Structural basis of lipoprotein recognition by the bacterial Lol trafficking chaperone LolA

Significance Lipoproteins in gram-negative bacteria underpin the formation and maintenance of the outer membrane that constitutes a vital protective barrier against antibiotics and other noxious molecules. An essential transport system comprising the LolABCDE proteins is required to traffic lipoproteins to the outer membrane. Following maturation on the inner membrane and extraction by the LolCDE transporter, lipoproteins are passed to the chaperone LolA that carries them across the periplasm prior to insertion into the outer membrane by the LolB receptor. Here, we report the molecular details of lipoprotein interaction with the chaperone LolA, a key intermediate located at the heart of the Lol pathway. The structure provides valuable insights into this important system and could be exploited to develop new antimicrobials.


Supplementary Figures
. Sequences of the N-terminal region of selected E. coli outer membrane lipoproteins. Protein sequences are aligned on the invariant cysteine (highlighted in green) which constitutes the first residue of the lipoprotein after maturation by the three inner-membrane proteins Lgt, Lsp and Lnt (1). Uniprot entry codes are displayed in gray. The corresponding conservation plot is presented in Figure 1B.    Zoomed-out view displaying the path that lipoprotein acyl chains have to follow to reach LolA with β10, β11 strands and the loop joining β8 and β9 colored in blue, yellow and red, respectively. (C) Accessible channel between the LolC Hook and LolA β-barrel with corresponding distance plot shown on the right. The vertical blue line represents the van der Waals diameter of carbon (3.4 Å), only the region located outside of this line (> 3.4 Å) would allow the simultaneous transfer of two proximal acyl chains. Analysis was performed with the MOLE server (3) using the LolC-LolA structure (6F3Z) and mutating residues located between the LolC Hook and the LolA β-barrel into alanine.   Representative fits and thermograms are presented in Figure 6 and Figure S4.

Construction of strains and plasmids
Primer sequences and constructs used in this study are detailed in Tables S4 and S5 respectively.
To construct an arabinose-inducible lolA conditional knockout strain (BW49), the lolA locus including the ribosome binding site was amplified using primers P1 and P2 and cloned into the EcoRI-XbaI sites of pBAD18 (13). The region encompassing the araC gene, pBAD promoter, lolA and the downstream terminator was amplified using primers P3 and P4, digested XbaI-SacI and cloned into the integration vector pLDR9 digested with the same enzymes. The construct was integrated into the lambda attB site of E. coli BW25113 according to a previously described protocol (8). The native copy of lolA was replaced with a kanamycin resistance cassette by amplifying pKD13 with primers P5 and P6 using the λ Red recombinase system as described (9) except that pSIM5 (10) was used for recombinase expression. Deletions were confirmed by PCR of the gene locus with primers P7 and P8.
To enable complementation of BW49, plasmid pAC80-LolA was constructed. A synthetic fragment encoding the lac T5 promoter derived from pQE80 (Qiagen) followed by the RBS, C-terminally strep-tagged lolA gene and a transcriptional terminator was synthesized (IDT DNA). This was then amplified with primers (P9/P10) and introduced by Gibson assembly (14) into pACYCDuet digested with EcoNI and AvrII to remove the T7 promoter and multiple cloning sites. LolA variants were created by two-step PCR reactions. In the first step, two reactions consisting of P9/lolA mutant_R and P10/lolA mutant_F primers were set up with pAC80-LolA as a template. Following purification, a mixture of these products was used as a template for a final PCR using primers P9 and P10. The resultant product was then introduced by Gibson assembly into digested pACYCDuet as described for the wild-type gene.
For periplasmic co-expression of LolA and Pal, full-length LolA (residues 1-203) was amplified and an internal NdeI site simultaneously removed by a two-step PCR from E. coli M1655 genomic DNA using primers P11/P14 and P12/P13. The products were purified, mixed and used as a template for a subsequent PCR reaction using primers P11 and P12. Following digestion with BspHI-BamHI, this product was inserted into the first multiple cloning site (MCS) of pCDFDuet (Novagen) digested NcoI-BamHI, resulting in pCDF-LolA. The pal gene was amplified from MG1655 genomic DNA using primers P15/P16, digested NdeI-XhoI and inserted into the 2 nd MCS of pCDF-LolA. The resultant vector pCDF-LolA-PalWT, encodes C-terminally Strep-tagged LolA and C-terminally His-tagged Pal. To enable the removal of the C-terminal His-tag on Pal, a fragment encoding a GS linker and TEV site (amino acid sequence: ENLYFQS) between the C-terminus of Pal and an octa-histidine tag was synthesized (IDT DNA) and cloned NdeI-XhoI into the 2 nd MCS of pCDF-LolA resulting in pCDF-LolA-PalWT(octa). To enable removal of the globular domain of Pal, constructs in which a TEV site was introduced at residues 34 and 49 of full-length Pal (residues 13 and 28, respectively, in the mature sequence) of full-length Pal were synthesized (IDT DNA) and cloned into the 2 nd site of pCDF-LolA using NdeI-XhoI resulting in LolA-PalTEV/FL and LolA-PalTEV/FL2 respectively. To permit removal of the His-tag from PalTEV/FL, a thrombin site was introduced immediately before the His-tag by Quikchange mutagenesis with primers P17 and P18 resulting in pCDF-LolA-PalTEV/FL(thrb). Plasmids pET28-LolA and pET28-mLolB expressing the mature domain of LolA (residues 22-203) and mLolB (residues 23-207) with an N-terminal His-tag were previously described (7). The R43L mutation was introduced into pCDF-LolA by Quikchange mutagenesis using primers P21 and P22. All constructs were verified by DNA sequencing (Source Bioscience).

