- Email: [email protected]
Experiments in membrane protein crystallization
Journal of Crystal Growth 90 (1988) 193—200 North-Holland, Amsterdam
193
EXPERIMENTS IN MEMBRANE PROTEIN CRYSTALLIZATION P. GROS, H. GROENDIJK, J. DRENTH and W.G.J. HOL Laboratory of Chemical Physics, University of Groningen, Nijenborgh 16, 9747 AG Groningen, The Netherlands
Received 25 July 1987; manuscript received in final form 16 February 1988
Experiments have been carried out aimed at obtaining crystals of three integral membrane proteins: phospholipase A, cytochrome-c reductase and cytochrome-c oxidase. Crystals, in the form of platelets and bipyramids, as well as several micro-crystal forms have been obtained of phospholipase A from the outer membrane of Escherichia coli. The best crystals obtained so far were thin plates having spacegroup C222 1 with a = 72.0 A, b 107 A, c = 75.8 A and one molecule per asymmetric unit. Reflections were observed Out to 2.7 A resolution. Long red needles have been obtained of the respiratory chain complex-Ill, or cytochrome-c reductase, from beefheart.
1. Introduction The three-dimensional structure determination of integral membrane and membrane associated proteins is an important prerequisite for solving several major biochemical problems such as proton gradient formation, ATP production, protein and metabolite translocation, cross-membrane signal transfer, cell recognition and lipid metabolism, Improved isolation and purification methods, in combination with the use of overproducing bacterial strains, have made the search for successful crystallization conditions feasible for an increasing number of membrane proteins. Development of a general strategy for crystallization of amphiphilic proteins will lead eventually to detailed insight into architecture and structure—function relationships for this important class of proteins. It is a tremendous help for interpretation and understanding of spectroscopic, kinetic and biochemical data to know the three-dimensional structure of the protein and the orientation and position of its ligands, if any. This is illustrated by the elucidation of the structure of the photosynthetic reaction centre [1]. The respiratory chain is a fascinating energy transfer system, containing four protein complexes with numerous chromophores, that has evolved
into an efficient electron transport system. Chemical energy is initially stored as a proton gradient across a membrane as first proposed by Mitchell [2]. The two respiratory chain complexes discussed in this paper, cytochrome-c reductase and cytochrome-c oxidase from beefheart, have been studied in great detail by numerous investigators (see, e.g., refs. [3—7]).The cytochrome-c reductase monomer is a mitochondrial protein complex of 230 kD and consists of 11 subunits. The second mitochondrial complex studied, cytochrome-c oxidase, is equally intricate. It has a molecular weight of 200 kD and consists of 13 subunits. It is clearly of great importance for fundamental biochemistry to know the architecture of these cornplexes at the atomic level. The much simpler integral membrane protein phospholipase A from the outer membrane of Escherichia coli, with a molecular weight of 29 kD [8], confronts us with other interesting problems. For instance, how is this lipid degrading protein, which is embedded in its substrate, regulated? This brings us directly to lipid metabolism and its important metabolites such as arachidomc acid. As the structures Of several water soluble phospholipase A2 have been determined [9—14],it will also be interesting to see how nature has designed non-homologous proteins to perform the same
0022-0248/88/$03.50 © Elsevier Science Publishers B.V. (North-Holland Physics Publishing Division)
194
P. Gros et al.
/ Experiments
in membrane protein crystallization
lipid hydrolysis. No significant sequence identity has been found between the membrane phospholipase and the water-soluble equivalents [15]. Also, fundamental questions that concern folding and stability of integral membrane proteins can be studied once the three-dimensional structures of these amphiphilic molecules are known. However, the difficult obstacle of obtaining high quality crystals must first be overcome. 2. Membrane protein crystallization techniques In recent years several reports on membrane protein crystallization have been published. A small number of membrane proteins has been crystallized and studied by X-ray diffraction, i.e. the photosynthetic reaction centre from Rhodopseudomonas (Rps.) viridis [16], the reaction centre from Rps. sphaeroides [17,18], porins from E. co/i [19,20], the light-harvesting complex B800-850 from Rps. capsulata [21] and bacteriorhodopsin from Halobacterium halobium [22]. In the first three cases a resolution of 2.3—6.0 A has been obtained; in the latter two 6—10 A resolution has been reported. Reports on the crystallization of more membrane proteins, or protein complexes, have appeared, but these have not been studied by X-ray diffraction [23—25]. General ideas, not yet fully established, have emerged from these experiences in crystallizing amphiphilic biomacromolecules. After solubilization, the membrane protein can in a first approximation be treated as if it was a water-soluble protein. Good crystal packing would result from interactions between hydrophilic parts, those parts of the protein that protrude outside the membrane, leaving the hydrophobic surfaces embedded in amphiphilic molecules. Therefore the crystal contacts should resemble those of water-soluble protein crystals. For solubilization small detergents and, if necessary, additional small amphiphilic molecules have been used. Large detergent molecules might prevent crystallization of the protein, by disrupting interactions between protein molecules by steric hindrance [26—28]. However, small crystals of photosystem I were obtained using a larger detergent, Triton X-100
[25]. In spite of the experience in handling the amphiphilic character of the membrane protein, the problem of finding the proper conditions for forming well-ordered crystals remains formidable in almost all cases.
