Crystal structure of the Yersinia enterocolitica type III secretion chaperone SycT.

Several Gram-negative pathogens deploy type III secretion systems (TTSSs) as molecular syringes to inject effector proteins into host cells. Prior to secretion, some of these effectors are accompanied by specific type III secretion chaperones. The Yersinia enterocolitica TTSS chaperone SycT escorts the effector YopT, a cysteine protease that inactivates the small GTPase RhoA of targeted host cells. We solved the crystal structure of SycT at 2.5 angstroms resolution. Despite limited sequence similarity among TTSS chaperones, the SycT structure revealed a global fold similar to that exhibited by other structurally solved TTSS chaperones. The dimerization domain of SycT, however, differed from that of all other known TTSS chaperone structures. Thus, the dimerization domain of TTSS chaperones does not likely serve as a general recognition pattern for downstream processing of effector/chaperone complexes. Yersinia Yop effectors are bound to their specific Syc chaperones close to the Yop N termini, distinct from their catalytic domains. Here, we showed that the catalytically inactive YopT(C139S) is reduced in its ability to bind SycT, suggesting an ancillary interaction between YopT and SycT. This interaction could maintain the protease inactive prior to secretion or could influence the secretion competence and folding of YopT.

Type three secretion systems (TTSSs) 1 are utilized by various Gram-negative bacteria, pathogens of animals and plants, to inject effector proteins into host cells, where they manipulate cellular functions (1,2). TTSSs of pathogens share a common ancestor with the most widespread TTSS, the bacterial flagellum (3). The core components of TTSS transport machineries are highly conserved, and some are functionally exchangeable among TTSSs (4,5). In contrast, TTSS effectors, some of them accompanied by dedicated chaperones, are relatively speciesspecific because of the particularities of every host-pathogen relationship. TTSS chaperones are characterized as small, acidic proteins that lack ATPase activity and associate with one or several TTSS secretion substrates within bacterial cells (6 -8). Several functions have been attributed to TTSS chaperones. Phenotypically, these chaperones are required for efficient translocation of their assisted effectors. However, their mode of action remains elusive. Some effectors are poorly soluble in the absence of their chaperones and are apparently stabilized upon their binding. Further, the prevention of unproductive interactions and the maintenance of secretion-competent folding states are discussed. In addition, it was proposed that effector/chaperone complexes could constitute three-dimensional secretion signals (9) and that chaperones could govern a hierarchical order of effector secretion (9 -11). A recent study suggests the guidance of effectors by their chaperones toward the TTSS ATPase (12).
TTSS chaperones were recently categorized into three classes (8). The class I chaperones specific to a single effector belong to the subgroup IA. Class IB comprises the promiscuous chaperones assisting more than one effector. Chaperones of class II assist the TTSS translocators, and the flagellar TTSS chaperones constitute class III according to this classification. Crystal structures of representatives of class IA (Salmonella SicP (13) and SigE (14); Yersinia SycE (15), SycH (16), and SycN/YscB (17); Escherichia coli CesT (14); Pseudomonas syringae AvrPphF ORF1 (18)), class IB (Shigella flexneri Spa15 (19)), and class III (Aquifex aeolicus FliS (20)) support this classification. The class IA chaperones SicP, SigE, SycE, CesT, and AvrPphF ORF1, although not similar on a sequence level, all form homodimers and share a common fold. The SycN/YscB complex is an exception in that SycN and YscB form a heterodimer (17) with a fold similar to that of the typical homodimers. The SycH fold also resembles that of the aforementioned chaperones (16); however, its biologically relevant oligomerization state is not unambiguously clear (21). The fold of the class IB chaperone Spa15 is very similar to that of class IA chaperones; dimer formation, however, is different, with the subunits rotated relative to each other (19). The crystal structure of the flagellar secretion chaperone FliS (class III) reveals a fold distinct from that of pathogenicity-related TTSS secretion chaperones (20). This suggests that in contrast to the conserved transport machineries of flagella and the TTSSs of pathogens, their accessory chaperones have different evolutionary origins. Structures of class II chaperones have not been solved so far, but the identification of a tetratricopeptide-like repeat motif in Yersinia SycD/LcrH (22) supports the view of a fold different from that of class I representatives. Recently, another distinct class of TTSS-related chaperones was disclosed with the crystallization of the EspA filament protein complexed to its antipolymerization chaperone CesA of enteropathogenic E. coli (23). CesA exhibits a highly extended three-helix hairpin forming extensive coiled-coil interactions with EspA.
