Tse1 (Tse is type VI secretion exported), an effector protein produced by Pseudomonas aeruginosa, is an amidase that hydrolyses the γ-D-glutamyl-DAP (γ-D-glutamyl-L-meso-diaminopimelic acid) linkage of the peptide bridge of peptidoglycan. P. aeruginosa injects Tse1 into the periplasm of recipient cells, degrading their peptidoglycan, thereby helping itself to compete with other bacteria. Meanwhile, to protect itself from injury by Tse1, P. aeruginosa expresses the cognate immunity protein Tsi1 (Tsi is type VI secretion immunity) in its own periplasm to inactivate Tse1. In the present paper, we report the crystal structures of Tse1 and the Tse1-(6–148)–Tsi1-(20-end) complex at 1.4 Å and 1.6 Å (1 Å=0.1 nm) resolutions respectively. The Tse1 structure adopts a classical papain-like α+β fold. A cysteine–histidine catalytic diad is identified in the reaction centre of Tse1 by structural comparison and mutagenesis studies. Tsi1 binds Tse1 tightly. The HI loop (middle finger tip) from Tsi1 inserts into the large pocket of the Y-shaped groove on the surface of Tse1, and CD, EF, JK and LM loops (thumb, index finger, ring finger and little finger tips) interact with Tse1, thus blocking the binding of enzyme to peptidoglycan. The catalytic and inhibition mechanisms provide new insights into how P. aeruginosa competes with others and protects itself.
- interbacterial competition
- type VI secretion exported 1 (Tse1)
- type VI secretion immunity 1 (Tsi1)
- type VI secretion system (T6SS)
To manipulate their invasion into host cells or the environment, pathogenic bacteria have developed several secretion systems which transport corresponding effector proteins to the exterior of bacteria. At least six types of secretion machines, termed T1SS–T6SS (type I to type VI secretion systems) respectively, have been identified in the Gram-negative bacterial pathogens of animals and plants [1–3]. These secretion machines have versatile structural components and substrates, and are crucial for the interactions between host and bacteria, as well as different types of bacteria [4,5]. T1SS, T2SS and T4SS usually transport effectors into the bacterial cell surface or into the milieu [6,7]. The T3SS and T4SS secretion systems of many bacterial pathogens take part in delivering secretion proteins into eukaryotic cells, yielding spectacular cellular responses, such as reshuffling of host cell actin and hijacking of host signalling transduction, and eventually they subvert the host response [8–12].
The characterized T6SS has emerged as a novel molecular syringe in many bacterial species [13–15] and plays roles in virulence-related processes, interbacterial competition and biofilm formation [1,16–20]. Although T6SSs are involved in numerous processes of bacterial life cycles, their underlying mechanisms are still unknown. Additionally, the secreted effector proteins, which directly interact with host cell or competitive bacteria, have become increasingly attractive for the attention of biologists because they can cause so many physiological consequences to cells.
Three effectors have been identified in the opportunistic pathogen Pseudomonas aeruginosa, namely Tse1, Tse2 and Tse3 (Tse is type VI secretion exported). These three effectors are controlled by the H1-T6SS of P. aeruginosa and injected into the cytoplasm (Tse2) and periplasm (Tse1 and Tse3) of the recipient. Tse2 is a toxic effector. Its main physiological function is correlated with growth arrest in prokaryotic cells, providing a competitive advantage for P. aeruginosa during fierce niche competition. To avoid inhibiting itself, an immunity protein Tsi2 (Tsi is type VI secretion immunity) in the cytoplasm of P. aeruginosa acts as a Tse2 inhibitor. Bacteria lacking Tsi2 are vulnerable and are easily excluded during intrabacterial competition . Tse1 has amidase activity hydrolysing the peptides between γ-D-glutamyl-DAP (γ-D-glutamyl-L-meso-diaminopimelic acid) of the murein, and Tse3 has muramidase activity cutting the chain of glycan of the murein. They are both lytic enzymes that degrade peptidoglycan. To protect itself from accidental injury via contact with other P. aeruginosa cells, the bacterium secretes two T6SS immunity proteins (Tsi1 and Tsi3) into its own periplasm, binding and neutralizing the cognate toxin. These three effectors and their cognate immunity proteins exist exclusively in P. aeruginosa. Therefore they inhibit the growth of competitor cells without causing accidental injury to P. aeruginosa, providing fitness advantages in the microbial jungle competition [20,21]. In the present paper, we report the crystal structures of Tse1 and the Tse1-(6–148)–Tsi1-(20-end) complex. The structural data, combined with biochemical data, provide further insight into the mechanism by which Tse1 functions as an amidase and how it is controlled by its cognate protein Tsi1.
