Lateral gating mechanism and plasticity of the β-barrel assembly machinery complex in micelles and Escherichia coli

Abstract The β-barrel assembly machinery (BAM) mediates the folding and insertion of the majority of outer membrane proteins (OMPs) in gram-negative bacteria. BAM is a penta-heterooligomeric complex consisting of the central β-barrel BamA and four interacting lipoproteins BamB, C, D, and E. The conformational switching of BamA between inward-open (IO) and lateral-open (LO) conformations is required for substrate recognition and folding. However, the mechanism for the lateral gating or how the structural details observed in vitro correspond with the cellular environment remains elusive. In this study, we addressed these questions by characterizing the conformational heterogeneity of BamAB, BamACDE, and BamABCDE complexes in detergent micelles and/or Escherichia coli using pulsed dipolar electron spin resonance spectroscopy (PDS). We show that the binding of BamB does not induce any visible changes in BamA, and the BamAB complex exists in the IO conformation. The BamCDE complex induces an IO to LO transition through a coordinated movement along the BamA barrel. However, the extracellular loop 6 (L6) is unaffected by the presence of lipoproteins and exhibits large segmental dynamics extending to the exit pore. PDS experiments with the BamABCDE complex in intact E. coli confirmed the dynamic behavior of both the lateral gate and the L6 in the native environment. Our results demonstrate that the BamCDE complex plays a key role in the function by regulating lateral gating in BamA.

).The background labeling is shown by the Cys-less BamABCDE (overlaid in red) when available.Spectra are normalized to the same cell density (OD600) value for a quantitative comparison.The overall spin concentration obtained was in the range between 30-100 µM (see the methods section on expression and spin labeling in E. coli).From the modulation depths of the PELDOR/DEER data (having a maximum value between 30 -35%), an effective labeling efficiency between 30-50% is estimated in E. coli.The corresponding distance distribution (when available) in detergent micelles (dark grey, from Fig. 3B and 3C) are overlaid.The error bounds show the variation in the probability amplitude due to uncertainties in background function as described in Table S2.The color coding for the distance distribution is as explained in Fig. 6.S2.The color code for the probability distribution is as explained in Fig. 6.Corresponding distances in micelles (DDM, when available) and simulations for the IO (salmon, PDB 5D0O), LO (light violet, PDB 5LJO) and LO BamA-bound conformation (green, PDB 6V05) are overlaid in dotted lines.

Figure S1 .
Figure S1.Characterization of full-and sub-complexes of BAM.(A) SDS-PAGE of the purified protein complexes, BamAB (lane 1), BamACDE (lane 2) and BamABCDE (lane 3) in detergent micelles.(B) Size-exclusion chromatogram of BamABCDE (dark grey), BamACDE (salmon) and BamAB (green).(C) LILBID mass spectrometry of BAM complexes.BamAB in the upper panel, BamACDE in middle panel and BamABCDE in the lower panel.Peaks were assigned based on the predicted molecular mass (indicated as theo:) of the complexes.

Figure S2 .
Figure S2.Colony growth assay for BamA cysteine variants in E. coli JCM166 cells.The cysteine variants expressed from the corresponding plasmids showed similar growth in the presence (on the left) and absence of arabinose (on the right).Cells were diluted for each variant as indicated.Cells transformed with the empty pCDFDuet-1 vector as a negative control did not grow in the absence of arabinose (see the Methods section for details).

Figure S3 .
Figure S3.Room temperature CW ESR spectra of the spin labelled double cysteine variants in DDM micelles.The different oligomeric states BamAB (green), BamACDE (salmon) and BamABCDE (dark grey) for each variant is overlaid.Spectra are normalized for a direct comparison of the shape and spectra for the other variants are shown in Fig. 1.

Figure S4 .
Figure S4.Structures of the BAM complex in different conformational states with the key positions highlighted.The structures show (A) inward-open (B) lateral-open and (C-D) lateral-open substrate bound states as indicated.BamA (grey), BamB (salmon), BamC (wheat), BamD (light green) and BamE (deep blue) are shown.The lateral gate region (red) is closed in the inward-open state, opens in the lateral-open state and the opening further increases in presence of substrates (BamA or EspP).Loop 3 (orange), loop 6 (black) and the POTRA 5 (violet) of BamA are highlighted.The positions investigated in Fig. 2B (434-801) and Fig. 5C (359-780) are shown in blue spheres.The change in distance between the Cα atoms of the original amino acid pairs (solid blue lines) is given.