The effector domains bound by class I chaperones typically encompass about 50 -100 amino acid residues that are localized at the N terminus directly following the putative secretion signal. Co-crystallization of TTSS chaperones together with effector fragments shed light on the binding mode of these chaperones (9,13,16). The effector domain is wrapped around the chaperone dimer in an expanded form retaining secondary structure elements. Work from several groups further suggests that the C-terminally located effector domains are folded and catalytically active in the presence of their respective chaperones (9,14,21,24). Only recently, the structure of the secretable regulatory TTSS component YopN from Yersinia in complex with the heterodimeric chaperone SycN/YscB was presented (17). This structure unambiguously confirms that the influence of chaperone binding on folding of the secretion substrate is confined to the chaperone-binding site and does not extend globally.
The TTSS of pathogenic Yersinia species is used to paralyze host cells such as macrophages by injection of effectors, called Yops (2). One of these effectors, YopT, is the representative of a novel family of cysteine proteases with a catalytic triad formed by residues Cys-139, His-258, and Asp-274 (25,26). In the Yersinia cytosol, YopT is accompanied by the specific Yop chaperone SycT (27). Here, we report on the Yersinia enterocolitica SycT structure, a representative of the class IA TTSS chaperones. The only close homologue of SycT (56% identity) is found in the entomopathogenic Photorhabdus luminescens (28). SycT is 20% identical to Yersinia SycE, which has been structurally solved (15,29,30).
The SycT-binding site of YopT was located within the first 124 amino acid residues (27). We demonstrated that a C139S mutation of YopT reduces its SycT binding efficiency, suggesting an accessory interaction between SycT and the catalytic domain of YopT.

MATERIALS AND METHODS
Recombinant DNA Techniques-The cloning of sycT was as follows. The SycT coding sequence was amplified by PCR using primers 5Ј-CAT-ATGCAGACAACCTTCACAGAACTTATGCA-3Ј and 5Ј-GTCGACTCAG-ATGAATAATATAGGTGATGTCG-3Ј, thereby introducing flanking NdeI and SalI restriction sites (underlined). Bacterial lysate from Y. enterocolitica strain WA-314 served as template. The PCR product was subcloned into TOPO TA cloning vector (Invitrogen), and the sycT fragment was cut out with NdeI and SalI and inserted into NdeI-SalI-cleaved pWS. pWS is a derivative of pMS470⌬8 (32) generated by cleavage with NdeI and HindIII and insertion of a linker hybridized from 5Ј-phosphorylated oligonucleotides 5Ј-TATGAAGCTTAGATCTGTCGACGGATC-3Ј and 5Ј-AGCTGATCCGTCGACAGATCTAAGCTTCA-3Ј.
Expression of GST-fused proteins in E. coli (BL21) was induced with 0.4 mM isopropyl-1-thio-␤-D-galactopyranoside for 3 h in cells growing at 37°C when A 600 nm reached 0.5-0.7. Cells were lysed in phosphatebuffered saline lysis buffer supplemented with 1 mM DTT, 100 M phenylmethylsulfonyl fluoride, and lysozyme (200 g/ml final concentration). The GST-fused proteins were purified from bacterial lysates by affinity chromatography using glutathione-coupled Sepharose beads (Amersham Biosciences). Recombinant proteins were eluted off the glutathione beads with 20 mM glutathione in 100 mM Tris-HCl (pH 7.4) and 50 mM NaCl.

Analysis of SycT Binding to GST-YopT Fusion
Variants-Glutathione-Sepharose resin was loaded with equal amounts of GST-YopT/SycT, GST-YopT C139S /SycT, or GST-YopT ⌬1-22 /SycT. Loaded beads were incubated overnight (gently rotating) in 1 or 50 ml of phosphate-buffered saline. Supernatant was removed, and beads were eluted in SDS sample buffer. The eluted samples were resolved on SDS-PAGE and analyzed by Western blotting with antibodies raised against YopT and SycT.
