Structural Investigation of the Interaction between LolA and LolB Using NMR
2009; Elsevier BV; Volume: 284; Issue: 36 Linguagem: Inglês
10.1074/jbc.m109.001149
ISSN1083-351X
AutoresS Nakada, Masayoshi Sakakura, Hideo Takahashi, Suguru Okuda, Hajime Tokuda, Ichio Shimada,
Tópico(s)Clostridium difficile and Clostridium perfringens research
ResumoLipoproteins that play critical roles in various cellular functions of Gram-negative bacteria are localized in the cells inner and outer membranes. Lol proteins (LolA, LolB, LolC, LolD, and LolE) are involved in the transportation of outer membrane-directed lipoproteins from the inner to the outer membrane. LolA is a periplasmic chaperone that transports lipoproteins, and LolB is an outer membrane receptor that accepts lipoproteins. To clarify the structural basis for the lipoprotein transfer from LolA to LolB, we examined the interaction between LolA and mLolB, a soluble mutant of LolB, using solution NMR spectroscopy. We determined the interaction mode between LolA and mLolB with conformational changes of LolA. Based upon the observations, we propose that the LolA·LolB complex forms a tunnel-like structure, where the hydrophobic insides of LolA and LolB are connected, which enables lipoproteins to transfer from LolA to LolB. Lipoproteins that play critical roles in various cellular functions of Gram-negative bacteria are localized in the cells inner and outer membranes. Lol proteins (LolA, LolB, LolC, LolD, and LolE) are involved in the transportation of outer membrane-directed lipoproteins from the inner to the outer membrane. LolA is a periplasmic chaperone that transports lipoproteins, and LolB is an outer membrane receptor that accepts lipoproteins. To clarify the structural basis for the lipoprotein transfer from LolA to LolB, we examined the interaction between LolA and mLolB, a soluble mutant of LolB, using solution NMR spectroscopy. We determined the interaction mode between LolA and mLolB with conformational changes of LolA. Based upon the observations, we propose that the LolA·LolB complex forms a tunnel-like structure, where the hydrophobic insides of LolA and LolB are connected, which enables lipoproteins to transfer from LolA to LolB. Gram-negative bacteria express lipid-modified proteins, lipoproteins, which are anchored to the cellular membrane via acyl chains attached to N-terminal cysteine residues of the lipoproteins. Putative lipoproteins have been found in various bacteria. For example, Escherichia coli has at least 90 types of lipoproteins (1.Tokuda H. Matsuyama S. Tanaka-Masuda K. Ehrmann M. The Periplasm. American Society for Microbiology, Washington, D. C.2007: 67-79Google Scholar), and the Lyme disease spirochete Borrelia burgdorferi has 105 putative lipoproteins (2.Fraser C.M. Casjens S. Huang W.M. Sutton G.G. Clayton R. Lathigra R. White O. Ketchum K.A. Dodson R. Hickey E.K. Gwinn M. Dougherty B. Tomb J.F. Fleischmann R.D. Richardson D. Peterson J. Kerlavage A.R. Quackenbush J. Salzberg S. Hanson M. van Vugt R. Palmer N. Adams M.D. Gocayne J. Weidman J. Utterback T. Watthey L. McDonald L. Artiach P. Bowman C. Garland S. Fuji C. Cotton M.D. Horst K. Roberts K. Hatch B. Smith H.O. Venter J.C. Nature. 1997; 390: 580-586Crossref PubMed Scopus (1722) Google Scholar). Although little is known about the functions of the majority of lipoproteins, some of the lipoproteins play essential roles in various cellular functions of Gram-negative bacteria, such as cell surface structure stabilization, cell shape maintenance, substrate transport, cell growth, and cell signaling (3.Braun V. Wu H.C. Ghuysen J.