Implications for Membrane Protein Crystallization - American

Mar 29, 2007 - for Membrane Protein Crystallization. Hugh O'Neill,*,† William T. Heller,†,‡ Katherine E. Helton,† Volker S. Urban,*,†,‡ an...
1 downloads 0 Views 182KB Size
J. Phys. Chem. B 2007, 111, 4211-4219

4211

Small-Angle X-ray Scattering Study of Photosystem I-Detergent Complexes: Implications for Membrane Protein Crystallization Hugh O’Neill,*,† William T. Heller,†,‡ Katherine E. Helton,† Volker S. Urban,*,†,‡ and Elias Greenbaum† Chemical Sciences DiVision and Center for Structural Molecular Biology, Oak Ridge National Laboratory, Oak Ridge, Tennessee 37831 ReceiVed: NoVember 10, 2006; In Final Form: February 7, 2007

Small-angle X-ray scattering (SAXS) was used to investigate the structure of isolated photosystem I (PSI) complexes stabilized in detergent solution. Two different types of PSI preparation were investigated. In the first preparation, thylakoid membranes were solubilized with Triton X100 and purified by density gradient centrifugation. SAXS data indicated large scattering objects or microphases that can be described as sheets with ∼68 Å thickness and a virtually infinite lateral extension. The observed thickness agreed well with the dimension of a PSI molecule across the thylakoid membrane. In the second preparation, PSI was isolated as before but was further purified by anion exchange chromatography resulting in functional complexes consisting of single PSI units with attached surfactant as evidenced by the particle volume and gyration radius extracted from the SAXS data. Several approaches were used to model the solution conformation of the complex. Three different ellipsoidal modeling approaches, a uniform density ellipsoid of revolution, a triaxial solid ellipsoid, and a core-shell model, found extended structures with dimensions that were not consistent with the PSI crystal structure (Ben-Shem, A.; et al. Nature 2003, 426, 630-635). Additionally, the SAXS data could not be modeled using the crystal structure embedded in a disk of detergent. The final approach considered the possibility that protein was partially unfolded by the detergent. The data were modeled using a “beadson-a-string” approach that describes detergent micelles associated with the unfolded polypeptide chains. This model reproduced the position and relative amplitude of a peak present in the SAXS data at 0.16 Å-1 but was not consistent with the data at larger length scales. We conclude that the polypeptide subunits at the periphery of the PSI complex were partially unfolded and associated with detergent micelles while the catalytically active core of the PSI complex remained structurally intact. This interpretation of the solution structure of isolated PSI complexes has broader implications for the investigation of the interactions of detergents and protein, especially for crystallization studies.

Introduction Transmembrane proteins play a vital role as cellular gatekeepers by performing a variety of diverse tasks such as energy conversion, cell signaling, and metabolite transport. Although it is predicted that they comprise 20-30% of the genome, less than 1% of the structures in the protein data bank are membrane proteins.1,2 High-resolution structural studies of membrane proteins have been hindered by their recalcitrance to produce crystals suitable for diffraction studies. The main difficulty associated with their crystallization is the tendency of membrane proteins to aggregate during crystallization perhaps due to the fact that they bear hydrophobic and amphiphilic structural features.3 One important factor governing crystallization is the choice of detergent used during solublization of the cell membranes to release the protein of interest. These amphiphiles can also interfere with the structural properties of the proteins and together with the loss of surrounding lipid can promote denaturation and aggregation. This investigation explores the interaction of the detergent Triton X-100 with spinach photo* To whom correspondence should be addressed. Phone: 865-574-5004 (H.O’N.); 865-576-2578 (V.S.U.). Fax: 865-574-1275 (H.O’N.); 865-5748363 (V.S.U.). E-mail: [email protected] (H.O’N.); [email protected] (V.S.U.). † Chemical Sciences Division. ‡ Center for Structural Molecular Biology.

system I complexes in solution as a potential model for membrane protein-detergent interactions. The oxygenic photosynthetic apparatus of higher plants, algae, and cyanobacteria is located in the thylakoid membrane of the chloroplast organelle and is comprised of the multisubunit protein complexes photosystems I and II (PSI and PSII), lightharvesting complex I and II (LHC I and LHCII), and the cytochrome b6f complex (cyt b6f). These membrane-associated protein complexes are responsible for the conversion of solar energy into chemical energy. PSI carries out the second lightdriven reaction of the so-called Z scheme of photosynthesis in which it accepts reducing equivalents from the oxidation of water by PSII via cyt b6f and the soluble electron relay protein plastocyanin. Photon absorption induces a charge separation in the reaction center of PSI resulting in the translocation of an electron across the thylakoid membrane. This vectorial electron transfer generates a 1 V potential4 between the special chlorophyll pair (P700) at the lumenal end of the molecule and the Fe-S cluster at the stromal end of PSI, a process that completes within 150 ns.5 The net result of the photosynthetic process is the conversion of light energy into chemical energy in the form of NADPH. Structural biology continues to play a prominent role in the elucidation of the mechanism of action of PSI. A low-resolution,

10.1021/jp067463x CCC: $37.00 © 2007 American Chemical Society Published on Web 03/29/2007