Complementation assay of LolA variants
Strain BW49 bearing pAC80-LolA or the indicated variant was cultured in LB supplemented with 0.2% arabinose and appropriate antibiotics. The next day, 1 mL of overnight culture was centrifuged at 7000 g for 2 mins, washed twice in LB supplemented with 0.1% D-fucose and then diluted 1/10000 into fresh LB containing 0.1% fucose. Cells were then grown to an OD600 of 0.5, pelleted at 7000 g, washed twice in LB and then serial tenfold dilutions performed in LB. Dilutions were plated out on LB agar with no arabinose and grown overnight at 37 °C before imaging the next day. To assess expression level of the variant proteins, whole cell samples were resolved on SDS-PAGE, transferred to PVDF membrane, and immunoblotted with an anti-strep (IBA) and a dye conjugated Donkey anti-mouse secondary (Licor) antibodies. Immunoblots were revealed using an Odyssey Licor fluorescence imager.

Purification of wild-type LolA complexed with Pal
Cultures of E. coli C43 (DE3) carrying pCDF-LolA-PalWT(octa), pCDF-LolA-PalTEV/FL or pCDF-LolA-PalTEV/FL(thrb) as appropriate were grown in 2YT media supplemented with 50 μg/mL streptomycin and 0.5% glycerol at 37 °C until an OD600 of 0.4 was attained. Protein expression was then induced with 1 mM IPTG and growth continued for a further 5 hours. Cells were pelleted at 3250 g for 20 min and carefully resuspended in TSE buffer (200 mM Tris pH 8.0, 1 mM EDTA, 20% sucrose) supplemented with EDTA-free mini protease inhibitor tablets (Roche). Lysozyme (0.5 mg/mL final concentration) was then added and cells incubated on ice for 1 hour. The resulting spheroplasts were removed by centrifugation at 20000 g, 4 °C, 30 minutes prior to clarification of the supernatant by centrifugation (1h at 115000 g, 4 °C). The resulting periplasmic fraction was then diluted with an equal volume of water to reduce the sucrose concentration to 10%, supplemented with 5 mM MgCl2, 300 mM NaCl and 20 mM imidazole and loaded on a 5 mL FF Histrap column (GE Healthcare). The column was washed with 75 mL of 25 mM HEPES pH 7.5, 300 mM NaCl, 20 mM imidazole and then eluted with 10 mL of the same buffer containing 250 mM imidazole. Fractions were then dialyzed against 25 mM HEPES pH 7.5, 200 mM NaCl for 4 hours. To remove the His-tag, the protein was supplemented with 2 mM TCEP and cleaved with a 5:1 molar ratio of LolA-Pal complex:protease using TEV protease produced from plasmid pSH24-TEV (15). Thrombin cleavage used the Thrombin CleanCleave kit (Sigma) following the manufacturer's instructions. After digestion, the mixture was applied to 1 mL His trap FF column (GE Healthcare) with LolA-Pal complex found in the flow-through. Where required, a further purification step utilizing the C-terminal strep-tag on LolA was performed using Strep resin (Strep-Tactin XT 4Flow, IBA). Complexes were concentrated, snap frozen in liquid nitrogen and stored at -80 ºC.

