explain why this mechanism of proteolysis is unique to AIDA-I. Our model also differs from the two previously described intramolecular mechanisms of cleavage for autotransporters; that is, the processing mediated either by endogenous serine protease domains or SPATE and pertactin-like autotransporters.The cleavage mechanism for several other autotransporters, including the Ag43 aggregation factor, remains unknown. As Ag43 and AIDA-I are closely related in sequence and possess similar functional features (18Klemm P. Vejborg R.M. Sherlock O. Int. J. Med. Microbiol. 2006; 296: 187-195Crossref PubMed Scopus (78) Google Scholar), it is tempting to hypothesize that a similar mechanism of processing involving carboxylic amino acids exists for Ag43. However, Ag43 is cleaved at a different site compared with the one of AIDA-I (43Klemm P. Hjerrild L. Gjermansen M. Schembri M.A. Mol. Microbiol. 2004; 51: 283-296Crossref PubMed Scopus (115) Google Scholar) and in a region that has only poor homologies with known autotransporter structures.The question of the function of AIDA-I cleavage remains. In autotransporters that are cleaved the processing can play different roles, such as the release of cytotoxic domains in the extracellular milieu (44Navarro-García F. Canizalez-Roman A. Luna J. Sears C. Nataro J.P. Infect. Immun. 2001; 69: 1053-1060Crossref PubMed Scopus (58) Google Scholar), keeping a protein localized at a pole (45d'Hauteville H. Dufourcq Lagelouse R. Nato F. Sansonetti P.J. Infect. Immun. 1996; 64: 511-517Crossref PubMed Google Scholar), or controlling the expression level of an adhesin (46Fink D.L. St. Geme 3rd, J.W. J. Bacteriol. 2003; 185: 1608-1615Crossref PubMed Scopus (32) Google Scholar). The intramolecular nature of AIDA-I processing excludes the possibility that it serves to regulate the protein level at the surface of the bacteria. In addition, we did not observe that cleavage is important for function, as uncleaved mutants resulting from mutations of the catalytic residues or mutations of the residues of the cleaved peptidic bond are as functional as the wild-type protein. This puzzling observation is at least consistent with the fact that we could not find any conditions that lead to the release of mature AIDA-I in the extracellular milieu (15Charbonneau M.E. Berthiaume F. Mourez M. J. Bacteriol. 2006; 188: 8504-8512Crossref PubMed Scopus (34) Google Scholar). We can hypothesize that in the context of an animal infection, a specific condition may cause the release of mature AIDA-I, which would be beneficial for the bacteria, for example to escape the immune response. Alternatively, cleavage could generate a new unique surface representing an interaction site with other unknown proteins or host factors, which could play a role during infection. This mechanism was recently observed with the EscU protein, a structural component of the inner membrane ring of the type-three secretion system in E. coli (47Zarivach R. Deng W. Vuckovic M. Felise H.B. Nguyen H.V. Miller S.I. Finlay B.B. Strynadka N.C. Nature. 2008; 453: 124-127Crossref PubMed Scopus (111) Google Scholar). This last hypothesis would be consistent with the small differences we observed between CD spectra for the D878N mutant and the WT protein in their native conformations or after in vitro folding.Self-processing of autotransporters is a frequent feature and occurs by a variety of different mechanisms, with our study providing yet another one. This diversity is quite surprising and strongly suggests that this processing plays an important role. Further investigation is, therefore, required to establish what this role might be in the case of AIDA-I. In this study we proved the autocatalytic nature of the cleavage of the AIDA-I autotransporter by following the in vitro refolding of a mutant lacking its C-terminal α-helix and β-barrel. We noted that in vitro cleavage was less efficient that processing in vivo. This could be because of incomplete folding. Indeed, we observed that folding and processing occur at similar rates, which strongly suggests that the two events are closely related. This observation is also evidence against the cleavage, resulting from a contaminant protease, as the folding of AIDA-I is strikingly slow, and an exogenous protease would most likely cleave AIDA-I faster. Furthermore, we observed that the in vitro refolding of AIDA-I is dramatically pH-dependent and without folding (pH 10), or with an aberrant folding (pH 4) the protein remains uncleaved, showing more evidence of a strong association between folding and cleavage. We identified by deletion analysis a 286-amino acid region directly involved in the processing of AIDA-I. This region includes the junction domain, which ends just before the cleavage site and the C-terminal part of the AIDA-I functional domain. We chose to mutate acidic residues of this region because of the observation that in vitro refolding and processing of AIDA-I occurred even in the presence of serine and cysteine protease inhibitors. Some acidic proteases, like the omptin family, rely on catalytic aspartate residues that cannot be identified by classical consensus motifs (42Hritonenko V. Stathopoulos C. Mol. Memb. Biol. 2007; 24: 395-406Crossref PubMed Scopus (61) Google Scholar). We found that Asp-878 and Glu-897 in the junction region are involved in the processing of AIDA-I. The complete elimination of cleavage when these residues are mutated strongly suggests that they play a catalytic role. Mutations that preserve the acidic side chain were tolerated, giving more evidence that the carboxylic acid groups are directly involved in catalysis. The requirement of Asp-878 for catalytic activity was confirmed by in vitro refolding experiments with the D878N mutant, showing that it can refold but remains uncleaved. We observed that the uncleaved mutant presents a normal global conformation, as this mutant is as functional and as stable as the wild-type AIDA-I. Last, we confirmed that the cleavage of AIDA-I relies on an intramolecular mechanism of processing. As shown in Fig. 9, our results are consistent with a three-dimensional structural model of the junction region and the AIDA-I cleavage site (the amino acids between Ile-807 and Leu-956), which we generated using the known crystal structure of B. pertussis pertactin (39Emsley P. Charles I.G. Fairweather N.F. Isaacs N.W. Nature. 