-M. Hakenbeck R. Bacterial Cell Wall. Elsevier Science Publishers B. V., Amsterdam1994: 319-341Google Scholar). Lipoproteins are located at three cellular membrane sites; they are the periplasmic side of the inner membrane, the periplasmic side of the outer membrane, and the outside of the outer membrane (4.Pugsley A.P. Microbiol. Rev. 1993; 57: 50-108Crossref PubMed Google Scholar). In E. coli most of the lipoproteins are anchored to the periplasmic side of the outer membrane, whereas others are anchored to that of the inner membrane (1.Tokuda H. Matsuyama S. Tanaka-Masuda K. Ehrmann M. The Periplasm. American Society for Microbiology, Washington, D. C.2007: 67-79Google Scholar). Therefore, the transportation of the lipoproteins to the outer membrane is essential for E. coli. Five Lol proteins, LolA, LolB, LolC, LolD, and LolE, play central roles in the outer membrane-directed lipoprotein localization. The Lol·CDE complex, which is anchored to the inner membrane, transfers the lipoproteins from the membrane to a soluble monomer periplasmic protein, LolA (182 amino acids) in an ATP-dependent manner (5.Yakushi T. Yokota N. Matsuyama S. Tokuda H. J. Biol. Chem. 1998; 273: 32576-32581Abstract Full Text Full Text PDF PubMed Scopus (52) Google Scholar, 6.Yakushi T. Masuda K. Narita S. Matsuyama S. Tokuda H. Nat. Cell Biol. 2000; 2: 212-218Crossref PubMed Scopus (204) Google Scholar, 7.Narita S. Tanaka K. Matsuyama S. Tokuda H. J. Bacteriol. 2002; 184: 1417-1422Crossref PubMed Scopus (63) Google Scholar). LolA transports the lipoproteins from the inner membrane through the periplasmic space to the outer membrane and transfers them to an outer membrane lipoprotein, LolB (186 amino acids). LolB is anchored to the membrane by acyl chains attached to its N-terminal cysteine, and it finally inserts the lipoproteins into the outer membrane (8.Matsuyama S. Yokota N. Tokuda H. EMBO J. 1997; 16: 6947-6955Crossref PubMed Scopus (171) Google Scholar, 9.Yokota N. Kuroda T. Matsuyama S. Tokuda H. J. Biol. Chem. 1999; 274: 30995-30999Abstract Full Text Full Text PDF PubMed Scopus (78) Google Scholar, 10.Tanaka K. Matsuyama S.I. Tokuda H. J. Bacteriol. 2001; 183: 6538-6542Crossref PubMed Scopus (70) Google Scholar). Among the Lol proteins the crystal structures of LolA and LolB have been solved. As for LolB, the soluble mutant of LolB, mLolB, in which the N-terminal cysteine residue was replaced with an alanine residue, was used for the structural analysis. Although LolA and mLolB share only 8% primary sequence identity, their tertiary structures are similar to each other (11.Takeda K. Miyatake H. Yokota N. Matsuyama S. Tokuda H. Miki K. EMBO J. 2003; 22: 3199-3209Crossref PubMed Scopus (107) Google Scholar). The structures of both LolA and mLolB resemble an open β-barrel with a lid. The convex side of the β-barrel is fully solvent-exposed, whereas the concave side is partly exposed (supplemental Fig. S1). The open β-barrels of LolA and LolB comprise 11 antiparallel β-strands (β1–β11) and an extra β-strand, β12 for LolA and β11′ for LolB. The lid is composed of three α-helices (α1–α3) and is embedded in the concave side of the β-barrel. The concave sides of LolA and LolB contain many hydrophobic residues. Therefore, this concave side of the proteins is speculated to be the binding site for the hydrophobic acyl chains of lipoproteins. Interestingly, one of the crystal structures of LolB accommodated a molecule of polyethylene glycol 2000 monomethyl ether, PEGMME2000, on the hydrophobic surface of the concave side (supplemental Fig. S1). The specific interaction between LolA and LolB is a decisive step in correctly sorting lipoproteins from LolA via LolB to the outer membrane. However, the structural aspects of the interaction, which would clarify how LolA transfers lipoproteins to LolB, remain unknown. To address this issue, we focused on the interaction between LolA and LolB. Here we investigated the interaction of LolA with LolB by NMR spectroscopy. We used LolA with a His6 tag and mLolB, which retain the biological activities similar to those of the wild type protein (8.Matsuyama S. Yokota N. Tokuda H. EMBO J. 1997; 16: 6947-6955Crossref PubMed Scopus (171) Google Scholar, 12.Narita S. Kanamaru K. Matsuyama S. Tokuda H. Mol. Microbiol. 2003; 49: 167-177Crossref PubMed Scopus (31) Google Scholar). By exploiting the cross-saturation and paramagnetic relaxation enhancement (PRE) 2The abbreviations used are: PREparamagnetic relaxation enhancementMTSL(1-oxyl-2,2,5,5-tetramethyl-Δ3-pyrroline-3-methyl)methanethiosulfonateTROSYtransverse relaxation-optimized spectroscopy. 