Simultaneous Removal of Thiolated Membrane Proteins Resulting in

May 10, 2006 - ... major parts of the mesoscopic self-assembled protein structures can be .... the step to react the PM with gold nanoparticles first ...
0 downloads 0 Views 815KB Size
Langmuir 2006, 22, 5213-5216

5213

Simultaneous Removal of Thiolated Membrane Proteins Resulting in Nanostructured Lipid Layers Aiguo Wu,† Zhihong Jia,‡ Andreas Schaper,‡ Frank Noll,† and Norbert A. Hampp*,†,‡ Faculty of Chemistry and Materials Sciences Center, UniVersity of Marburg, Hans-Meerwein-Strasse, D-35032 Marburg, Germany ReceiVed NoVember 22, 2005. In Final Form: April 7, 2006 Self-organization of membrane-embedded peptides and proteins causes the formation of lipid mesostructures in the membranes. One example is purple membranes (PM), which consist of lipids and bacteriorhodopsin (BR) as the only protein component. The BRs form a hexagonal crystalline lattice. A complementary structure is formed by the lipids. Employing BR and PM as an example, we report a method where major parts of the mesoscopic self-assembled protein structures can be extracted from the lipid bilayer membrane. A complementary lipid nanostructure remains on the substrate. To remove such a large number of thiolated proteins simultaneously by applying a mechanical force, they are first reacted at physiological conditions with gold nanoparticles, and then a thin gold film is sputtered onto them that fuses with the gold nanoparticles forming a uniform layer, which finally can be lifted off. In this step, all of the previously gold-labeled proteins are pulled out of the membrane simultaneously. A stable lipid nanostructure is obtained on the mica substrate. Its stability is due to either binding of the lipids to the substrate through ionic bonds or to enough residual proteins to stabilize the lipid nanostructure against reorganization. This method may be applied easily and efficiently wherever thiolated proteins or peptides are employed as self-assembling and structure-inducing units in lipid membranes.

Introduction Lipid membranes are fluid at room temperature, and any damage (e.g., cracks or holes) will be repaired rapidly and disappear. At room temperature, one is unlikely to observe a stable lipid pattern on a substrate if supporting structures are not employed (e.g., grids of chromium lines1) or if a strong interaction of the lipid headgroups with the substrate does not exist. In general, covalent bonding is believed to be necessary to stabilize a lipid structure (e.g., a thiol-gold linkage). One way to obtain lipid nanostructures is to use lipid-coated AFM tips, but again suitably prepared and prestabilized lipid areas are needed.2 We introduce here a method that has two applications. First, it allows us to remove a large number of thiolated proteins from a lipid membrane simultaneously, and second, it enables us to obtain a nanoscopically structured supported lipid pattern. Thiolated membrane proteins, or other synthetic compounds with suitable properties and structure, that form supramolecular structures inside a lipid bilayer are mechanically pulled off of the lipid layer in a way that a negative image of the protein assembly remains in the lipid layer. This method is introduced using purple membranes (PM) as an example. PMs are cell membrane areas isolated from Halobacterium salinarum, which consist of lipids and bacteriorhodopsin (BR) only.3,4 The BRs are arranged in trimers that form a dense hexagonal crystalline lattice within the PMs. The stoichiometric ratio between lipids and BR is 10:1. The mass contribution of lipids in the PM is about 30%.5-8 Five lipid molecules are located at the extracellular side, three at the * To whom correspondence should be addressed. E-mail: hampp@ staff.uni-marburg.de. Phone: +49-6421-282-5775. Fax: +49-6421-2825798. † Faculty of Chemistry. ‡ Materials Sciences Center. (1) Jackson, B. L.; Groves, J. T. J. Am. Chem. Soc. 2004, 126, 13878. (2) Carlson, J. W.; Bayburt, T.; Sligar, S. G. Langmuir 2000, 16, 3927. (3) Oesterhelt, D.; Stoeckenius, W. Nat. New Biol. 1971, 233, 149. (4) Frederix, P. L. T. M.; Akiyama, T.; Staufer, C.; Gerber, C.; Fotiadis, D.; Mu¨ller, D. J.; Engel, A. Curr. Opin. Chem. Biol. 2003, 7, 641. (5) Essen, L. O.; Siegert, R.; Lehmann, W. D.; Oesterhelt, D. Proc. Natl. Acad. Sci. U.S.A. 1998, 95, 11673. (6) Hendler, R. W.; Dracheva, S. Biochemistry (Moscow) 2001, 66, 1311.

