Supplementary Information HIV 1 Nef membrane association depends on charge, curvature, composition and sequence Holger Gerlach 1, Vanessa Laumann 1, Sascha Martens 2, Christian F W Becker 1,3, Roger S Goody 1 & Matthias Geyer 1 1 Max-Planck-Institut für molekulare Physiologie, Abteilung Physikalische Biochemie, Otto-Hahn-Strasse 11, 44227 Dortmund, Germany. 2 Max F. Perutz Laboratories, University of Vienna, Dr. Bohr-Gasse 9/3, 1030 Vienna, Austria. 3 Present address: Department of Chemistry, TU München, Lichtenbergstrasse 4, 85747 Garching, Germany. Correspondence should be addressed to M.G. (matthias.geyer@mpi-dortmund.mpg.de). SUPPLEMENTARY RESULTS Sensitivity of Nef to lipid composition In another experimental series the preference of Nef for specific membrane constituents was analyzed. To this end we tested the binding kinetics of Nef to membranes enriched with cholesterol, a lipid component that influences the fluidity of membranes, with phosphatidylinositol-4,5- bisphosphate (PtdIns(4,5)P 2 ), which is a marker for the plasma membrane, and with sphingomyelin, which has a ceramid lipid backbone that is a marker for the extracellular side of eukaryotic cells (Supplementary Fig. 4a). As membrane scaffold PC/PG (70:30) liposomes of 50 nm radius were used. Cholesterol, which replaced the DOPC portion, was tested in two different concentrations of 10% and 30%. Similarly, sphingomyelin used at 30% concentration replaced the DOPC portion, because of their common phosphocholine head group. The concentration of PtdIns(4,5)P 2 was set to 1% of the lipid constituents, but here the content of DOPG was reduced to 25% in order to compensate for the doubly negatively charged inositol phosphate groups and the phospholipid ester. Surprisingly, neither the association nor the dissociation kinetics of myristoylated wildtype Nef showed significant changes in the binding to these various membrane compositions (Supplementary Fig. 4b+c). A mixture of 15% cholesterol and 15% sphingomyelin showed a reduced apparent onrate. The observation that Nef is not targeted by PtdIns(4,5)P 2 to membranes is supported by results from NMR chemical shift perturbation experiments. Titration of phytic acid (IP 6 ) in steps up to a tenfold higher concentration than 15 N-labeled Nef (45-210) did not induce changes in the chemical shift pattern of Nef, suggesting the lack of interactions with inositol phosphates (data not shown). In addition, analysis of the interaction to lipids on phosphoinositol lipid strips (PIP strips) showed that binding to highly negatively charged lipids are required to keep Nef associated to the membrane strip (Supplementary Fig. 4d). From these observations we conclude that neither the flexible N-terminal domain of Nef nor the well folded core domain contains a phosphoinositol recognition site that would contribute to the interaction beyond the electrostatic attraction. Membrane binding kinetics of an N-terminal Nef peptide In addition to the kinetic and structural experiments with native Nef protein, a peptide of the N- terminal Nef sequence (2-27) was synthesized, both in the myristoylated and non-myristoylated form,
to study its cytosolic and membrane bound conformation (Supplementary Fig. 6). While the myristoylated Nef peptide appeared poorly soluble in aqueous water solution (Supplementary Fig. 7c), the unmodified form readily associated with liposomes and again showed a significant increase in helicity upon addition of PC/PG liposomes (Supplementary Fig. 7a). According to its small size an increase from 4% to 45% in helix content was calculated for the Nef peptide from a reference fit while the Nef counterpart (28-210) did not associate to lipids and hence remained unchanged (Table 2 and Supplementary Fig. 7b). A kinetic analysis of the peptide liposome interactions based on FRET transfer revealed again a biphasic association reaction with a fast lipid concentration dependent rate and a second concentration