2The abbreviations used are: PREparamagnetic relaxation enhancementMTSL(1-oxyl-2,2,5,5-tetramethyl-Δ3-pyrroline-3-methyl)methanethiosulfonateTROSYtransverse relaxation-optimized spectroscopy. techniques, we successfully determined the interfacial residues of LolA and mLolB and the relative orientation of the two molecules in the complex. In addition, we identified the binding sites of an acyl chain analogue, decanoate, on LolA and mLolB. The results obtained from the present study not only explain how LolA might achieve lipoprotein transfer to LolB but also may provide new insights into the structural and functional aspects of other fatty acid-binding proteins. paramagnetic relaxation enhancement (1-oxyl-2,2,5,5-tetramethyl-Δ3-pyrroline-3-methyl)methanethiosulfonate transverse relaxation-optimized spectroscopy. paramagnetic relaxation enhancement (1-oxyl-2,2,5,5-tetramethyl-Δ3-pyrroline-3-methyl)methanethiosulfonate transverse relaxation-optimized spectroscopy. E. coli BL21 (DE3) Gold (Stratagene) cells were transfected with the plasmid bearing the LolA or mLolB gene (13.Nakada S. Takahashi H. Sakakura M. Kurono M. Shimada I. Biomol. NMR Assign. 2007; 1: 125-127Crossref PubMed Scopus (2) Google Scholar, 14.Nakada S. Sakakura M. Takahashi H. Tokuda H. Shimada I. Biomol. NMR. Assign. 2007; 1: 121-123Crossref PubMed Scopus (2) Google Scholar). For unlabeled LolA or mLolB, the cells were grown in Luria-Bertani broth. For proteins uniformly labeled with 2H and 15N and/or 13C, the cells were grown in M9 minimal medium containing 15NH4Cl (1 g liter−1; Spectra Stable Isotopes) and [2H6]glucose or [13C6/2H6]glucose (98% labeled; 3 g liter−1; Spectra Stable Isotopes) in 99% D2O. LolA and mLolB were purified as described elsewhere (13.Nakada S. Takahashi H. Sakakura M. Kurono M. Shimada I. Biomol. NMR Assign. 2007; 1: 125-127Crossref PubMed Scopus (2) Google Scholar, 14.Nakada S. Sakakura M. Takahashi H. Tokuda H. Shimada I. Biomol. NMR. Assign. 2007; 1: 121-123Crossref PubMed Scopus (2) Google Scholar). The inhibitory effect of decanoate (Sigma) on the incorporation of lipoproteins into the outer membrane was examined. A spheroplast supernatant containing the LolA·[35S]L10P complex was mixed with 0.2 mg ml−1 LolB-depleted outer membranes, as described previously (5.Yakushi T. Yokota N. Matsuyama S. Tokuda H. J. Biol. Chem. 1998; 273: 32576-32581Abstract Full Text Full Text PDF PubMed Scopus (52) Google Scholar, 15.Matsuyama S. Tajima T. Tokuda H. EMBO J. 1995; 14: 3365-3372Crossref PubMed Scopus (164) Google Scholar, 16.Yakushi T. Tajima T. Matsuyama S. Tokuda H. J. Bacteriol. 1997; 179: 2857-2862Crossref PubMed Google Scholar, 17.Terada M. Kuroda T. Matsuyama S.I. Tokuda H. J. Biol. Chem. 2001; 276: 47690-47694Abstract Full Text Full Text PDF PubMed Scopus (95) Google Scholar, 18.Wada R. Matsuyama S. Tokuda H. Biochem. Biophys. Res. Commun. 2004; 323: 1069-1074Crossref PubMed Scopus (6) Google Scholar). mLolB (0.08 μg) was pretreated with the specified concentrations of decanoate in 20 mm Tris-HCl (pH 7.5) for 3 h on ice. A 10-μl aliquot of this decanoate-treated mLolB solution was added to 90 μl of a mixture containing the LolA·[35S]L10P complex and outer membranes followed by incubation at 30 °C for 1 h. The membrane incorporation of [35S]L10P was examined by SDS-PAGE and fluorography as described (15.Matsuyama S. Tajima T. Tokuda H. EMBO J. 1995; 14: 3365-3372Crossref PubMed Scopus (164) Google Scholar) after fractionation into pellet and supernatant by centrifugation at 100,000 × g for 30 min. We conducted chemical shift perturbation experiments of LolA and mLolB upon the addition of decanoate. Decanoate was titrated to 15N-labeled samples of LolA (0.28 mm) or mLolB (0.28 mm), and each 1H,15N TROSY spectrum was recorded. The weighted sum of the 1H and 15N chemical shift changes for each residue was calculated with the equation Δδwei = Δδ(1HN) + 0.2 Δδ(15N) upon the addition of 5 mm decanoate. To determine the dissociation constant (Kd) of the LolA·mLolB complex, we conducted isothermal titration calorimetry measurements using a VP-ITC (MicroCal) instrument at 25 °C. LolA