12376
J. Phys. Chem. B 2010, 114, 12376–12382
Static and Dynamic Characterization of Nanodiscs with Apolipoprotein A-I and Its Model Peptide Masakazu Miyazaki, Yoko Tajima, Tetsurou Handa,† and Minoru Nakano* Graduate School of Pharmaceutical Sciences, Kyoto UniVersity, Sakyo-ku, Kyoto 606-8501, Japan ReceiVed: March 8, 2010; ReVised Manuscript ReceiVed: July 30, 2010
The class A amphipathic R-helical peptide 18A is known to form discoidal phospholipid complexes (nanodiscs) similar to that formed by apolipoprotein A-I (apoA-I). To reveal the structural differences in nanodiscs formed with this protein and peptide, we prepared nanodiscs with 1-palmitoyl-2-oleoyl-sn-glycero-3-phosphocholine and applied fluorescence techniques to these nanoparticles. Fluorescence resonance energy transfer between 1,2-dioleoyl-sn-glycero-3-phosphoethanolamine-N-(7-nitro-2-1,3-benzoxadiazol-4-yl) and 1,2-dioleoyl-snglycero-3-phosphoethanolamine-N-(lissamine rhodamine B sulfonyl) in nanodiscs revealed that lipid exchange with 18A nanodiscs is mediated by collisions between nanodiscs. The fluorescence lifetime of dansyl phosphatidylethanolamine and excimer fluorescence of 1,2-bis(1-pyrenedecanoyl)-sn-glycero-3-phosphocholine showed that the degree of hydration of the membrane surface and lateral pressure of acyl chains in 18A nanodiscs are independent of the disc size, suggesting that 18A nanodiscs form planar lipid bilayers irrespective of their size, which differs from apoA-I nanodiscs, whose bilayer deforms to a saddle surface with decreasing size. These results suggest that the flexible structure of a chain of helices in apoA-I is crucial for the formation of saddle surfaces in nanodiscs. Introduction Apolipoprotein A-I (apoA-I), a major protein component of high-density lipoprotein (HDL), is a 28 kDa polypeptide that consists of a series of highly homologous 11- and 22-residue amphipathic R-helices.1 In the neogenesis of HDL, apoA-I interacts with the transmembrane ATP-binding cassette transporter A1 (ABCA1)2-4 and forms discoidal HDL in which apoA-I molecules wrap around the edges of the lipid bilayer. The discoidal nanoparticles are then transformed into a spherical form by the conversion of cholesterol in these particles to cholesteryl ester by lecitin:cholesterol acyltransferase (LCAT).5,6 Mature spherical HDL transports cholesterol in peripheral tissues to the liver for recycling and secretion into bile.7 This “reverse cholesterol transport” is a crucial pathway in cholesterol homeostasis. The class A amphipathic R-helical peptide 18A (AcDWLKAFYDKVAEKLKEAF-NH2) has been used as a model of lipid-binding sites for apoA-I and other exchangeable apolipoproteins.8,9 This amphipathic peptide has been designed to possess a secondary structural motif of R-helices represented by the clustered charge distribution with positively charged residues at the polar/nonpolar boundary and negatively charged residues at the center of the polar face (Figure 1).10 When the amphipathic peptide binds to the phospholipid bilayer, the positive charges of the peptide interact with the negative charges of the phosphate group of the phospholipids, which enables the peptides to penetrate deeply into the bilayer and solubilize the membrane.11 The formation of discoidal phospholipid particle (nanodisc) with 18A has been confirmed in previous studies.10,12-15 * To whom correspondence should be addressed. Mailing address: Graduate School of Pharmaceutical Sciences, Kyoto University, Kyoto 6068501, Japan. Tel.: +81-75-753-4565. Fax: +81-75-753-4601. E-mail:
[email protected]. † Present address: Faculty of Pharmaceutical Sciences, Suzuka University of Medical Science, Suzuka University of Medical Science, 3500-3 MinamiTamagaki-cho, Suzuka, Mie 513-8670, Japan.
Figure 1. Helical wheel representation of 18A. The amphipathic helix has positively charged (Lys) residues near the hydrophilic-hydrophobic interface and negatively charged (Asp and Glu) residues at the center of the hydrophilic face.