4212 J. Phys. Chem. B, Vol. 111, No. 16, 2007 three-dimensional (3D) structure obtained by electron crystallography from two-dimensional crystals formed at the grana margins of thylakoid membranes shows the dimensions of PSI to be 150-160 Å × 110-125 Å in the plane of the membrane and 80-90 Å along the direction of vectorial electron transport.6,7 The structure has a ridge on the stromal end and a distinct central depression on the lumenal side that is postulated to be the binding site for plastocyanin. More recently, a 4.4 Å resolution X-ray crystallography structure of PSI from Pisum satiVum (var. Alaska) shows that the structure consists of 12 core subunits and 4 light-harvesting complexes (LHCI) assembled in a half-moon shape on one side of the core that give rise to 45 transmembrane helices, 167 chlorophylls, 3 ironsulfur clusters, and 2 phylloquinones.8 The molecular architecture of PSI from the cyanobacteria Synechococcus elongatus that was also resolved at 2.5 Å by X-ray crystallography reveals that the protein monomer is composed of 12 subunits and 127 cofactors (96 chlorophyll a, 22 caratenoids, 2 phylloquinones, 3 Fe4-S4 clusters, and 4 lipids).9 Both PSI isoforms share a high degree of structural similarity, but the cyanobacterial protein is smaller than its plant counterpart and does not have any peripheral antenna in the form of light-harvesting complex directly associated with the reaction center. In addition, it is postulated to exist as a trimeric oligomer in the cyanobacterial thylakoid membrane. Small-angle scattering (SAS) techniques, using X-ray or neutrons, provide another approach to the investigation of the structural properties of proteins. These methods provide information about the conformation and aggregation state of proteins in solution. Several of the molecular components of the photosynthetic apparatus have been investigated by SAS techniques. Hydrodynamic and small-angle X-ray scattering (SAXS) measurements show that the manganese-stabilizing protein subunit of plant PSII has an extended, prolate ellipsoid shape in solution.10,11 The solution conformation of the integral membrane protein light-harvesting complex LH2 from Rhodobacter spheroides 2.4.1 has also been determined by SAXS.12 In this case, the molecular shape of the LH2 complex deviates from the ringlike crystal structure, having the shape of an oblate plate with an eccentricity of 0.59 in solution. In this study, we report the first investigation by SAXS of the solution properties of active PSI complexes isolated from Spinacia oleracea by two different purification strategies. Activity studies indicate that the complex, which was solubilized with Triton X-100 detergent, remains functional under the conditions used for the SAXS measurements. The SAXS results demonstrate that the different preparations of the PSI/detergent system adopt two different states in solution, one of which suggests that the PSI complexes are surrounded by a cloud of attached detergent micelles. The other state is more consistent with a sheetlike structure typical of a bilayer membrane. These results have important implications for the study of other detergent-solubilized membrane proteins and complexes and for membrane protein crystallization. Materials and Methods PSI Isolation. Commercially obtained market spinach was the starting material. The thylakoid preparation (1 mg of chlorophyll/mL) was solubilized using Triton X-100 at a final concentration of 0.8% (w/v) at 20 °C for 30 min. All other steps were performed as described previously.13 The PSI purified by density gradient centrifugation (designated prep A) was further fractionated using anion exchange chromatography (designated prep B). The first steps were dialysis versus 10 mM Tris-HCl

O’Neill et al. pH 7.6 followed by ultracentrifugation (Beckman VTi50 rotor, 40 000 rpm) for 1 h. The pelleted PSI complexes were resuspended in a minimum volume of 20 mM MES, pH 6.4, 0.02% (w/v) Triton X-100. Resupension was aided by using a Dounce homogenizer to disperse large aggregates. The Triton X-100 concentration was increased to 0.1% (w/v), and the sample was allowed to equilibrate at 4 °C for 20 h with gentle shaking. Anion exchange chromatography was carried out on a Hi-Trap Q Sepharose (Amersham Biosciences) column attached to a Pharmacia FPLC system. The equilibration buffer A was 20 mM MES pH 6.4 and 0.02% (w/v) Triton X-100, and the limit buffer B was 20 mM MES pH 6.4, 400 mM magnesium sulfate, and 0.02% (w/v) Triton X-100. Prior to sample loading, the column was equilibrated with 6 column volumes of buffer A. After loading, the column was washed with a further 3 volumes of buffer A. PSI was eluted using a gradient of 10100% buffer B. A substantial chlorophyll containing fraction eluted during the washing step. PSI eluted as two separate peaks, one at 25% buffer B and the other at 35% buffer B. These differed slightly in their chlorophyll content. Deriphat PAGE Electrophoresis. A modification of previously described procedures was used.14,15 Polyacrylamide resolving gels, comprised of 8.6% acrylamide (38.7% acrylamide and 1.3% bis(acrylamide)) containing 0.32 M Tris-HCl, pH 8.8, 9.0% glycerol, and 0.03% Deriphat, were polymerized with 0.05% ammonium persulfate and 0.05% TEMED. The stacking gels were comprised of 5.3% acrylamide (from 40% (w/v) acrylamide stock, 29:1 acrylamide and bis(acryalmide)) containing 0.125 M Tris-HCl, pH 6.8, polymerized with 0.08% ammonium persulfate and 0.2% TEMED. The upper electrophoresis buffer was 25 mM Tris-HCl, pH 8.3, 192 mM glycine, and 0.2% Deriphat. The lower reservoir buffer was the same except that Deriphat was omitted. Electrophoresis was performed at a constant 200 V at 20 °C. Analytical Procedures. Protein concentration determination was carried out using a modified Lowry assay protocol (Pierce) with bovine serum albumin as the standard protein.16 The concentration of chlorophyll was determined in 80% acetone as described by Porra et al.17 The molar concentration of PSI was determined by measurement of P700 using a chemical assay procedure.18 The photochemical activity of PSI was determined from the light-induced P700 absorption changes at 810 nm using the dual wavelength emitter detector unit ED-P700DW-E connected to a PAM 101 fluorometer (Walz, Effeltrich, Germany). This technique was used to identify the anion exchange chromatography fractions that had active PSI complexes. The samples were dark adapted before illumination with white light (1400 (µmol/m2)/s, Schott KL 1500 light source) to excite P700 over a 1 min period. The actinic light was then extinguished, and the rereduction of P700+ was followed for an additional 1 min. Small-Angle X-ray Scattering. The SAXS experiments were performed using the BESSRC-CAT19 and BIO-CAT20 beamlines at the Advanced Photon Source at Argonne National Laboratory. PSI from prep A having protein concentrations of 0.46, 0.38, 0.30, 0.23, 0.15, and 0.08 mg/mL in 20 mM Tris-HCl pH 7.5 and 0.5% (w/v) Triton X-100 were used to collect scattering data on the BESSRC-CAT beam line. PSI from prep B having protein concentrations of 1.16, 0.93, 0.70, 0.46, and 0.23 mg/ mL in 20 mM Tris-HCl pH 7.5 and 0.1% (w/v) Triton X-100 were used to collect X-ray scattering data on the BIO-CAT beam line. The appropriate buffer solution without protein was used to record the baseline scattering curves. A glass capillary tube (i.d. ∼ 1 mm) served as the active window in the beam line.