and mLolB samples were prepared in a buffer containing 66 mm sodium phosphate (pH 6.9) and 100 mm NaCl. mLolB was injected into LolA in 29 aliquots of 10 μl each at 240-s intervals. The data acquired from the mLolB injections into the buffer were subtracted from the experimental data. The data were analyzed using the MicroCal Origin software. NMR samples (0.3–0.5 ml) were prepared using 0.3–2.0 mm LolA and/or mLolB in the following buffers. For the assignments of LolA in the LolA·mLolB complex and the chemical shift perturbation experiments of LolA with mLolB or decanoate, NMR samples (0.3–0.5 ml) using buffer A (66 mm sodium phosphate (pH 6.9), 100 mm NaCl) were used. For other NMR measurements buffer B (38 mm 2-morpholinoethanesulfonic acid (pH 6.9) was used. For cross-saturation experiments buffer B contained 60% 2H2O. For other NMR experiments the buffers contained 10% 2H2O. For the dimensional NMR experiments of the LolA assignments in the complex state with mLolB, a Bruker AVANCE 800 MHz spectrometer was used. For the cross-saturation experiments, a Bruker AVANCE 600 MHz spectrometer with a cryogenic probe was used. For the other NMR experiments, a Varian UNITY-INOVA 500 MHz spectrometer was used. All NMR spectra were recorded at 37 °C. Backbone resonance assignments of LolA and mLolB in the free state were performed as described elsewhere (13.Nakada S. Takahashi H. Sakakura M. Kurono M. Shimada I. Biomol. NMR Assign. 2007; 1: 125-127Crossref PubMed Scopus (2) Google Scholar, 14.Nakada S. Sakakura M. Takahashi H. Tokuda H. Shimada I. Biomol. NMR. Assign. 2007; 1: 121-123Crossref PubMed Scopus (2) Google Scholar). The assignments of LolA and mLolB in the LolA·mLolB complex were derived from titration experiments, and the assignments of free state LolA and mLolB were supported by three-dimensional NMR experiments. In the titration experiments 1H,15N TROSY spectra of the labeled LolA or mLolB with a series of concentrations of the unlabeled binding partner were measured. The molar ratios of unlabeled mLolB or LolA to labeled LolA or mLolB were varied from 0 to 5.5. At the molar ratio of 3:1 (unlabeled protein/labeled protein), where no more chemical shift changes of the labeled proteins occurred, the weighted sum of 1H and 15N chemical shift changes was calculated with the equation Δδwei = Δδ(1HN) + 0.2 Δδ(15N). In the cross-saturation experiments (19.Takahashi H. Nakanishi T. Kami K. Arata Y. Shimada I. Nat. Struct. Biol. 2000; 7: 220-223Crossref PubMed Scopus (278) Google Scholar, 20.Nakanishi T. Miyazawa M. Sakakura M. Terasawa H. Takahashi H. Shimada I. J. Mol. Biol. 2002; 318: 245-249Crossref PubMed Scopus (100) Google Scholar) we prepared a sample containing LolA or mLolB labeled with 2H and 15N and the unlabeled binding partner. The molar ratio of the unlabeled protein to the labeled protein was set to 4:1. Saturation transfer from the unlabeled protein to the labeled protein was made using the WURST decoupling scheme with the saturation frequency set at 0.833 ppm. The saturation time was 2.0 s, and relaxation delay was 2.0 s. All spectra were processed using nmrPipe (21.Delaglio F. Grzesiek S. Vuister G.W. Zhu G. Pfeifer J. Bax A. J. Biomol. NMR. 1995; 6: 277-293Crossref PubMed Scopus (11529) Google Scholar). Data analysis was facilitated by the Sparky software (T. D. Goddard and D. G. Kneller, Sparky 3, University of California, San Francisco). For site-directed spin labeling, five single Cys substituted mutants of LolA, i.e. V24C, V32C, I58C, L59C, and Q145C, were prepared. The proteins were modified with the spin-label reagent MTSL ((1-oxyl-2,2,5,5-tetramethyl-Δ3-pyrroline-3-methyl)methanethiosulfonate)(Toronto Research Chemicals), which attaches the nitroxide spin label via a disulfide bond to the single cysteine (22.Berliner L.J. Grunwald J. Hankovszky H.O. Hideg K. Anal. Biochem. 1982; 119: 450-455Crossref PubMed Scopus (261) Google Scholar). The spin label reagent MTSL in acetonitrile was added to the solution of each LolA mutant at a molar ratio of 7–8:1 (MTSL:LolA) and incubated at 25 °C for 20–24 h. After the reaction, the buffer was changed to buffer B, and the excess MTSL was removed. The molar ratio of each spin-labeled LolA to 15N-labeled mLolB was set to 3:1, where no more chemical shift changes of 15N-labeled mLolB occurred. 