3. Materials and methods The general procedure employed by us for crystallization of a membrane protein consists of several screening levels. First, depending on the amount of protein available, a few different types of detergents were chosen. The detergents used most frequently were /3-octylglucoside (Bohringer, Mannheim, Fed. Rep. of Germany), N,N-dimethyl-dodecylamine-N-oxide (Fluka, Buchs, Switzerland), lauryl maltoside (Bohringer, Mannheim, Fed. Rep. of Germany), octyl oligooxyethelene (Kwant, Industrieweg 27, Bedum, The Netherlands) and N-dodecyl-N,N dimethylammonio-3-propanesulfonate (dodecyl-DA PS) (Serva, Heidelberg, Fed. Rep. of Germany). Secondly, various buffers at different pH’s were selected. To exchange the detergent and the buffer, the protein sample was dialyzed for about one week. The dialyzed protein samples were used for crystallization trials, varying pH, temperature, precipitating agent, additives, etc., as is done usually for soluble proteins. Care was taken that the detergent concentration did not fall below its actual critical micellar concentration (CMC) in that particular mixture of protein, detergent and, possibly, lipids. The crystallization methods so far have been restricted to the hanging drop technique and the liquid—liquid diffusion, or two-layer, method. After visual inspection of the experiments, the most promising conditions were taken as a starting point for reproduction and/or improvement of the crystals obtained. The search for crystallization conditions for phospholipase A was conducted with lipid-contaming protein as well as delipidated protein. Protein isolation and purification was done by H. Verheij and co-workers (University of Utrecht, Utrecht, The Netherlands) [15]. After purification, the protein was dissolved in 2.5mM dodecylDAPS, 10mM histidine buffer at pH 6.0 and 2mM
P. Gros et aL
/ Experiments in
ethylenedinitrilotetraacetic acid (EDTA). The protein concentration used in the trials ranged from 4 to about 20 mg/mi. Beefheart cytochrome-c reductase was isolated and purified by J. Berden and co-workers from the B.C.P. Janssen Institute, University of Amsterdam, Amsterdam, The Netherlands [4]. In this case the sample was not specifically delipidated prior to the crystallization trials. The enzyme cornplex was solubilized in 1% cholate, 56 mM tris(hydroxymethyl)-aminomethane(Tris). HC1, pH 8.0 and 10%-saturated ammonium sulphate. For crystallization trials at pH-values below 6.5 the protein sample was first dialyzed to a non-cholate detergent at a pH higher than 6.5, because of the low solubility of cholate below pH 6.5. It is not definitely established that this procedure removes cholate completely. Beefheart cytochrome-c oxidase was isolated and purified by B. van Gelder and co-workers from the B.C.P. Janssen Institute, University of Amsterdam, The Netherlands [6]. The protein buffer consists of 1% cholate, 60mM Tris-H2S04, pH 8.0 and 36%-saturated (NH4)2S04. The cytochrome-c oxidase was treated similarly to cytochrome-c reductase. For both complexes a sys-
membrane protein crystallization
195
tematic search of 2000 experiments was set up. Parameters varied were: (i) the type of detergent, i.e. /3-octyl glucoside, lauryl maltoside, lauryl dimethyl amine oxide, nonanoyl-N-methyl-glucamide (Oxyl, Bobingen, Fed. Rep. of Germany), dodecyl-DAPS and octyl hexaoxyethylene (Kwant, Bedum, The Netherlands); (ii) pH varying from 6.5 to 9.0 in steps of 0.5 pH units; (iii) the nature and concentration of the precipitant (ammonium sulphate, 2-methyl-2,4-pentanediol, polyethyleneglycol-2000, 4000, and 6000); (iv) the temperature, either room temperature or 4°C. In all cases the two-layer method was employed with 5 pl protein solution (with a detergent concentration of more than 2 CMC) and 5 p~lprecipitant solution in capillaries with 1.3 mm diameter. Besides this systematic search a smaller amount of hanging drops were set up, varying the same parameters. Crystals of sufficient size were tested for their diffraction quality. Al glassware used for mounting crystals needed to be siliconized. The presence of detergents lowers the liquid surface tension which made the mounting process almost impossible without siliconization. Preliminary X-ray studies were done on a rotating anode, Elliott GX-6; oscillation photographs were made at the EMBL
4
__
Fig. 1. Needle-like crystals of cvtochrome-c reductase grown in a hanging drop at 4°C. 100mM K~HPO4,KH~PO4,pH 6.7, 0.35% w/v nonyl oligo-oxyethylene and 20% polyethylene glycol-6000.
196
P. Gros et aL
/ Experiments
in membrane protein crystallization
outstation in Hamburg (Fed. Rep. of Germany) using synchrotron radiation.
4. Results and discussion 4.1. Cytochrome-c reductase and cytochrome-c oxidase
Dark coloured, reddish, needle-like crystals (up to 0.7 mm in length) of cytochrome-c reductase were obtained at a number of conditions, using the hanging drop method at 4°C. A photograph of one experiment is shown in fig. 1. Needles were grown in experiments with 1.0% w/v ~8-octylglucoside or 0.35% w/v nonyl oligo-oxyethylene (C9E ~)with 100mM potassium phosphate buffer at pH 6.7 and 20 to 30% w/v polyethylene glycol4000 or 6000. Equivalent experiments with decyl oligo-oxyethylene (C10E ~ (0.35% w/v) as detergent gave needles in experiments with 20% w/v -
polyethylene glycol-6000 and also in experiments with 20 to 30%-saturated ammonium sulphate. Some similarity was found between the results using different types of detergents. However, this is not a generally observed feature. Precipitation points of the protein can change drastically when another detergent is used for solubilization. Large numbers of colourless bipyramidal mlcrocrystals, with dimensions up to 50 X 30 X 30 ~sm3,have been observed in liquid—liquid diffusion experiments of cytochrome-c oxidase. Conditions at the beginning of the liquid—liquid diffusion experiment were as follows. The upper layer contained 10mM maleic acid buffer pH 6.5, 1mM NaN 3, 5% w/v polyethylene glycol-2000, 4000 or 6000, or 10% w/v polyethylene glycol-4000. The lower consisted of 15 mg/ml protein solution with 0.2% w/v octyl hexa oxy ethylene, 10mM maleic acid, pH 6.5 and 1mM NaN3. It remains to be established, which subunits of cytochrome-c oxidase are present in the micro-
Table 1 Crystallization of phospholipase A Experiment
Protein solution
Additive solution
Precipitant solution
Results
Exp. A
2.5 fil of: 10mM/3-00 0.9% BES•NaOH, pH 6.8 12 mg/mI PLA — 50 lipids/PLA’)
1.091 of: 30mM CaCI,
2.591 of: 10mM 20% v/vBESNaOH, MPD pH 6.8
Large number of crystals with3. sizesallupintergrown but toO.4X0.4X0.05 mm
Exp. B
See exp. A
1.0 91 of: 15mM CaCl
See exp. A
One single crystal.