Crystallization and Structure Determination-Crystals of SycT were grown at 20°C within 4 days to their final size of 500 ϫ 100 ϫ 50 m 3 by using the hanging drop vapor diffusion method. The drops contained equal volumes of protein (40 mg/ml) and reservoir solution (1.2 M DL-malic acid, pH 7.0, and 100 mM Bis-Tris propane, pH 7.0). Before exposure to X-rays, crystals were soaked in a solution of 1.2 M DL-malic acid, pH 7.0, and 100 mM Bis-Tris propane, pH 7.0, 25% glycerol for 30 s and subsequently frozen in a stream of cold nitrogen gas at 100 K (Oxford Cryosystems). Multiple anomalous dispersion methods were performed using synchrotron radiation at the BW6 beamline at the Deutsches Elektronen Synchrotron Center (DESY) in Hamburg, Germany. However, crystals severely suffered during their exposure with synchrotron radiation, which resulted in increasing R merge values upon their exposure time. Furthermore, SycT crystals showed high anisotropy, in particular at high resolution, which caused tremendous difficulties in obtaining suitable data sets. Molecular replacement using the structure of the monomer or the dimer of SycE (Protein Data Bank ID: 1JYA) as a poly-Ala-model failed, most likely because: 1) the SycT crystals have high anisotropy; 2) the structure of SycT shows significant differences in its dimerization domain, as compared with other Syc proteins; and 3) the primary sequences of both molecules show only 20% sequence identity.
Testing a broad variety of SycT crystals, incubated with various heavy metal atom solutions, it turned out that crystals treated with K 2 PtCl 6 for 6 h were most suitable for structure elucidation. After a successful wavelength scan for the platinum absorption edge, data sets were collected on a MAR CCD detector for peak (1.0719 Å), inflection point (1.0723 Å), and remote (0.98 Å) wavelength. Images in frames of 1°were recorded over a range of 360°, resulting in complete anomalous data sets. The space group of ScyT crystals was C2 with unit cell dimensions of a ϭ 91 Å, b ϭ 46 Å, c ϭ 34 Å, and ␤ ϭ 105°. The images were processed with DENZO and SCALEPACK (34) and scaled further with TRUNCATE, CAD, and SCALEIT of the CCP4 software package (35). The Pt 4ϩ position in the asymmetric unit cell was localized by a combination of direct and difference Patterson search methods using ShelXD (36). Initial phase angles were calculated with MLPHARE, and the electron density was calculated by Fourier transformation and improved by solvent flattening (35). Hereby, the hand ambiguity could be solved. However, the quality of the multiple anomalous dispersionelectron density was rather poor, which most likely found its reason in the radiation damage of the crystals during their exposure. Calculating phase angles at the single Pt 4ϩ peak wavelength (single wavelength anomalous dispersion) and using the program SHARP (37) resulted in an electron density that allowed interpreting the secondary structure elements of SycT. Conceivable ␤-strands and helices were traced as polyalanine residues. Subsequent phase combination using the preliminary model and the experimental Pt 4ϩ phase angles was performed with SHARP (37). Using these improved phase angles, we looked for further derivatives of our measured data sets. We were able to identify a second derivative by difference Fourier analysis and could locate two Pb ϩ heavy metal atom-binding sites in a data set, in which crystals were soaked for 6 h with Pb(CH3) 3 Cl. However, the quality of the single wavelength anomalous dispersion peak-Pb ϩ data set was rather weak as compared with the Pt 4ϩ data set, revealing a figure of merit of 0.62/0.5 (centric/acentric) and a phasing power of 1.4; the Pt 4ϩ data set data showed a figure of merit of 0.77/0.68 and a phasing power of 3.4 (Table I). Although the phasing power of the Pb ϩ derivative was quite low (mainly caused by the low occupancy of the heavy metal atoms), it could successfully be used to enhance the signal-to-noise ratio by averaging the Pt 4ϩ and Pb ϩ electron densities, which allowed us to interpret and to complete the SycT model (both data sets, Pt 4ϩ and Pb ϩ , showed an acceptable estimation in their isomorphous differences). The calculated electron density allowed identifying characteristic side chain residues, thus completing the structure. Continuous model building and refinement was performed with the interactive three-dimensional graphics program MAIN (38) and REFMAC5 (39).