MATERIALS AND METHODS
Protein expression, purification and site-directed mutagenesis
The genes encoding the full-length and fragments of Tse1 and Tsi1 were PCR-amplified from P. aeruginosa PAO1 genomic DNA using oligonucleotide primers with the restriction recognition sites NdeI and XhoI located at 5′ and 3′ ends. The full-length Tse1 and its fragment Tse1-(6–148) were cloned into pET-21b(+) and pET-28bTEV [with the thrombin protease cutting site replaced with TEV (tobacco etch virus) protease; Novagen] expression vectors respectively and expressed in Escherichia coli BL21(DE3) cells. Tsi1-(20-end) was cloned into pET-21b(+) vector and expressed in E. coli Origami (DE3) cells.
To examine the influence of Tse1 and its mutants on bacterial growth (to detect the amidase activity of Tse1 and its mutants), we used the same method as described previously . Briefly, Tse1 was constructed into pET-22b(+) between BamHI and XhoI restriction recognition sites and thus expressed in the periplasm of the bacteria to degrade peptidoglycan of host cells. Three Tse1 mutants (C30A, H91A and C110A) with Tse1-pET-22b as a template were produced using the QuikChange® Site-Directed Mutagenesis kit (Strategene), and the mutants were confirmed by sequencing.
BL21(DE3) cells harbouring the plasmid of recombinant full-length Tse1 protein were grown in LB (Luria–Bertani) medium supplemented with 100 μg/ml ampicillin. When the bacterial concentration reached a D600 of 0.8, the incubation temperature was decreased to 16°C, and IPTG (isopropyl β-D-thiogalactopyranoside) at a final concentration of 0.2 mM was added to induce protein expression. Approximately 12 h later, the cells were harvested by centrifugation (4200 g for 30 min at 4°C). The precipitated cells were resuspended in the lysis buffer [20 mM Tris/HCl (pH 8.0) and 200 mM NaCl] and lysed by sonication. Soluble fractions containing the recombinant Tse1 were obtained by centrifugation at 28500 g for 45 min and then applied to a Ni2+-chelating Sepharose affinity column (GE Healthcare) pre-equilibrated using lysis buffer. The affinity column was washed with buffer [20 mM Tris/HCl (pH 8.0), 150 mM NaCl and 10 mM imidazole]. Full-length Tse1 was eluted with elution buffer [20 mM Tris/HCl (pH 8.0), 100 mM NaCl and 200 mM imidazole]. The eluate was further loaded on to an ion-exchange column (Source 15S, GE Healthcare) and then eluted using a linear 150 ml gradient of 0–0.5 M NaCl. Finally, Tse1 was purified using a Superdex-200 size-exclusion column (GE Healthcare) in 20 mM Hepes (pH 7.5) and 100 mM NaCl. The fractions containing the purified Tse1 were collected according to protein purity, analysed by SDS/PAGE, and the final protein concentration was 10 mg/ml for cystallization. The SeMet (selenomethionine)-labelled Tse1 was expressed in E. coli BL21(DE3) cells which were grown in SeMet-containing medium through the metabolic inhibition pathway , and was purified using the same procedure to that of Tse1. SeMet-labelled Tse1 was concentrated to 10 mg/ml in buffer containing 20 mM Hepes (pH 7.5) and 100 mM NaCl for crystallization.
To purify the Tse1–Tsi1 complex, BL21(DE3) cells carrying Tse1-(6–148) and Origami (DE3) cells containing Tsi1-(20-end) were separately incubated in LB medium. The protein induction conditions were the same as that for BL21(DE3) cells carrying full-length Tse1. When the two kinds of E. coli cells were harvested by centrifugation, they were mixed together. The procedures for cell lysis and for Ni2+-chelating column binding and washing were the same as that for Tse1. After the Ni2+-affinity column was washed using wash buffer, the TEV protease was added to the resin and incubated with the Tse1–Tsi1 complex at 4°C for 12 h to cleave the N-terminal histidine tag of Tse1-(6–148). Accordingly, the Tse1–Tsi1 complex was eluted with lysis buffer. The eluate was then purified using an ion-exchange column (Source 15Q, GE Healthcare) and a sizeexclusion column (Superdex-200, GE Healthcare). The purified complex was concentrated to 8 mg/ml in buffer containing 10 mM Tris/HCl (pH 8.0) and 100 mM NaCl for crystallization. The purification procedure of Tsi1 used in the SPR (surface plasmon resonance) experiment was the same as that for the Tse1–Tsi1 complex, except for the incubation with TEV protease.