Purification of R43L LolA-Pal complex
The complex was produced from the LolA R43L-PalTEV/FL construct where the TEV cleavage site was introduced at position 49 of full-length Pal (position 28 in the mature sequence). Proteins were produced in a similar manner to the wild-type LolA-Pal complex but with the following modifications. E. coli cells were induced at an OD600 of 0.9 for 2h only and pelleted at 3250 g for 30 min. The spheroplast treatment was carried for 1h at RT, and proteins were desalted in 20 mM HEPES pH 7.5, 200 mM NaCl after IMAC. No TCEP was added to the protein and TEV cleavage was performed with a molar ratio of 1:35 protease: complex overnight at 4 ºC, before removing the His-tagged Pal mature domain and the TEV protease with an excess of Ni resin. The resultant complex containing LolA R43L associated with the 28 N-terminal residues of Pal was concentrated to 6.4 mg/mL, snap frozen in liquid nitrogen and stored at -80 ºC.
Transfer of lipoproteins from LolA to mLolB. LolA-Pal complexes from which the His-tag had been cleaved (see above) were mixed at a 2:1 ratio to 15 µM of His-tagged soluble mLolB (7)  Fractions were analyzed on SDS-PAGE, transferred to a PVDF membrane and probed with anti-His (Qiagen) and a dye-conjugated Goat anti-mouse secondary (LI-COR). LolB or inner membrane protein AcrA (16) were detected using polyclonal rabbit antibodies and dye conjugated Donkey anti-rabbit secondary antibody. Bands were visualized using a LI-COR Odyssey imaging system.

Crystallization and structure determination
All crystals were grown at 15 °C by the sitting-drop vapor-diffusion method over a reservoir of 80 µL in MRC 2-drop plates (Molecular Dimensions).

Wild-type LolA-lipoprotein complex
Complexes of LolA and lipoprotein, derived from the N-terminal 13 residues of mature Pal lipoprotein, were concentrated to 11 mg/mL and mixed at a 2:1 protein:reservoir ratio in 1 µL final volume. Initial hits were obtained in 2.1 M DL-Malic acid pH 7.0 but diffracting crystals were grown at pH 6.0. Crystals were cryoprotected in the reservoir solution supplemented with 20% glycerol before being flash frozen in liquid nitrogen. Diffraction data were collected on beamline I04-1 at Diamond synchrotron, indexed and reduced with iMosflm (17), scaled with Aimless (4). The structure was solved by molecular replacement using Phaser (18) and LolA (1UA8). The model was further improved by several rounds of Refmac (5) and manual building in Coot (19). The final structure was validated with Rampage (6) and Procheck (20). Density inside the cavity of LolA enabled the location of the lipoprotein three acyl chains and corresponding triacylated cysteine but not of the other residues. Structural data were collected for another eight crystal structures from this construct as well as a complex containing wild-type LolA and the N-terminal 28 residues of mature Pal lipoprotein; all revealed identical positioning of the lipoprotein acyl chains.

LolA R43L-lipoprotein complex
Complexes of LolA R43L and lipoprotein, derived from the N-terminal 28 residues of mature Pal lipoprotein (LolA R43L-Pal28), were concentrated to 6.5 mg/mL and mixed at a 1:1 protein:reservoir ratio in 1 µL final volume. Diffracting crystals grew in 30% PEG3350, 50 mM bis-Tris pH 6.0 with seeds of crystals obtained in 10% 2-Propanol (v/v), 0.1 M HEPES pH 7.5 and 0.2 M NaCl. Crystals directly flash frozen in liquid nitrogen and data were collected under cryogenic conditions on beamline X06SA at SLS (Switzerland) on a EIGER 16M detector. The structure was solved as described for wild-type LolA-lipoprotein complex using the apo R43L LolA structure (2ZPD) for molecular replacement. As for the wild-type LolA complex, there was no density to model the protein sequence after the triacylated +1 cysteine.

Structure PDB depositions
Coordinates and structure factors were deposited in the Protein Data Bank under accession codes 7Z6W (LolA wild-type -lipoprotein), and 7Z6X (LolA R43L -lipoprotein).

Evaluation of A22 and MAC13243 inhibitors effect on Lol proteins
MAC13243 (Cambridge bioscience) and A22 (Sigma Aldrich) compounds were prepared in DMSO at 100 mM and tested at 0, 0.5, 1 or 2 mM in 2% DMSO final concentration.