SAXS of Photosystem I-Detergent Complexes

J. Phys. Chem. B, Vol. 111, No. 16, 2007 4213

The sample (50-100 µL) was passed though the capillary at approximately 5 µL/min. This experimental setup minimized the effect of radiation damage to the protein sample. The scattered X-ray intensities were recorded on a CCD area detector with a sample to detector distance of 2 m (BESSRCCAT) and 965 mm (BIO-CAT). The data accumulation time was 1 s, and 5 separate exposures were collected for each analytical sample. All individual exposures were visually inspected and generally excellent reproducibility was found. Occasionally, individual exposures deviated significantly from the rest of a set, which is likely due to rare passing of bubbles through the capillary. Reproducible data were averaged while the outliers were discarded. Data reduction followed standard procedures to correct for instrumental background, detector efficiency, and scatter from the buffer solution using software provided at the respective beamline.19,20 Two approaches to buffer background subtraction were tried: (a) including and (b) excluding subtraction of scatter from Triton X-100 micelles. Both methods produced very similar data due to the very low signal of Triton X-100 micelles relative to the scatter from the PSI containing samples. Method (a) was selected for further data analysis because a comparison of micelle volume concentration and available PSI surface suggests that the majority of Triton X-100 should remain present as Triton X-100 micelles in all studied PSI preparations. Data were azimuthally averaged to produce the 1D scattering function as a function of modulus of the scattering vector Q ) (4π sin θ)/λ, where 2θ is the scattering angle from the incident beam and λ is the wavelength of the X-rays. The wavelength of the monochromatic X-ray beam used in each experimental setup was 1.0 Å. The final data were scaled into absolute units of 1/cm taking into account the measured sample transmission21 and sample thickness and using purified water as calibration standard for the measured scattering intensities. SAXS Data Analysis. The one-dimensional scattering intensity, I(Q), of an isotropic solution of particles can be described by

P(R) )

I(Q) ) Np(∆F)2Vp2|F(Q)|2 S(Q)

(1)

where the number density of scattering particles is Np, the difference in scattering length density between the particles and solvent is ∆F, and Vp is the volume of the particle. F(Q) and S(Q) correspond to the particle form factor, which is the scattering from the particle and the interparticle interference function, respectively. The form factor for an isotropic solution represents the time and ensemble averaged integration over the particle volume that is averaged over all rotations with respect to the incident beam. S(Q) results from interparticle correlations that exist between particles in solution, such as would result from an electrostatic repulsion. For a dilute solution of noninteracting particles, S(Q) is unity. The average radius of gyration (Rg) is related to the size of a particle in solution22 and can be calculated from the so-called “Guinier regime” of small Q values (QRg e 1-1.3) using the approximation in eq 2:23

I(Q) ) I(0) exp(-Q2Rg2/3)

(2)

The value of Rg is determined from the slope of a linear fit of ln(I(Q)) versus Q2. I(Q) can also be related to P(R), the intraparticle distance distribution function by the transform defined in eq 3:

1 2π2

∫0∞ I(Q)QR sin(QR)dQ

(3)

This function provides information about the shape of the scattering particle and is very sensitive to asymmetry and domain structure within scattering particles. In this study, the program GNOM24 was used to determine P(R) from the experimental SAXS data. The results of the fitting provide the maximum linear dimension of the particle, Dmax. Model Fitting. The scattering data of PSI from prep A was modeled by the form factor of a large sheet of thickness L.25 This is a cylindrical disk form factor and can be simplified considerably in the case L , R, where R is the radius of the disk. For practically infinite sheets or disks, where QR . 1 for all observed Q, the equation further simplifies to

|F(Q)|2 )

[

]

sin(QL/2) 2 2 QL/2 (QR)

2

(4)

Fitting with eq 4 yields the sheet thickness L. Note that R in this case has no influence on the shape of the curve and merely acts as a normalization factor that cancels out with the large volume prefactor of the large disk (eq 1). From the absolute value of the prefactor, the mass/unit area of the sheet can in principle be calculated. The determination of this parameter was not attempted in this study, because of the uncertainty in concentration and contrast due to the mixed composition of scattering objects composed of protein, surfactant, and lipid. The complete large disk form factor as given by Pedersen26 was also used for the data fitting and was found to be equivalent to using eq 4 for values of R > 1000 Å. Different approaches were employed to obtain structural models of the protein-detergent complex from prep B. The first approach employed the use of analytical form factors, F(Q), for a solid sphere and an ellipsoid of revolution.26 When these models are fitted, it is assumed that the structure factor S(Q) is unity. The second approach applied to the prep B samples was to fit both a solid triaxial ellipsoid and a core-shell triaxial ellipsoid to the data using the software ELLSTAT,27 which calculates averages and standard deviations for the structural parameters found to fit the data. The density of the core was set to the average electron density of protein (∼0.442 electrons/ Å3), while the shell was given the average value for Triton X-100 (0.205 electrons/Å3).28 Both applications of ELLSTAT used the same range of ellipsoidal semiaxes. A third modeling approach applied to prep B PSI complexes used the medium-resolution crystal structure of PSI. This structure was placed in the center of a circular disk having a thickness appropriate for a lipid membrane (60 Å). Radii were tested every 0.5 Å from 60 to 200 Å. To appropriately fill in the undetermined portions of the medium-resolution crystal structure with the correct density, the scattering length density assigned to the disk was a function of the distance from the axis of the disk. All volume lying less than 60 Å from the disk axis was assigned the electron density of the protein while the volume lying further than 60 Å from the disk axis was assigned the average electron density of the Triton X-100. Any volume occupied by the atoms provided in the medium-resolution crystal structure were assigned the average electron density of the protein. The intensities were calculated from the structure using the Monte Carlo method implemented in ELLSTAT.27 P(R) determined from the Monte Carlo sampling is transformed to I(Q) for comparison with the input data using

4214 J. Phys. Chem. B, Vol. 111, No. 16, 2007

I(Q) ) 4π

∫0∞ P(R)

sin(QR) dR QR

O’Neill et al.