There was no electron density visible for the 2 N-terminal residues and 18 C-terminal residues, whereas residues 3-112 could be built in the defined electron density (see Fig. 2D). The protein model was refined by REFMAC5 using TLS (TLS parameters describe anisotropic motion; an anisotropic U factor is derived for each atom in the TLS group) (40), conventional crystallographic rigid body, and positional and anisotropic temperature factor refinements (39), resulting in the current crystallographic values of R cryst ϭ 24.3%, R free ϭ 25.9%, and root mean square difference of bonds as follows: root mean square difference Ϫ bond ϭ 0.007 Å and root mean square difference Ϫ angle ϭ 1.25° ( 41). The slightly increased R-values for this resolution are due to the anisotropic crystalline order causing deterioration in data quality. The current SycT model comprises 910 non-hydrogen atoms and 22 water molecules/asymmetric unit cell. Data of SycT have been deposited in the RCSB Protein Data Bank with Protein Data Bank ID code 2bho.

Co-expression and Purification of YopT/SycT
Complexes-Work from several groups suggests that the influence of class I TTSS chaperones on the folding of their substrates is restricted to the binding site generally encompassing about 50 -100 amino acid residues (9,14,17,21,24). To understand the functioning of these chaperones, high resolution structural data on complete type III effectors in complex with their respective chaperones are required. We established a co-expression and co-purification protocol to facilitate crystallization trials with YopT/SycT complexes. The bicistronic organization of yopT and sycT on the Yersinia virulence plasmid was utilized to construct a plasmid for translational fusion of YopT to GST, allowing the simultaneous production of SycT. The yield was ϳ15 mg of YopT/SycT per liter of bacterial culture. We did not succeed in growing any YopT/SycT crystals. However, we ob- e Cullis R-factor is the lack of closure residual/isomorphous difference. acent./cent., acentric/centric. f Phasing power ϭ root mean square. F H /lack of closure, where FH is the calculated heavy atom contribution. ano. acent., anomalous acentric. g Figure of merit ϭ ͚͗ ␣ P(␣)e ix /͚ ␣ P(␣), after density modification (35), where ␣ is the phase and P(␣) is the phase probability distribution. acent./cent., acentric/centric. h r ϭ ͚ hkl ʈF obs ͉ Ϫ ͉F calc ʈ/͚ hkl ͉F obs ͉, where R free (39) is calculated without a sigma cutoff for a randomly chosen 10% of reflections, which were not used for structure refinement, and R work is calculated for the remaining reflections.
i Deviations from ideal bond lengths/angles. j Number of residues in favored region/allowed region/outlier region.
served some degree of instability of YopT after cleaving off GST, which could interfere with crystallization. Assuming autoproteolytic cleavage, we introduced a C139S mutation into YopT, which renders the protease inactive (25). Further, an N-terminally truncated form of YopT was produced with the first 22 amino acid residues being deleted. Expression and purification of both variant YopT/SycT complexes proved feasible and resulted in homogeneous material. However, up to now, we have not succeeded in yielding crystals from this material. Catalytically Inactive YopT C139S Is Reduced in Its Ability to Bind SycT-When we accomplished the YopT C139S /SycT purification, we observed that the amount of co-purified SycT was reduced as compared with YopT/SycT or YopT ⌬1-22 /SycT preparations. The effect was rather subtle but could be intensified by prolonged incubation of the GST-YopT C139S /SycT complexes in dilute solutions followed by affinity purification on glutathione-Sepharose resin. The phenomenon is illustrated in Fig. 1. Glutathione-Sepharose resin was loaded with equal amounts of purified GST-YopT/SycT, GST-YopT C139S /SycT, and GST-YopT ⌬1-22 /SycT. Loaded beads were incubated overnight in 1 (A) or 50 ml (B) of phosphate-buffered saline. Beads were centrifuged, eluted with SDS loading buffer, and subjected to SDS-PAGE. The immunoblot was developed with antisera specific to YopT and SycT. The intensified dissociation of SycT from GST-YopT C139S upon dilution can be observed (B). In contrast, deletion of the YopT N terminus (⌬1-22) did not interfere with SycT binding. We thus conclude that the catalytic center of YopT also interacts with SycT. This finding suggested SycT properties undescribed for TTSS chaperones and motivated us to recombinantly express SycT in E. coli to crystallize it.
