Activity and Crystal Structure of the Adherent-Invasive Escherichia coli Tle3/Tli3 T6SS Effector/Immunity Complex Determined Using an AlphaFold2 Predicted Model

The type VI secretion system (T6SS) delivers enzymatic effectors into target cells to destroy them. Cells of the same strain protect themselves against effectors with immunity proteins that specifically inhibit effectors. Here, we report the identification and characterization of a Tle3 phospholipase effector and its cognate immunity protein Tli3—an outer membrane lipoprotein from adherent-invasive Escherichia coli (AIEC). Enzymatic assays demonstrate that purified Tle3AIEC has a phospholipase A1, and not A2, activity and that its toxicity is neutralized by the cognate immunity protein Tli3AIEC. Tli3AIEC binds Tle3 in a 1:1 stoichiometric ratio. Tle3AIEC, Tli3AIEC and the Tle3AIEC-Tli3AIEC complex were purified and subjected to crystallization. The Tle3AIEC-Tli3AIEC complex structure could not be solved by SeMet phasing, but only by molecular replacement when using an AlphaFold2 prediction model. Tle3AIEC exhibits an α/β-hydrolase fold decorated by two protruding segments, including a N-terminus loop. Tli3AIEC displays a new fold of three stacked β-sheets and a protruding loop that inserts in Tle3AIECcatalytic crevice. We showed, experimentally, that Tle3AIEC interacts with the VgrG AIEC cargo protein and AlphaFold2 prediction of the VgrGAIEC-Tle3AIEC complex reveals a strong interaction between the VgrGAIEC C-terminus adaptor and Tle3AIEC N-terminal loop.


Introduction
The type VI secretion system (T6SS) is a macromolecular machine anchored in the envelope of Gram-negative bacteria. This nanomachine allows the injection of effectors or toxins into eukaryotic or bacterial target cells and is involved in virulence and antibacterial competition. T6SS + cells that deliver antibacterial effectors also produce specific immunity proteins to protect themselves from the toxicity of these effectors.
The subunits required to assemble a functional T6SS machine are usually encoded within gene clusters, together with additional toxin and immunity genes. These subunits assemble a membrane complex embedded in the envelope that anchors a cytoplasmic contractile tail structure [1][2][3][4]. The contractile tail is an internal tube made of Hcp protein hexamers, surrounded by a sheath-like structure and tipped by a needle-spike. The spike is built by a trimeric protein named VgrG, topped by a conical protein named PAAR [5,6]. The contractile tail assembles on a baseplate platform in an elongated conformation [7][8][9]. Upon contact with the target bacterium, the contraction of the sheath outwardly propels the tube and spike directly into the target cell. Because the toxins can be fused to or interact with the Hcp tube component or the VgrG/PAAR spike proteins, the contraction of the sheath leads to the secretion and delivery of the effectors into the target cell.
Antibacterial toxins' injection into the cytoplasm or the periplasm of target bacteria leads to bacterial cell lysis or growth inhibition. Cytoplasmic-acting antibacterial effectors include DNases, deaminases, NAD(P) + glycohydrolase toxins, p(p)pApp synthetase and post-translational modification enzymes, such as ADP-ribosyltransferases [10,11]. Periplasmic-acting antibacterial effectors include enzymes degrading the peptidoglycan, such as amidases or glycosides hydrolases, or toxins targeting the membrane, such as pore-forming toxins or phospholipases [10,11].
We previously characterized Tle1 and Tli1, a toxin/immunity couple of enteroaggregative Escherichia coli (EAEC) [17]. Tle1 is a periplasmic-acting toxin with phospholipase A1/A2 activity that mediates the antibacterial activity of EAEC T6SS-1. Tle1 EAEC activity is counteracted by Tli1 EAEC , a small outer membrane lipoprotein encoded downstream from the tle1 gene. We further showed that the delivery of Tle1 EAEC into bacteria requires a direct interaction of Tle1 EAEC with the VgrG1 protein [17]. The EAEC VgrG1 is comprised of a gp27-like base, gp5 β-helix and a C-terminal extension with a transthyretin (TTR) fold. We recently determined the high-resolution structure of the VgrG1/Tle1 complex by cryo-EM, demonstrating that Tle1 EAEC binds to the VgrG1 TTR domain and that the interaction is stabilized by additional contacts with the VgrG β-helix and base [18].
The presence of phospholipase/immunity genes is a conserved feature of T6SS-1 clusters in E. coli strains [19]. However, these phospholipase effectors belong to different families (Tle1 for EAEC, Tle4 for avian pathogenic E. coli and a predicted Tle3 for uropathogenic E. coli CFT073 or adherent-invasive E. coli (AIEC) [12,17,19,20]). AIEC is implicated as a determinant of human Crohn's disease (CD) development, as ileal gut mucosa colonization by AIEC is often observed in CD patients [21,22].
Here, we biochemically and structurally characterized Tle3 from AIEC and its cognate Tli3 immunity protein. We showed that Tle3 AIEC has PLA1 activity that is specifically counteracted by the Tli3 AIEC immunity protein. We solved the structure of the Tle3 AIEC /Tli3 AIEC complex by X-ray crystallography, with the help of Alphafold2 structure predictions [23,24]. Finally, a model of a Tle3 AIEC /VgrG AIEC complex was proposed.