(5)

The quality of the fit was evaluated using the reduced χ2 parameter.29 For these models, S(Q) is assumed to be unity, meaning that there are assumed to be no interparticle interactions. The final modeling approach applied to the data considered the possibility that the detergent denatured the protein. Previous small-angle scattering of protein-detergent complexes have been analyzed by a “beads-on-a-string” model. This model assumes that the polypeptide chain is in an unfolded, random state in solution with detergent micelles distributed along the polypeptide chain. In this case, the intensity is modeled as described in eq 1, up to a scaling factor. The F(Q) of the micelles were modeled as core-shell ellipsoids of revolution using an analytical form for F(Q)26 with the electron densities of the core and shell being 0.175 and 0.379 electrons/Å3, respectively, as determined by previous workers.28 In the “beads-on-a-string” model, S(Q) accounts for interparticle correlations that exist between the micelles. Previous workers have derived an analytical function for S(Q),30,31 which is given in eq 6:

S(Q) ) 1 +

DΓ(D - 1) 1 × D (Qr0) [1 + 1/(Q2ξ2)](D-1)/2 sin[(D - 1) tan-1(Qξ)] (6)

D is the fractal dimension of the packing of the micelles. If D ) 3, then the particles are in a compact arrangement. For values of D less than 3, the micelles are in a more open structure. ξ is the correlation length of the fractal system. The term r0 is the equivalent micellar radius. In the model, the interaction between the micelles is assumed to have a finite range and therefore provides a measure of the extent of unfolding of the polypeptide chain. Γ(x) is the gamma function. The free parameters for the model fitting were the ellipsoidal semiaxes, and the D and ξ parameters, from S(Q). Again, the quality of the fit was evaluated using the reduced χ2 parameter.29 Results Properties of Photosystem I. Photosystem I complexes were isolated from spinach leaves by partial solubilization of the thylakoid membranes with Triton X-100.32 They were then separated from other solubilized membrane components by ultracentrifugation over a 0.1-1.0 M sucrose gradient and formed a layer at the interface of the gradient and the 2.0 M sucrose cushion. This preparation was designated prep A. PSI isolated by this method was purified further by anion exchange chromatography using a linear magnesium sulfate gradient (0400 mM) that contained 0.02% Triton X-100 to fractionate the protein mixture. This preparation was designated prep B. Chlorophyll-containing peaks that were assigned to PSI, by measurement of light-induced spectral changes in the protein, eluted at 100 mM MgSO4 and 140 mM MgSO4. Comparison of the preparations by native polyacrylamide gel electrophoresis showed that there was an additional chlorophyll binding band in the PSI preparation purified by sucrose density gradient centrifugation that was not present after anion exchange chromatography (Figure 1A). This band was previously assigned to LHC II,15 which is the most abundant light-harvesting complex in the thylakoid membrane and is known to associate with PSI complexes.33 Hence, the gel analysis shows that LHC II was removed during anion exchange chromatography and that the size of the PSI complex was unchanged after this procedure.

The physicochemical characteristics of the two types of PSI preparation are summarized in Table 1. The chlorophyll content of the PSI preparation after anion exchange chromatography is approximately 5-fold lower than complexes prepared by density gradient centrifugation alone, and the chlorophyll a/b ratio increases from approximately 5 to 7. The first PSI peak that elutes from the anion exchange column has a higher content of chlorophyll a than the second peak. Conversely, the second peak has a lower amount of total chlorophyll associated with the core complex. The PSI obtained from the first peak was used for X-ray scattering experiments. The visible absorption spectra of the PSI preparations are very similar, as shown in Figure 1B, and almost identical with those published in other studies.32,34 The 675 nm peak is slightly red-shifted (by 2 nm) in the density gradient centrifugation preparation. Most notable is that the shoulder at 461 nm is largely absent after subjecting the PSI preparation to anion exchange chromatography. This shoulder is associated with the chlorophyll b absorption in LHC II, which binds both chlorophyll a and chlorophyll b in a ratio of 1.2-1.4.35,36 The small differences in this region of the spectra of the two fractions of PSI that eluted from the ion-exchange column support the chlorophyll a/b ratio data reported in Table 1. The P700 chemical assay demonstrated that the intramolecular electrontransfer chain from the chlorophyll pair (P700) to the terminal iron-sulfur cluster was intact. This was further supported by measurement of the photochemical activity of the PSI preparations by near infrared absorption spectroscopy. The difference in absorbance at 810 nm, due to the P700+ cation, and at 860 nm, an isosbestic region of the spectrum, was recorded during actinic illumination of the sample (Figure S1).37 PAGE analysis, spectrophotometric data, and physicochemical data all support the conclusion that the major difference between the two types of PSI preparation is the removal of chlorophyll b in the form of LHC II. SAXS Results. SAXS data collected on the sample prepared by density gradient centrifugation (prep A) are shown in Figure 2 on a log-log scale. The very low Q data approximate a straight line up to a Q-value of about 0.01 Å-1 following a power law with an exponent of -2.04. Beyond this point, the slope of the curve increases and shows a second linear region up to Q ∼ 0.04 Å-1, having a power-law exponent of -2.6. At larger Q, the scattering data decline sharply before transitioning into a broad peak centered near 0.13 Å-1. A fit to the data using eq 4 is consistent with large (in-plane dimension >1000 Å), thin sheetlike structures at very low (Q < 0.01 Å-1) and high (Q > 0.04 Å-1) values of the scattering vector. The thickness of the sheets determined from the fitting is 68 Å. The central region of the data, approximately 0.012 < Q < 0.045 Å-1, deviates from the sheet model fit and is dominated by powerlaw scattering with a power-law exponent of -2.6. This intermediate Q range corresponds to the characteristic external dimensions of individual PSI complexes. SAXS analysis of the anion exchange preparation was carried out with PSI that was obtained from peak 1 (Table 1). A concentration series of PSI in 0.1% (w/v) Triton X-100 is shown in Figure 3 with an intensity profile collected for 0.1% (w/v) Triton X-100 in the absence of protein. The measured intensity scales well with protein concentration. The inset plot of Figure 3 shows the PSI data normalized for concentration and corrected for the 0.1% (w/v) Triton X-100 background. The overlaid curves agree well at larger Q values. However, some deviation in the low-Q region is observed at the highest and lowest concentrations. This is most probably an artifact of the sample