1H,15N TROSY spectra of 15N-labeled mLolB with each spin-labeled LolA were recorded before and after the MTSL was reduced by an incubation with ascorbate (Wako) at room temperature for 2 h. The analyses of lipoprotein interactions with LolA and LolB are important to elucidate the molecular mechanism of lipoprotein localization. However, no structural information about the interaction is currently available because of the low aqueous solubility of the lipoproteins due to their acyl chains. In the present study we used a saturated fatty acid as an analogue of the acyl chains of lipoproteins. In the major lipoproteins a saturated hydrocarbon with 16 carbon atoms is most frequently adopted as the acyl chains (23.Hantke K. Braun V. Eur. J. Biochem. 1973; 34: 284-296Crossref PubMed Scopus (277) Google Scholar, 24.Sankaran K. Wu H.C. J. Biol. Chem. 1994; 269: 19701-19706Abstract Full Text PDF PubMed Google Scholar). However, the aqueous solubility of the saturated fatty acid with 16 carbon atoms, palmitic acid, was less than 0.5 mm at room temperature, which was too low to perform NMR analyses. On the other hand, decanoate, a saturated fatty acid with 10 carbon atoms, possesses the sufficient solubility of more than 200 mm and is suitable for the NMR experiments. To assess the validity of the use of decanoate as the analogue of an acyl chain of the lipoprotein, we examined the inhibitory effect of decanoate on the lipoprotein (L10P) transfer activity of mLolB (Fig. 1). The amount of L10P incorporated into the outer membrane decreased depending on the concentration of decanoate preincubated with mLolB. These results suggested that the lipoprotein transfer reactions from LolA to mLolB and/or mLolB to outer membranes are sensitive to decanoate. To determine the decanoate binding sites on LolA and mLolB, chemical shift perturbation experiments were performed (Fig. 2) (supplemental Fig. S2). Here we calculated the weighted chemical shift changes (Δδwei) using the equation (25.Hajduk P.J. Dinges J. Miknis G.F. Merlock M. Middleton T. Kempf D.J. Egan D.A. Walter K.A. Robins T.S. Shuker S.B. Holzman T.F. Fesik S.W. J. Med. Chem. 1997; 40: 3144-3150Crossref PubMed Scopus (154) Google Scholar, 26.Meyer B. Peters T. Angew. Chem. Int. Ed. Engl. 2003; 42: 864-890Crossref PubMed Scopus (893) Google Scholar) Δδwei = Δδ(1HN) + 0.2 Δδ(15N). Upon the addition of 5 mm decanoate, 11 residues (supplemental Table S1) of LolA showed Δδwei of >0.1 ppm. Among the affected residues, Gly-86 is located on the loop preceding α-helix 2. Phe-90 and Met-91 are located on α-helix 2, embedded in the hydrophobic cavity. The other affected residues are located on β-strands 1–2, 9, and 11 (Fig. 2A). Among the 11 affected residues, 6 hydrophobic residues are located in the core of the molecule, and their side chains are directed toward the concave side of the β-barrel. Although the other affected residues, Gln-33, Gly-86, Thr-128, Lys-155, and Ser-156, direct their side chains toward the outside of the protein surface, the amide groups of Gly-86, Thr-128, Lys-155, and Ser-156 are in close proximity to the affected hydrophobic residues. The observation that the residues with substantial chemical shift changes were mainly located inside the concave surface indicates that LolA accommodates decanoate on the concave side. Upon the addition of decanoate, 12 residues (supplemental Table S1) of LolB showed Δδwei of >0.4 ppm. These residues are located on α-helices 2–3, β-strands 1–2, 9, and 11 (Fig. 2B). Among the 12 affected residues, Ale-38 and Thr-55 direct their side chains to the convex side of the β-barrel, whereas the other residues direct their side chains toward the concave side of the β-barrel. Interestingly, the side chains of Thr-105, Met-107, Leu-114, and Ile-118 are located in close proximity to the binding site of PEGMME2000 found in the crystal structure, suggesting that decanoate and PEGMME2000 