Genomic Context and Bioinformatic Analyses of AIEC T6SS-1 Phospholipase
Effector/Immunity E. coli T6SS-1 clusters encode similar machines, but different effector/immunity pairs are encoded in between the vgrG and PAAR genes. The AIEC reference strain LF82 encodes a putative Tli3/Tle3 immunity/toxin couple ( Figure 1A). The tle3 AIEC gene (LF82_435, NCBI Gene Identifier (GI): 222034514) is predicted to encode a 658-aminoacid (73.9 kDa) cytoplasmic protein identical to the uncharacterized c3395 protein from E. coli CFT073, classified by Russell et al. as a member of the Tle3 family of T6SS effectors [12]. Alignment of Tle3 members reveals the presence of a conserved XHSQG motif (G/S/C/A-H-S-Q-G motif in Tle3) ( Figure S1).
In T6SS, protection against kin cells that produce toxic effectors is ensured by the production of immunity proteins that specifically bind and inhibit their corresponding toxin. Effector/immunity genes are found in tandem in genomes, and immunity proteins of Tle effectors, named Tli (Type VI lipase immunity), are usually periplasmic proteins or lipoproteins, as Tle are periplasmic-acting toxins. Computer analysis of the LF82_434 gene product (GI: 222034513) using SignalP 6.0 suggests that LF82_434 is a lipoprotein with a signal sequence bearing a characteristic lipobox motif with an isoleucine as +2 residue, suggesting an outer membrane localization ( Figure S2; [25]). A fractionation experiment confirmed that LF82_434 co-localizes with the membrane fraction ( Figure S2). Its localization and genomic context close to tle3 suggest it could correspond to the Tle3 immunity protein; therefore, it was named Tli3. and LF82_435 genes encoding Tli3 AIEC and Tle3 AIEC are indicated in green and blue, respectively. The clpV, vgrG and PAAR T6SS-1 genes are indicated in grey. (B) The SDS-PAGE analysis of the purified Tle3 AIEC and Tli3 AIEC proteins (10 µ g) after Coomassie blue staining. Molecular weight markers from Euromedex (in kDa) are indicated on the left. (C) Specific phospholipase A1 (PLA1) and A2 (PLA2) activity measurements of the Tle3 AIEC protein using fluorescent phospholipids. Specific activities were calculated from the velocity slope obtained for 25 min using 20 µ g of purified protein. Data are expressed as mean of three, from three independent assays (CV% < 5%). (D) Inhibition studies with Tli3 AIEC were performed by incubating 30 µ g of Tle3 AIEC with various molar ratio of Tli3 AIEC (xi = 0, 0.25, 0.5, 1 and 2). The rate of hydrolysis of PED-A1 by purified Tle3 AIEC at 20 °C in presence of increasing concentrations of Tli3 AIEC was plotted against the molar excess of Tli3. (E) Cross-inhibition studies with Tli1 EAEC were performed by incubating 20 µ g of Tle3 AIEC ( ) with a molar ratio of The SDS-PAGE analysis of the purified Tle3 AIEC and Tli3 AIEC proteins (10 µg) after Coomassie blue staining. Molecular weight markers from Euromedex (in kDa) are indicated on the left. (C) Specific phospholipase A1 (PLA1) and A2 (PLA2) activity measurements of the Tle3 AIEC protein using fluorescent phospholipids. Specific activities were calculated from the velocity slope obtained for 25 min using 20 µg of purified protein. Data are expressed as mean of three, from three independent assays (CV% < 5%). (D) Inhibition studies with Tli3 AIEC were performed by incubating 30 µg of Tle3 AIEC with various molar ratio of Tli3 AIEC (x i = 0, 0.25, 0.5, 1 and 2). The rate of hydrolysis of PED-A1 by purified Tle3 AIEC at 20 • C in presence of increasing concentrations of Tli3 AIEC was plotted against the molar excess of Tli3. (E) Cross-inhibition studies with Tli1 EAEC were performed by incubating 20 µg of Tle3 AIEC ( cytoplasmic protein identical to the uncharacterized c3395 protein from E. coli CF classified by Russell et al. as a member of the Tle3 family of T6SS effectors [12]. Align of Tle3 members reveals the presence of a conserved XHSQG motif (G/S/C/A-H-S motif in Tle3) ( Figure S1).
In T6SS, protection against kin cells that produce toxic effectors is ensured b production of immunity proteins that specifically bind and inhibit their correspon toxin. Effector/immunity genes are found in tandem in genomes, and immunity pro of Tle effectors, named Tli (Type VI lipase immunity), are usually periplasmic protei lipoproteins, as Tle are periplasmic-acting toxins. Computer analysis of the LF82_434 product (GI: 222034513) using SignalP 6.0 suggests that LF82_434 is a lipoprotein w signal sequence bearing a characteristic lipobox motif with an isoleucine as +2 res suggesting an outer membrane localization ( Figure S2; [25]). A fractionation experi confirmed that LF82_434 co-localizes with the membrane fraction ( Figure S2). Its loca tion and genomic context close to tle3 suggest it could correspond to the Tle3 imm protein; therefore, it was named Tli3. of 10, 20 µg of Tle3 AIEC alone ( ) or incubated with Tli3 AIEC ( ) at molar ratio of 2, were us controls. (F) Cytotoxic activities of Tle1 EAEC and Tle3 AIEC towards eukaryotic cells. Cytotoxici periments were carried with approximately 1 × 10 5 RAW264.7 cells/well pulsed with Tle1 EAE Tle3 AIEC recombinant proteins (30 µg) incubated for 16 h. Tle1 EAEC phospholipase A2 and lysis b containing 1% Triton X-100 were used as positive controls. Alternatively, cells were incubated medium containing the vehicle buffer or the Tle1 EAEC -Tli1 EAEC complex as additional negative trols. Viability was evaluated using resazurin assay. Results are displayed as box-plots with viduals' raw data. Black dots were added to highlight the mean of each condition. Results ar resentative from n = 2 biologically independent experiments performed in technical quadrupli Statistical analysis was performed using Wilcoxon signed-rank test, where vehicle treatmen used as reference condition. Exact adjusted p-value are indicated on the graph and were consi significant when p ≤ 0.05.
The AIEC Tle3 protein, Tle3 AIEC , was purified ( Figure 1B), and its activity was tested u fluorogenic PLA1-and PLA2-specific phospholipid substrates. As shown in Figure   ) or incubated with Tli3 AIEC ( Int. J. Mol. Sci. 2023, 24, x FOR PEER REVIEW 4 of 10, 20 µg of Tle3 AIEC alone ( ) or incubated with Tli3 AIEC ( ) at molar ratio of 2, were us controls. (F) Cytotoxic activities of Tle1 EAEC and Tle3 AIEC towards eukaryotic cells. Cytotoxicit periments were carried with approximately 1 × 10 5 RAW264.7 cells/well pulsed with Tle1 EAEC Tle3 AIEC recombinant proteins (30 µg) incubated for 16 h. Tle1 EAEC phospholipase A2 and lysis b containing 1% Triton X-100 were used as positive controls. Alternatively, cells were incubated medium containing the vehicle buffer or the Tle1 EAEC -Tli1 EAEC complex as additional negative trols. Viability was evaluated using resazurin assay. Results are displayed as box-plots with viduals' raw data. Black dots were added to highlight the mean of each condition. Results are resentative from n = 2 biologically independent experiments performed in technical quadruplic Statistical analysis was performed using Wilcoxon signed-rank test, where vehicle treatmen used as reference condition. Exact adjusted p-value are indicated on the graph and were consid significant when p ≤ 0.05.
The AIEC Tle3 protein, Tle3 AIEC , was purified ( Figure 1B), and its activity was tested u fluorogenic PLA1-and PLA2-specific phospholipid substrates. As shown in Figure   ) at molar ratio of 2, were used as controls. (F) Cytotoxic activities of Tle1 EAEC and Tle3 AIEC towards eukaryotic cells. Cytotoxicity experiments were carried with approximately 1 × 10 5 RAW264.7 cells/well pulsed with Tle1 EAEC and Tle3 AIEC recombinant proteins (30 µg) incubated for 16 h. Tle1 EAEC phospholipase A2 and lysis buffer containing 1% Triton X-100 were used as positive controls. Alternatively, cells were incubated with medium containing the vehicle buffer or the Tle1 EAEC -Tli1 EAEC complex as additional negative controls. Viability was evaluated using resazurin assay. Results are displayed as box-plots with individuals' raw data. Black dots were added to highlight the mean of each condition. Results are representative from n = 2 biologically independent experiments performed in technical quadruplicates. Statistical analysis was performed using Wilcoxon signed-rank test, where vehicle treatment was used as reference condition. Exact adjusted p-value are indicated on the graph and were considered significant when p ≤ 0.05.
In T6SS, protection against kin cells that produce toxic effectors is ensured by the production of immunity proteins that specifically bind and inhibit their corresponding toxin. Effector/immunity genes are found in tandem in genomes, and immunity proteins of Tle effectors, named Tli (Type VI lipase immunity), are usually periplasmic proteins or lipoproteins, as Tle are periplasmic-acting toxins. Computer analysis of the LF82_434 gene product (GI: 222034513) using SignalP 6.0 suggests that LF82_434 is a lipoprotein with a signal sequence bearing a characteristic lipobox motif with an isoleucine as +2 residue, suggesting an outer membrane localization ( Figure S2; [25]). A fractionation experiment confirmed that LF82_434 co-localizes with the membrane fraction ( Figure S2). Its localization and genomic context close to tle3 suggest it could correspond to the Tle3 immunity protein; therefore, it was named Tli3.
The Tli3 AIEC putative immunity, devoid of is signal sequence and N-terminal cysteine residue, was purified ( Figure 1B), and tested for its ability to interfere with Tle3 AIEC activity. Thirty micrograms of Tle3 AIEC were incubated with various molar ratios of Tli3 AIEC (xi = 0, 0.25, 0.5, 1 and 2), and the activity was measured immediately or after 30 min incubation at room temperature. Figure 1D shows that the Tle3 AIEC phospholipase activity was inhibited by Tli3 AIEC in a dose-dependent manner. Interestingly, the Tle3 AIEC activity was nearly abolished with a Tle3 AIEC :Tli3 AIEC molecular ratio of 1:2 ( Figure 1D). Tli3 AIEC is thus likely to interact strongly with Tle3 AIEC , probably inside or in a close vicinity to the active site, since no substrate can have access to the active site to be hydrolyzed. These results indicate that Tli3 AIEC is an immunity protein that inhibits the phospholipase activity of Tle3 AIEC .
Because AIEC is a human pathogen, we then investigated whether Tle3 AIEC is cytotoxic in mammalian cells ( Figure 1F). Many bacterial enzymes presenting phospholipase A2/A1 activity possess cytotoxic activity against mammalian cells [26,27]. In order to estimate the effect of Tle3 AIEC on a RAW 264.7 cell, we used the well-established resazurin-based cell viability assay. As controls, lysis buffer containing Triton X-100 leads to 100% cellular lysis while the untreated and buffer-only conditions maintained cell viability. The addition of the Tle3 AIEC PLA1 did not show any significant cytotoxicity (~15%) towards mouse macrophages. In contrast, the incubation with the Tle1 EAEC PLA1/PLA2 led to~50% lysis of the RAW 264.7 cells in 16 h. This cytotoxic effect can be inhibited by preincubating Tle1 EAEC with Tli1 EAEC , its cognate and specific immunity. Our results are in agreement with the literature, supporting the idea that the PLA2 activity is required for a cytotoxic effect towards mammalian cells [28]. Finally, these results confirm the biochemical properties of Tle3 AIEC as a PLA1 enzyme.