independent rate (Supplementary Fig. 7d-f). Due to the reduced mass of the peptide to approximately 1/8 of the native protein and its very polar sequence (pi 12.0) the association rates were significantly higher (140 to 220 s -1 ) but also the off rate (k -1 ) was about ten times increased potentially due to the missing hydrophobic interactions mediated by the myristate. Kinetic model of Nef membrane binding The kinetics of Nef membrane binding is explained using a two step model. However, it seems likely that the first observable phase includes 2 distinct processes, since there is strong evidence for an electrostatic component (K el ) as well as for lipid insertion (K ins ) occurring in this phase (Fig. 6). A plausible mechanism involves an initial rapid electrostatic interaction followed by insertion of the lipid into the membrane, with both events occurring in the observed rapid phase. The apparent second order rate constant for the first phase in binding would then be given by the product of K el, the equilibrium association constant from the electrostatic interaction, and k +ins, the rate of lipid insertion (i.e. k +1 = K el k +ins ). The y-axis intercepts of the plots of pseudo-first order rate constant for the rapid phase would then give the value of k -ins, the rate of spontaneous lipid extraction. SUPPLEMENTARY METHODS Liposome preparation. Synthetic phosphatidylcholine (PC), phosphatidylethanolamine (PE), phosphatidylglycerol (PG), phosphatidylserine (PS) and phosphatidylinositol-(4,5)-bisphosphate (PtdIns(4,5)P 2 ) all stored in chloroform as well as polycarbonate filters with pore diameters of 0.05, 0.08, 0.1, 0.2 and 0.4 µm were purchased from Avanti Polar Lipids. Cholesterol (C3045) and bovine brain sphingomyelin (S7004) were purchased from Sigma-Aldrich. The fluorescence dye β-dph- HPC (D476) was purchased from Invitrogen. The size distribution of the liposomes was controlled by dynamic light scattering (DLS) and the concentration of phospholipids was determined by a standard phosphate assay according to a protocol of Avanti Polar Lipids and as described 49. In brief, a calibration line for the photometric absorbance at 820 nm was generated using a phosphate standard (Sigma). Phospholipid samples were spiked with 0.45 ml of 8.9 N H 2 SO 4 and incubated for 25 minutes at 215 C. 150 µl H 2 O 2 were added to the tubes and heating was continued for another 30 minutes. The samples were cooled to room temperature and 3.9 ml deionized water, 0.5 ml ammonium molybdate (VI) tetrahydrate solution and 0.5 ml ascorbic acid were added. The tubes were heated to 100 C for 7 minutes before the absorbance at 820 nm of each sample was determined. PIP-Strip assay. The phosphatidylinositol phosphate (PIP) strips were purchased from Echelon Bioscience (P-6001), the blocking reagent from Roche and the chemiluminescence detection kit from Amersham Biosciences. The PIP-strips were blocked with 10% blocking reagent in 0.05% PBS- Tween at room temperature for 1 h and then incubated with 1 µg/ml myristoylated Nef wildtype in 10% blocking solution at 4 C o/n. After removal of the protein solution and three times of washing - 2 -
with PBS-Tween, the membrane was incubated with an anti-his-antibody for 1 h at room temperature. Another three washing steps and incubation with an anti-rabbit antibody (conjugated with horseradish peroxidase) followed. After three final washing steps, the chemiluminescence detection kit was used to yield the signal. Circular dichroism spectroscopy. CD spectra were recorded on a Jasco J-815 spectropolarimeter at 25 C using a quartz cuvette with a path length of 2 mm. The myristoylated Nef protein was dialyzed against 5 mm KP i buffer (ph 8.0) and measured at a concentration of 3.9 µm. SUV liposomes contained 70% of DOPC and 30% of DOPG. Spectra were recorded with a step size of 1 nm and an integration time of 1 s. The secondary structure content was analyzed using the program DICHROWEB 50 by fitting the CD spectra to a reference data set. Transmission electron microscopy. 