bind to LolB in a similar manner. Using isothermal titration calorimetry analyses, we determined the dissociation constant Kd of LolA for mLolB. The analysis revealed a Kd value of 30.7 × 10−6m and a stoichiometry of 1.1 mol mLolB/mol of LolA. These results indicate that LolA and mLolB can specifically interact with each other even in the absence of a lipoprotein. By using LolA uniformly labeled with 2H, 13C, and 15N ([13C/2H/5N]LolA) and mLolB labeled with 2H and 15N ([2H/15N]mLolB), we recorded 1H,15N TROSY spectra. The superimposed 1H,15N TROSY spectra of [13C/2H/5N]LolA (Fig. 3A) and [2H/15N]mLolB (Fig. 3B) with various amounts of the binding partner are shown. We previously established the backbone resonance assignments of [13C/2H/5N]LolA and [2H/15N]mLolB in their free states (13.Nakada S. Takahashi H. Sakakura M. Kurono M. Shimada I. Biomol. NMR Assign. 2007; 1: 125-127Crossref PubMed Scopus (2) Google Scholar, 14.Nakada S. Sakakura M. Takahashi H. Tokuda H. Shimada I. Biomol. NMR. Assign. 2007; 1: 121-123Crossref PubMed Scopus (2) Google Scholar). The backbone resonance assignments of [13C/2H/5N]LolA and [2H/15N]mLolB complexed with each binding partner were determined by following the chemical shift changes upon the titration of each binding partner. Triple-resonance experiments were also performed to confirm the assignments of the complexed forms. Of 175 non-proline residues in LolA and mLolB, 82 and 84% of the resonances originating from the backbone amide groups could be assigned in their bound states, respectively, although the other resonance assignments could not be established, mainly because of line-broadening upon the addition of the binding partner. Upon the addition of a 3-fold molar ratio of mLolB, 34 residues of LolA exhibited Δδwei of >0.3 ppm, and 20 residues of LolA showed Δδwei of 0.2–0.3 ppm (supplemental Table S2). In addition, the cross-peaks originating from 16 residues of LolA disappeared upon the addition of mLolB. Upon the addition of a 3-fold molar ratio of LolA, 12 residues of mLolB exhibited Δδwei of >0.3 ppm, and 9 residues of mLolB showed Δδwei of 0.2–0.3 ppm (supplemental Table S2). In addition, the cross-peaks originating from nine residues of mLolB disappeared upon the addition of LolA. To identify the residues located on the interface between LolA and mLolB in the complex, we used cross-saturation methods. Either LolA or mLolB labeled with 2H, 15N, and/or 13C was bound to its unlabeled binding partner. LolA/mLolB molecule-selective saturation was achieved by applying the radio frequency irradiation centered at 0.833 ppm, which corresponds to the CH2/CH3 protons within the non-labeled protein. Residue-selective signal intensity reductions were observed with an irradiation length of 2.0 s. Among the 131 analyzed signals originating from the main chain amide groups of LolA, three signals from Val-32, Gln-75, and Gln-145 exhibited intensity reduction ratios of >0.5, and signals from 25 residues showed intensity reduction ratios from 0.3 to 0.5 (Fig. 4A) (residue lists in supplemental Table S3). The residues affected by the saturation are indicated on the crystal structure of LolA (Fig. 4B). Most of the affected residues were located along the edge of the β-barrel. Specifically, the N-terminal ends of β-strands 2, 6, 9, and 11, the C-terminal ends of β-strands 1 and 10, the middle regions of β-strands 3 and 5, and all of β-strand 4 are involved in the LolB recognition. In addition to the affected β-barrel residues, 2 residues on α-helix 2, Leu-92 and Ala-94, exhibited remarkable signal intensity reductions, indicating that α-helix 2 is part of the mLolB binding surface. Among the affected residues, five acidic residues, Glu-34, Glu-56, Asp-146, Asp-147, and Asp-178, and one basic residue, Arg-149, are located along the edge of the β-barrel structure. Among the 123 analyzed signals originating from the main chain amide groups of mLolB, signals from 8 residues exhibited intensity reduction ratios of >0.5, and signals from 14 residues showed intensity reduction ratios from 0.3 to 0.5 (Fig. 5A) (residue lists are shown in