Tle3 AIEC -Tli3 AIEC Heterodimer Crystallization and Data Collection
The purified Tle3 AIEC and Tli3 AIEC proteins were mixed with a molar ratio of 1:1.3. The complex was then collected by gel filtration chromatography (Supplementary Figure S3). The Tle3 AIEC -Tli3 AIEC complex has an apparent molecular mass of 100 kDa (Supplementary Figure S3C). Considering its theoretical molecular weight of 101.9 kDa, it is likely to be a 1:1 Tle3 AIEC -Tli3 AIEC heterodimer in solution.
Highly purified protein fractions were pooled and concentrated to 5.6 mg/mL. The Tle3 AIEC -Tli3 AIEC SeMet purified complex was then submitted to crystallization. Sharp crystals appeared within 10 days to 3 weeks, reaching maximum dimensions of 100 × 40 × 40 µm within 2 months. The best crystals were exposed at the Soleil synchrotron (Saint-Aubin, France) and diffracted to 3.6-Å resolution ( Table 1). Tle3 AIEC -Tli3 AIEC crystals belong to space group P21, with unit-cell parameters a = 67.9 Å, b = 449.1 Å, c = 116.2 Å and β = 91.2 • ( Table 1). The crystals exhibit a Vm of 17.4 Å 3 /Da for 1 complex per asymmetric unit. Based on the molecular weight of 101,9 kDa, a total of 4 to 8 complexes per asymmetric unit should be expected, as this would lead to more reasonable values of Vm, in the range of 4.35-2.17 Å 3 /Da, and solvent contents of 72-43%. However, due to the low redundancy of the collected data set, data resolution, large cell and 4-8 complexes of~900 amino acids in the asymmetric unit, the structure could not be solved by SeMet phasing nor with molecular replacement, and the project was abandoned for 6 years.