Figure S5 .
Figure S5.Structures of the BAM complex in different conformational states with the key positions highlighted.The structures show (A) inward-open (B) lateral-open and (C-D) lateral-open substrate bound states as indicated.BamA (grey), BamB (salmon), BamC (wheat), BamD (light green) and BamE (deep blue) are shown.The lateral gate region (red) is closed in the inward-open state, opens in the lateral-open state and the opening further increases in presence of substrates (BamA or EspP).Loop 3 (orange), loop 6 (black) and the POTRA 5 (violet) of BamA are highlighted.The positions investigated in Fig. 2C (434-796), 3B (501-796) and 5B (452-732) are shown in blue spheres.The change in distance between the Cα atoms of the original amino acid pairs (solid blue lines) is given.

Figure S6 .
Figure S6.Room temperature CW ESR spectra of the spin labeled single cysteine (A) and double cysteine (B) variants of the BamABCDE complex in E. coli.The labeled positions and the corresponding loops are indicated.The single cysteine variants were used as the control(s) for DEER/PELDOR experiments with double cysteine variants (see Figs. S7-S10).The background labeling is shown by the Cys-less BamABCDE (overlaid in red) when available.Spectra are normalized to the same cell density (OD600) value for a quantitative comparison.The overall spin concentration obtained was in the range between 30-100 µM (see the methods section on expression and spin labeling in E. coli).From the modulation depths of the PELDOR/DEER data (having a maximum value between 30 -35%), an effective labeling efficiency between 30-50% is estimated in E. coli.

Figure S7 .
Figure S7.DEER/PELDOR measurement of spin labeled single cysteine variants of the BamABCDE complex in E. coli.(A) The primary data (blue) and the fit (grey) obtained using a stretched exponential decay corresponding to the dimensionality for spin distribution (d) as indicated.(B) The primary data (blue) and the fit (grey) obtained by fitting the decay to a polynomial function (n=5, which corresponded to a stretched exponential decay with d = 2.0 -2.5).

Figure S8 .
Figure S8.DEER/PELDOR data for BamABCDE in E. coli analyzed using Tikhonov regularization.Corresponding analysis using the DEERNet program is shown in Fig. 6.Primary data (blue) with the intermolecular contribution (grey) obtained using the DeerAnalysis (1) program are shown in the left panels.The middle panels indicate the form factor obtained after intermolecular background correction.The distance distribution obtained using Tikhonov regularization are shown on the right.Simulations for the IO (salmon, PDB 5D0O), LO (light violet, PDB 5LJO) and LO BamA-bound conformation (green, PDB 6V05) are overlaid in dotted lines.The corresponding distance distribution (when available) in detergent micelles (dark grey, from Fig.3B and 3C) are overlaid.The error bounds show the variation in the probability amplitude due to uncertainties in background function as described in TableS2.The color coding for the distance distribution is as explained in Fig.6.

Figure S9 .
Figure S9.DEER/PELDOR data for the replicate samples of BamABCDE in E. coli.A similar data set is shown in Fig. 6 (and Fig. S8).Primary data (blue) with the intermolecular contribution (grey) obtained from the DEERNet program (A, (2)) or from the Tikhonov regularization using the DeerAnalysis program (B) as described in Fig. 6G are shown in the left panels.The middle panels show the form factors and the obtained distance distributions are shown on the right.The error bounds show the variation in distances due to uncertainties in the intermolecular function.The color code for the probability distribution is as explained in Fig. 6.Corresponding distances in micelles (DDM, when available) and simulations for the IO (salmon, PDB 5D0O), LO (light violet, PDB 5LJO) and LO BamA-bound conformation (green, PDB 6V05) are overlaid in dotted lines.

Figure S10 .
Figure S10.DEER/PELDOR data of replicate samples of BamABCDE in E. coli analyzed using Tikhonov regularization.The DeerNet analysis of the data are shown in Fig. S9.Primary data (blue) with the intermolecular contribution (grey) obtained using the DeerAnalysis program are shown in the left panels.The middle panels show the corresponding form factors.The distance distributions are shown on the right.The error bounds show the variation for the probability amplitudes due to uncertainties in background function as described in TableS2.The color code for the probability distribution is as explained in Fig.6.Corresponding distances in micelles (DDM, when available) and simulations for the IO (salmon, PDB 5D0O), LO (light violet, PDB 5LJO) and LO BamA-bound conformation (green, PDB 6V05) are overlaid in dotted lines.