Mishra et al. have reported the results of high-resolution nuclear magnetic resonance (NMR) studies of 1,2-dimyristoyl-snglycero-3-phosphocholine (DMPC)-18A nanodiscs.14 They demonstrated that the peptide helices are arranged in an antiparallel head-to-tail fashion to cover the edge of the nanodiscs. Previously, we studied the structure of nanodiscs (also referred as reconstituted HDLs) consisting of apoA-I and 1-palmitoyl2-oleoyl-sn-glycero-3-phosphocholine (POPC) using two fluorescence techniques:16 excimer formation of 1,2-bis(1-pyrenedecanoyl)-sn-glycero-3-phosphocholine (C10dipyPC), which correlates to the lateral pressure in the acyl chain region, and fluorescence lifetime of dansyl phosphatidylethanolamine (dansyl PE), which is sensitive to headgroup packing. Using these approaches, we demonstrated that the smaller apoA-I nanodiscs with mean hydrodynamic diameters of 7.9 and 9.0 nm have a curved lipid bilayer and form a saddle surface, while the 9.6 nm nanodiscs have a planar bilayer. In this study, we prepared POPC-18A nanodiscs and analyzed the structure of the lipid bilayer using C10dipyPC and dansyl PE. In addition, we evaluated the lipid exchange dynamics by using fluorescence resonance energy transfer (FRET) from 1,2dioleoyl-sn-glycero-3-phosphoethanolamine-N-(7-nitro-2-1,3-
10.1021/jp102074b 2010 American Chemical Society Published on Web 09/02/2010
Structural Differences of 18A and apoA-I Nanodiscs
J. Phys. Chem. B, Vol. 114, No. 38, 2010 12377
benzoxadiazol-4-yl) (NBD-DOPE) to 1,2-dioleoyl-sn-glycero3-phosphoethanolamine-N-(lissamine rhodamine B sulfonyl) (Rho-DOPE). These approaches revealed differences in the structure and dynamics of lipid exchange between 18A- and apoA-I-containing nanodiscs. Experimental Methods Materials. We purchased POPC from Sigma-Aldrich (St. Louis, MO, USA) and NBD-DOPE, Rho-DOPE, and dansyl PE from Avanti Polar Lipids (Alabaster, AL, USA). 1-Hexadecanoyl-2-(1-pyrenedecanoyl)-sn-glycero-3-phosphocholine (βpy-C10-HPC) and C10dipyPC were purchased from Invitrogen (Eugene, OR, USA) and 18A from Hayashi Kasei (Osaka, Japan). We isolated ApoA-I from pig plasma using procedures described previously.16 The purity of apoA-I, as determined by sodium dodecyl sulfate polyacrylamide gel electrophoresis (SDS-PAGE), was greater than 95%. All other chemicals were of the highest reagent grade. Sample Preparation. To prepare large unilamellar vesicles (LUVs), required amounts of a chloroform-methanol solution of POPC and fluorescence probes were mixed in a roundbottomed glass flask. After the organic solvent was removed by evaporation, the sample was dried overnight in vacuum. TrisHCl buffer (10 mM Tris, 150 mM NaCl, 1 mM EDTA, and 0.01 g/mL NaN3; pH 7.4) was added to the dried lipid mixture. The mixture was vortexed, freeze-thawed five times, and extruded through a polycarbonate filter with a pore size of 100 nm. We prepared POPC-apoA-I nanodiscs by using the sodium cholate dialysis method.17,18 The dried lipid mixture was hydrated with Tris-HCl buffer. Sodium deoxycholate (Cholate) was added at a POPC:cholate ratio of 1:2. The mixture was then vortexed and incubated until clear. Next, apoA-I was added to the mixture at a molar ratio of POPC:apoA-I of 100:1 or 25:1. After overnight incubation, the mixture was dialyzed at 4 °C for 2 days against Tris-HCl buffer to remove the cholate. The samples were fractionated into each subclass by gel filtration chromatography on a Superdex 200 column. To prepare POPC-18A nanodiscs, POPC LUVs (300 µM) were mixed with 18A at a molar ratio of POPC:18A of 4:1 or 2:1 and incubated at 25 °C. The microsolubilization of the LUVs by 18A was confirmed