Crystal Structure Determination Using Molecular Replacement with AlphaFold2 Models
Recently (15 July 2021) AlphaFold2 (AF2) [23], a highly efficient protein structure prediction program, was released. With the idea to resuscitate the project, we submitted the sequences of Tle3 AIEC and Tli3 AIEC to AlphaFold2 and obtained models with satisfying statistics (pLDDT > 80; Supplementary Figure S4). However, since some loops or stretches of residues were poorly predicted, we removed them from the starting models used for molecular replacement (Supplementary Figure S5). Molecular replacement with Molrep [29] using the AF2 "cleaned" model yielded four Tle3 AIEC molecules in a first pass. This starting model was refined with Phenix [30] and a second pass of molecular replacement yielded three more molecules. After refinement, the seven-molecule ensemble was used as the fixed model for molecular replacement with Tli3 AIEC , yielding two Tli3 AIEC molecules. The Tle3 AIEC -Tli3 AIEC complex was substituted on the six remaining Tle3 AIEC , and the seven complexes were subjected to refinement with Phenix. After cycles of model building with Coot [31] and Phenix refinement [30], an eight-complex ensemble was identified and a new cycle of refinement was performed. Finally, the 8 complexes exhibited a reasonable geometry for a 3.6 Å resolution with R and R free factors of 23.4% and 26.8%, respectively (Table 1).

Structure of Tle3 AIEC and Tli3 AIEC
The Tle3 AIEC phospholipase chain is 658 residues long, of which 525, on average, have been identified in the electron density map. Tle3 AIEC structure is bulky, with overall dimensions of~75 Å × 65 Å × 55 Å (Figure 2A,B). Stretches in the amino acid chain are lacking at the N-and C-termini, as well as in some more or less extended loops. Tle3 AIEC shares the classical core of an α/β-hydrolase fold [32], with a β-sheet of six parallel β-strands in the order β 2 β 1 β 3 β 4 β 5 β 6 . The β-sheet is covered by two α-helices on one face, and five helices on the other face. Very large decorations, in which structured elements are a minority, complete the structure (Figure 2A,B). A cation, probably Ca 2+ , is observed at bonding distance from three acidic side chains from Asp444, Asp484 and Asp573 and the C=O group of Asp377 ( Figure 2, inset 1; Supplementary Figure S6a). In enzymes belonging to the α/β-hydrolase fold, the nucleophilic serine is located at the tip of a sharp β-strand/α-helix kink with a GxSxG consensus sequence. In Tle3 AIEC , the catalytic serine was identified in a short loop between a β-strand and a α-helix, with the sequence SHSQG at position 265. Notably, the first glycine of the consensus kink sequence GxSxG is replaced here by a serine. As in all α/β-hydrolase fold, the Ser exhibits a special position in the Ramachandran graph, with Phi and Psi angles of 60 • and −135 • , respectively. The Ser265 hydroxyl moiety is implicated in a hydrogen bound with the Ne2 atom of His574 which is, in turn, hydrogen-bound by its Nd1 atom to the Asp377 carboxyl group, forming the catalytic triad (Figure 2, inset 2; Supplementary Figure S6b). An Asn side chain further orientates the carboxyl moiety of Asp377. Notably, Asp377 C=O group participates in the binding of the Ca 2+ cation, whose function might be to stabilize the catalytic triad geometry. The oxyanion hole, a device that stabilizes the oxyanion transition state via hydrogen bonds with two main-chain nitrogens, is often evidenced by using organophosphate or organophosphonate complexes. Two main-chain NH groups establish short hydrogen bonds with an oxygen belonging to the phosphonate group. In Tle3 AIEC , in the absence of bound inhibitors, the oxyanion hole has been identified on the basis of comparisons with other α/β-hydrolases. Within the α/β hydrolase fold family, one obligate component of the oxyanion hole is the NH group of the residue following the nucleophile Ser (Gln266 in Tle3 AIEC ). The second NH group occupies a different position in sequence but always originates from the same spatial region. Here, by comparison with gastric lipase in complex with a phosphonate inhibitor ( [33], PDB id: 1k8q), we identified Val55 as the second NH group involved in the oxyanion hole. The Tli3 AIEC immunity protein is predicted to be translated as a 274-residue long pr The catalytic triad is located at the bottom of a crevice~30-Å deep and~20 × 15 Å (Figure 2A). Surprisingly, the walls of the crevice, as well as the surface residues around it, are not particularly hydrophobic. A Dali search with Tle3 AIEC structure returned hits with a palmitoyl thioesterase (PDB id 1pja; Z = 12.1; r.m.s.d. = 3.3 Å; aligned 187) and a lipase (PDB id 5h6g; Z = 11.4; r.m.s.d. = 3.1; aligned 176). Indeed, most residues belonging to the α/β-hydrolase fold core, i.e., the six β-strands and the seven α-helices, are aligned. Comparison with the apo-and phosphonate-complexed gastric lipase suggests that the oxyanion hole is preformed, even without substrate.
The Tli3 AIEC immunity protein is predicted to be translated as a 274-residue long protein, with 24 residues cleaved upon processing and acylated at the N-terminal cysteine residue of the mature form ( Figure S2). A total of 30-33 residues at the N-terminus of the mature form, which may correspond to a linker, as well as 0-5 residues at the Cterminus, are not observed in the electron density map. The Tli3 AIEC structure possesses overall dimensions of 60 × 45 × 45 Å. (Figure 2C,D). The core of Tli3 AIEC is assembled in three stacked β-sheets of 4, 3 and 4 β-strands, respectively, with topology of β 1 β 2 β 3 β 4 , β 6 β 7 β 8 , β 5 β 10 β 11 β 12 . A long loop of 17 residues is observed between strands β6 and β7 (residues 133-150). A large loop after strand β 12 is followed by a very long extended C-terminal segment that runs antiparallel to the extended N-terminus. The extremities of these segments point outwards from the core. Dali returned several hits with small Z values (~6-6.4) and a relatively low number of residues aligned (~120). These hits cover many different functions and reflect only resemblances between the β structures. Overall, the fold of Tli3 AIEC seems unique.