Crystal Structure of SycT-SycT was heterologously produced in E. coli in its outright form without an affinity tag and purified to near homogeneity. We managed to grow SycT crystals and solved the structure at 2.5 Å resolution by multiple wavelength anomalous dispersion methods using a platinum derivative. Our model included residues 3-112 and lacks 18 disordered residues at the C terminus. SycT has a ␣-(␤) 6 -␣ fold and appears as a dimer, although the asymmetric unit cell comprises only one subunit (Fig. 2, A and B). Dimer formation of SycT was also in accordance with analytical size exclusion chromatography. 2 Basically, the SycT fold is very similar to that of all other class I chaperones crystallized so far (13-19). However, in SycT ␣-helix H2, which mediates dimerization of all so far structurally analyzed class I TTSS chaperones, was replaced by an outstretched loop-like structure. Among the TTSS chaperones structurally solved, SycE was the one closest to SycT (20% identical to SycE). A superimposition of the SycT and SycE structural models (Fig. 2C) illustrates the characteristic similarities as well as the differences. The N-and Cterminal ␣-helices of both chaperones (H1 and H3) were almost identically oriented in the monomer. An antiparallel ␤-sheet formed of five ␤-strands in SycE was found in a very similar orientation in the SycT structure. Strands S1-S5 of SycE match well with strands S1-S5 of SycT (Fig. 2, C and D). An interesting difference between SycT and SycE concerns the orientation of the loop connecting ␣-helix H1 and ␤-strand S1. In the SycT structure, an additional small ␤-strand, designated S0, cut in on this connecting loop. The major difference between the structures of SycT and SycE, however, was found at the dimerization interface. The dimerization of SycT is mainly brought about by hydrophobic interactions (Fig. 3, A-C). In particular, Trp-47 of one subunit was in close contact with Trp-69 and Pro-70 of the second subunit, and Phe-65 makes contact with Gln-49 and Trp-84. Trp-69 is in contact with Tyr-43, His-44, Trp-47, Gln-49, and Gln-86. Finally, Trp-84 interacts with Phe-65 and Ala-71. The dimer interface of SycE, also mainly stabilized via hydrophobic contacts (15), is arranged differently, as depicted in Fig. 3D. The differences in the organization of the dimer contacts of SycT and SycE result in a different tilting of the respective subunits (Figs. 2C and 3, C and D, see the limited congruence of the left-hand side subunits).

DISCUSSION
The Three-dimensional Secretion Signal-Birtalan et al. (9) have recently suggested that chaperone/effector complexes may function as three-dimensional secretion signals targeting the secretion substrates toward the type III transport machinery. This hypothesis is supported by the impressive conservation of folds among class I chaperones of TTSSs given their low sequence similarity. Further, it is supported by the structural conservation of the mode of effector/chaperone interaction, which is not based on a high degree of sequence similarity (9,13,16,17). This functional conservation of chaperone activity is underscored by the interchangeability of effector/chaperone pairs among TTSSs of different species (5,42). Recently, Gauthier and Finlay (12) could show that the ATPase of the TTSS from enteropathogenic E. coli, EscN, binds to the chaperone CesT directly and independently of the belonging effector Tir. This suggests that it is the chaperone that directs the effector toward the ATPase. This finding is in line with our recently presented model of type III secretion, predicting that TTSS ATPases act as unfoldases using the TTSS chaperones to encounter the secretion substrates (21,24,43). After displacement of the chaperone by the ATPase, the chaperone-binding site of the effector, lacking tertiary structure, is distinguished as an ideal starting point for unraveling the effector. Which part of the chaperone structure could represent the common recognition pattern? In view of the SycT structure, we can now exclude ␣-helix H2 and the dimerization domain as part of this recognition pattern. Further, when comparing all available structures, it seems unlikely that the dimer as a whole is recognized. Rather, the most conserved parts of the monomer should be considered. A superimposition of SycE, SycH, SycN/ YscB, and SycT ( Fig. 2C) (16,17) displays that ␣-helices H1 and H3 as well as ␤-strands S1-S5 colocalize very nicely but do so in only one subunit because of considerable variation of subunit tilting. The latter phenomenon is most pronounced in the structure of S. flexneri Spa15 (19). The Spa15 monomer, however, exhibited the same overall fold as class IA chaperones. Further, it has been shown that Yersinia, Salmonella, and Shigella TTSSs are functionally conserved (5). Taken together, we concluded that the recognition pattern is represented by structural features of the chaperone monomer. Furthermore, the model of a recognition pattern as part of the monomer may provide a clue to understanding of the Nterminal secretion signal. Why should there be an N-terminal secretion signal if there is a three-dimensional recognition pattern? We suggest that the N terminus may serve as the starting point for chaperone displacement followed by substrate unfolding and that the ATPase therefore recognizes the one subunit of the chaperone that accommodates the N terminus. This model could also help to explain the conflicting data on the nature of the N-terminal secretion signal (see Refs. 44 and 45 for reviews). If the function of the N terminus is not only to target the secretion substrate but also to serve as an initiation site for chaperone displacement and substrate unfolding, the characteristics of the N terminus would follow completely different constraints than previously thought. The structures of SycE, SycH, SycN/YscB, and SycT should now provide a basis for detailed mutational analyses of these chaperones to reveal the residues and structural elements critical for recognition.