1 mg Folch lipids were prepared as described above. 10 µl liposomes were incubated with myrnef at a final concentration of 20 µm for 30 minutes. For the myrnef only control 40 µm protein was incubated in buffer (20 mm Tris ph 8, 100 mm NaCl, 1 mm DTT) alone for 30 minutes. 10 µl were spotted on a glow discharged formvar carbon coated copper grid and stained with 2% uran acetate. Cell culture and reagents. COS-7 cells were routinely maintained in DMEM supplemented with 10% fetal bovine serum. For life cell microscopy, cells were cultured on 35-mm glass bottom dishes (MatTek) and transferred to low bicarbonate DMEM without phenol red supplemented with 25 mm HEPES ph 7.4. The pegfp expression plasmids of wildtype HIV-1 SF2 Nef and Nef(G2A) were a kind gift from Oliver Fackler. Transient expression of plasmid DNA was achieved by transfecting COS-7 cells using Fugene (Roche). Transfected cells were serum starved for 5-6 hours before all experiments. Fluorescence imaging. Confocal laser scanning microscopy was performed on a Leica TCS SP2 microscopy equipped with a 63X/1.3 NA oil immersion lens and a temperature controlled chamber at 37 C. pegfp was excited by using the 488 nm Ar laser. The fluorescence was detected using a dichroic beamsplitter (Q530 LP, Chroma Tech. Corp.) and a narrow-band emission filter (HQ538/25; Chroma). Confocal images were processed with AdobePhotoshop CS2 software (Adobe Systems). SUPPLEMENTARY REFERENCES 49. Nickel, W. & Wieland, F.T. Receptor-dependent formation of COPI-coated vesicles from chemically defined donor liposomes. Methods Enzymol. 329, 388 404 (2001). 50. Whitmore, L. & Wallace, B.A. DICHROWEB, an online server for protein secondary structure analyses from circular dichroism spectroscopic data. Nucleic Acids Res. 32, W668 73 (2004). - 3 -
Supplementary Table 1: Observed rate constants of the binding series of myristoylated Nef to liposomes of different sizes. liposome radius 48 nm lipid concentration 22.5 µm 39.4 µm 56.3 µm 84.5 µm 112.6 µm obs. rate constant k 1 (s -1 ) 17.0 20.1 24.8 32.7 43.9 obs. rate constant k 2 (s -1 ) 2.5 3.7 3.5 2.6 resulting fit for the fast association process a : k on = 295,400 M -1 s -1 ; k off = 9.18 s -1 liposome radius 67 nm lipid concentration 22.5 µm 39.4 µm 56.3 µm 84.5 µm 112.6 µm obs. rate constant k 1 (s -1 ) 15.8 17.9 20.6 25.6 32.2 obs. rate constant k 2 (s -1 ) 2.5 3.2 3.6 3.4 resulting fit for the fast association process a : k on = 182,600 M -1 s -1 ; k off = 10.9 s -1 liposome radius 85 nm lipid concentration 23.8 µm 41.7 µm 59.5 µm 89.3 µm 119.0 µm obs. rate constant k 1 (s -1 ) 16.4 17.5 19.0 23.2 28.4 obs. rate constant k 2 (s -1 ) 2.4 3.3 3.5 4.5 3.8 resulting fit for the fast association process a : k on = 127,800 M -1 s -1 ; k off = 12.37 s -1 liposome radius 113 nm lipid concentration 25 µm 43.8 µm 62.5 µm 93.8 µm 125 µm obs. rate constant k 1 (s -1 ) 15.2 16.1 17.4 20.3 24.2 obs. rate constant k 2 (s -1 ) 2.1 3.0 3.4 3.0 2.3 resulting fit for the fast association process a : k on = 90,200 M -1 s -1 ; k off = 12.32 s -1 a For calculation of the apparent binding constant K 1 of the fast association process, k on has to be multiplied by the number of lipids that form a Nef binding site (k on x 25). - 4 -
Supplementary Table 2: Observed rate constants for a series of Nef mutants to 70:30 PC/PG liposomes of 50 nm radius. lipid concentration Nef protein 15.6 µm 31.3 µm 46.9 µm 62.5 µm 93.8 µm 125 µm observed fast rate constant k 1 (s -1 ) Nef(wt) 22.9 23.2 27.1 31.1 39.1 37.4 Nef(R4) 15.6 14.7 14.7 15.8 18.6 19.2 Nef(WMW) 15.9 18.1 20.3 21.3 25.0 27.8 Nef(WR) 25.7 27.5 24.5 25.3 24.2 Nef(G2A) 14.8 17.4 13.8 13.4 16.2 12.6 Nef59+ observed slow rate constant k 2 (s -1 ) Nef(wt) 2.6 2.8 2.8 2.8 2.8 2.5 Nef(R4) 2.3 2.7 2.6 2.5 2.6 2.2 Nef(WMW) 1.4 1.2 1.2 1.3 1.4 Nef(WR) 2.6 1.3 2.0 1.4 1.7 1.0 Nef(G2A) 1.7 1.9 1.3 1.4 1.4 1.1 Nef59+ Results from fitting of the fast rate constants: Nef(wt) k on = 161 10 3 M -1 s -1, k off = 20.1 s -1 ; Nef(R4) k on = 44 10 3 M -1 s -1, k off = 13.7 s -1 ; Nef(WMW) k on = 107 10 3 M -1 s -1, k off = 14.7 s -1 ; with k on to be multiplied x25 to correct for the number of lipids that form a Nef binding site. - 5 -