Comparison of the X-ray-and AlphaFold2-Predicted Structures
Superposition of AF2 prediction of Tle3 AIEC onto the crystal structure yields an r.m.s.d of 1.12 Å (528/555 residues) ( Figure 4A). Two predicted loops are absent in the crystal structure between residues 203-215 and residues 537-552. The AF2 prediction of Tle3 AIEC exhibits two protruding domains at opposite sides of Tle3 AIEC core that are not visible in the crystal structure, probably due to dynamic disorder. One of these extensions is the above-mentioned loop, 537-552, that starts and terminates with two antiparallel β-strands. The second is formed by the N-terminus of Tle3 AIEC (residues 1-43), including a poorly predicted stretch of~10 residues followed by a well-predicted structure of 3 extended stretches interlocked by 4 β-strands ( Figure 4A). The tip of these two extensions, as well as loop 203-215, display lower prediction statistics (Supplementary Figure S4), which together with their absence in the electron density map of the X-ray structure, may indicate that they are flexible. Superposition of AF2 prediction of Tli3 AIEC onto the crystal structure yields an r.m.s.d. of 1.4 Å (205/211 residues) ( Figure 4B). The Tle3 AIEC -Tli3 AIEC complex was predicted by AlphaFold2 multimer [24] (Figure 4C). Although the complex is roughly predicted, the position of Tli3 AIEC relative to Tle3 AIEC is displaced by~5 Å outside of the cavity in the predicted structure of the complex, leading to reduced predicted interactions.

Tle3 AIEC Interacts with VgrG AIEC
In EAEC, the Tle1 phospholipase was shown to be transported through its association to the VgrG1 spike protein. As for EAEC VgrG1 [18], the VgrG protein from AIEC (VgrG AIEC ) carries a C-terminal extension (COG 4532 domain) comprising a DUF2345/TTR domain separated from the VgrG base by two helices. This domain was shown to be involved, together with the base of the VgrG needle β-prism, in the interaction of VgrG1 with Tle1 from EAEC [17,18]. The predicted structural organization of VgrG AIEC and its genetic proximity to tle3 suggest a similar mode of binding. To test whether Tle3 AIEC associates with VgrG AIEC , we used a pull-down assay. Strep-tagII VgrG AIEC ( S VgrG) and Tle3 AIEC 6-His (Tle3 H )-tagged proteins were produced separately and cell lysates were mixed and loaded on a Streptactin resin. SDS-PAGE followed by Coomassie blue staining and Western blot analysis of the eluted material showed that Tle3 H co-precipitated with S VgrG (Figure 5), suggesting that Tle3 AIEC indeed interacts with VgrG AIEC .

Tle3 AIEC Interacts with VgrG AIEC
In EAEC, the Tle1 phospholipase was shown to be transported through its association to the VgrG1 spike protein. As for EAEC VgrG1 [18], the VgrG protein from AIEC (VgrG AIEC ) carries a C-terminal extension (COG 4532 domain) comprising a DUF2345/TTR domain separated from the VgrG base by two helices. This domain was shown to be involved, together with the base of the VgrG needle β-prism, in the interaction of VgrG1 with Tle1 from EAEC [17,18]. The predicted structural organization of VgrG AIEC and its genetic proximity to tle3 suggest a similar mode of binding. To test whether Tle3 AIEC associates with VgrG AIEC , we used a pull-down assay. Strep-tagII VgrG AIEC ( S VgrG) and Tle3 AIEC 6-His (Tle3 H )-tagged proteins were produced separately and cell lysates were mixed and loaded on a Streptactin resin. SDS-PAGE followed by Coomassie blue staining and Western blot analysis of the eluted material showed that Tle3 H co-precipitated with S VgrG (Figure 5), suggesting that Tle3 AIEC indeed interacts with VgrG AIEC .