by the decrease in right-angle light scattering with a wavelength of 650 nm, monitored with a Hitachi F-2500 fluorescence spectrometer (Tokyo, Japan). The mean particle diameter was determined by using the Dynapro 99 dynamic light scattering (DLS) instrument (Protein Solutions Ltd., UK).19 The concentration of PC was determined by using an enzyme assay kit from Wako (Osaka, Japan). Fluorescence Measurements. Lipid exchange between the POPC-18A nanodiscs or POPC-apoA-I nanodiscs was evaluated by means of FRET. Both fluorescence-labeled (0.5 mol % NBDDOPE and 2.0 mol % Rho-DOPE) and nonlabeled nanodiscs were prepared and mixed at a molar ratio of 1:9. Fluorescence was monitored at 530 nm (Idonor) and 590 nm (Iacceptor) with excitation at 460 nm at 25 °C on a Hitachi F-2500. FRET efficiency, FRET(t), is given by the following equation:
FRET(t) ) (F(t) - F(∞))/(F(0) - F(∞))
(1)
where F(t) is given by Iacceptor(t)/Idonor(t). We measured F(0) and F(∞) using labeled nanodiscs and nanodiscs containing 0.05 mol % NBD-DOPE and 0.2 mol % Rho-DOPE, respectively.
Figure 2. Reduction observed in the right-angle light-scattering intensity of POPC LUVs (300 µM) by 18A at 25 °C. The intensity, I(t), was detected at 650 nm and normalized by the initial intensity before the addition of 18A (I(0)). The POPC:18A ratios were set to 2:1 (squares) and 4:1 (circles).
For structural evaluation of the POPC bilayers, POPC-18A nanodiscs, POPC-apoA-I nanodiscs, and POPC LUVs containing 0.1 mol % dansyl PE were prepared. The fluorescence lifetime of dansyl PE in these samples was measured with a Horiba NAES-550 ns fluorometer (Kyoto, Japan) with a pulsed hydrogen lamp (full width at half-maximum: ∼2 ns). The samples were excited through a Hoya U350 filter, and their fluorescence was detected through a Hoya Y48 filter at 25 °C. The mean fluorescence lifetime, 〈τ〉, was determined as described previously.16 In addition, POPC-18A nanodiscs, POPC-apoA-I nanodiscs, and POPC LUVs containing either 0.1 mol % C10dipyPC or 0.1 mol % β-py-C10-HPC were prepared, and the fluorescence was detected at 378 nm (monomer) and 478 nm (excimer) with excitation at 345 nm at 25 °C on a Hitachi F-2500 to determine the excimer-to-monomer fluorescence intensity ratio (Ie/Im). For 18A nanodiscs, Ie/Im values were calculated after background subtraction using nonfluorescent 18A nanodiscs, since these samples, differently from apoA-I nanodiscs and LUVs, brought about Raman scattering around 378 nm. Results Analysis of the Size of Nanodiscs. We prepared POPC-18A nanodiscs by incubating POPC LUVs and 18A at 25 °C. To confirm the microsolubilization of LUVs by 18A, the change in right-angle light scattering was monitored. The reduction in the intensity of light scattering can be attributed to the transformation of LUVs into small discoidal particles.20 As shown in Figure 2, the reduction in scattering was almost complete within 3 min. At POPC:18A molar ratios of 4:1 and 2:1, the hydrodynamic diameter of 18A nanodiscs determined by DLS was 8.5 and 5.9 nm, respectively (Table 1), suggesting that the lipid:peptide ratio determines the size of the nanodiscs. Histograms showed the unimodal size distribution and absence of vesicles (∼100 nm) (Figure 3). The POPC-apoA-I nanodiscs were prepared by using the sodium cholate dialysis method. (see Experimental Methods). We previously showed that preparations made at POPC:apoA-I molar ratios of 100:1 and 25:1 resulted in peaks in gel filtration chromatography that were maximum at the positions that corresponded to the diameters of 9.6 and 7.9 nm, respectively.16
12378
J. Phys. Chem. B, Vol. 114, No. 38, 2010
Miyazaki et al.