Effect of inhibitors on
LolA-Pal complex. E. coli phospholipid bilayer coated silica beads (100 µL), prepared as previously described (21), were incubated with 15 µM LolA-Pal complex from which the C-terminal His-tag of Pal had been removed and A22 or MAC13243 compound in a buffer containing 50 mM Tris pH 8.5, 200 mM NaCl and 2% DMSO supplemented with 50 µM bovine serum albumin (Thermofisher) to reduce non-specific binding in a final volume of 250 µL. After 30 min of incubation under gentle agitation, the mixture was centrifuged at 16000 g for 30 seconds, the supernatant removed and the beads washed three times with 1 mL of the same buffer before elution in 25 µL of 2% SDS, 0.5 mM EDTA, 1 mM TCEP, 6 M urea, 50 mM Tris pH 9.5, and bromophenol blue. For each fraction, 1 µL of the SDS preparation was loaded on a gradient SDS-PAGE gel before imaging with an Odyssey (LI-COR) system.

Effect of inhibitors on LolC-LolA association and LolA to LolB lipoprotein transfer.
LolC-LolA interaction in the presence of A22 and MAC13243 was assessed as follows: tag-free LolA at 15 µM final concentration was pre-incubated with A22 or MAC13243 (0, 0.5, 1 or 2mM final concentrations) for 30 min in a buffer containing 25 mM HEPES pH 7.5, 150 mM NaCl and 2% DMSO. The same concentration of His-tagged LolC periplasmic domain (7) was added to the mixture and loaded in microbatch spin columns (Generon) with 100 µL nickel resin (Biorad) for 5 min. The resin was washed three times with 500 µL of the same buffer before elution of resinbound proteins with 250 µL of buffer containing 250 mM of imidazole. Eluted proteins were analyzed on gradient SDS-PAGE gels, imaged with an Odyssey (LI-COR) system and the results normalized to the values obtained without compound. Transfer of Pal from LolA to LolB in the presence of A22 or MAC13243 inhibitor was assessed in a similar manner by pre-incubating His-tagged soluble LolB (7) with the compounds prior to addition of tag-free LolA-Pal complex.

Isothermal titration calorimetry (ITC)
ITC experiments were performed with a VP-ITC calorimeter (Malvern Panalytical) at 25 °C in 20 mM HEPES pH 7.5, 200 mM NaCl. Injections were 10 µL except the first which was 5 µL and occurred every 200 s until the syringe was empty. Cell stirring speed was 300 rpm, reference power 25 µcal/s and initial delay 60 s. Association of wild-type LolA or the R43L variant with mLolB was measured by injecting 250 to 500 µM of LolA from the syringe into the cell containing mLolB at 25 or 30 µM. Binding between the LolC periplasmic domain and LolA R43L was assessed at a protein concentration of 450 µM in the syringe and 40 or 45 µM in the cell. LolA wild-type at 300 µM was injected onto LolC periplasmic domain at 25 µM, while the reverse data are from (7). For each ITC run, a control experiment was performed by injecting protein into a cell containing only buffer. Corresponding data were then subtracted from the protein-protein interaction as a linear fit. Raw data were fitted with PEAQ-ITC (Malvern Panalytical) software using the singlesite binding model for LolA wild-type or R43L interaction with mLolB and LolC while the twosite binding model was used for the LolA R43L-LolC data. Good fits were obtained for the biphasic data but inspection of reduced Chi-squared statistics by offsetting several values indicated that ΔH and -TΔS parameters were poorly defined by the fit.

Fluorescence spectroscopy
Fluorescence measurements of the fluorescent fatty acid probe 11-(dansylamino)undecanoic acid (DAUDA, Cayman Chemical) was performed in a FluoroLog (Horiba) spectrometer at room temperature in a final volume of 1.2 mL using a 1 cm quartz cuvette. DAUDA stock solution was prepared at a concentration of 1 mM in DMSO. Fluorescence was recorded between 400 and 600 nm in 20 mM HEPES pH 7.5, 150 mM NaCl in the presence of 10 µM DAUDA, after excitation at 335 nm. His-tagged LolA WT or R43L and tag-free LolC periplasmic domain proteins were mixed in a LolC:LolA ratio of 2:1. Measurements were performed with LolA concentrations of 0, 5, 10, 15, 20, 30 and 40 µM after briefly stirring the solution with a magnet.

Molecular Docking
The protein-protein association between lipoprotein-bound LolA after removal of the ligand (7Z6W) and mLolB (1IWM, chain A) was predicted with the ClusPro 2.0 server (22). We selected the docking solution corresponding to the second most populated cluster according to ClusPro