SAXS of Photosystem I-Detergent Complexes

J. Phys. Chem. B, Vol. 111, No. 16, 2007 4215

Figure 1. Physical properties of PSI preparations. (a) Native Deriphat PAGE electrophoresis of PSI preparations: lane 1, PSI complexes after density gradient centrifugation; lanes 2 and 3, PSI in fractions peak I and peak 2 after anion exchange chromatography, respectively. (b) Visible absorption spectra of PSI preparations: prep A, PSI after density gradient centrifugation (‚‚‚); prep B, PSI from peak 1 (s) and peak 2 (- - -) after anion exchange chromatography.

TABLE 1: Physicochemical Properties of Purified PSI Complexes PSI sample

Chl (µM)

P700 (µM)

Chl/P700

Chla/Chlb

prep A prep B, peak 1 prep B, peak 2

281.9 307.0 146.4

0.46 1.78 1.34

619.6 172.5 109.3

4.97 7.71 7.03

preparation and suggests the possibility that the samples contain a small population of higher order aggregates. Guinier plots23 (ln(I(Q)) versus Q2) are shown in Figure 4 for two different data sets, yielding an Rg of 63.2 ( 0.1 and 63.5 ( 0.1 Å. The intraparticle distance distribution function, P(R), is a measure of the frequency of the interatomic vector lengths within a scattering particle. The P(R) determined using the program GNOM24 is shown in Figure 5. The maximum particle dimension, Dmax, determined for the PSI-detergent system is 250 Å, while the Rg determined from the fitting, being the second moment of P(R), is 64.6 Å. The curve for the complex has a main peak near 48 Å that trails off with a very long tail to a Dmax of 250 Å. The extended tail is consistent with an extended structure, such as long rodlike or thin disklike structures. The uncertainty on Dmax and Rg determined from P(R) was evaluated

Figure 2. Experimental SAXS data of PSI purified by density gradient centrifugation (prep A). The experimental curve (O) was fitted to a large disk function (s). A line indicating Q-2 power-law scattering is also shown (- - -).

Figure 3. SAXS curves of a concentration series of PSI complexes purified by anion exchange chromatography (prep B). The 0.1% (w/v) Triton X-100 has the lowest scattering intensity followed by a concentration series of PSI from 0.23 to 1.16 mg of protein/mL. The intensity of scattering was standardized to a buffer baseline. In the inset graph, the data were corrected for the scattering of detergent micelles and protein concentration.

by repeating the GNOM program several times using a range of maximum dimension from 150 to 300 Å for evaluating P(R). Using a Dmax < 230 Å produced a significantly worse fit to the data, while increasing Dmax above 250 Å did not change the fit quality or the shape of P(R) and especially did not lead to significant frequency measure of larger distances. An estimate for the uncertainty on Dmax may therefore be placed at Dmax ) 250 ( 20 Å. Correspondingly, the variation in Rg using 230 Å < Dmax < 250 Å leads to Rg ) 64.6 ( 2 Å. A shoulder can be seen at very short vector lengths (∼15 Å), which may be due to the detergent micelles. This feature persists when the P(R) calculation is repeated while changing the minimum and maximum Q used by the GNOM program. The shoulder only disappears when the maximum Q is reduced to about 0.1 Å-1, that is, when the peak in I(Q) at 0.16 Å-1 is not included in the analysis, indicating that the peak in I(Q) is related to the shoulder in P(R).

4216 J. Phys. Chem. B, Vol. 111, No. 16, 2007

Figure 4. Guinier analysis of prep B PSI scattering curves. PSI concentration was 0.7 mg/mL (+) and 0.46 mg/mL (O) yielding an Rg of 63.6 and 63.2 Å, respectively.

Figure 5. Distance distribution function (P(R)) of prep B PSI complexes in solution. The concentration of PSI was 1.16 mg/mL. The fit to the data yielded an Rg of 65 Å and Dmax of 250 Å.

The Guinier and P(R) analyses suggested that the samples consisted of large monodisperse scattering particles. Several different approaches were employed to model the solution structure of prep B PSI complexes as a single large shape. The simplest rigid body shape, a sphere, was initially tried (Figure 6). The best fit of this analytical form factor to the data produced a sphere with a radius of 52.2 Å that does not reproduce the shape of the scattering curve. The comparison of the sphere form factor with the data indicates that a less isometric shape is required for an adequate description of the scattering signal in accordance with the observation of the tail in P(R) as discussed above. Using the analytical form factor for a prolate ellipsoid of revolution26 with an adjustable baseline, a large extended ellipsoid having semiaxes of 39.33, 39.33, and 120.66 Å was found. The curve fitted the experimental data up to 0.09 Å-1, as can be seen in Figure 6. The length of the ellipsoid is in agreement with P(R). The peak shown in the data is not well reproduced by the fit. The peak width and amplitude relative to the I(0) of the data are too large for the model intensity profile of the ellipsoid. Assuming an average protein density of 1.35 g/cm3,38 a molecular mass of 633 kDa is deduced from the volume of the ellipsoid. This value agrees well with a mass of 525 kDa published for the crystal structure.8 The observed differences may be attributed to the Triton X-100 associated

O’Neill et al.