TABLE 1: Mean Hydrodynamic Diameters of 18A- and apoA-I-Containing Nanodiscs diameter (nm)a nanodisc
b
18A apoA-I
molar ratio
planed saddled
4:1 2:1b 78:1e 25:1e
gel filtration c
N.D. N.D.c 9.6 ( 0.1 7.9 ( 0.1
DLS 8.5 ( 0.5 5.9 ( 0.5 10.5 ( 0.2 8.2 ( 0.2
a The data represent mean ( SD for at least four experiments. POPC:18A molar ratio. c Not detectable. d Shape of apoA-I nanodiscs suggested by previous study.16 e POPC:apoA-I molar ratio.16 b
These fractions were separated and analyzed by DLS. As shown in Table 1, DLS yielded similar (but slightly larger) diameters in gel filtration chromatography; these results are in good agreement with those of a previous study.21 Lipid Exchange between Nanodiscs. To better understand the dynamics of lipid exchange in nanodiscs, FRET experiments were performed using nanodiscs labeled with both a FRET donor (NBD-DOPE) and an acceptor (Rho-DOPE). In the fluorescence emission spectrum obtained from nanodiscs labeled with the NBD-DOPE and Rho-DOPE pair, the fluorescence of Rho-DOPE (Iacceptor) due to FRET was observed at 590 nm in addition to the fluorescence of NBD-DOPE (Idonor) at 530 nm (data not shown). Fluorescence-labeled and nonlabeled nanodiscs were mixed at a molar ratio of 1:9 at time zero, and the FRET efficiency (i.e., Iacceptor/Idonor) ratio was monitored. Because the lipid exchange between labeled and nonlabeled nanodiscs reduces the local concentration of the fluorescence pair, it can be detected as a decrease in the FRET efficiency. As shown in Figure 4A and B, FRET efficiency for 18A nanodiscs decreased after being mixed with the nonlabeled ones and decayed more steeply with an increase in the concentration of nanodiscs. These results suggest that lipid exchange is mediated by a mechanism that involves collisions between nanodiscs. In addition, the decay for 5.9 nm 18A nanodiscs was faster than that for 8.5 nm nanodiscs at the same POPC concentration. This is presumably because of the higher volume fraction of nanodiscs (POPC + 18A) in the dispersed media, which results in an increase in the collision frequency. On the other hand, the FRET efficiency of apoA-I nanodiscs decreased little (Figure 4C and D), which suggests that the collision-mediated lipid exchange did not occur. All particle sizes were constant at different concentrations and even after the FRET experiments (data not shown).
Packing of Phospholipid Headgroups in Nanodiscs and LUVs. Dansyl PE has a fluorophore at the phospholipid headgroup, and the measurement of the fluorescence lifetime enables evaluation of the membrane surface packing state.22 Previously, we observed significantly longer lifetimes for the 7.9 nm apoA-I nanodiscs than for the 9.6 nm nanodiscs. We ascribed this tighter surface packing in the former particles to the deformation into a saddle-shape. In this study, the mean fluorescence lifetime (〈τ〉) for 18A nanodiscs with 0.1 mol % dansyl PE was determined. The results are shown in Figure 5 and Table 2. The 〈τ〉 values for 18A nanodiscs were independent of size and were lower than the values for LUVs. The 〈τ〉 values did not depend on the concentration of nanodiscs (data not shown). Interestingly, 18A nanodiscs had 〈τ〉 values similar to those of planar apoA-I nanodiscs; however, these values were significantly lower than those of the saddle-shaped nanodiscs. To further confirm the difference in the degree of hydration, similar lifetime experiments