supplemental Table S3). The affected residues are predominantly located on β-strands 1 and 2 and the loop connecting β-strands 5 and 6 (Fig. 5B). These residues are clustered in a limited area on the convex surface of the β-barrel of mLolB. The interfacial surface of mLolB includes six basic residues, whereas no acidic residue is present. Among the six basic residues, Arg-34, Lys-45, Arg-49, Arg-91, and Lys-177 are located on the β-strands. Lys-88 is located in the loop connecting β-strands 5 and 6. In contrast to LolA, the α-helices of mLolB in the complex are not affected, which indicates that the convex side of mLolB is involved in the interaction. To determine the relative orientations of LolA and mLolB, we introduced site-directed spin-labeling and performed PRE experiments. This technique provides distance information from the spin center up to 20–25 Å (27.Gillespie J.R. Shortle D. J. Mol. Biol. 1997; 268: 158-169Crossref PubMed Scopus (272) Google Scholar). Spin labels were introduced at each of positions 24, 32, 58, 59, and 145 of LolA (Fig. 6A). The residues at these positions were identified as the interfacial residues by the cross-saturation method. The residues at positions 24, 32, and 145 are on the edge of the β-barrel. On the other hand, the residues at positions 58 and 59 are in the middle of the β-barrel. The side chain at position 58 is directed toward the convex side of the β-barrel, whereas the side chain at position 59 points toward the concave side of the β-barrel. We recorded the 1H,15N TROSY spectra of 15N-labeled mLolB complexed with LolA spin-labeled with MTSL in the presence and absence of a reducing agent, ascorbic acid, which reduces the spin-label agent covalently attached to each of the five positions of LolA. The chemical shifts of the amide resonances originating from mLolB complexed with each spin-labeled LolA in the reduced states were almost identical to those observed for mLolB complexed with the wild type LolA, suggesting that the cysteine mutations and the MTSL labeling for LolA exerted minimal influence on the conformations of LolA and mLolB and the binding mode between LolA and mLolB. We compared the signal intensities of the spectra with and without ascorbic acid. The mLolB residues that showed marked signal intensity reductions are mapped on the crystal structure (Figs. 6, B–F) (complete lists of affected residues are shown in supplemental Table S4). Upon complex formation with LolA spin-labeled at position 24, the backbone amide signals originating from 11 residues were broadened to undetectable levels. The signal intensities of six residues were remarkably reduced (intensity ratio <0.3) (Fig. 6B). Upon complex formation with LolA spin-labeled at position 32, the signals originating from 18 residues disappeared. The signals from seven residues showed reduced intensities (intensity ratio <0.3) (Fig. 6C). Upon complex formation with LolA spin-labeled at position 59, the signals originating from eight residues were broadened to undetectable levels. The signal intensity of one residue, Tyr-39 was reduced (intensity ratio <0.3) (Fig. 6E). Upon complex formation with LolA spin-labeled at position 145, the signals originating from 10 residues were broadened to undetectable levels. The signal intensities of seven residues were reduced (intensity ratio <0.3) (Fig. 6F). In contrast, upon complex formation with LolA spin-labeled at position 58, only one residue on mLolB, Tyr-47, displayed a weak signal intensity reduction (intensity ratio 0.3 ppm, respectively. Red bars indicate residues that disappeared upon complex formation. B, the affected residues in the chemical shift perturbation experiments are mapped on the crystal structures of mLolB (ribbon representations are shown in the left panel, and surface representations are shown in the right panel). The same coloring system is used as panel A. Proline residues and unassigned residues in a free state are colored cyan. The other residues that are assigned but unaffected are white.View Large Image Figure ViewerDownload Hi-res image Download (PPT) The residues that displayed