Crystallization and data collection
Crystals of Tse1, SeMet–Tse1 and Tse1-(6–148)–Tsi1-(20-end) were obtained at 20°C using the hanging-drop diffusion method. Both Tse1 and SeMet–Tse1 were crystallized by mixing equal volumes (2 μl/2 μl) of protein with reservoir solution containing 0.1 M Tris/HCl (pH 8.5) and 20% PEG [poly(ethylene glycol)] 3350. Tse1-(6–148)–Tsi1-(20-end) was crystallized by mixing equal volumes (2 μl/2 μl) of reservoir solution containing 0.1 M BisTris (pH 6.5) and 20% PEG3350. All crystal diffraction data were collected at the SSRF (Shanghai Synchrotron Radiation facility), beamline BL17u1. To prevent radiation damage, all crystals were flash frozen in a 100 K nitrogen stream in the presence of reservoir buffer containing 15% glycerol. As for Tse1 and SeMet–Tse1 crystals, the best data were obtained from a crystal of SeMet–Tse1. The crystal belongs to space group P212121 with the unit cell dimensions a=37.130 Å (1 Å=0.1 nm), b=61.028 Å, c=64.098 Å and β=90.00°. The Tse1-(6–148)–Tsi1-(20-end) crystal belongs to space group P21 with the unit cell dimensions a=48.514 Å, b=99.204 Å, c=56.333 Å and β=98.82°. All datasets were processed using the HKL2000 software package .
Structure determination and refinement
The Tse1 structure was determined using SAD (single anomalous dispersion) phasing. Three selenium atom sites were identified with the program SOLVE . The initial single anomalous dispersion phases were further improved by the program RESOLVE , and the protein chain was automatically traced using the program ARP/WARP . The atomic model was built using COOT  and refinement was carried out with PHENIX .
The structure of the Tse1-(6–148)–Tsi1-(20-end) complex was determined through molecular replacement using PHASER  from the CCP4  software package. The Tse1 structure was used as the search model. Since the occupancy ratio of the residues of Tse1 is less than 50% in the total residues of the complex, the program Oasis was carried out to optimize the phases . The atomic model of the Tse1-(6–148)–Tsi1-(20-end) complex was built using COOT and further refined with PHENIX. All data collection and structure refinement statistics about SeMet–Tse1 and Tse1-(6–148)–Tsi1-(20-end) are given in Table 1. All of the molecular graphics Figures were produced using PyMol (http://www.pymol.org).
SPR (surface plasmon resonance)
SPR experiments were performed using Biacore3000 (GE Healthcare) to measure the binding affinity of Tse1 with Tsi1. The experiment was carried out at 25°C using 20 mM Hepes (pH 7.5) and 100 mM NaCl as running buffer. Tse1 (2 μg/ml) in sodium acetate (10 mM, pH 4.0) was immobilized to the CM5 sensorchip using the Amine Coupling kit (GE Healthcare). Immobilization densities of 300 RU (resonance units; 1 RU=1 pg·mm−2) were attained. Protein-captured surfaces were subsequently blocked with 1 M ethanolamine (pH 8.5). Tsi1 protein was serially diluted (2-fold) in SPR-binding buffer and injected for a 2 min contact time at 30 μl/min and then allowed to dissociate for 4 min. The sensorchip was regenerated by 10 mM Glycine-HCl (pH 2.0) buffer. The data were analysed with BIAevaluation 4.1 software by fitting to a 1:1 binding model.