cytoplasmic side, and two inside the BR trimer.9 The lateral dimensions of PMs range from several hundred nanometers up to more than a micrometer, but all PMs have a uniform thickness of about 5.2 nm, the thickness of a monomolecular layer of BRs. It should be mentioned that PMs are quite different from a lipid membrane in general. The crystalline array formed from BR and the archaebacterial lipid composition indicates that this membrane cannot be considered to be a typical fluid lipid membrane. Furthermore, the composition of the archaebacterial membrane lipids is unique,10 and the specific properties of these lipids may be important to the results presented here. Experimental Section Purple Membranes. PMs containing the mutant D36C (PMD36C) were a gift from MIB (Leuna, Germany). The material was subjected to density gradient centrifugation. The main band was used for the experiments. PM-D36C differs from wild-type BR by the replacement of aspartic acid with cysteine at position 36. Preparation of PM Sheets on Mica. PMs are adsorbed onto mica surfaces using appropriate salt concentrations and pH values.11 Usually, 100 µL of a 0.1 mg/mL of purple membrane suspension was cast onto freshly cleaved mica substrates (about 2 cm × 6 cm) and rinsed three times with water in order to remove nonadsorbed purple membranes and inorganic salts. The samples were dried in air. Procedure for Gold Decoration of PM-D36C. First, the mutant PM-D36C is reacted with a colloidal solution of 5 nm gold nanoparticles (Sigma-Aldrich, Steinheim, Germany, 0.01% HAuCl4). The PM-D36C sample is washed twice with water to remove mercaptoethanol, which is required during storage to protect the cysteine groups from oxidation. Then it is dissolved in 3 mM phosphate buffer at pH 7 to a final concentration of PM-D36C corresponding to an absorbance of OD570 ) 3.5 to 4.0. Then, 6 µL (7) Corcelli, A.; Lattanzio, V. M. T.; Mascolo, G.; Papadia, P.; Fanizzi, F. J. J. Lipid Res. 2002, 43, 132. (8) Baudry, J.; Tajkhorshid, E.; Molnar, F.; Phillips, J.; Schulten, K. J. Phys. Chem. B 2001, 105, 905. (9) Mitsuoka, K.; Hirai, T.; Murata, K.; Miyazawa, A.; Kidera, A.; Kimura, Y.; Fujiyoshi, Y. J. Mol. Biol. 1999, 286, 861. (10) Renner, C.; Kessler, B.; Oesterhelt, D. J. Lipid Res. 2005, 46, 1755. (11) Mu¨ller, D. J.; Amrein, M.; Engel, A. J. Struct. Biol. 1997, 119, 172.

10.1021/la053162n CCC: $33.50 © 2006 American Chemical Society Published on Web 05/10/2006

5214 Langmuir, Vol. 22, No. 12, 2006

Letters

Figure 2. Formation of nanoscaled lipid structures by the extraction of mesoscopic protein assemblies. Here, PM sheets containing BRD36C were used. (I) Gold nanoparticles (5 nm) are coupled to the protein thiols. (II) The gold nanoparticles may bind to one, two, or three of the BR molecules within a BR trimer. (III) Gold evaporation onto the assembly causes fusion of the evaporated gold with the gold nanoparticles already attached, and a homogeneous layer is formed. (IV) Mechanical removal of the gold layer pulls the bound thiolated proteins off of the lipid layer. Figure 1. (A) AFM image of a typical PM patch on mica obtained in tapping mode. The thickness is 5.2 nm. (Inset) Height profile along the marked line. (B) Each PM patch is a crystalline array of BR trimers. A 5 nm Au particle is schematically shown for means of comparison. (C) The amino acids of the BR molecules are arranged in seven R-helical transmembrane helices. The location of Cys-36 on the cytoplasmic side is marked in red. of tributylphosphine solution is added, and the sample is incubated for 1 h at 37 °C. To 1300 µL of this solution 400 µL of the colloidal gold solution is added, and the sample is incubated for 24 h at 40 °C. The resulting material is washed twice in a centrifuge (15 min at 13 000 rpm, Biofuge 13, Heraeus), and the supernatant is discarded. The pellet is dissolved in 200 µL of water and collected for the experiments. The gold nanoparticles are somewhat smaller than the 6.3 nm periodicity of the BR lattice, and statistically each BR trimer may react with one gold nanoparticle but the number of bonds formed may range from one to three. After washing, the gold-modified PMs were adsorbed onto mica in the conventional manner. Excess material was removed by suctioning into filter paper. Gold Evaporation on the PM Sheets Adsorbed onto Mica. The samples were placed into the vacuum evaporating system. The gold was heated to 300-400 °C. Similar to the method reported in ref 13, the mica with PM sheets adsorbed was evaporated with gold. The mica/PM substrate was not heated. AFM Imaging. Mica used in these experiments was freshly cleaved and treated with a 10 mM MgCl2 solution for 5 min, and