See exp. A
3 0.3x0.3x0.01 mm Two single crystals,
2.0 ~l of: 10mM BES•NaOH, pH 6.8 15% w/v PEG 4000
3 0.3 >< 0.3 )< 0.025 mm Large number of hipyramidal single crystals up to 0.1 xO.l X0.l mm3
2.5 91 of:
One large crystal.
pH 6.8 ethanolaniine, 10mM 50% v/v MPD
3, consisting of three 0.4x0.4X0.4 mm intergrown crystals
Exp. B’
See exp. A, but with 20 lipids/PLA
Exp. C
1.0 91 of: 10mM BESNaOH, pH 6.8 0.9% /3-00 = 50 lipids/PLA
Exp. D
2.5 ~l of: 10mM ethanolamineNaOH, p1-I 10.0 0.9% /3-00 50 lipids/PLA
2
1.0 /11 of: 15mM CaCI
2
1.0 91 of: 30mM CaCl 0.9M KI 2
PLA: phospholipase A. BES: N-N-bis[2-hydroxyethyl]-2-aminoethane sulfonic acid. MPD: 2-methyl-2,4-pentane diol. PEG: polyethyleneglycol. The amount of lipid molecules per protein molecule present is measured after purification.
P. Gros et al.
/ Experiments in pnembrane protein crystallization
crystals described above. The same holds for the exact composition of the needle-shaped crystals of cytochrome-c reductase. These crystals, however, have the same colour as the native protein, mdicating that at least the chromophores are present in the crystals.
197
4.2. Phospholipase A
A number of microcrystals, platelets, needles and bipyramids, were obtained in the first crystallization experiments, testing mainly two pH’s, 10mM histidine-buffer pH 6.0 and pH 9.3; four different detergents: /3-octyiglucoside, octyl
aiTh.ii~ii~IIip,mp
I,
-~
~iTI11i1tII,N~U~IIilflu1
_____
Fig. 2. (a) Crystal obtained from experiment B, as described in table 1. (b) Side view of a crystal obtained from experiment B’, i.e. using 60% less lipid molecules per phospholipase A molecule compared to experiment B.
198
P. Gros et a!.
/
Experiments in membrane protein crystallization
oligo-oxyethylene, dodecyl-DAPS and lauryl dimethyl amine oxide; and five different precipitants: ammonium sulphate, 2-methyl-2,4-pentane diol, ethanol, polyethylene glycol-4000 and 6000; with CaCI2 as an additive. Calcium chloride was added because of its necessity for the activity of the protein [15]. From crystallization of watersoluble phospholipases A2 it is known that CaCl2 has a distinct effect on the crystallization behaviour [10]. After the first series of experiments, attention was centred mainly on two crystallization conditions, i.e. conditions roughly similar to experiments A and C in table 1. Finally, large crystals were obtained for three crystallization conditions. Table 1 shows the result of the suecessful experiments, while in fig. 2 photographs of crystals obtained are shown. Experiments C and D clearly showed phase separation in the drop, while phase separation was not visible in experiments A and B. In experiments analogous to A and B, phase separation could be induced by increasing the precipitant concentration. The different detergent states between C, D and A, B can be correlated with the difference in crystal stability during crystal handling. Crystals from experiments C and D were extremely unstable and dissolved during mounting. Crystals from A and B were stable and have been mounted and used for
preliminary X-ray diffraction studies. The problem of instability for crystals grown in vapour diffusion systems [27] may be limited to those cases in which crystals have been grown at or near detergent phase separation conditions. This coincides with the idea of micelle interaction in the latter crystals and firmer crystal contacts through electrostatic and hydrogen bond interactions between the protein molecules in crystals grown without phase separation. Although a crystal from experiment A diffracted up to 2.7 A resolution (fig. 3), it is clear that these crystals are not suitable for X-ray data collection. In experiment B (see table 1), the problem of twinning is overcome, but, unfortunately, these crystals are very thin. Diffraction experiments have been performed at the EMBL outstation in Hamburg (Fed. Rep. of Germany) using synchrotron radiation. These experiments were not yet satisfactory, mainly due to curving of the thin crystals by the capillary wall. A lipid-free protein sample has been used to set up new experiments equivalent to experiment B. In experiment B, as described in table 1, the concentration of /3-octyl glucoside is below its CMC. Due to large amounts of lipids present, micelles can be formed and the protein is still soluble. However, in a lipid-free sample the pro-
•0
Fig. 3. Procession photographs (~= 12°),collected on a rotating anode, of a phospholipase A crystal from experiment A. The crystal diffracts to 2.7 A resolution. X-ray damage was1~M noticeable = 2.5 A3/dalton after 1 week corresponding and 60 h with of X-ray one molecule exposure. per Theasymmetric crystal has unit. spacegroup As the C2221: aof lipid number 72.0 and A. 6=107 detergent A,molecules c = 75.8 A perand protein molecule have not been determined, the V~ refers to volume per protein mass only.