Tse1, Tsi1 and the Tse1–Tsi1 complex (full-length) were subjected to gel-filtration analysis (Superdex-200 10/300 GL column; GE Healthcare) with buffer containing 10 mM Tris/HCl (pH 8.0) and 100 mM NaCl. The assays were performed with a flow rate of 0.5 ml·min−1 and an injection volume of 0.5 ml of buffer containing Tse1, Tsi1 and the Tse1–Tsi1 complex (full-length) (approximately 1.8 mg/ml) at 4°C. The elution volumes of Tse1, Tsi1 and Tse1–Tsi1 complex (full-length) were 17.8 ml, 16.5 ml and 15.5 ml respectively. The proteins were visualized by SDS/PAGE followed by Coomassie Blue staining.
DLS (dynamic light scattering)
Experiments were carried out in a Dynapro DLS instrument (Protein Solutions). The proteins were diluted to 1 mg/ml in 10 mM Tris/HCl (pH 8.0) and 100 mM NaCl. Data were acquired at 20°C. Regularization histogram analyses of DLS results were carried out using the software DYNAMICS v.5.25.44.
Growth curve measurements
BL21(DE3) pLysS cells containing expression plasmids [pET-22b(+) with pelB signal peptide targeting the proteins into the periplasm] were grown overnight at 37°C in liquid LB medium in a shaking incubator. The harvested cells were diluted to a D600 of 0.01–0.02 to start a new incubation respectively. When the cultures has been grown for 170 min, IPTG was added to induce protein expression. To measure the growth curves, samples were taken from the induced cultures at 30 min intervals.
Recombinant Tsi1 forms a 1:1 complex with Tse1 in vitro
To facilitate the structural study of the Tse1–Tsi1 complex, full-length Tse1 and Tsi1 were purified to homogeneity. Then their interaction was studied via a gel-filtration and DLS analysis assay. Tse1 formed a stable complex with Tsi1 in solution, as indicated by the comigration of the two proteins (Figure 1A). DLS analysis indicates that Tse1 and Tsi1 are monomers and Tse1–Tsi1 is a heterodimer (Supplementary Figure S1 at http://www.BiochemJ.org/bj/448/bj4480201add.htm). The binding affinity of Tse1 and Tsi1 was measured using SPR. The SPR results showed that full-length Tse1 interacts with Tsi1 tightly with a dissociation constant (Kd) of 2.42 nM (Figure 1B).
Overall structure of Tse1
The final model of Tse1 contains one Tse1 molecule (residues 2–154) in the asymmetric unit. Tse1 shows an α+β fold with five α-helices and four β-strands. The four antiparallel strands β1–β4 produce a central β-sheet. Helices α1– α 3 and α4–α5 locate at each side of the β-sheet respectively (Figures 2A and 2B). Structural alignment using the DALI server shows that Tse1 shares a similar structural fold with several NlpC/P60 family proteins in spite of low sequence identity, including YkfC from Bacillus cereus (PDB code 3H41), Spr from E. coli (PDB code 2K1G) and AvPCP from Anabaena variabilis (PDB code 2HBW) [DALI Z scores of 9.7 for 3H41, 8.8 for 2K1G and 8.2 for 2HBW; Cα RMSDs (root mean square deviations) of 2.9 Å for 3H41, 3.0 Å for 2K1G and 3.2 Å for 2HBW]. In the structure of Tse1, one extended long groove occurred on the surface of the protein (Figure 2C), which is supposed to be the active-site cleft by which the whole protein is divided into two subdomains (the right and left domain) (Figures 2D and 2E). The left domain consists of α1, α2 and β3, whereas α3, α4, α5, β1, β2 and β4 form the right domain. One disulfide bond (Cys7–Cys148) locates at the bottom of the protein, cross-linking these two subdomains together. This type of domain constitution is widely spread in the cysteine peptidases . The right subdomain is also divided by a short groove located between the β1β2 loop and the α5β4 loop and is vertical to the β3α5 loop (Figure 2E). These three loops appear like the petals, and the two loops (α1α2 loop and β2β3 loop) look like the calyces. Together they form the flower-like structure, and the two grooves form a Y-shaped active-site cleft (Figures 2B and 2C). Unlike the V-shaped cleft of the other papain-like proteins which binds the extended peptide chain of substrates, Tse1 presumably uses this type of groove to accommodate a peptidoglycan mesh (glycan and peptide side chains).