after rinsing three times with water, it was dried and stored until used. AFM imaging was performed on a Nanoscope IV system (Veeco, Santa Barbara, CA). Pyramidal, oxide-sharpened Si3N4 tips attached to a V-shaped substrate (model NP-STT20, with a spring constant of either 0.06 or 0.32 N/m, Veeco, Santa Barbara, CA) were used for all studies in contact mode, both in air as well as in liquid. For tapping mode in air, silicon tips with a spring constant of 20-40 N/m were used.

Results and Discussion Individual BRs have been pulled out from PM patches in a modified atomic force microscope (AFM) when a force in the range of 100-200 pN was applied.12 We report a novel method here that enables the extraction of BR proteins from larger membrane areas simultaneously. Astonishingly, we found that the remaining lipid layer does not diffuse and does not reorganize on the surface of the substrate in order to form a uniform layer, but a stable nanoporous lipid structure is obtained. The PM samples (Figure 1) were prepared on mica from a solution comprising 0.4 mg/mL PM, 70 mM KCl, and 5 mM MgCl2 in a 5 mM Tris buffer solution at pH 8.2. The BR trimers show a periodicity of 6.3 nm. The loops (12) Oesterhelt, F.; Oesterhelt, D.; Pfeiffer, M.; Engel, A.; Gaub, H. E.; Mu¨ller, D. J. Science 2000, 288, 143. (13) Ulman, A. Chem. ReV. 1996, 96, 1533.

Letters

Langmuir, Vol. 22, No. 12, 2006 5215

Figure 3. TEM images of gold nanoparticle-modified PM-D36C sheets. (A) Gold nanoparticles seem to be bound statistically to the PM surface. (B) Areas are found where the gold nanoparticles reflect the underlying protein lattice. The different sizes of the membrane-attached gold nanoparticles are probably caused by a different number of linkages formed between the gold nanoparticles and the BRs within the BR trimers (i.e., one, two, or three). The substructure of the gold nanoparticles can be seen at the resolution shown. (C) Image B superimposed with a properly scaled sketch of the arrangement of BR trimers in PM.

connecting the seven R-helices of BR are accessible from the medium, and the thiol group of the cysteine amino acid exposed to the surface reacts with gold.13 The method comprises four steps (Figure 2): (I) the reaction of thiolated proteins with gold nanoparticles, (II) adsorption of the membrane to a substrate, (III) careful evaporation of a thin gold layer onto the assembly that causes the gold nanoparticles and the evaporated gold to fuse into a single layer, and (IV) peeling off the gold layer and pulling out the bound proteins. The method is demonstrated using PM sheets containing BR-D36C, which carries an extramembranous cysteine, but the method should also be suitable for a wide range of proteins and other types of membranes. First, the decoration of PM-D36C with 5 nm gold nanoparticles was examined by TEM (Figure 3). As a reference, wild-type PM (PM-WT) was used, but hardly any gold nanoparticles were observed to react with PM-WT. Quite different results are obtained with PM-D36C. Numerous gold nanoparticles are observed on the surface of PM-D36C, but they are not arranged in a regular structure with long-range order. This is due to the fact that, on average, only about 40% of the cysteine groups of all BRs in a membrane are reactive. However, quite a number of areas are found on the PM patches where a hexagonal grouping of the nanoparticles, reflecting the underlying crystalline lattice of the BR trimers, is observed. The nanoparticles bound to the surface seem to have different sizes. Of course we checked the uniformity of the purchased 5 nm gold particles by TEM prior to use. We assume that the different binding situationsssingle, double, or triple thiol linkages between a gold nanoparticle and the BRs inside a trimer occursaffect their charging characteristics during TEM analysis. Gold-modified PM patches are adsorbed preferentially with the extracellular side facing the mica substrates. A 0.1 mg/mL gold nanoparticle-modified PM-D36C sample (100 µL) was allowed to react with the mica substrate. After drying in air, the mica sheets were placed into a vaccum system, and gold was evaporated onto the samples to form a polycrystalline gold layer.14 The thickness of the gold film was about 1 µm. A steel disk was mounted onto the gold film using an epoxy glue (Araldit, ForboCTU, Scho¨nenwerd, Switzerland), and the gold layer was carefully peeled off. During this step, attached BR molecules were pulled out from the PM patches. On the mica substrate, a structured nanoporous lipid array is obtained that is stable for months (Figure 4). (14) Wagner, P.; Kernen, P.; Hegner, M.; Ungewickell, E.; Semenza, G. FEBS Lett. 1994, 356, 267.