P. Gros eta!.
/
Experiments in membrane protein crystallization
tein tends to denature if the detergent concentration drops below its CMC [15]. The first crystallization trials with lipid free material using condilion B, with double and triple amounts of detergent, resulted in intergrown microcrystals, with dimensions of 50 X 50 X 10 p.m3. This result is very similar to the early, non-optimized, result around condition B. From these preliminary results it may be concluded that this membrane protein can be crystallized both from lipid-containing and lipid-free samples.
5. Concluding remarks The main results of the crystallization experiments are the crystals of phospholipase A from E. co/i. The single crystals obtained have two dimensions up to about 0.3 mm. The third dimension, however, is still a problem. A second result of the phospholipase experiments is the growth of crystals from both lipid-containing and lipid-free protein. Furthermore we have shown that membrane protein crystals from hanging drops can be mounted and used for X-ray diffraction studies. However, if phase separation had occurred in the hanging drop, the crystals obtained dissolved during mounting. Other promising results are the needle-like crystals of cytochrome-c reductase, that have the colour of the native protein. However, those crystals still need further improvement. One way of performing large numbers of crystallization trials around these cytochrome-c reductase crystallization conditions, as described above, is to apply automation. A simple, but efficient, way to automate crystallization trials has recently been described by Kelders et al. [29].
Acknowledgements It is a pleasure to acknowledge the stimulating cooperation with Drs. H. Verhey, P. de Geus, A. Horrevoets (PLA), J. Berden, S. de Vnes (Complex-Ill), B.F. van Gelder, T. Muyser, T. Hakvoort and H. Dekkers (Complex-W). We furthermore like to thank Drs. R. Wierenga, R. Read, M.
199
Swarte and H. Kelders for discussions and Mrs. R. Hogenkamp for preparing the manuscript. This research was supported by The Netherlands Foundation for Chemical Research (SON) with financial aid from The Netherlands Organization for the Advancement of Pure Research (ZWO).
References [1] J. Deisenhofer, 0. Epp, K. Miki, R. Huber and H. Michel, J. Mol. Biol. 180 (1984) 385. 121 P. Mitchell, Nature 191 (1961) 144. [31Y. Hafeti, AG. Haavik and P. Jurtshuk, Biophys. Acta 52 (1961) 106. [4] Y. Hafeti, Methods Enzymol. 53 (1978) 35. [5] B.D. Nelson and P. Gellerfors, Methods Enzymol. 53 (1978) 80. [6] C.R. Hartzell, H. Beinert, B.F. van Gelder and T.E. King, Methods Enzymol. 53 (1978) 54. [7] R.A. Capaldi, F. Malatesta and V.-M. Darley-Usmar, Biochem. Biophys. Acta 726 (1983) 135. [8] P. de Geus, I. van Die, H. Bergmans, J. Tommassen and G. de Haas, Mol. Gen. Genetics 190 (1983) 150. [91B.W. Dijkstra, K.H. Kalk, W.G.J. Hol and J. Drenth, J. Mol. Biol. 147 (1981) 93. [10] B.W. Dijkstra, G.J.H. van Nes, K.H. Kalk, NP. Brandenburg, W.G.J. Hol and J. Drenth, Acta Cryst. B38 (1982) 793. [11] B.W. Dijkstra, R. Renetseder, K.H. Kalk, W.G.J. HoI and J. Drenth, J. Mol. Biol. 168 (1983) 163. [12] B.W. Dijkstra, K.H. Kalk, J. Drenth, G.H. de Haas, MR. Egmond and A.J. Slotboom. Biochemistry 23 (1984) 2759. [13] C. Keith, D.S. Feldman, S. Degemello. J. Glick, KB. Ward, GD. Jones and PB. Sigler, J. Biol. Chem. 256 (1981) 8602. [14] S. Brume, J. Bolin. D. Gewirth and PB. Sigler. J. Biol. Chem. 260 (1985) 9742. [15] P. de Geus, Thesis, University of Utrecht (1986). [16] H. Michel, J. Mol. Biol. 158 (1982) 567. [17] J.P. Allen and 0. Feher, Proc. NatI. Acad. Sci. USA 81 (1984) 4795. [18] C.H. Chang, M. Schiffer, D. Tiede, U. Smith and J. Norris, J. Mol. Biol. 186 (1985) 201. [19] R.M. Garavito, J. Jenkins, J.N. Jansonius. R. Karlsson and J.P. Rosenbusch, J. Mol. Biol. 164 (1983) 313. [20] R.M. Garavito, U. Hinz and J.M. Neuhaus, J. Biol. Chem. 259 (1984) 4254. [21] W. Welte, T. Wacker, M. Leiss, W. Kreutz, J. Shiozawa, N. Gadón and G. Drews, FEBS Letters 182 (1985) 260. [22] H. Michel and D. Oesterhelt, Proc. NatI. Acad. Sci. USA 77 (1980) 1283. [23] W. Kuehlbrandt, Nature 307 (1984) 478.
200
P. Gros et aL
/ Experiments
[24] T. Wacker, N. GadOn, A. Becker, W. Mantele, W. Kreutz, 0. Drews and W. Welte, FEBS Letters 197 (1986) 267. [25] R.C. Ford, D. Picot and R.M. Garavito, EMBO J. 6 (1987) 1581. [26] H. Michel, Trends Biochem. Sci. 8 (1983) 56. [27] R.M. Garavito and J.P. Rosenbusch, Methods Enzymol. 125 (1986) 309.
in membrane protein crystallization
[28] R.M. Garavito, Z. Markovic-Housley and J.A. Jenkins, J. Crystal Growth 76 (1986) 701. [29] HA. Kelders, K.H. Kalk, P. Gros and W.G.J. Hol, Protein Eng. 1(1987) 301.