Active sites and catalytic centre
The catalytic residues of the reaction centre were identified by superimposing the structure of Tse1 with Spr from E. coli (PDB code 2K1G) and with AvPCP from A. variabilis (PDB code 2HBW). Cys30 and His91 superimposed well with the corresponding residues at the catalytic reaction centre of these two proteins (Figure 3). Cys30 resides in the N-terminus of helix α2 in the left domain, and His91 occurs in β2 located in the right domain (Figures 2D and 2E). Mutagenesis studies showed that the growth of E. coli strain BL21(DE3)pLysS was not inhibited by Tse1 mutants with substitution of these two residues expressed in periplasm. The result indicated that point mutation of these two residues caused dysfunction of Tse1 since they lose their amidase activity to damage the peptidoglycan of host cells (Supplementary Figure S2 at http://www.BiochemJ.org/bj/448/bj4480201add.htm). The results of the present study are also consistent with previous research . The third residue in Tse1 is Cys110 orienting from β3, instead of the predicted Gln103 which occurs on the surface of Tse1. Cysteine, as the third residue of the catalytic centre in the protease, has not been reported previously. Site-directed mutagenesis was performed to test the function of Cys110. Our data show that when Cys110 was mutated to alanine, which cannot form hydrogen bonds with His91, the catalytic activity was not obviously affected. Thus this cysteine residue is not absolutely necessary for the catalytic activity of Tse1. The hydrogen bond length between the thiol group of Cys110 and the Nδ1 of His91 is 3.53Å, much weaker than the other type of hydrogen bond. In this regard, Cys110 may not efficiently orient the imidazole ring during catalysis. Hence the function of the corresponding cysteine residue in the Tse1 reaction centre needs to be rethought. In fact, this structural catalytic triad can be regarded as a functional catalytic diad.
The Y-shaped substrate-binding site is divided into two parts (V-shaped at the top and I-shaped at the bottom, i.e. large pocket and small pocket respectively) by Tyr89 from the β1β2 loop and Ile113 from the α5β4 loop (Figure 2C). These two residues reside directly above the catalytic Cys30. This type of configuration is similar to the Trp116 of AvPCP, which is considered to play an important role in releasing products . Presumably they serve as a switch that regulates the binding and release of the substrate in Tse1.
Overall structure of the Tse1–Tsi1 complex
The final model of the Tse1–Tsi1 complex contains four protein molecules (two Tse1 molecules and two Tsi1 molecules) in the asymmetric unit, which forms two Tse1–Tsi1 heterodimeric complexes (Figure 4). The Tse1 molecule in the Tse1–Tsi1 complex includes residues 6–148 of the entire Tse1, and Tsi1 contains residues from 21 to the end (Figure 5A). Superimposition of two Tse1 monomers from Tse1 and the Tse1–Tsi1 complex gives an RMSD of approximately 0.212 Å for all corresponding Cα atoms. Structural variations mainly occurred in the α1α2 loop, β1β2 loop, β3α5 loop and α5β4 loop, all of which participate in the formation of the protein interface between Tse1 and Tsi1. The three flower petals, the β1β2 loop, β3α5 loop and α5β4 loop, move outwards when one middle finger (consisting of βI+HI-loop+βH) protrudes from Tsi1 and the α1α2 loop moves towards Tsi1 (Supplementary Figure S3 at http://www.BiochemJ.org/bj/448/bj4480201add.htm).
The structure of Tsi1 in the complex contains 13 antiparallel β-strands (A–M), forming three β-sheets (left, middle and right sheets) (Figure 5B). The middle β-sheet, consisting of βA and βF–βJ, is surrounded by the two other β-sheets comprising βB–βE (left sheet) and βK–βM (right sheet) respectively. Three disulfide bonds appear in Tsi1, including Cys22–Cys167 cross-linking the N- and C-terminus of Tsi1, Cys79–Cys121 linking the βE and IG loops, and Cys147–Cys155 linking βL and βM. These disulfide bonds hold these structural elements together and contribute to the stabilization of Tsi1. A DALI search using Tsi1 revealed several structural homologues, with the β-propeller domain of the PEP (prolyl endopeptidase)  (PDB code 1IUN, Z score 7.3) and the SLP (surface layer protein)  (PDB code 1L0Q, Z score 7.2) as the top two hits. However, Tsi1 is different from the β-propeller domain. First, Tsi1 is only half of the circle (Figure 5C), whereas β-propeller structures have a highly symmetrical structure with circular configuration (Figure 5D). Secondly, the β-propeller has four to ten repeats of a four-stranded antiparallel β-sheet motif. Tsi1 has three antiparallel β-sheet motifs, and only one β-sheet (left sheet) adopts a stringent four-stranded simple antiparallel topology. At the middle β-sheet, the topological connectivity is more complex, with five-stranded blades (the F and G β-strands were considered as one). A strand exchange was observed between the left β-sheet, which is usually seen in the structural termini as part of the velcro closure helping to circularize the propeller . The right β-sheet has only three β-strands (Figure 5E).