were performed in buffer containing 70 vol % D2O. We observed a longer lifetime in D2O-containing buffer (Table 2), which suggests that proton transfer between the fluorophore and the water is involved in the fluorescence quenching mechanism and is retarded in D2O.23 This isotopic effect is considered more prominent when dansyl group is more exposed to water. Differences between the reciprocal lifetimes in D2O and H2O (1/〈τ〉H2O - 1/〈τ〉D2O), which can be a measure of hydration, are shown in Table 2. For 18A nanodiscs the value of 1/〈τ〉H2O - 1/〈τ〉D2O was independent of the size and was close to that of planar apoA-I nanodiscs, but significantly larger than that of saddle-shaped apoA-I nanodiscs. These results indicate that the membrane surface of 18A nanodiscs allows a similar level of water penetration as the planar apoA-I nanodiscs, but it is not so tightly packed as the saddle-shaped ones. Lateral Pressure in Nanodiscs and LUVs. In our previous study, the lateral pressure of the acyl chain in apoA-I nanodiscs was evaluated by using the excimer fluorescence of C10dipyPC. An increase in the lateral pressure forces two pyrene moieties of C10dipyPC to close with each other and enhances the frequency of intramolecular excimer formation. Therefore, the ratio of excimer (Ie) to monomer (Im) fluorescence intensity (Ie/ Im) is sensitive to the acyl chains’ packing,24,25 and it detects the lamellar-nonlamellar phase transition,25 that is, the change in the curvature. We have previously shown that the Ie/Im for 7.9 nm apoA-I nanodiscs was much lower than that for 9.6 nm nanodiscs.16 This suggests that C10dipyPC increases the distance between the two pyrene moieties in the former nanodiscs by
Figure 3. Size distribution histograms of POPC-18A nanodiscs by DLS at 25 °C. The POPC:18A ratios were set to 2:1 (A) and 4:1 (B). The size was measured after the reduction in the right-angle scattering was confirmed.
Structural Differences of 18A and apoA-I Nanodiscs
J. Phys. Chem. B, Vol. 114, No. 38, 2010 12379
Figure 4. Lipid exchange assay by FRET. Changes in FRET efficiency were monitored at 25 °C after fluorescence-labeled (0.5 mol % NBDDOPE and 2.0 mol % Rho-DOPE) and nonlabeled nanodiscs were mixed at a molar ratio of 1:9 with three different total POPC concentrations: 75 (open squares), 150 (closed diamonds), and 300 µM (open circles). The profiles are for 5.9 nm 18A nanodiscs (A), 8.5 nm 18A nanodiscs (B), planar apoA-I nanodiscs (C), and saddle-shaped apoA-I nanodiscs (D).
TABLE 2: Mean Fluorescence Lifetime (〈τ〉) of Dansyl PE in Particles at 25 °C 〈τ〉 (ns)a
LUV 18A nanodisc (8.5 nm) 18A nanodisc (5.9 nm) apoA-I nanodisc (planar) apoA-I nanodisc (saddle)
1/〈τ〉H2O - 1/〈τ〉D2O
in H2O
in D2Ob
(10-3 ns-1)
15.72 ( 0.12 15.10 ( 0.10 14.96 ( 0.06 14.93 ( 0.07 16.38 ( 0.15
17.43 ( 0.10 16.92 ( 0.11 16.73 ( 0.09 16.64 ( 0.05 17.83 ( 0.07
6.23 7.15 7.07 6.92 4.99
The data represent mean ( SD for at least three experiments. Measured in buffer with 70 vol % D2O. a
b
Figure 5. Mean fluorescence lifetime 〈τ〉 of 0.1 mol % dansyl PE in POPC nanoparticles. Dansyl PE in LUV, 8.5 and 5.9 nm 18A nanodiscs, and planar and saddle-shaped apoA-I nanodiscs was excited by pulsed excitation light passing through a Hoya U350 filter. Fluorescence decay was detected by using a Hoya Y48 filter at 25 °C. The data for apoA-I nanodiscs are from ref 16. The data represent mean ( SD for at least six experiments.