193
EXPERIMENTS IN MEMBRANE PROTEIN CRYSTALLIZATION P. GROS, H. GROENDIJK, J. DRENTH and W.G.J. HOL Laboratory of Chemical Physics, University of Groningen, Nijenborgh 16, 9747 AG Groningen, The Netherlands
Received 25 July 1987; manuscript received in final form 16 February 1988
Experiments have been carried out aimed at obtaining crystals of three integral membrane proteins: phospholipase A, cytochrome-c reductase and cytochrome-c oxidase. Crystals, in the form of platelets and bipyramids, as well as several micro-crystal forms have been obtained of phospholipase A from the outer membrane of Escherichia coli. The best crystals obtained so far were thin plates having spacegroup C222 1 with a = 72.0 A, b 107 A, c = 75.8 A and one molecule per asymmetric unit. Reflections were observed Out to 2.7 A resolution. Long red needles have been obtained of the respiratory chain complex-Ill, or cytochrome-c reductase, from beefheart.
1. Introduction The three-dimensional structure determination of integral membrane and membrane associated proteins is an important prerequisite for solving several major biochemical problems such as proton gradient formation, ATP production, protein and metabolite translocation, cross-membrane signal transfer, cell recognition and lipid metabolism, Improved isolation and purification methods, in combination with the use of overproducing bacterial strains, have made the search for successful crystallization conditions feasible for an increasing number of membrane proteins. Development of a general strategy for crystallization of amphiphilic proteins will lead eventually to detailed insight into architecture and structure—function relationships for this important class of proteins. It is a tremendous help for interpretation and understanding of spectroscopic, kinetic and biochemical data to know the three-dimensional structure of the protein and the orientation and position of its ligands, if any. This is illustrated by the elucidation of the structure of the photosynthetic reaction centre [1]. The respiratory chain is a fascinating energy transfer system, containing four protein complexes with numerous chromophores, that has evolved
into an efficient electron transport system. Chemical energy is initially stored as a proton gradient across a membrane as first proposed by Mitchell [2]. The two respiratory chain complexes discussed in this paper, cytochrome-c reductase and cytochrome-c oxidase from beefheart, have been studied in great detail by numerous investigators (see, e.g., refs. [3—7]).The cytochrome-c reductase monomer is a mitochondrial protein complex of 230 kD and consists of 11 subunits. The second mitochondrial complex studied, cytochrome-c oxidase, is equally intricate. It has a molecular weight of 200 kD and consists of 13 subunits. It is clearly of great importance for fundamental biochemistry to know the architecture of these cornplexes at the atomic level. The much simpler integral membrane protein phospholipase A from the outer membrane of Escherichia coli, with a molecular weight of 29 kD [8], confronts us with other interesting problems. For instance, how is this lipid degrading protein, which is embedded in its substrate, regulated? This brings us directly to lipid metabolism and its important metabolites such as arachidomc acid. As the structures Of several water soluble phospholipase A2 have been determined [9—14],it will also be interesting to see how nature has designed non-homologous proteins to perform the same
0022-0248/88/$03.50 © Elsevier Science Publishers B.V. (North-Holland Physics Publishing Division)
194
P. Gros et al.
/ Experiments
in membrane protein crystallization
lipid hydrolysis. No significant sequence identity has been found between the membrane phospholipase and the water-soluble equivalents [15]. Also, fundamental questions that concern folding and stability of integral membrane proteins can be studied once the three-dimensional structures of these amphiphilic molecules are known. However, the difficult obstacle of obtaining high quality crystals must first be overcome. 2. Membrane protein crystallization techniques In recent years several reports on membrane protein crystallization have been published. A small number of membrane proteins has been crystallized and studied by X-ray diffraction, i.e. the photosynthetic reaction centre from Rhodopseudomonas (Rps.) viridis [16], the reaction centre from Rps. sphaeroides [17,18], porins from E. co/i [19,20], the light-harvesting complex B800-850 from Rps. capsulata [21] and bacteriorhodopsin from Halobacterium halobium [22]. In the first three cases a resolution of 2.3—6.0 A has been obtained; in the latter two 6—10 A resolution has been reported. Reports on the crystallization of more membrane proteins, or protein complexes, have appeared, but these have not been studied by X-ray diffraction [23—25]. General ideas, not yet fully established, have emerged from these experiences in crystallizing amphiphilic biomacromolecules. After solubilization, the membrane protein can in a first approximation be treated as if it was a water-soluble protein. Good crystal packing would result from interactions between hydrophilic parts, those parts of the protein that protrude outside the membrane, leaving the hydrophobic surfaces embedded in amphiphilic molecules. Therefore the crystal contacts should resemble those of water-soluble protein crystals. For solubilization small detergents and, if necessary, additional small amphiphilic molecules have been used. Large detergent molecules might prevent crystallization of the protein, by disrupting interactions between protein molecules by steric hindrance [26—28]. However, small crystals of photosystem I were obtained using a larger detergent, Triton X-100
[25]. In spite of the experience in handling the amphiphilic character of the membrane protein, the problem of finding the proper conditions for forming well-ordered crystals remains formidable in almost all cases.