Tse1–Tsi1 protein interface
Protein interface analysis with the PISA program  in the CCP4i software package showed that the interface area between monomers Tse1 and Tsi1 is 1000.2 Å2, which covers approximately 13.8% of the solvent-accessible surface area (7272.4 Å2) of Tse1 and 12.3% of the solvent-accessible surface area (8121.7 Å2) of Tsi1.
The protein surface of Tse1 surrounding the Y-shaped groove is extensively negative charged and interacts tightly with the Tsi1 interface with a broad positively charged area (Supplementary Figure S4 at http://www.BiochemJ.org/bj/448/bj4480201add.htm). The dimer interface of the Tse1-(6–148)–Tsi1-(20-end) complex primarily consists of the α1α2 loop, β1β2 loop, β2β3 loop, β4α5 loop, and α5 and α5β5 loop of the Tse1, which were packed against the CD loop, EF loop, βF loop, HI loop, JK loop and LM loop of the Tsi1. An extensive network of hydrogen bonds, van der waals forces, and water-mediated hydrogen bonds occurred at the dimer interface of the Tse1–Tsi1 complex. A total of 14 Tse1 residues (Asp28, Cys30, His91, Tyr101, Arg102, Ser112, Ile113, Ala116, Lys124, Gln128, Val129, Asn131, Asp134 and Arg135) and ten Tsi1 residues (Glu53, Asp54, Asp61, Glu65, Gly85, Ser107, Gly108, Ser109, Arg133 and Glu150) are involved in the hydrogen bond network. Meanwhile, several hydrophobic interactions formed by three Tse1 residues (Val117, Val129 and Trp130) and three Tsi1 residues (Ile86, Phe106 and Gly108) also contribute to the formation of the Tse1–Tsi1 complex (Figure 4).
The mechanism of Tsi1 inhibits Tse1
On the basis of the crystal structure of the Tse1-(6–148)–Tsi1-(20-end) complex, Tsi1 using the HI loop, which adopts a middle finger tip-like structure, inserts into the large pockets of the Y-shaped groove in the subsites of Tse1. At the same time, Tsi1 also employs the CD loop and EF loop on one side, and the JK loop and LM loop on the other side mimicking the rest of the four finger tips to contact the three petals (β1β2 loop, β3α5 loop and α5β4 loop) and the two calyces (α1α2 loop and β2β3 loop) of Tse1. Thus this type of architecture can prevent the peptide chain of peptidoglycan from entering into the Y-shaped groove (Figure 4A). The HI loop is a β-hairpin structure and consists of Ser107, Gly108, Ser109 and Ala110. Ser107 is hydrogen-bonded with Asp134 and Asn131 from Tse1. The N-amide of Gly108 is hydrogen-bonded with the carbonyl group of Val129. The carbonyl group in Gly108 is hydrogen-bonded with the hydroxy group of Ser112 from Tse1 (Figures 4B and 4D). Interestingly, the hydroxy group of Ser109 of Tsi1 directly forms hydrogen bonds with the Nϵ2 of His91 of Tse1 with a bond length of 2.73Å, as well as with the thiol group of Cys30 with a bond length of 3.5Å (Figure 6). This type of hydrogen-bond network may directly inactivate the reaction centre.
