Discussion

However, Ricci et al. (2012) could demonstrate interactions between BamA and BamD. These were greatly destabilized, when the negatively charged amino acid residue E373 in BamA was substituted by a positively charged residue. The mutation dissociated the two subcomplexes of BamAB and BamCDE, whereby BamD was inactivated. Then, this profoundly impaired the BAM complex, what resulted in a restricted insertion and assembly of OMPs. It was also proposed that the interaction with the periplasmic POTRA domain 5 of BamA leads to the activation of BamD.

The amino acid residue E373 in POTRA domain 5 was found to play an important role for the interaction with BamD and was expected to be involved in a direct manner in the activation of BamD. Unfolded BamA was also previously demonstrated to be copurified with and to bind to a soluble his-tagged BamD construct, that lacks the N-terminal lipid acylation site in vitro (Hagan et al., 2013; Hagan et al., 2015) Intermolecular FRET could be observed and was measured between fluorescence donor and acceptor pairs of IAEDANS-labeled single cysteine BamD mutants and 5-IAF-labeled single cysteine mutants of PD-BamA, with a much longer R0 of the IAEDANS-5-IAF pairs than the R0 of the tryptophan-IAEDANS pairs. The distances, that were obtained for the PD-BamA and BamD FRET pairs, were in the range from 73.4 Å to 162.9 Å. For these FRET experiments, three single cysteine mutants of PD-BamA were created for site-specific FRET studies with four different single cysteine mutants of BamD, whereby 12 donor-acceptor pairs of PD-BamA and BamD were used and the fluorescence spectra were recorded followed by their analysis. The FRET results for IAEDANS-labeled C230-BamD to the three mutants of PD-BamA showed notable energy transfer to the 5-IAF of the respective PD-BamA mutant. The spectra of the donor-IAEDANS were quenched with a corresponding reduction in the IAEDANS fluorescence intensity as a consequence of energy transfer to the acceptor-5-IAF, which was then excited and increasingly fluorescent. The donor-acceptor distances observed for C230-BamD and the three different mutants of PD-BamA, namely C363, C390 and C403, were in the range from 75.3 Å to 87.1 Å. Mutant C230-BamD is located on the C-terminal α-helix of TPR segment 5, so the obtained FRET and the donor-acceptor distances confirmed the predicted binding site on α10 of BamD (Dong et al., 2012). The charge distribution around α10 was found to be amphoteric with negatively charged residues located on one side close to position 230 and with positively charged residues located on the other side. The results obtained by

one side of POTRA domain 5, while negatively charged residues are clustered on the opposite side, leading to the suggestion that helix α10 is linked with the positively charged region of POTRA domain 5. FRET to all three mutants of PD-BamA was also observed for C181-BamD, located on the C-terminal long α7 of the BamD C-terminal part, with donor-acceptor distances ranging from 81.5 Å to 94.5 Å. The similarity of the distances obtained for these BamD mutants did not allow the determination of specific orientations for PD-BamA to BamD. Much more information would be available from the analysis of the donor nanosecond emission decay in time-resolved FRET experiments, whereby the analysis of donor-acceptor distance distributions could resolve the structural orientation and the flexibility of proteins in a solution. Steady-state measurements of FRET, therefore, provide the average distance between donor and acceptor (Klostermeier and Millar, 2001).

Conversely, however, only the FRET spectra obtained for PD-BamA mutant C363 showed transfer of energy from all four IAEDANS-labeled BamD mutants to its acceptor-5-IAF. This single cysteine PD-BamA mutant was designed to be located close to the proposed E373 interaction site (Ricci et al., 2012), that itself is located in the positively charged region of POTRA domain 5. No FRET was observed for PD-BamA mutants C390 and C403 with the mutants C75 and C114 of BamD. The energy transfer was calculated to be zero and therefore too low for an accurate estimation of the distance r. Since the efficiency of the resonance energy transfer is a sensitive function of r only over the range 0.5 R0 < r > 1.9 R0 (Fairclough and Cantor, 1978), with R0 that was calculated to be ~46 Å and the distances were estimated to be longer than 87.4 Å. BamD mutants C75 and C114 are located on the N-terminal helices of TPR2 and 3, respectively. The N-terminal region of BamD was shown to be structural similar to other proteins containing TPR motifs, that interact with C-terminal protein targeting sequences of unfolded OMP substrates for the BAM complex (Dong et al., 2012). Crystal structures of BamD revealed a cavity in the TPR1-3 scaffold that was proposed to be a substrate-binding pocket, playing a regulatory role in OMP biogenesis (Kim et al., 2011; Albrecht and Zeth, 2011; Sandoval et al., 2011).