Figure 6. Fitting of prep B PSI scattering data. The concentration of PSI was 1.16 mg/mL. Curve fitting with a solid ellipsoid of rotation (- -) yielded a perpendicular semiaxis of 39.3 ( 0.002 Å and parallel semiaxis of 120.4 ( 0.033 Å with a radius of gyration was 59.32 ( 0.030 Å. Triaxial ellipsoid shapes were fitted with ELLSTAT.27 The semiaxes of solid triaxial ellipsoid were found to be 121.1 ( 4.2, 46.6 ( 2.6, and 32.4 ( 1.9 Å (‚‚‚), and the core-shell ellipsoid model generated yielded semiaxes of 119.2, 39.7, and 25.8 Å with a detergent shell thickness of 17.0 Å (- ‚ ‚). The fit of experimental data with a model built using the crystal structure of PSI8 embedded in a detergent disk (- - -) resulted in a best fit disk radius of 70 Å. Curve fitting with a solid sphere form factor (radius 52.2 Å) is also shown (s).

with the protein in solution that contributes to the solution dimensions of the scattering particles. Triaxial ellipsoidal fits using a solid model and a core-shell model were generated using the program ELLSTAT,27 the results of which are also shown in Figure 6. In the case of the solid ellipsoid, the ellipsoidal semiaxes were found to be 121.1 ( 4.2, 46.6 ( 2.6, and 32.4 ( 1.9 Å. The core-shell model fitting found the ellipsoidal semiaxes to be 119.2 ( 2.4, 39.7 ( 0.7, and 25.8 ( 0.7 Å with a 17.0 ( 0.5 Å shell of detergent. In both cases, the standard deviations in these average structural parameters calculated by ELLSTAT and analysis of the values used to calculate the averages suggest a well-defined solution space rather than a multimodal solution space composed of multiple distinct structures. While the intensity profiles of the ellipsoidal models all reproduce the low-angle data very well, the characteristic peak near 0.16 Å-1 is reproduced far better by the core-shell model than by either the solid ellipsoid of revolution or the solid triaxial ellipsoid, which also fit the data better than the ellipsoid of revolution. The difference in scattering length density between the core and the shell enhance the peak in the model data, resulting in a better fit to the data than is possible for either of the solid ellipsoids. However, the position of the peak and the associated local minimum are not exactly reproduced by the core-shell model. The attempt to fit the experimental data with a model in which the medium-resolution crystal structure of PSI8 is embedded in a detergent disk produced interesting results. The fit profile is also shown in Figure 6. The fit of the model intensity profile to the data is significantly worse than the two ellipsoidal models but is slightly better at wide angles than the spherical model. The first peak in the model intensity is located at the same Q as the minimum in the measured profile, and there is a second peak in the model profile at roughly the same Q-value as the peak in the measured profile. The best fit disk radius was found to be 70 Å, which means that the longest dimension of the model

SAXS of Photosystem I-Detergent Complexes

Figure 7. Analysis of prep B PSI SAXS curve using “beads on a string” fractal model. The concentration of PSI was 1.16 mg/mL, and the scattering curve was corrected for the scattering of the detergent micelles. The fits on the fractal model to the data at Qmin values of 0.03 Å-1 (- -), 0.045 Å-1 (- - -), and 0.06 Å-1 (‚‚‚) are shown.

is significantly shorter than that determined from the P(R) fitting or the ellipsoidal model fitting. The fit demonstrates that the medium-resolution crystal structure embedded in a membranelike disk is not the most appropriate model for the system. The physicochemical data indicated that the system was functional and the large Rg determined from the Guinier fitting suggested that the system was properly folded into a large, single shape, but the peak near 0.16 Å-1 was only reasonably reproduced by the triaxial core-shell ellipsoid having dimensions inconsistent with the medium-resolution crystal structure. These analyses suggested a second possibility, namely that the peripheral subunits of PSI were associated with smaller detergent micelles while the catalytic core remained properly folded. The resulting scattering profile for such a system would be a superposition of the scattering of the core and the numerically more abundant, micelle-solubilized protein, which can be described by the “beads-on-a-string” model. The larger core dominates the scattering at low Q, while the micelle scattering dominates at larger Q. For this reason, different values of the minimum Q were used for the “beads-on-a-string” model fitting, specifically 0.030, 0.045, and 0.060 Å-1, all of which are greater than the value used for fitting with a single, larger structure described above (0.015 Å-1). This choice of Q-range limited the tendency of the fitting to gravitate to a large core-shell ellipsoid similar to that found above. Restricting the fit to these higher Q-values implies that the correlations between detergent micelles, described by a “beads-on-a-string” structure factor may be limited to distances up to 100-200 Å. The fits to the experimental data using intensities calculated from ellipsoidal core-shell models of revolution using Triton X-100 densities as previously determined28 multiplied by the S(Q) from eq 6 are shown in Figure 7. The “beads on a string” fits performed in the current work have micelles with an average semiaxis of revolution of 29.6 and 37.4 Å for the independent semiaxis with an average shell thickness of 19.6 Å. Previous structural studies of Triton X-100 micelles found sizes smaller than the crystal structure of PSI.28 The fractal dimension when Qmin is 0.030 Å-1 is 1.27, while the fit using a Qmin of 0.045 Å-1 yielded a value of D of 1.13, and with Qmin of 0.060 Å-1, the value was 1.03. The values of ξ are 149, 184, and 167 Å, respectively, for the three values of Qmin. Previous studies of proteins interacting with surfactants, such as lithium dodecyl sulfate,31 sodium dodecyl sulfate,39-41 and cetyltrimethylam-