The structure of Tse1 reveals that it is a member of the NlpC/P60 family, with an evolutionarily conserved catalytic centre and a primitive structural core. The NlpC/P60 family belongs to the CHAP (cysteine, histidine-dependent amidohydrolases/peptidases) superfamily , playing important roles in cell division and pathological processes. The conserved Cys30 is located at the N-terminal of the second helix α2, followed by a C-terminal conserved region with a characterized His91 residing in the β3. In between there is also a conserved Gly67. This type of constellation is the marker for the classical NlpC/P60 family and indicates that Tse1 is a new member of this protein family . However, Tse1 has a unique characteristic which distinguishes it from other family members. First, it has four anti-parallel β strands compared with five in the NlpC/P60 family. Instead of consisting of the core structure of the protein, the fifth β strand in Tse1 becomes a long loop residing in petal 2 which interacts with Tsi1 in bacterial self-protection (Figure 2B). Secondly, the third residue orienting the catalytic histidine residue is a cysteine residue which has not been observed in this protein family before, although the function of cysteine seems to be dispensable. Thus the reaction centre in Tse1 becomes a catalytic diad, compared with the classical catalytic triad in NlpC/P60 family. Thirdly, there is an insertion which forms one helix following the conserved Gly67 (Figure 3C), which is not observed in the rest of the family members. These characteristics indicate that Tse1 originated from the NlpC/P60 family and acquired certain mutations and insertions during evolution. The family members usually fuse with auxiliary domains, such as the SH3 (Src homology 3), LysM (lysine motif) and choline-binding domains, which are suggested to be the targeting domains localizing them to the cell wall for their physiological purpose [39,40]. Tse1 is considered to be a toxic component for the rival cells, and has no other regulative domains which may regulate their activity at different spatial and temporal terms, and, therefore, it is more promiscuous and destructive.
The structure of Tsi1 is a whole β-strands fold, and three β-sheets constitute half of the β-propeller-like form. Tsi1 shares little sequence similarity with the papain-like protease inhibitor of the cystatin superfamily. The cystatin superfamily consists of stefin, cystatin, kininogen and cystatin-related protease inhibitors . Unlike kininogen, Tsi1 is not a glycoprotein. Tsi1 is rich in disulfide bonds and does not contain the conserved QXVXG region in the central part of the molecule and the proline–tryptophan pair in the C-terminal part of cystatins . The Tsi1 fold is different from the cystatin fold which is based on a five-stranded antiparallel β-sheet wrapped around the central N-terminal helix . It is also different from P41 which is homologous with the thyroglobulin type-1 domain fold  chagasin, which is similar to the immologlobulin fold , and staphostatin which displays a β-barrel structure . Five loops (CD loop, EF loop, HI loop, JK loop and LM loop) of Tsi1 mimic five finger tips. The middle finger tip (the HI loop) inserts into the large pocket of Tse1 and, meanwhile, Ser109 from the HI loop directly interacts with reaction centre (Cys30 and His91), whereas the rest of the four finger tips touch the three pedals and the two calyces of flower-like Tse1. This type of binding mode is unique to the cysteine peptidase inhibitor family. In contrast with cystatins, which are non-specific inhibitors of cysteine proteases, Tsi1 targets to Tse1 specifically and should not inhibit other members of the NlpC/P60 family existing in the periplasm of P. aeruginosa. Because Tsi1 is expressed in its own periplasm, if Tsi1 can inhibit other NlpC/P60 family members which play important role in peptidoglycan recycling, it may interfere with cell cycles. The contrast between the Tse1–Tsi1 complex and cystatin–peptidase (take stefin A–cathepsin H for example) complexes in binding specificity could be illustrated from two points of view. First, 12 out of the 14 residues involved in the hydrogen-bond network between Tse1 and Tsi1 are not conserved in the alignment with its closely related NlpC/P60 family members (Figure 3C). The hydrogen-bond network is mainly formed between the 14 residues of Tse1 and the five loops of Tsi1 (Figure 4B). The only two conserved residues are catalytic residues (Cys30 and His91). The non-conservation of the residues forming the binding interface leads to a specific interaction with Tsi1. In the stefin A–cathepsin H complex, five residues from cathepsin H are involved in the intermolecular hydrogen-bond network. These five residues are highly conserved in the cathepsin family and papain. In addition, many other residues in their interface are also conserved (Supplementary Figure S5 at http://www.BiochemJ.org/bj/448/bj4480201add.htm). Consequently, stefin A is able to inhibit many members of papain-like-fold family. The second aspect is from their interface configuration. The shape of the active cleft of Tse1 is different from its closely related members. Tse1 adopts a deep and long Y-shaped active-site groove, but other NlpC/P60 members (such as Spr and Ykfc) show a narrow and shallow groove, and usually their fusion domains contribute to formation of their substrate-binding sites. With the unique structural feature of five loops, Tsi1 cannot only match the binding surface of Tse1, but also distinguish Tse1 from other NlpC/P60 family members. The half β-propeller-like Tsi1 can only insert into the large pocket of the Y-shaped groove of Tse1, as is shown in Sumpplementary Figure S4. Papain-like peptidases (such as cathepsins and papain) all exhibit long and extended V-shaped clefts which divides the protein into two parts, left and right domains, so that their wedge-shaped inhibitors, which are wide at one end and narrow at the other end, could insert into clefts and interact with evolutionarily conserved residues which reside in the clefts. So the interaction between papain-like peptidases (such as cathespins) and their inhibitors is not highly specific. In brief, we can compare Tsi1 to the key of the Tse1 and stefin to the skeleton key of papain-like peptidases (like cathespins).