3. Materials and methods The general procedure employed by us for crystallization of a membrane protein consists of several screening levels. First, depending on the amount of protein available, a few different types of detergents were chosen. The detergents used most frequently were /3-octylglucoside (Bohringer, Mannheim, Fed. Rep. of Germany), N,N-dimethyl-dodecylamine-N-oxide (Fluka, Buchs, Switzerland), lauryl maltoside (Bohringer, Mannheim, Fed. Rep. of Germany), octyl oligooxyethelene (Kwant, Industrieweg 27, Bedum, The Netherlands) and N-dodecyl-N,N dimethylammonio-3-propanesulfonate (dodecyl-DA PS) (Serva, Heidelberg, Fed. Rep. of Germany). Secondly, various buffers at different pH’s were selected. To exchange the detergent and the buffer, the protein sample was dialyzed for about one week. The dialyzed protein samples were used for crystallization trials, varying pH, temperature, precipitating agent, additives, etc., as is done usually for soluble proteins. Care was taken that the detergent concentration did not fall below its actual critical micellar concentration (CMC) in that particular mixture of protein, detergent and, possibly, lipids. The crystallization methods so far have been restricted to the hanging drop technique and the liquid—liquid diffusion, or two-layer, method. After visual inspection of the experiments, the most promising conditions were taken as a starting point for reproduction and/or improvement of the crystals obtained. The search for crystallization conditions for phospholipase A was conducted with lipid-contaming protein as well as delipidated protein. Protein isolation and purification was done by H. Verheij and co-workers (University of Utrecht, Utrecht, The Netherlands) [15]. After purification, the protein was dissolved in 2.5mM dodecylDAPS, 10mM histidine buffer at pH 6.0 and 2mM
P. Gros et aL
/ Experiments in
ethylenedinitrilotetraacetic acid (EDTA). The protein concentration used in the trials ranged from 4 to about 20 mg/mi. Beefheart cytochrome-c reductase was isolated and purified by J. Berden and co-workers from the B.C.P. Janssen Institute, University of Amsterdam, Amsterdam, The Netherlands [4]. In this case the sample was not specifically delipidated prior to the crystallization trials. The enzyme cornplex was solubilized in 1% cholate, 56 mM tris(hydroxymethyl)-aminomethane(Tris). HC1, pH 8.0 and 10%-saturated ammonium sulphate. For crystallization trials at pH-values below 6.5 the protein sample was first dialyzed to a non-cholate detergent at a pH higher than 6.5, because of the low solubility of cholate below pH 6.5. It is not definitely established that this procedure removes cholate completely. Beefheart cytochrome-c oxidase was isolated and purified by B. van Gelder and co-workers from the B.C.P. Janssen Institute, University of Amsterdam, The Netherlands [6]. The protein buffer consists of 1% cholate, 60mM Tris-H2S04, pH 8.0 and 36%-saturated (NH4)2S04. The cytochrome-c oxidase was treated similarly to cytochrome-c reductase. For both complexes a sys-
membrane protein crystallization
195
tematic search of 2000 experiments was set up. Parameters varied were: (i) the type of detergent, i.e. /3-octyl glucoside, lauryl maltoside, lauryl dimethyl amine oxide, nonanoyl-N-methyl-glucamide (Oxyl, Bobingen, Fed. Rep. of Germany), dodecyl-DAPS and octyl hexaoxyethylene (Kwant, Bedum, The Netherlands); (ii) pH varying from 6.5 to 9.0 in steps of 0.5 pH units; (iii) the nature and concentration of the precipitant (ammonium sulphate, 2-methyl-2,4-pentanediol, polyethyleneglycol-2000, 4000, and 6000); (iv) the temperature, either room temperature or 4°C. In all cases the two-layer method was employed with 5 pl protein solution (with a detergent concentration of more than 2 CMC) and 5 p~lprecipitant solution in capillaries with 1.3 mm diameter. Besides this systematic search a smaller amount of hanging drops were set up, varying the same parameters. Crystals of sufficient size were tested for their diffraction quality. Al glassware used for mounting crystals needed to be siliconized. The presence of detergents lowers the liquid surface tension which made the mounting process almost impossible without siliconization. Preliminary X-ray studies were done on a rotating anode, Elliott GX-6; oscillation photographs were made at the EMBL
4
__
Fig. 1. Needle-like crystals of cvtochrome-c reductase grown in a hanging drop at 4°C. 100mM K~HPO4,KH~PO4,pH 6.7, 0.35% w/v nonyl oligo-oxyethylene and 20% polyethylene glycol-6000.
196
P. Gros et aL
/ Experiments
in membrane protein crystallization
outstation in Hamburg (Fed. Rep. of Germany) using synchrotron radiation.
4. Results and discussion 4.1. Cytochrome-c reductase and cytochrome-c oxidase
Dark coloured, reddish, needle-like crystals (up to 0.7 mm in length) of cytochrome-c reductase were obtained at a number of conditions, using the hanging drop method at 4°C. A photograph of one experiment is shown in fig. 1. Needles were grown in experiments with 1.0% w/v ~8-octylglucoside or 0.35% w/v nonyl oligo-oxyethylene (C9E ~)with 100mM potassium phosphate buffer at pH 6.7 and 20 to 30% w/v polyethylene glycol4000 or 6000. Equivalent experiments with decyl oligo-oxyethylene (C10E ~ (0.35% w/v) as detergent gave needles in experiments with 20% w/v -
polyethylene glycol-6000 and also in experiments with 20 to 30%-saturated ammonium sulphate. Some similarity was found between the results using different types of detergents. However, this is not a generally observed feature. Precipitation points of the protein can change drastically when another detergent is used for solubilization. Large numbers of colourless bipyramidal mlcrocrystals, with dimensions up to 50 X 30 X 30 ~sm3,have been observed in liquid—liquid diffusion experiments of cytochrome-c oxidase. Conditions at the beginning of the liquid—liquid diffusion experiment were as follows. The upper layer contained 10mM maleic acid buffer pH 6.5, 1mM NaN 3, 5% w/v polyethylene glycol-2000, 4000 or 6000, or 10% w/v polyethylene glycol-4000. The lower consisted of 15 mg/ml protein solution with 0.2% w/v octyl hexa oxy ethylene, 10mM maleic acid, pH 6.5 and 1mM NaN3. It remains to be established, which subunits of cytochrome-c oxidase are present in the micro-
Table 1 Crystallization of phospholipase A Experiment
Protein solution
Additive solution
Precipitant solution
Results
Exp. A
2.5 fil of: 10mM/3-00 0.9% BES•NaOH, pH 6.8 12 mg/mI PLA — 50 lipids/PLA’)
1.091 of: 30mM CaCI,
2.591 of: 10mM 20% v/vBESNaOH, MPD pH 6.8
Large number of crystals with3. sizesallupintergrown but toO.4X0.4X0.05 mm
Exp. B
See exp. A
1.0 91 of: 15mM CaCl
See exp. A
One single crystal.