Predicted Model of VgrG AIEC and Docking with Tle3 AIEC
Comparison of the crystal structure of Tle3 AIEC ( Figure 4A) with the cryoEM structure of Tle1 EAEC ( Figure 6A) indicates that both structures share a α/β-hydrolase fold core of six β-strands surrounded by α-helices. Besides this core, the extra-parts of the structures diverge deeply. Two extensions that are absent in Tle3 AIEC crystal structure are observed in Tle1 EAEC at opposites sides ( Figure 4A). However, these Tle1 EAEC extensions are present in the Tle3 AIEC AF2 prediction ( Figure 4A) at similar positions; one of the two extensions also involves its N-terminus. Considering that the Tle1 EAEC extensions interact with VgrG EAEC [18] ( Figure 6A) and that we have shown that Tle3 AIEC interacts with VgrG AIEC , we hypothesized that a similar mode of interaction may occur for Tle3 AIEC -VgrG AIEC . To explore this hypothesis, we co-predicted the VgrG AIEC (residues 421-Ct) and Tle3 AIEC structures with AF2-multimer. As expected from the sequence similarity, both VgrGs exhibit close structures ( Figure 6B). After a non-predicted linker (750-767), the predicted structure of VgrG AIEC C-terminus domain, the adaptor (or TTR domain, 768-824), forms a complex with the N-terminus loop of Tle3 AIEC (Figure 6B-inset). It completes the two β-sheets of the Tle3 AIEC N-terminal loop by providing two β-strands to each of them. However, AF2 was not able to identify the second anchoring point of the complex, and the position of Tle3 AIEC was adjusted manually to resemble that of Tle1 EAEC in the complex with its VgrG ( Figure 6B).   Ribbon representation of the cryoEM structure of Tle1 EAEC in complex with VgrG EAEC . VgrG EAEC is colored magenta, violet and green. Tle1 EAEC is colored beige. Surfaces are colored white and transparent. (B) Surface representation of the predicted structure of Tle3 AIEC and VgrG AIEC , and a possible mode of interaction in their complex; inset: details of the interaction of VgrG AIEC C-terminal adaptator with Tle3 AIEC N-terminal loop. VgrG AIEC trimer is colored magenta, blue and green. Tle3 AIEC is colored yellow. The blue dot square identify the interaction between VgrG EAEC TTR domain with the N-terminus loop of Tle3 AIEC . The red question mark means that the interaction between the bVgrG E-AEC and Tle3 AIEC cores remains unknown.

Discussions
In this study, we provide the first biochemical and structural characterization of a type VI phospholipase effector from the Tle3 family and its immunity protein Tli3. Tle3 effectors were identified by Russell et al. by bioinformatics [12]. Pseudomonas aeruginosa Tle3 PA was previously shown to be a periplasmic antibacterial T6SS effector, but its biochemical activity was not determined [35,36]. Here we show that Tle3 AIEC demonstrates PLA1 activity, inhibited by Tli3 AIEC .
Tle3 AIEC was soluble and could be readily purified from the cytoplasm of E. coli. The absence of toxicity of Tle3 AIEC when produced in the cytoplasm of a host cell is in agreement with its predicted periplasmic activity. Indeed, Tli3 AIEC co-fractionates with the membrane fraction as a predicted outer membrane lipoprotein, suggesting a periplasmic activity for Tle3 AIEC that is common for Tle effectors.
The structure of the Tle3 AIEC /Tli3 AIEC complex presented here is only the second structure of a Tle effector in complex with its immunity protein. The structure of Tle3 AIEC /Tli3 AIEC reveals the inhibition mechanism, as a loop of Tli3 AIEC inserts deeply in the Tle3 AIEC substrate-binding crevice. This mechanism is very different from the "crabclaw like mechanism" used by Tli4 immunity protein to maintain Tle4 from P. aeruginosa in a closed conformation [37].
In EAEC, three Tle1s bind to the VgrG trimeric protein that serves as cargo for their delivery into the target cell. Each Tle1 interacts with the C-terminal transthyretin (TTR) extension of each VgrG monomer through its N-terminal domain extension, as well as on two additional zones of contact on the side and at the base of the VgrG needle. As expected Figure 6. Structure of Tle1 EAEC and Tle3 AIEC , and their interaction with VgrG EAEC and VgrG AIEC . (A) Ribbon representation of the cryoEM structure of Tle1 EAEC in complex with VgrG EAEC . VgrG EAEC is colored magenta, violet and green. Tle1 EAEC is colored beige. Surfaces are colored white and transparent. (B) Surface representation of the predicted structure of Tle3 AIEC and VgrG AIEC , and a possible mode of interaction in their complex; inset: details of the interaction of VgrG AIEC C-terminal adaptator with Tle3 AIEC N-terminal loop. VgrG AIEC trimer is colored magenta, blue and green. Tle3 AIEC is colored yellow. The blue dot square identify the interaction between VgrG EAEC TTR domain with the N-terminus loop of Tle3 AIEC . The red question mark means that the interaction between the bVgrG EAEC and Tle3 AIEC cores remains unknown.

