TolC partner identified by cryo-EM.

(A) Cryo-EM density map (left) resolved here and model (right) for the tripartite pump MacA-MacB-TolC (PDB 5NIL). Density for the MacB dimer is less well defined. (B) Close-up view on the TolC region showing additional density in green. TolC domain organisation is indicated. (C) Modelling of secondary structural elements fitting in the additional density. (D) The refined structure of the TolC partner, YbjP. After lipoprotein maturation, the N-terminal cysteine carries a lipid modification which is not shown here. Alignment of the YbjP structure from cryo-EM and corresponding AlphaFold prediction model is shown as inset. (E) Sequence and secondary structure alignment annotation of mature YbjP. The tri-acylated cysteine (C19) conserved in lipoproteins is underlined.

Interactions in the TolC-YbjP complex.

(A) Superposition of TolC from the MacAB-TolC-YbjP assembly reported this paper in blue (TolC) and yellow (YbjP) (PDB 9QGY) and from the previously reported MacAB-TolC complex in green (PDB 5NIK). (B) Close-up view on the YbjP-TolC interface with TolC residues involved in the interaction shown as sticks. (C) Overview of protein-protein interactions with close-up views presented in panels 1, 2 and 3. The YbjP lipoprotein (yellow) contacts two adjacent protomers of the TolC trimer (blue). Lipoprotein acyl modification and residues involved in intermolecular contacts are shown in stick representation. (D) Residue variability analysis of YbjP highlighting the clustering of conserved residues at the TolC interface. Analysis was performed with the CONSURF server. YbjP is shown as spheres with the lipidation in yellow.

In vitro and in vivo validation of the TolC-YbjP interface.

(A) In vitro binding assay between TolC and soluble YbjP that is missing the lipid modification. TolC-FLAG was mixed with N-terminally His-tagged soluble YbjP (YbjPs) or soluble LolB (LolBs) prior immobilization on IMAC resin. After several washes, the elution fractions (Elu) were analysed by SDS-PAGE. Molecular masses of protein standards (M) are indicated. (B) Isothermal titration calorimetry (ITC) profile for the interaction of TolC and YbjPs. Background-corrected heats of injection are shown together with a fitted binding curve in red. The raw thermograms corresponding to the injection of YbjPs into a TolC-containing cell or into buffer are shown in upper and lower insets, respectively. ITC parameters are listed in Supplementary Table S3. (C) In vivo validation of the interaction between YbjPFL and TolC by photo-crosslinking. The experimental procedure was validated using another periplasmic lipoprotein, AcrA, which associates with TolC. As shown underneath, AcrA residues Q136 and Y137, proximal to TolC in the structure of the AcrABZ-TolC pump (PDB 5NG5), were replaced by pBPA. For YbjP, the two residues N113 and T110 proximal to TolC in the MacAB-TolC-YbjP complex (PDB 9QGY) and the three residues N43, N90 and H104 distal to TolC were mutated. In brief, E. coli C43 cells expressing YbjPFL,His, or AcrAHis mutants carrying a photoactivatable pBPA group at indicated positions were grown in the presence (+) or absence (-) of pBPA and irradiated (+) or not (-) with UV. Cells were then lysed by successive freeze-thaw cycles, solubilized with DDM and lysates purified by Ni-NTA chromatography. Eluted proteins were analysed by SDS-PAGE and immunoblotting using anti-His antibodies. Purified YbjPs was loaded as a benchmark control. Presence of crosslinked products are indicated with a black arrow. Note that FL YbjP migrates faster than its predicted molecular weight (∼19 kDa), likely due to the presence of the lipoprotein lipid chains. A double detection of YbjPFL,His and endogenous TolC proteins for the N113 sample is shown in Supplementary Figure S2.

The structural comparison of AcrABZ-TolC-YbjP and MacAB-TolC-YbjP.

(A) Cryo-EM maps and models of AcrABZ-TolC-YbjP and MacAB-TolC-YbjP reconstituted in peptidisc. (B) Superposition of TolC from the AcrABZ-TolC-YbjP and MacAB-TolC-YbjP assemblies reported here (red: TolC, green: YbjP, PDB 9TG4; blue: TolC and yellow: YbjP, PDB 9QGY). (C) Close-up views on the YbjP-TolC interface in the AcrABZ-TolC-YbjP (top) and MacAB-TolC-YbjP (bottom) structures.

YbjP evolution and co-evolution with TolC.

(A) Structural alignment of YbjP with other DUF3828-containing proteins: Tai3 (PDB 4HZ9, purple) and YqhG (AlphaFold2 Q46858, blue). A 90° rotation along the x-axis is shown underneath. (B) A cladogram of DUF3828-containing proteins. The reviewed proteins (Tai3, YbjP and YqhG) are marked by red arrows. (C) Surface conservation of TolC proteins from Enterobacterales containing or not containing YbjP. Results are displayed on E. coli TolC on the left and on P. atrosepticum TolC (AlphaFold model, Q6DAC5) on the right. For clarity, YbjP is displayed on the left panel but not on the right. Analysis was performed with the CONSURF server.

YbjP lipoprotein is not required for TolC trafficking in E. coli.