J. Phys. Chem. B, Vol. 111, No. 16, 2007 4217 monium chloride,42 found comparable values for D. The values found for ξ were lower, being near 100 Å, which implies a stronger correlation, presumably because the surfactants employed were ionic, unlike Triton X-100. The trend of decreasing D with increasing Qmin and variability in ξ may also be due to the influences of the different length scales sampled as the broader fitting ranges include increasingly large length scales where the scattering would be dominated by the larger, folded protein core of PSI instead of by the micelles. In both fits, the position of the minima and the position and relative height of the peak are consistent with the data. The minima observed in the fit curves are more pronounced than in the data. A possible interpretation is that the sizes of the micelles may vary significantly, which will blur the local minima in the intensity. Alternatively, the scattering from the folded PSI core complex contributes to the scattering in this Q-range, resulting in an effective blurring of the minima due to the superposition of the two contributions to the scattering. Discussion Triton X-100 is widely used for the isolation and stabilization of membrane proteins and is by far the most popular detergent for the purification of plant PSI complexes from thylakoid membranes. Prep A PSI complexes were made using a method described by Mullet et al.32 This purification strategy was also used to determine the subunit stoichiometry of plant PSI complexes and identified most of the low molecular mass subunits in the complex43 giving good confidence that Triton X-100 is very compatible with PSI complexes. In this approach, the PSI complexes are separated from other solubilized membrane components during density gradient centrifugation because the density of the environment surrounding the protein complex increases and the density of the complex changes as less detergent remains bound. Some self-aggregation of PSI also occurs due to an increase in hydrophobic interactions between the complexes.32 The concentration of Triton X-100 and the length of time of the thylakoid membrane solubilization reaction are optimized for the starting material used. Optimization of these parameters is important because of seasonal variations and the origin of the spinach leaves and time elapsed since harvesting is largely unknown. Too much Triton X-100 can result in the release of a large amount of free chlorophyll from the membranes and low yield of PSI.32 The most common contaminant in PSI preparations obtained by density gradient centrifugation is chlorophyll b containing light-harvesting complex II.36 The prep B type PSI complexes were obtained by fractionation of prep A PSI by anion exchange chromatography using a procedure similar to those described for the crystallization of plant and cyanobacterial PSI complexes.44,45 The solution structures of the two types of preparations of detergent-associated PSI preparations were investigated by SAXS analysis and produced unexpected results. The SAXS curves of prep A samples showed that the PSI-detergent complexes were not monodisperse. A power-law fit to the low-Q data yielded an exponent of -2.04, which is in agreement with a sheetlike structure; the data at wider angles deviated from this character, having an exponent of -2.6. The absolute value of the exponent of a power-law fit to scattering data is related to the fractal dimension of the sample over a range of length scales. A recent report on the analysis of mass fractal dimension and the compactness of proteins concluded that the average fractal dimension of proteins is approximately 2.5.46 This suggests that the structure appears as a thin sheet at large length scales with a well-defined thickness that dominates the scattering

4218 J. Phys. Chem. B, Vol. 111, No. 16, 2007 at short length scales. However, at intermediate length scales, the inhomogeneous composition of the sheet leads to a deviation from the simple sheet model and is dominated by the scattering from the fractal dimension of typical internal packing of the subunits in a protein.46 The analysis suggests that the samples contain large, membranelike structures having a thickness of 68 Å. This thickness is in reasonable agreement with a typical membrane thickness (∼60 Å) and the transmembrane thickness of PSI (80-90 Å).8 This density gradient centrifugation PSI preparation (prep A) was subjected to anion exchange chromatography (prep B) for further investigation by SAXS. Although two peaks with PSI activity were resolved during chromatography, the first peak was chosen for the SAXS study because its chlorophyll to P700 ratio was 172 (Table 1) which was very similar to a P700 ratio of 167 reported for the PSI preparation used in the crystallization of the membrane protein complex.8 The most notable difference between the two PSI preparations was that LHCII was removed during anion exchange chromatography. As described above, LHC II was identified as a major contaminant that was removed during anion exchange chromatography by spectrophotometry and polyacrylamide gel electrophoresis. Given the hydrophobic nature of LHC II, it is likely that lipid components accompanied its removal. The role of LHC II in the formation of the large ordered lamellar structures is supported by an earlier study that showed the solution properties of monogalactosyldiacylglycerol, a major component of the thylakoid membrane, could be changed from an inverted hexagonal (II) phase to a lamellar phase by the addition of LHC II to the lipid mixture.47 It is thus likely that the sheetlike structures observed in the density gradient centrifugation preparation are induced by LHC II and its removal led to a disruption of the lamellar sheet and formation of monodisperse PSI-detergent complexes. The PSI in prep B produced data consistent with monodisperse particles, based on the linear Guinier region and P(R) fitting. The use of large, simple shapes to model the solution structure of the complex resulted in model intensity profiles that fit the data reasonably well, but the structures were not consistent with the shape of the known crystal structure.8 Although the solid ellipsoidal model produced a structure with dimensions similar to those generated using the triaxial solid and core-shell ellipsoid models, neither solid structure reproduced the peak at 0.16 Å-1 as well as the core-shell model. In the case of the core-shell model, which assumes that the detergent is uniformly distributed around the surface of the PSI complex, the dimensions are similar to the solid triaxial ellipsoid, but the 17 Å thick detergent shell results in a protein volume that is too small to accommodate a PSI molecule and hence cannot be considered to be a physically realistic interpretation of the data. Models built of the crystal structure embedded in a disk of detergent did not fit the data as well as the ellipsoid models. The difficulty of fitting the data with a large, single structure suggested that the solution state of PSI was more complicated. Therefore, it was necessary to consider the possibility that the structure was not intact. However, it is important to note that the physicochemical data give good confidence that PSI preparations used in this study were catalytically active. The fact that PSI remains functional after either preparation suggests that a significant portion of the complex remains properly folded. This is also supported by other studies that have shown that PSI in 0.05-0.1% Triton X100 is stable and catalytically active.48,49 Indeed, the photocatalytic properties of a PSI core complex with 9 chlorophylls/P700 obtained by harsh treatment