the chemical shift perturbation upon complex formation are indicated on the crystal structures (Figs. 7B and 8B). The affected LolA residues are spread all over the β-barrel (Fig. 7B). In contrast, the residues of mLolB with the substantial chemical shift perturbations upon LolA binding were almost identical to the interfacial residues identified in the cross-saturation experiments (Figs. 5B and 8B). Considering the fact that the chemical shift perturbations reflect secondary effects, which are the effects of the conformational changes caused by binding as well as the direct effects of the binding, we conclude that LolA undergoes a conformational change of its β-barrel to an open form upon mLolB binding as represented in Fig. 9, whereas the structure of mLolB does not. We also observed a substantial difference in the hydrogen/deuterium (H/D) exchange rates of the amide groups between LolA and mLolB in the free state (supplemental “Experimental Procedures” and Fig. S4). This observation suggests that LolA exhibits higher plasticity than mLolB, supporting the conformational changes of LolA upon complex formation. Furthermore, we measured 13Cα and 13Cβ chemical shifts of labeled LolA in the presence and absence of unlabeled mLolB. The small but substantial changes in the 13Cα and 13Cβ chemical shifts of LolA were observed upon binding of mLolB (supplemental Fig. S5), suggesting that some conformational changes occur. In the LolA·mLolB binding mode suggested by the present study, a contiguous hydrophobic surface, which is composed of part of the concave side of LolA and mLolB, is generated upon complex formation. On the basis of the NMR analyses, we propose the following mechanism of lipoprotein transfer from LolA to LolB (Fig. 9). 1) LolA receives a lipoprotein from the Lol·CDE complex in the inner membrane. In the LolA-lipoprotein complex, LolA accommodates the acyl chains of the lipoprotein on its hydrophobic concave side and undergoes some conformational changes (29.Watanabe S. Oguchi Y. Takeda K. Miki K. Tokuda H. J. Biol. Chem. 2008; 283: 25421-25427Abstract Full Text Full Text PDF PubMed Scopus (14) Google Scholar, 30.Oguchi Y. Takeda K. Watanabe S. Yokota N. Miki K. Tokuda H. J. Biol. Chem. 2008; 283: 25414-25420Abstract Full Text Full Text PDF PubMed Scopus (31) Google Scholar). LolA then transports the lipoprotein from the inner to the outer membrane. 2) The LolA-lipoprotein complex interacts with LolB, which is anchored to the inner side of the outer membrane. In the LolA·LolB complex, the concave side of the β-barrel of LolA contacts with the convex side of the β-barrel of LolB. Accordingly, the hydrophobic insides of LolA and LolB are connected, and a hydrophobic tunnel-like structure is formed. 3) The acyl chains in LolA are smoothly transferred through the hydrophobic tunnel, from LolA to LolB. The driving force of the lipoprotein transfer from LolA to LolB might simply be the higher affinity of a lipoprotein for LolB than for LolA. Good evidence for this can be found in the previous study (31.Taniguchi N. Matsuyama S. Tokuda H. J. Biol. Chem. 2005; 280: 34481-34488Abstract Full Text Full Text PDF PubMed Scopus (48) Google Scholar). They revealed that an outer membrane-directed lipoprotein, Pal, forms a tighter complex with mLolB than with LolA. Therefore, the difference in the affinities between LolA and LolB is responsible for the lipoprotein transfer in the complex. Although several fatty acid-protein interactions have been investigated based on the structural analysis (32.Hamilton J.A. Prog. Lipid. Res. 2004; 43: 177-199Crossref PubMed Scopus (125) Google Scholar), only limited structural information about protein-protein or protein-membrane interactions involving the fatty acid-binding proteins has been available. Our clarification of the interactions between the lipoprotein-binding proteins may provide new insights into these aspects of fatty acid-interacting proteins. We thank Mr. Iwao Fujiwara, Dr. Yukio Tominaga, and Dr. Masuo Kurono for continuous support and helpful discussions. Download .pdf (.78 MB) Help with pdf files
Referência(s)