See exp. A
3 0.3x0.3x0.01 mm Two single crystals,
2.0 ~l of: 10mM BES•NaOH, pH 6.8 15% w/v PEG 4000
3 0.3 >< 0.3 )< 0.025 mm Large number of hipyramidal single crystals up to 0.1 xO.l X0.l mm3
2.5 91 of:
One large crystal.
pH 6.8 ethanolaniine, 10mM 50% v/v MPD
3, consisting of three 0.4x0.4X0.4 mm intergrown crystals
Exp. B’
See exp. A, but with 20 lipids/PLA
Exp. C
1.0 91 of: 10mM BESNaOH, pH 6.8 0.9% /3-00 = 50 lipids/PLA
Exp. D
2.5 ~l of: 10mM ethanolamineNaOH, p1-I 10.0 0.9% /3-00 50 lipids/PLA
2
1.0 /11 of: 15mM CaCI
2
1.0 91 of: 30mM CaCl 0.9M KI 2
PLA: phospholipase A. BES: N-N-bis[2-hydroxyethyl]-2-aminoethane sulfonic acid. MPD: 2-methyl-2,4-pentane diol. PEG: polyethyleneglycol. The amount of lipid molecules per protein molecule present is measured after purification.
P. Gros et al.
/ Experiments in pnembrane protein crystallization
crystals described above. The same holds for the exact composition of the needle-shaped crystals of cytochrome-c reductase. These crystals, however, have the same colour as the native protein, mdicating that at least the chromophores are present in the crystals.
197
4.2. Phospholipase A
A number of microcrystals, platelets, needles and bipyramids, were obtained in the first crystallization experiments, testing mainly two pH’s, 10mM histidine-buffer pH 6.0 and pH 9.3; four different detergents: /3-octyiglucoside, octyl
aiTh.ii~ii~IIip,mp
I,
-~
~iTI11i1tII,N~U~IIilflu1
_____
Fig. 2. (a) Crystal obtained from experiment B, as described in table 1. (b) Side view of a crystal obtained from experiment B’, i.e. using 60% less lipid molecules per phospholipase A molecule compared to experiment B.
198
P. Gros et a!.
/
Experiments in membrane protein crystallization
oligo-oxyethylene, dodecyl-DAPS and lauryl dimethyl amine oxide; and five different precipitants: ammonium sulphate, 2-methyl-2,4-pentane diol, ethanol, polyethylene glycol-4000 and 6000; with CaCI2 as an additive. Calcium chloride was added because of its necessity for the activity of the protein [15]. From crystallization of watersoluble phospholipases A2 it is known that CaCl2 has a distinct effect on the crystallization behaviour [10]. After the first series of experiments, attention was centred mainly on two crystallization conditions, i.e. conditions roughly similar to experiments A and C in table 1. Finally, large crystals were obtained for three crystallization conditions. Table 1 shows the result of the suecessful experiments, while in fig. 2 photographs of crystals obtained are shown. Experiments C and D clearly showed phase separation in the drop, while phase separation was not visible in experiments A and B. In experiments analogous to A and B, phase separation could be induced by increasing the precipitant concentration. The different detergent states between C, D and A, B can be correlated with the difference in crystal stability during crystal handling. Crystals from experiments C and D were extremely unstable and dissolved during mounting. Crystals from A and B were stable and have been mounted and used for
preliminary X-ray diffraction studies. The problem of instability for crystals grown in vapour diffusion systems [27] may be limited to those cases in which crystals have been grown at or near detergent phase separation conditions. This coincides with the idea of micelle interaction in the latter crystals and firmer crystal contacts through electrostatic and hydrogen bond interactions between the protein molecules in crystals grown without phase separation. Although a crystal from experiment A diffracted up to 2.7 A resolution (fig. 3), it is clear that these crystals are not suitable for X-ray data collection. In experiment B (see table 1), the problem of twinning is overcome, but, unfortunately, these crystals are very thin. Diffraction experiments have been performed at the EMBL outstation in Hamburg (Fed. Rep. of Germany) using synchrotron radiation. These experiments were not yet satisfactory, mainly due to curving of the thin crystals by the capillary wall. A lipid-free protein sample has been used to set up new experiments equivalent to experiment B. In experiment B, as described in table 1, the concentration of /3-octyl glucoside is below its CMC. Due to large amounts of lipids present, micelles can be formed and the protein is still soluble. However, in a lipid-free sample the pro-
•0
Fig. 3. Procession photographs (~= 12°),collected on a rotating anode, of a phospholipase A crystal from experiment A. The crystal diffracts to 2.7 A resolution. X-ray damage was1~M noticeable = 2.5 A3/dalton after 1 week corresponding and 60 h with of X-ray one molecule exposure. per Theasymmetric crystal has unit. spacegroup As the C2221: aof lipid number 72.0 and A. 6=107 detergent A,molecules c = 75.8 A perand protein molecule have not been determined, the V~ refers to volume per protein mass only.