Figure 4. AFM image, taken in contact mode, of the nanoporous lipid lattice on mica obtained upon extraction of BR molecules from the PM patch. (Inset) Close-up (1 µm2) of a region on the left side of the membrane patch. The lipid structure is very sensitive. Minor distortions of the structure due to the AFM imaging process may have occurred.

The 5 nm gold-labeled BRs leave with the gold layer when it is lifted off of the substrate (Figure 2). The attached proteins (Figure 5A) appear on the gold mold. They can be easily scratched away from the gold by means of the AFM tip. A negative mold of the PM is obtained in the gold (Figure 5B) having exactly the depth of the PM. The question may arise as to whether the nanostructures obtained on the mica are really lipids. As soon as forces in AFM imaging are increased, the structures first become less sharp and upon further increase are irreversibly damaged. This excludes the possibility that any type of nonsoft matter is imaged. Energy-dispersive X-ray (EDX) analysis shows S, O, and C on the gold (Figure 5A, inset). Because the only sulfur-containing groups are the thiols in the cysteine of BR-D36C and some methionines, these measurements indicate that the protein moiety is attached to the gold substrate. The results obtained with PM-D36C were checked by doing numerous control experiments. PM-WT was used as a reference because it does not have any surface-exposed cysteines such as PM-D36C. Also, the step to react the PM with gold nanoparticles first was skipped, and gold was evaporated directly onto micaadsorbed PMs, PM-WT, and PM-D36C. In Table 1, the results of these experiments are summarized.

5216 Langmuir, Vol. 22, No. 12, 2006

Letters

Figure 5. Gold layer with the BRs attached that were extracted simultaneously from a PM-D36C. (A) BRs are attached to the gold mold. (Inset) EDX spectrum taken from the gold area where the proteins are attached. The presence of carbon (C) and oxygen (O) and in particular sulfur (S) clearly indicates that the extracted proteins are bound to the surface. (B) The proteins were scratched away using the AFM tip. A mold of the membrane patch, showing the exact membrane thickness, remains. (Inset) Height profile along the line indicated. Table 1. Dependence of the Protein Extraction Efficiency on the Presence or Absence of Surface-Exposed Thiol Groupsa

direct gold evaporation reaction with nanogold followed by gold evaporation a

PM-WT

PM-D36C cytoplasmic side up (cysteine exposed)

PM-D36C extracellular side up (cysteine hidden)

none none

little very good

none none

See also Supporting Information.

Neither PM-WT nor PM-D36C, which is adsorbed to the substrate facing the cytoplasmic side to the mica, reacts with gold nanoparticles. The direct evaporation of gold, as described in the Experimental Section, also does not cause any reaction with gold. It is reported15 that at higher temperatures a reaction of surfaceexposed methionines with gold is observed, but such temperatures were not employed here. The effect with PMs was quite negligible, probably because the PM structure is lost because of melting and denaturation at the temperatures required for this process. The striking advantage of the reaction of the cysteines with gold nanoparticles at ambient conditions over direct gold evaporation is that the cysteines more easily react in liquid and because of (15) http://www.ajinomoto.co.jp/amino/e_aminoscience/bc/amino_12.html

the longer reaction time they obviously react to a much greater extent.

Summary and Conclusions The presented method allows one to pull huge numbers of thiolated protein molecules from a lipid bilayer simultaneously. Mesostructures formed by self-organization are preserved during this process. The lipids, as well as nonreacted and nonextracted proteins, are left on the mica. With PM as an example, micrometersized nanoporous lipid structures are obtained, but the method is not restricted to PM and BR. Proteins with surface-exposed cysteines are required. Because membrane proteins often form assemblies or nanostructures inside lipid membranes, this method benefits from their self-organization properties in order to prepare nanostructured lipid templates that are much larger and more complex than those obtained by other methods. Potential applications of this method may be in the analysis of protein distributions and interactions in lipid membranes but also in the fabrication of nanoporous structures that are formed from a matrix material and embedded molecules with surface-exposed thiols. Acknowledgment. We thank A. Scho¨nafinger for critical discussions and MIB for donating the PM samples. This work was supported through grant DFG HA2906/4-1. Supporting Information Available: AFM images from control experiments. This material is available free of charge via the Internet at http://pubs.acs.org. LA053162N