Supplementary Fig. 1 Experimental setup of the stopped flow Nef membrane binding assay. (a) Tryptophans in Nef are used to transfer fluorescence energy in a spatial dependence to a dye in the lipid membrane. The Nef structure was modeled based on the NMR structures of the core (PDB: 2NEF) and the myristoylated anchor domain (1QA5). (b) Spectral characterization of the FRET pair used. The areas of tryptophan excitation, spectral overlap and recording of β-dph-hpc emission are shaded. - 6 -
Supplementary Fig. 2 FRET based association and dissociation experiments of Nef from liposomes of different sizes. (a) Comparison of association curves of 0.2 µm myrnef(wt) with liposomes of 70:30 PC/PG content and 48, 67, 85 or 113 nm radius. Displayed are time curves at 31.3 µm lipid concentrations. The line fit of a bi-exponential association curve is indicated (red lines). (b) Display of the slow rate constant k 2 for the association of Nef to differently curved liposomes. The concentration independent slow rate process is not significantly affected by variations of the liposome size. Error bars indicate the standard deviation of the line averaging assuming a first order binding kinetics. (c) Dissociation curves of Nef upon ten-fold excess of unlabelled liposomes with variable radius from 48 to 113 nm. The time course was monitored over 5 s and could be fitted by a biexponential curve (red lines). The dissociation rate constants of the slow membrane binding process correspond to 0.551 s -1 at 48 nm, 0.437 s -1 (67 nm), 0.443 s -1 (85 nm) and 0.389 s -1 (113 nm) and are enclosed as k -2 values in the calculation of the Nef membrane binding kinetics (Table 1). - 7 -
Supplementary Fig. 3 Membrane association of Nef correlates with liposome curvature. (a) myrnef induces positive membrane curvature seen as tubulation of Folch liposomes. The tubules had an average diameter of 13 nm. No tubular structures were seen with myrnef or liposomes alone. The concentration of phospholipids within the Folch fraction corresponded to approx. 400 µm; myristoylated Nef was used at 20 µm concentration. Scale bars: 100 nm. (b) Co-sedimentation assay showing that myristoylated Nef efficiently binds to Folch liposomes. No sedimentation was seen in the absence of liposomes. Lipids and protein were used at a final concentration of approx. 600 µm and 4 µm, respectively, in the spin assay. (c) Localization of fluorescent tagged Nef.GFP in transfected COS-7 cells. Confocal fluorescence microscopy images of GFP alone, myristoylated Nef.GFP and non-myristoylated Nef(G2A).GFP are shown from left to right. -8-
Supplementary Fig. 4 Composition dependence of the Nef-membrane interaction. (a) Schematic display of the lipid components tested. To compensate for the fivefold net charge of PIP 2, the PG content was reduced in the experiments accordingly. (b) Membrane binding kinetics of myristoylated wild type Nef with liposomes of 50 nm radius and varying lipid compositions. No significant change in the fast association process could be observed that would indicate a preference for a particular membrane composition. (c) Dissociation experiments of Nef with liposomes of 40/30/30 PC/PG/cholesterol composition, 40/30/30 PC/PG/sphingomyelin composition and 40/30/15/15 PC/PG/Cholesterol/sphingomyelin composition using ten-fold excess of unlabelled liposomes. (d) Screen of Nef binding to phosphoinositol-phosphates using a PIP strip assay. While Nef binds all single, double and triple inositol-phosphates and phosphatidic acid to a comparable level, it does not bind the non-charged lipids PC, PE and LPC. Single chained lipids as LPA and sphingosine 1- phosphate (S1P) are also not bound as are PI and PS, which carry only one negative net charge. Overall, the PIP strip screen suggests that Nef has no preference for a specific inositol phosphate head group. C-terminally His-tagged Nef was recognized by an anti-his antibody. - 9 -