P. Gros eta!.
/
Experiments in membrane protein crystallization
tein tends to denature if the detergent concentration drops below its CMC [15]. The first crystallization trials with lipid free material using condilion B, with double and triple amounts of detergent, resulted in intergrown microcrystals, with dimensions of 50 X 50 X 10 p.m3. This result is very similar to the early, non-optimized, result around condition B. From these preliminary results it may be concluded that this membrane protein can be crystallized both from lipid-containing and lipid-free samples.
5. Concluding remarks The main results of the crystallization experiments are the crystals of phospholipase A from E. co/i. The single crystals obtained have two dimensions up to about 0.3 mm. The third dimension, however, is still a problem. A second result of the phospholipase experiments is the growth of crystals from both lipid-containing and lipid-free protein. Furthermore we have shown that membrane protein crystals from hanging drops can be mounted and used for X-ray diffraction studies. However, if phase separation had occurred in the hanging drop, the crystals obtained dissolved during mounting. Other promising results are the needle-like crystals of cytochrome-c reductase, that have the colour of the native protein. However, those crystals still need further improvement. One way of performing large numbers of crystallization trials around these cytochrome-c reductase crystallization conditions, as described above, is to apply automation. A simple, but efficient, way to automate crystallization trials has recently been described by Kelders et al. [29].
Acknowledgements It is a pleasure to acknowledge the stimulating cooperation with Drs. H. Verhey, P. de Geus, A. Horrevoets (PLA), J. Berden, S. de Vnes (Complex-Ill), B.F. van Gelder, T. Muyser, T. Hakvoort and H. Dekkers (Complex-W). We furthermore like to thank Drs. R. Wierenga, R. Read, M.
199
Swarte and H. Kelders for discussions and Mrs. R. Hogenkamp for preparing the manuscript. This research was supported by The Netherlands Foundation for Chemical Research (SON) with financial aid from The Netherlands Organization for the Advancement of Pure Research (ZWO).
References [1] J. Deisenhofer, 0. Epp, K. Miki, R. Huber and H. Michel, J. Mol. Biol. 180 (1984) 385. 121 P. Mitchell, Nature 191 (1961) 144. [31Y. Hafeti, AG. Haavik and P. Jurtshuk, Biophys. Acta 52 (1961) 106. [4] Y. Hafeti, Methods Enzymol. 53 (1978) 35. [5] B.D. Nelson and P. Gellerfors, Methods Enzymol. 53 (1978) 80. [6] C.R. Hartzell, H. Beinert, B.F. van Gelder and T.E. King, Methods Enzymol. 53 (1978) 54. [7] R.A. Capaldi, F. Malatesta and V.-M. Darley-Usmar, Biochem. Biophys. Acta 726 (1983) 135. [8] P. de Geus, I. van Die, H. Bergmans, J. Tommassen and G. de Haas, Mol. Gen. Genetics 190 (1983) 150. [91B.W. Dijkstra, K.H. Kalk, W.G.J. Hol and J. Drenth, J. Mol. Biol. 147 (1981) 93. [10] B.W. Dijkstra, G.J.H. van Nes, K.H. Kalk, NP. Brandenburg, W.G.J. Hol and J. Drenth, Acta Cryst. B38 (1982) 793. [11] B.W. Dijkstra, R. Renetseder, K.H. Kalk, W.G.J. HoI and J. Drenth, J. Mol. Biol. 168 (1983) 163. [12] B.W. Dijkstra, K.H. Kalk, J. Drenth, G.H. de Haas, MR. Egmond and A.J. Slotboom. Biochemistry 23 (1984) 2759. [13] C. Keith, D.S. Feldman, S. Degemello. J. Glick, KB. Ward, GD. Jones and PB. Sigler, J. Biol. Chem. 256 (1981) 8602. [14] S. Brume, J. Bolin. D. Gewirth and PB. Sigler. J. Biol. Chem. 260 (1985) 9742. [15] P. de Geus, Thesis, University of Utrecht (1986). [16] H. Michel, J. Mol. Biol. 158 (1982) 567. [17] J.P. Allen and 0. Feher, Proc. NatI. Acad. Sci. USA 81 (1984) 4795. [18] C.H. Chang, M. Schiffer, D. Tiede, U. Smith and J. Norris, J. Mol. Biol. 186 (1985) 201. [19] R.M. Garavito, J. Jenkins, J.N. Jansonius. R. Karlsson and J.P. Rosenbusch, J. Mol. Biol. 164 (1983) 313. [20] R.M. Garavito, U. Hinz and J.M. Neuhaus, J. Biol. Chem. 259 (1984) 4254. [21] W. Welte, T. Wacker, M. Leiss, W. Kreutz, J. Shiozawa, N. Gadón and G. Drews, FEBS Letters 182 (1985) 260. [22] H. Michel and D. Oesterhelt, Proc. NatI. Acad. Sci. USA 77 (1980) 1283. [23] W. Kuehlbrandt, Nature 307 (1984) 478.
200
P. Gros et aL
/ Experiments
[24] T. Wacker, N. GadOn, A. Becker, W. Mantele, W. Kreutz, 0. Drews and W. Welte, FEBS Letters 197 (1986) 267. [25] R.C. Ford, D. Picot and R.M. Garavito, EMBO J. 6 (1987) 1581. [26] H. Michel, Trends Biochem. Sci. 8 (1983) 56. [27] R.M. Garavito and J.P. Rosenbusch, Methods Enzymol. 125 (1986) 309.
in membrane protein crystallization
[28] R.M. Garavito, Z. Markovic-Housley and J.A. Jenkins, J. Crystal Growth 76 (1986) 701. [29] HA. Kelders, K.H. Kalk, P. Gros and W.G.J. Hol, Protein Eng. 1(1987) 301.