The YopT/SycT Interaction-In the absence of the structure of a YopT/SycT complex, we can use the YopE 23-78 /SycE model for comparison. Basically, the YopE 23-78 peptide cannot be simply accommodated onto SycT; the peptide clashes with SycT at several positions (not shown). In Fig. 2D, 4 residues of SycE, which contribute significantly to binding of the YopE peptide, are indicated. Two of these residues were not conserved in SycT; one showed conservation, and only one was identical. Further, SycT owned an additional small ␤-strand, designated S0, which was protruding into that space occupied by the YopE peptide in the YopE 23-78 /SycE model. Therefore, ␤-strand S0 might be a putative interaction site for YopT. Finally, the different relative orientation of the subunits in the SycT and SycE dimer contributed to clashes when modeling YopE 23-78 binding to SycT. It would be interesting to test whether it is possible, due to the conserved fold of the chaperones, to mutate one of the chaperones in such a way that it is able to bind to a non-destined effector and mediate its type III-dependent transport.
A Novel Form of Chaperone/Effector Interaction-Work from several groups suggests that the influence of TTSS chaperones FIG. 2. a, crystal packing 3. a, stereo view of SycT dimerization. One subunit is drawn as a surface representation, and the other is drawn as a ribbon plot (blue). Specific residues, particularly those contributing to dimerization, are drawn as balls-and-sticks. Carbon atoms are drawn in green, and atoms of oxygen and nitrogen are shown in red and blue, respectively. b, an electron density map of the dimerization domain. Wire frame structure of the dimerization domain in SycT is shown. Interactions between subunits are particularly formed by hydrophobic side chains. Electron density is calculated with experimental phases calculated form the platinum and lead heavy metal atom derivatives. The map is contoured at 1 , with 2F o Ϫ F c coefficients. Temperature factors of residues in the dimerization domain are below the average temperature factor of the whole molecule, showing less flexibility in this region. c, a stereo view of the SycT dimerization domain, similarly oriented as in a. d, a stereo view of the SycE dimerization domain. The orientation of SycE is comparable with SycT and based on the structural superposition.
on the conformation of their substrates is restricted to the binding sites identified close to the N termini (9,14,17,21,24). Here, for the first time, an interaction between the catalytic domain of an effector, YopT, and its specific chaperone, SycT, was revealed. We demonstrated that catalytically inactive YopT C139S is reduced in its ability to bind SycT. It is not very likely that this conserved substitution causes dramatic conformational changes. Rather, this finding suggested that a direct interaction between the catalytic pocket and the chaperone was disturbed. We could imagine two plausible functions for such an interaction. Since YopT is a cysteine protease, the ancillary interaction with SycT could serve to prevent proteolytic activity inside the bacteria. Alternatively, the interaction could contribute to the secretion competence of YopT. However, we have scrutinized the latter hypothesis but could not demonstrate any difference in respect to transport-efficiency between YopT and YopT C139S . 2 Which part of SycT is likely to interact with the catalytic domain of YopT? Assuming that YopT is accommodated by SycT in a manner comparable with the YopE/SycE interaction, the obviously flexible and thus unresolved C-terminal end was the most likely candidate for this interaction. Class I chaperones differ considerably with respect to the Cterminal extension following ␣-helix H3. Several chaperones exhibit very small extensions of 1-5 residues (e.g. Spa15, SigE, AvrPphF ORF1), whereas others, such as the Yersinia Syc chaperones, have 12-18-residue extensions that are generally undefined in the structures and to which no function could be assigned so far. Truncated versions of these chaperones now have to be analyzed in vitro and in vivo to learn about the role of these extensions.