Discussions
In this study, we provide the first biochemical and structural characterization of a type VI phospholipase effector from the Tle3 family and its immunity protein Tli3. Tle3 effectors were identified by Russell et al. by bioinformatics [12]. Pseudomonas aeruginosa Tle3 PA was previously shown to be a periplasmic antibacterial T6SS effector, but its biochemical activity was not determined [35,36]. Here we show that Tle3 AIEC demonstrates PLA1 activity, inhibited by Tli3 AIEC .
Tle3 AIEC was soluble and could be readily purified from the cytoplasm of E. coli. The absence of toxicity of Tle3 AIEC when produced in the cytoplasm of a host cell is in agreement with its predicted periplasmic activity. Indeed, Tli3 AIEC co-fractionates with the membrane fraction as a predicted outer membrane lipoprotein, suggesting a periplasmic activity for Tle3 AIEC that is common for Tle effectors.
The structure of the Tle3 AIEC /Tli3 AIEC complex presented here is only the second structure of a Tle effector in complex with its immunity protein. The structure of Tle3 AIEC /Tli3 AIEC reveals the inhibition mechanism, as a loop of Tli3 AIEC inserts deeply in the Tle3 AIEC substrate-binding crevice. This mechanism is very different from the "crab-claw like mechanism" used by Tli4 immunity protein to maintain Tle4 from P. aeruginosa in a closed conformation [37].
In EAEC, three Tle1s bind to the VgrG trimeric protein that serves as cargo for their delivery into the target cell. Each Tle1 interacts with the C-terminal transthyretin (TTR) extension of each VgrG monomer through its N-terminal domain extension, as well as on two additional zones of contact on the side and at the base of the VgrG needle. As expected from their sequence similarity, an AlphaFold2 model of VgrG AIEC predicts a structure similar to VgrG EAEC , including two α-helices located alongside of the extended needle and a TTR domain, features corresponding to COG4532 domains found in several VgrG proteins [38]. Here we found that Tle1 EAEC and Tle3 AIEC share the small overall α/β mixed hydrolase fold core, and two divergent extensions, as modelled by AlphaFold2 for Tle3 AIEC . The docking simulations of Tle3 AIEC on VgrG AIEC suggest a similar mode of interaction, with an N-terminal domain/TTR adaptor interaction, and likely other contacts with the side of the needle as well as a contact with the base of the needle through the C-terminal extension of Tle3 AIEC . We propose that effectors could have evolved different extensions to allow interaction with VgrG proteins, especially with the COG4532 domain-containing VgrG protein family.

Plasmid Constructions
E. coli AIEC LF82 chromosomal DNA was used as a template for all PCRs. E. coli strain DH5α was used for cloning procedures. Polymerase chain reactions (PCR) were performed using a Biometra thermocycler using Q5 High Fidelity DNA Polymerase (New England Biolabs, Herts, UK) or PfuTurbo (Agilent, Santa Clara, CA, USA). The genes encoding Tle3 AIEC and Tli3 AIEC (without the N-terminal 24-residue signal peptide) were amplified from AIEC LF82 genomic DNA using the primers listed in Supplementary Table S1. Tle3 AIEC was cloned, using the gateway technology, into the pETG20A vector that incorporates a Trx fusion protein followed by N-terminal His tag and a TEV cleavage site. The tli3 AIEC gene sequence corresponding to the mature form of Tli3 AIEC was cloned into the pET22b vector (Novagen, Madison, Wis, USA), fused to the PelB signal sequence, with a C-terminal 6-His tag in order to be addressed to the periplasm. Plasmids pET-Duet-SVgrG AIEC and pRSF-Duet1-Tle3 H used for the pull-down experiments were constructed by a classical restriction/ligation method using the primers and restriction sites described in Supplementary Table S1. Plasmid pBAD33-Tli3 VSVG used for the fractionation experiment was constructed by restriction-free cloning, as previously described [39], using oligonucleotides introducing extensions annealing to the target vector (see Supplementary Table S1). The constructs were verified by sequencing (Eurofins, Nantes, France).

Expression, Production and Purification of Tle3 AIEC , Tli3 AIEC and Tle3 AIEC -Tli3 AIEC Complex
For Tle3 AIEC production, the expression plasmid pETG20A harboring the Tle3 AIEC coding sequence was transformed into E. coli T7 Iq pLys strain. Recombinant cells were grown in a 2 L Erlenmeyer flask at 37 • C in Terrific Broth medium supplemented with 100 µg/mL ampicillin. When the OD 600nm reached 0.8, the temperature was reduced to 17 • C and expression was induced overnight by the addition of 0.5 mM IPTG. Expression of selenomethionine (SeMet)-labelled Tle3 AIEC (Tle3 AIEC SeMet) was performed in E. coli T7 Iq pLys cells using the method of methionine biosynthesis pathway inhibition [40].
For Tli3 AIEC production, E. coli C41 (DE3) cells were transformed with pET22b-Tli3 AIEC for production in the periplasmic compartment and grown at 37 • C in Terrific Broth medium containing 0.1% glucose and 100 µg/mL ampicillin to an OD 600nm of approximately 0.8. Gene expression was then induced by the addition of 1 mM IPTG and the bacterial culture was further incubated for 16 h at 28 • C.
After induction, cells were harvested by centrifugation and stored at −80 • C. The Tle3 AIEC or Tle3 AIEC SeMet cell pellets were resuspended in Tris-HCl 20 mM pH 8.0, NaCl 300 mM, glycerol 5% (v/v), lysozyme (0.25 mg/mL), DNase (2 µg/mL) and MgSO 4 20 mM, and cells were lysed by ultrasonication on ice. The insoluble material was discarded by centrifugation at 20,000× g for 60 min at 4 • C. The soluble thioredoxin 6×His-tagged Tle3 AIEC or Tle3 AIEC SeMet fusion proteins were purified by affinity chromatography on a nickel-nitrilotriacetic acid resin (Bio-Rad) and the tag was removed after dialysis by overnight hydrolysis with the TEV protease and re-purified in presence of 10 mM imidazole. Pure fractions were concentrated by ultrafiltration using Amicon-Ultra 10-kDa cut-off and further purified by gel filtration chromatography (Superdex 200, 16/60 GE Healthcare) equilibrated in Hepes 10 mM pH 6.8, NaCl 150 mM using an AKTA purifier System (Amersham, Bath, UK).
Purification of Tli3 AIEC was performed from the periplasmic fraction prepared by osmotic shock, according to Skerra and Plückthun [41]. Briefly, the cell pellet was resuspended in 9 mL cold TES buffer (0.2 M Tris-HCl pH 8.0, 0.5 mM EDTA, 0.5 M sucrose) and kept on ice for 1 h. The periplasmic proteins were obtained by osmotic shock by addition of 13.5 mL cold TES diluted 1/4 into H 2 O, incubation for 1-2 h on ice, and w consecutive centrifugations at 20,000× g for 30 min at 4 • C. The Tli3 AIEC protein was then purified from the periplasmic fraction using the same procedure described above for Tle3 AIEC . Production and purification of recombinant Tle1 EAEC was performed as previously described [17,18].
For the purification of the Tle3 AIEC -Tli3 AIEC complex, the purified Tle3 AIEC and Tli3 AIEC were mixed with a molar ratio of 1:1.3. The complex was purified by gel filtration chromatography (Superdex 200, 16/60 GE Healthcare) equilibrated in a Hepes 10 mM pH 6.8, NaCl 150 mM buffer using an AKTA purifier System (Amersham, Bath, UK).