Supplementary Fig. 5 N-terminal residues in Nef sustain membrane association by formation of an amphipathic helix. (a) SDS PAGE analysis of the Nef proteins tested for membrane binding. (b) CD spectroscopy of myristoylated Nef revealed an increase of helical content upon addition of PC/PG liposomes, suggesting the formation of an amphipathic helix upon membrane binding. (c) In contrast, the CD spectra of Nef(WR) remained unchanged upon addition of PC/PG liposomes. - 10 -
Supplementary Fig. 6 Analysis on an N-terminal Nef peptide for lipid membrane binding. (a) Sequence of the Nef SF2 peptide encompassing residues 2-27 in a myristoylated and non-myristoylated form. The peptide contains eight positively charged residues but only two negatively charged residues resulting in a theoretical pi of 12.0. (b) ESI mass spectrometry analysis of the two peptides synthesized. The determined mass of 3018 Da for Nef SF2 (2-27) corresponds perfectly to the calculated mass of 3017.4 Da (upper panel). Similarly, the mass of 3229 Da for the myristoylated peptide matches well with the calculated mass of 3226.8 Da (lower panel). (c) Fluorescence emission spectra of the Nef (2-27) peptide in absence and presence of PC/PG liposomes (70:30). The peptide contains two highly conserved tryptophans at positions 5 and 13, whose emission maximum is at 356 nm wavelength in the absence of lipids. Addition of 94 µm or 500 µm lipids of 50 nm radius led to an increase in the fluorescence signal and a shift in the emission spectra to 342 nm. - 11 -
Supplementary Fig. 7 CD spectra of Nef peptides and association kinetics support formation of an N-terminal helix upon liposome binding. (a) Addition of 311 µm lipids (PC/PG 70:30) to 11 µm Nef (2-27) peptide revealed a large increase in α-helical content, as seen by a spectral intensity decrease at 208 and 222 nm wavelength. In contrast, the spectra of 14 µm Nef (2-27) in water indicate an unstructured peptide. (b) CD spectra of the remaining counterpart of Nef (28-210) indicate no conformational change about addition of liposomes, in line with the lack of interaction. (c) Similar experiment as in (a) using the myristoylated Nef (2-27) peptide. Due to the low solubility the spectra appear very noisy, precluding the quantitative examination of the secondary structure content. Part of the peptide was also precipitated. Addition of liposomes led again to an increase of helical content. Together, these experiments suggest that an N-terminal amphipathic helix is formed upon Nef membrane binding. (d) Kinetic data of a titration series of non-myristoylated Nef peptide (2-27) at 1 µm concentration to PC/PG lipids (70:30, 50 nm liposomal radius) from 2.5 to 62.5 µm concentration. As before, the fluorescence signal results from FRET of the tryptophans in the peptide to a dye embedded in the lipids. The association process was recorded over 500 ms time. The binding kinetics can be sufficiently described by a bi-exponential association process. Displayed are the first 100 ms of the recording that cover the apparent rate constants. (e) Representative fit of the bi-exponential association curve at 37.5 µm lipid concentration. Signal amplitudes correspond to 61% for the fast rate process and 39% for the slow rate process. (f) The evaluation of the binding kinetics revealed again a concentration dependent fast association rate (starting from 140 s -1 to 220 s -1 ) and a concentration independent constant rate, now at 50 s -1. Note that the molecular weight of the peptide is only about 1/8 of the full length protein while the calculated pi changed from 5.9 of the native protein to 12.0 for the N-terminal 27 residues. - 12 -