1996; 381: 90-92Crossref PubMed Scopus (260) Google Scholar). This model predicts that Asp-878 and Glu-897 are located close to the cleaved peptidic bond, which is found in a loop. This gives an explanation as to why folding and cleavage occur at similar rates; the junction domain of AIDA-I and the region encompassing the cleavage site have to be properly folded for the correct positioning of the catalytic residues in relation to the cleaved peptidic bond. The model also explains why mutations such as E845A or N849A lead to a decrease in the processing efficiency, as these changes would likely destabilize the loop bearing the bond that has to be cleaved. The mechanism of autocatalytic cleavage we describe here for AIDA-I is unique among the monomeric autotransporter family. Alignment of the junction region sequences of several autotransporters highlights that Asp-878 of AIDA-I is conserved among autotransporters whether they are processed or not (Fig. 9). However, Glu-897 is unique to AIDA-I among autocatalytic-cleaved autotransporters, which could explain why this mechanism of proteolysis is unique to AIDA-I. Our model also differs from the two previously described intramolecular mechanisms of cleavage for autotransporters; that is, the processing mediated either by endogenous serine protease domains or SPATE and pertactin-like autotransporters. The cleavage mechanism for several other autotransporters, including the Ag43 aggregation factor, remains unknown. As Ag43 and AIDA-I are closely related in sequence and possess similar functional features (18Klemm P. Vejborg R.M. Sherlock O. Int. J. Med. Microbiol. 2006; 296: 187-195Crossref PubMed Scopus (78) Google Scholar), it is tempting to hypothesize that a similar mechanism of processing involving carboxylic amino acids exists for Ag43. However, Ag43 is cleaved at a different site compared with the one of AIDA-I (43Klemm P. Hjerrild L. Gjermansen M. Schembri M.A. Mol. Microbiol. 2004; 51: 283-296Crossref PubMed Scopus (115) Google Scholar) and in a region that has only poor homologies with known autotransporter structures. The question of the function of AIDA-I cleavage remains. In autotransporters that are cleaved the processing can play different roles, such as the release of cytotoxic domains in the extracellular milieu (44Navarro-García F. Canizalez-Roman A. Luna J. Sears C. Nataro J.P. Infect. Immun. 2001; 69: 1053-1060Crossref PubMed Scopus (58) Google Scholar), keeping a protein localized at a pole (45d'Hauteville H. Dufourcq Lagelouse R. Nato F. Sansonetti P.J. Infect. Immun. 1996; 64: 511-517Crossref PubMed Google Scholar), or controlling the expression level of an adhesin (46Fink D.L. St. Geme 3rd, J.W. J. Bacteriol. 2003; 185: 1608-1615Crossref PubMed Scopus (32) Google Scholar). The intramolecular nature of AIDA-I processing excludes the possibility that it serves to regulate the protein level at the surface of the bacteria. In addition, we did not observe that cleavage is important for function, as uncleaved mutants resulting from mutations of the catalytic residues or mutations of the residues of the cleaved peptidic bond are as functional as the wild-type protein. This puzzling observation is at least consistent with the fact that we could not find any conditions that lead to the release of mature AIDA-I in the extracellular milieu (15Charbonneau M.E. Berthiaume F. Mourez M. J. Bacteriol. 2006; 188: 8504-8512Crossref PubMed Scopus (34) Google Scholar). We can hypothesize that in the context of an animal infection, a specific condition may cause the release of mature AIDA-I, which would be beneficial for the bacteria, for example to escape the immune response. Alternatively, cleavage could generate a new unique surface representing an interaction site with other unknown proteins or host factors, which could play a role during infection. This mechanism was recently observed with the EscU protein, a structural component of the inner membrane ring of the type-three secretion system in E. coli (47Zarivach R. Deng W. Vuckovic M. Felise H.B. Nguyen H.V. Miller S.I. Finlay B.B. Strynadka N.C. Nature. 2008; 453: 124-127Crossref PubMed Scopus (111) Google Scholar). This last hypothesis would be consistent with the small differences we observed between CD spectra for the D878N mutant and the WT protein in their native conformations or after in vitro folding. Self-processing of autotransporters is a frequent feature and occurs by a variety of different mechanisms, with our study providing yet another one. This diversity is quite surprising and strongly suggests that this processing plays an important role. Further investigation is, therefore, required to establish what this role might be in the case of AIDA-I. We thank François Lépine for performing the mass spectrometry experiments.