As the POTRA domain 5 of BamA was supposed to be involved in substrate binding or to have a function in substrate transfer to the barrel domain, fluorescence resonance energy transfer was also used to determine possible interactions between PD-BamA and the substrate OmpA. The FRET study was performed with 12 different donor-acceptor pairs, whereby the three single cysteine PD-BamA mutants were labeled

with IAEDANS and served as donors and four 5-IAF-labeled mutants of OmpA served as acceptors. The lowest FRET and longest donor-acceptor distances were observed for the single cysteine mutant C363-PD-BamA, for which FRET could be already shown in experiments with BamD. Shorter distances and more energy transfer was observed for mutant C403-PD-BamA, while the FRET results for the various 5-IAF-labeled OmpA mutants and the IAEDANS-labeled PD-BamA mutant C390 show a large energy transfer to the 5-IAF. For this mutant, the observed donor-acceptor distances were in the range from 48.9 Å to 59.2 Å.

In C403-PD-BamA, the mutation is located close to the β-barrel next to β-strands β1 and β16, which were suggested to be involved in the assembly of OMPs. Noinaj et al.

(2013) proposed a highly dynamic membrane environment. By comparing the crystal structures of the BamA proteins from Neisseria gonorrhoeae and Haemophilus ducreyi and also by using molecular dynamics (MD) simulations, they could show that BamA was able to perturb the membrane by a reduced hydrophobic surface close to the β-strand β16, that decreased the lipid order and the membrane thickness and destabilized the outer membrane to prepare the membrane for the insertion of OMPs.

They also could show a transient separation of β-strands β1 and β16 that caused a lateral opening in the β-barrel, whereby both β-strands acted as templates for OMP folding. A possible POTRA gating motion was assumed that included POTRA domain 5 for being involved in the regulation of the substrate access to the inside of the β-barrel. Moreover, previous studies suggested the functional importance of the conserved VRGF/Y motif of the loop L6 in members of the Omp85 superfamily. It has been proposed that the long extracellular loop L6 of BamA that is located within the lumen of the β-barrel is essential for the function of BamA (Browning et al., 2013;

Leonard-Rivera and Misra, 2012), similarly to loop L6 that is located within the β-barrel of FhaC, a protein of the two-partner secretion system in Bordetella pertussis, which was found to be essential for the secretion of its substrate protein FHA (Clantin et al., 2007; Delattre et al., 2010).

Recently, two more crystal structures of the E. coli BamA β-barrel were determined (Ni et al., 2014; Albrecht et al., 2014) and the integration of OMP β-strands into the OM was also suggested at the interface of β-strands β1 and β16. The soluble loops or domains were thought to be transported into the extracellular space through a potential surface cavity located at the top of the interface of the two β-strands, that

Conformational changes in the β-barrel of BamA resulting in the surface exposure of loop L6 upon activation of the BAM complex by an unfolded substrate were proposed by Rigel et al. (2013), whereby these transitions were suggested to be regulated by the lipoproteins BamD and BamE indirectly through interactions with POTRA domain 5.

POTRA domain 5 was suggested to interact with the lipid membrane. Intrinsic tryptophan fluorescence spectroscopy was performed with a range of single tryptophan mutants of PD-BamA to explore conformational changes in the periplasmic domain of BamA with focus on POTRA domain 5. Binding to lipid membranes revealed strong effects of PD-BamA mutants W372 and W376. In dependence on the environment around their respective tryptophan, these two mutants demonstrated the strongest blue-shift in the emission wavelength and the highest increase in the fluorescence intensity, respectively. By using NMR, Sinnige et al.

(2015) could demonstrate conformational plasticity in the highly conserved region of POTRA domain 5, that encompasses the amino acid residue E373. This residue was shown to be important for the interaction with BamD and the function of the BAM complex and the plasticity was supposed to be unique to the function of POTRA domain 5. Both mutations in PD-BamA, W372 and W376, are located close to E373.

It seems that this region comprising the two α-helices, orients towards the lipid membrane, whereby the β-strands are pointing to the opposite direction. Also PD-BamA mutant W205 showed an increase in the fluorescence intensity upon membrane binding and confirmed previous reports on the flexibility at the interface of POTRA domains 2 and 3 (Kim et al, 2007; Gatzeva-Topalova et al., 2008).

However, further evidence is needed to demonstrate the role of the periplasmic domain of BamA. Further elucidation of conformational dynamics in terms of interactions with lipoproteins and substrate OMPs will provide insight into the molecular mechanism for the biogenesis of β-barrel membrane proteins.