(A) Structural alignment of E. coli TolC (blue), CusC (green, PDB 3PIK,) and P. aeruginosa OprM (red, PDB 3D5K). The YbjP- or TolC-homologue linker is shown as transparent surface and modified cysteines are indicated by arrows. Dashes lines show the extension of the lipid group (partially-or non-modelled in the structures). (B) AlphaFold3 prediction (Abramson et al., 2024) between E. coli outer membrane efflux proteins and YbjP. The model with the highest ipTM and pTM scores is shown out of three technical repeats. (C) CLANS (CLuster ANalysis of Sequences) analysis of outer membrane efflux proteins (IPR003423) in Pseudomonadota. Clustering was performed in 2D until equilibrium. Outer membrane proteins in E. coli are shown in red, and other key reviewed proteins in black. Yellow, lipidation; blue, secreted; grey, other. (D) Distribution of the ybjP gene and lipidated TolC in Gammaproteobacteria. Repartitions are as follows: ybjP present (light blue, n=67) or absent (magenta, n=525); lipidated (yellow, n=5) or non-lipidated (dark blue, n=587) TolC.

YbjP lipoprotein deletion has modest impacts on the proteome in E. coli.

Volcano plots showing the proteome-wide differential expression relative to WT. Columns correspond to ΔybjP, ΔtolC and ΔybjP ΔtolC (left to right), and rows correspond to the global, exponential, stationary phases (top to bottom). Proteins with |log2 fold change| >1 and Bonferroni-Hochberg adjusted p-value < 0.05 are shown in blue. Significant hits in ΔybjP and selected genes are labelled.

Cryo-EM processing workflow for MacAB-TolC-YbjP complex.

(A) Cryo-EM processing workflow. (B) Gold-standard Fourier Shell Correlation (GS-FSC) curves for final reconstruction. (C) Local resolution distribution in final map. (D) Angular distribution plot for final map, left, and Fourier Shell Correlation between half maps of MacAB-TolC-YbjP, right. These data suggest that there is not a large orientational bias in the final map.

Double detection of YbjP and TolC proteins for the in vivo photo crosslinking.

Left, immunoblot showing purified N-terminally His-tagged soluble YbjP (YbjPs) and His-tagged TolC detected with anti-His (green) and anti-TolC (red) antibodies. Right, immunoblot of E. coli C43 cells expressing His-tagged full-length YbjP (YbjPFL) carrying a photoactivatable pBPA group at position N113 and endogenous, tag-free, TolC detected with anti-His (green) and anti-TolC (red) antibodies, after UV irradiation. An overlay of the two images is shown on the right. The double detection of YbjP and TolC proteins in the crosslinked sample is indicated by an arrow.

Cryo-EM workflow for AcrABZ-TolC-YbjP complex.

(A) Cryo-EM processing workflow. (B) Gold-standard Fourier Shell Correlation (GS-FSC) curves for final reconstruction. (C) Local resolution distribution in the final map. (D) Angular distribution plot of the final particles.

YbjP binds at the same position in the AcrABZ-TolC and MacAB-TolC tripartite pumps and may not interact with peptidoglycan.

(A) Cryo-EM map of AcrABZ-TolC-YbjP reconstituted in peptidoglycan. (B) Fourier Shell Correlation between half-maps (top) and angular distribution plot of the particles (bottom). Collection statistics for the AcrABZ-TolC-YbjP complex reconstituted in peptidoglycan are listed in Supplementary Table S2. (C) Peptidoglycan location based on the cryo-ET map described by Shi et al., 2019.

The cladogram of TolC (IPR010130) in Pseudomonadota.

The sequence that showed the highest pairwise similarity to E. coli TolC in each organism is selected.

Impact of tryptophan on growth in minimal media.

The growth curves of WT (blue), ΔybjP (orange), ΔtolC (green), ΔybjP ΔtolC (red) E. coli were recorded in M9 media supplemented with 0.4% glycerol and indicated L-tryptophan (Trp) concentrations. Results are shown as mean ± standard deviation from two biological replicates, each with three technical replicates.

Structural comparison of TolC3–YbjP3 and TolC3–SlyB11.

(A) AlphaFold3 structure of TolC-SlyB11 with 30 palmitic acid and 30 oleic acid molecules. The model is coloured by pLDDT score. (B) PAE plot of TolC3–SlyB11. (C) Overlay of TolC3–YbjP3 structure and TolC3–SlyB11 AlphaFold3 structure. The lipidation in YbjP is displayed as green spheres.

Cryo-EM data and refinement statistics for peptidisc-reconstituted MacAB-TolC-YbjP and AcrABZ-TolC-YbjP models.

Cryo-EM data collection statistics for AcrABZ-TolC-YbjP in peptidoglycan.

Thermodynamic parameters for the TolC-YbjPs, interaction.

Results from isothermal titration calorimetry (ITC) were obtained with DDM-purified TolC-FLAG and N-terminally His-tagged YbjPs. The mean ± spread of two independent measurements is shown in bold. Experiments were carried out at 25°C. Values of ΔG, ΔH and TΔS are in kcal mol-1. An example of one of the thermograms and fit is shown in Figure 3B. N corresponds to the stoichiometry of the interaction.

Criteria for classification of DUF3828-containing proteins

Minimum inhibitory concentration (MIC) values for wild-type and indicated knockout strains of E. coli BW25113

. Raw values are in bold and fold-changes values for indicated knockout strains compared to wild-type E. coli BW25113 are in brackets, under aerobic or microaerobic (5-6% O2, 8-10% CO2) conditions.

Swimming motility assay

. Colony diameters of wild-type and knockout strains of E. coli BW25113. Values are presented as mean ± standard deviation from at least 5 replicates.