The inhibitory mechanism for Tse1 by Tsi1 is slightly different from the cystatin family. Cystatin adopts a wedge-shaped structure using its wedged edge to contact the V-shaped active sites of the protease. There are three parts (N-terminal trunk and two hairpin loops) in cystatin involved in their interaction with peptidase . The N-terminal trunk displays a substrate-like manner binding in the active site and the conserved QXVXG positioned in the first hairpin binds near to the catalytic cysteine residue. Although P41 and chagasin have a different fold than cystatin, they also adopt the same three-part interaction manner as in cystatin, which is considered to be a converged evolution for cysteine protease inhibitors. Unlike this three finger tips interaction manner, Tsi1 uses a five finger tips contacting manner to bind Tse1. The middle finger tip (HI loop) directly inserts into the subsite of Tse1 and interacts with the reaction centre residues His91 and Cys30 and, thus, it is similar to the first hairpin in the cystatin family. The JK loop (ring finger tip) and LM loop (little finger tip) are comparable with the N-terminal elephant trunk in cystatin which adopts a substrate-like manner binding into the active site of papain-like cysteine. Instead of binding into the active site, these two loops interact with the α1α2 loop, β1β2 loop and β3α5 loop directly and indirectly, and are above the active centre. The CD loop (thumb tip) and ED loop (index finger tip) is like the second hairpin in the cystatin family and contacts the β2β3 loop, α5 loop and α5β4 loop. In this respect, this type of five finger tips inhibitory mode can be considered to be a variation of the classical three finger tips interaction mode, although some differences could be observed (Figure 7). In this regard, it is intriguing to suggest that the cysteine peptidase inhibitory mechanism displays a convergent evolution. The equipment of a five-finger is an advantage for Tsi1, since Tsi1 could use these fingers to enhance its specificity during the interaction with Tse1.
Guijun Shang, Xiuhua Liu, Defen Lu and Lichuan Gu designed the research; Guijun Shang, Xiuhua Liu, Defen Lu, Junbing Zhang, Ning Li, Chunyuan Zhu, Shiheng Liu, Qian Yu, Yanyu Zhao, Heqiao Zhang and Junqiang Hu performed the experiments; Guijun Shang, Xiuhua Liu, Defen Lu, Huaixing Cang, Sujuan Xu and Lichuan Gu analysed data and wrote the paper; all authors contributed to editing the paper before submission.
This work was supported by the State Key Laboratory of Microbial Technology, Shandong University and the Hi-Tech Research and Development Program of China [grant number 2006AA02A324].
We thank the staff at beamline BL17u1 at the Shanghai Synchrotron Radiation facility for support with data collection and Zheng Fan (Institute of Microbiology, Chinese Academy of Sciences) for assistance in Biacore experiments.
The atomic co-ordinates and structure factors have been deposited in the PDB with the accession codes 4EQ8 for Tse1 and 4EQA for the Tse1-(6–148)–Tsi1-(20-end) complex.
Abbreviations: DLS, dynamic light scattering; IPTG, isopropyl β-D-thiogalactopyranoside; LB, Luria–Bertani; PEG, poly(ethylene glycol); RMSD, root mean square deviation; SeMet, selenomethionine; SLP, surface layer protein; SPR, surface plasmon resonance; TEV, tobacco etch virus; Tse, type VI secretion exported; Tsi, type VI secretion immunity; T1SS etc., type I secretion system etc
- © The Authors Journal compilation © 2012 Biochemical Society