Cytotoxic Effects of Tle1 and Tle3 Phospholipases on Mouse Macrophage
The cytotoxicity of the Tle1 EAEC and Tle3 AIEC phospholipases against eukaryotic cells was measured based on the reduction of resazurin as reporter of cellular viability [44][45][46]. Murine macrophage cells RAW264.7 (ATCC ® TIB-71™) were cultured in Dulbecco's Modified Eagle's medium (DMEM; Gibco, Thermo Fischer Scientific, Waltham, MA, USA). containing 4.5 g/L glucose, L-glutamine and sodium pyruvate supplemented with 10% heat-inactivated fetal bovine serum (Lonza, BioWhittaker ® sera, Walkersville, MD, USA), in the presence of 5% CO 2 , at 37 • C. For viability assays, cells were seeded in 96-well plates to a density of 1 × 10 5 cells/well and, after 16 h, tightly attached macrophages were washed 1 time with 200 µL of complete culture medium. Then, fresh complete medium alone or fresh medium containing the purified Tle1 EAEC and Tle3 AIEC recombinant proteins (30 µg) was added into each well. After 16 h incubation, 20 µL of a 0.025% (w/v) resazurin solution was added to each well and fluorescence was measured following 4 h incubation at 37 • C and 5% CO 2 in the dark, by excitation at 530 nm and emission at 590 nm using a TECAN Spark 10 M. Similar control experiments were carried out by treating the cells with the vehicle buffer, the pre-incubated Tle1 EAEC -Tli1 EAEC inactive complex at stochiometric 1:1 ratio, as previously described [17], or with Triton X-100, used as lysis control.

VgrG AIEC -Tle3 AIEC Pull-Down Experiments
An amount of 100 milliliters of E. coli BL21 (DE3) culture cells transformed with pETDuet-S VgrG or pRSF-Tle3 H were grown to an OD 600 of 0.8, and the expression of S vgrG or tle3 H was induced with 0.5 mM of IPTG for 16 h at 16 • C. Cultures were centrifuged and resuspended in 1.5 mL buffer A (50 mM Tris-HCl pH8.5, 150 mM NaCl, 1 mM TCEP, cOmpleteTM™ protease inhibitors (Merck, Darmstadt, Germany ). Cells were lysed by sonication followed by a 30 min centrifugation at 20,000× g at 4 • C. Cleared lysates were then mixed (500 µL S VgrG+ 500 µL Tle3 H ) or not ( S VgrG alone, Tle3 H alone) for 1 h and then incubated with 100 µL of Strep-Tactin ® Sepharose ® resin (IBA Technologies) equilibrated with buffer A and gently mixed on a wheel at 4 • C. The resin was washed 5 times with buffer A and the proteins were eluted with buffer B (50 mM Tris-HCl pH 8.5, 250 mM NaCl, 1 mM TCEP, 2.5 mM desthiobiotin). In total, 1 µL of lysates and 7.5 µL of eluted fractions were analyzed by 10%-acrylamide SDS-PAGE followed by InstantBlue ® Commassie Blue Protein Stain (Abcam, Cambridge, UK) staining or immunoblotted using anti Strep (Mouse anti Strep-Tag Classic antibody, clone Strep-tag II, Biorad, MCA2489) or anti 6 × His (6 × His, His-Tag Monoclonal Antibody, Proteintech, 66005-1-Ig) antibodies and secondary antibodies coupled to phosphatase alkaline, and developed in alkaline buffer in presence of 5-bromo-4-chloro-3-indolylphosphate and nitroblue tetrazolium.

Data Collection and Processing
The SeMet Tle3 AIEC /Tli3 AIEC data were collected at Proxima 2 using a Pilatus 6M detector at a wavelength of 0.9791 Å. Before data collection, the crystals were soaked in reservoir solution supplemented with 12% (v/v) ethylene glycerol for a few seconds and then flash-cooled in liquid nitrogen. All data were processed using XDS [47]. The final data-collection and processing statistics are provided in Table 1. A total of 375,390 measured reflections were merged into 78,970 unique reflections. The merged dataset was 98.9% complete to 3.6 Å resolution. The relevant data collection statistics are provided in Table 1.
Funding: This work was supported by the Aix-Marseille Université (AMU), the Centre National de la Recherche Scientifique (CNRS) and a grant from the Agence Nationale de la Recherche (ANR-18-CE15-0013 to LJ). MB was supported by the ANR-18-CE15-0013.

Data Availability Statement:
The predicted coordinates of Tle3, Tli3 and Tle3/Tli3 structures will be deposited in the open data repository Zenodo. The X-ray structure has been deposited with the Protein Data Bank with identifier 8BOZ.