inducing a negative curvature (i.e., saddle surface). To determine the structure of the 18A nanodiscs, we applied a similar method, using 0.1 mol % C10dipyPC. Figure 6A shows that the Ie/Im values of 18A nanodiscs were concentration-dependent, and they decreased with an increase in particle concentration. For apoA-I nanodiscs, however, the Ie/Im values did not change with concentration (data not shown). These results imply that the
collisions of 18A nanodisc affect the excimer formation of C10dipyPC. Therefore, intrinsic Ie/Im values of 18A nanodiscs were determined by linear extrapolation to zero concentration (Figure 6A). As shown in Figure 6B, the intrinsic Ie/Im values of 18A nanodiscs were comparable to those of LUVs, but they were considerably higher than the Ie/Im value of saddle-shaped apoA-I nanodiscs. These results, together with the fluorescence lifetime data (Figure 5), suggest that, although 18A nanodiscs prepared (8.5 and 5.9 nm) are comparable to saddle-shaped apoA-I nanodisc in size (8.2 nm by DLS), their bilayer structures are largely different from the latter. Because the concentration of C10dipyPC is 0.1 mol % of total lipids and nanodisc comprises less than 160 lipids, each nanodisc contains one or less labeled PCs, and therefore the intermolecular excimer formation is negligible. This was confirmed using 18A nanodiscs containing
12380
J. Phys. Chem. B, Vol. 114, No. 38, 2010
Figure 6. Excimer-to-monomer fluorescence intensity ratios (Ie/Im) of 0.1 mol % C10dipyPC in POPC nanoparticles at 25 °C. (A) The lipid concentration dependence of the Ie/Im values of 18A nanodiscs with diameters of 5.9 and 8.5 nm. (B) The Ie/Im values for LUV, 8.5 and 5.9 nm 18A nanodiscs, and planar and saddle-shaped apoA-I nanodiscs. The Ie/Im values for 18A nanodiscs were determined by linear extrapolation in Figure 6A. The data for LUV and apoA-I nanodiscs are from ref 16. The data represent mean ( SD for at least three experiments. (C) Typical fluorescence spectra of C10dipyPC and β-pyC10-HPC in 18A nanodiscs. Fluorescent intensity is normalized by the monomer intensity at 378 nm.
0.1 mol % monofunctionalized β-py-C10-HPC, which represented no excimer fluorescence (Ie/Im < 0.02, Figure 6C). Discussion The amphipathic R-helical peptide 18A was designed to study the protein-lipid association in plasma lipoprotein.26 This
Miyazaki et al. peptide is known to mimic the properties of apoA-I. For example, 18A solubilizes phospholipid vesicles and forms lipid18A complexes. A negative-stain electron micrograph revealed that the complexes have a disc-like geometry and rouleaux formation as observed for lipid-apoA-I complexes.10,12,13 The helices of this peptide are arranged in an antiparallel manner on the edges of the nanodiscs. In addition, LCAT-activating ability of 18A has been shown for small unilamellar vesicles and discoidal complexes consisting of egg yolk phosphatidylcholine and cholesterol.13 In this study, we prepared POPC-18A nanodiscs. The DLS method showed the unimodal size distribution of the nanodiscs without large vesicles (Figure 3). Interestingly, the size distribution could not be determined by gel filtration chromatography or by nondenaturing gradient gel electrophoresis (NDGGE). Gel filtration showed no trace of nanodiscs, while NDGGE resulted in band broadening (data not shown). This implies that the 18A nanodiscs were unstable during interaction with gel substrates. We evaluated the dynamics of lipid exchange by using FRET between NBD-DOPE and Rho-DOPE in nanodiscs. The results showed that lipid mixing between 18A nanodiscs takes place during collisions of nanodiscs, while that between apoA-I nanodiscs does not occur (Figure 4). Addition of free 18A peptide to a solution of 18A nanodiscs resulted in a decrease in the mean diameter of nanodiscs with keeping the unimodal size distribution (data not shown), which suggests that lipid exchange is mediated by fusion-fission mechanism. The slight decrease in FRET efficiency for apoA-I nanodiscs might represent monomeric diffusion of fluorescence lipids in an aqueous medium. Using small-angle neutron scattering, we previously reported that DMPC-apoA-I nanodiscs showed about 20-fold faster lipid exchange than DMPC LUVs, but that had no collision-mediated exchange processes;27 these findings are in good agreement with the present data for POPC-apoA-I nanodiscs. The differences in dynamics between apoA-I- and 18Acontaining nanodiscs can be explained by the presence of “cleavages” on the belt. In apoA-I nanodiscs (