O’Neill et al. with organic solvents also retained its photocatalytic properties50 highlighting the structural stability of the PSI core complex. The “beads-on-a-string” structure factor30,31 often used for detergent-denatured proteins31,39,40,42 describes the interactions of micelles with unfolded polypeptides. Models built using this structure factor incorporating a small core-shell form factor accurately reproduce key features in the data, specifically the peak near 0.16 Å-1. This result supports the conclusion that the PSI is partially unfolded and associated with detergent micelles. In combination with the physiochemical data, the most appropriate model is likely a folded PSI core complex with detergent micelles associated with partially unfolded but attached peripheral subunits. Conclusions This investigation highlights important characteristics of the interaction of Triton X-100 and PSI complexes in solution. It was observed that the PSI molecules assemble into thin sheets one molecule in thickness in the presence of LHCII and that removing LHCII by anion exchange chromatography resulted in a monodisperse solution of scattering particles of an unexpected nature that is inconsistent with a properly folded structure embedded in a bilayer. We propose that the peripheral subunits remain associated with the protein core but are partially unfolded and interact with detergent micelles. It would be of interest to investigate other detergent-stabilized membrane proteins and complexes using SAXS to determine if partially unfolded structures are observed as shown in this study. Another important result of the work is that functional assays may not necessarily be a reliable indicator that the detergent-associated membrane proteins and complexes are in their native conformation. The partially folded state of PSI implied by the data has interesting implications for the crystallization of membrane proteins for high-resolution structure determination. This work suggests that some of the difficulty in crystallizing membrane proteins may arise from partial denaturation by detergent. In addition to driving the transition from a membrane-associated state to a crystalline lattice,3,51,52 it is necessary that the protein correctly and reproducibly folds. Therefore, systems that crystallize may have a stronger tendency to properly refold during condensation into the crystal lattice. The choice of detergent is very important during this process as its interaction with the polypeptide chain(s) influences the probability of crystallization occurring. Small-angle scattering could play an important role in the process by characterizing the detergent associated conformation of membrane proteins. Small-angle scattering measurements to determine the degree of sample polydispersity and to look for the presence of a micelle peak that based on the present investigation is indicative of protein unfolding could serve as a rapid screen of detergents for compatibility with membrane proteins as a prelude to crystallization trials. Abbreviations The following abbreviations are used: PSI, photosystem I; PSII, photosystem II; LHCI, light-harvesting complex I; LHCII, light-harvesting complex II; prep A, PSI density gradient centrifugation preparation; prep B, PSI anion exchange chromatography preparation; MES, 2-(N-morpholino)ethanesulfonic acid; TEMED, N,N,N′,N′-tetramethylethylenediamine; SAXS, small-angle X-ray scattering; Rg, radius of gyration; Dmax, maximum dimension; D, fractal dimension; ξ, correlation length. Acknowledgment. We thank Soenke Seifert of the BESSRCCAT and Elena Kondrashkina at BIO-CAT for their support

SAXS of Photosystem I-Detergent Complexes during the SAXS experiments. Use of the Advanced Photon Source was supported by the U.S. Department of Energy, Office of Science, Office of Basic Energy Sciences, under Contract No. W-31-109-Eng-38. This work was supported by the Offices of Basic Energy Sciences and Biological and Environmental Research, U.S. Department of Energy, under Contract No. DEAC05-00OR22725 with Oak Ridge National Laboratory, managed and operated by UT-Batelle, LLC. The submitted manuscript has been authored by a contractor of the U.S. Government under Contract DE-AC05-00OR22725. Supporting Information Available: A figure showing the P700 photooxidation and reduction profiles of PSI preparations. This material is available free of charge via the Internet at http:// pubs.acs.org. References and Notes (1) Krogh, A.; Larsson, B.; von Heijne, G.; Sonnhammer, E. L. L. J. Mol. Biol. 2001, 305, 567. (2) White, S. H. Protein Sci. 2004, 13, 1948. (3) Caffrey, M. J. Struct. Biol. 2003, 142, 108. (4) Lee, I.; Lee, J. W.; Stubna, A.; Greenbaum, E. J. Phys. Chem. B 2000, 104, 2439. (5) Brettel, K. Biochim. Biophys. Acta 1997, 1318, 322. (6) Kitmitto, A.; Mustafa, A. O.; Holzenburg, A.; Ford, R. C. J. Biol. Chem. 1998, 273, 29592. (7) Kitmitto, A.; Holzenburg, A.; Ford, R. C. J. Biol. Chem. 1997, 272, 19497. (8) Ben-Shem, A.; Frolow, F.; Nelson, N. Nature 2003, 426, 630. (9) Jordan, P.; Fromme, P.; Witt, H. T.; Klukas, O.; Saenger, W.; Krauss, N. Nature 2001, 411, 909. (10) Zubrzycki, I. Z.; Frankel, L. K.; Russo, P. S.; Bricker, T. M. Biochemistry 1998, 37, 13553. (11) Svensson, B.; Tiede, D. M.; Barry, B. A. J. Phys. Chem. B 2002, 106, 8485. (12) Hong, X. G.; Weng, Y. X.; Li, M. Biophys. J. 2004, 86, 1082. (13) O’Neill, H.; Greenbaum, E. Chem. Mater. 2005, 17, 2654. (14) Peter, G. F.; Thornber, J. P. J. Biol. Chem. 1991, 266, 16745. (15) Sarvari, E.; Nyitrai, P. Electrophoresis 1994, 15, 1068. (16) Lowry, O. H.; Rosebrough, N. J.; Farr, A. L.; Randall, R. J. J. Biol. Chem. 1951, 193, 265. (17) Porra, R. J.; Thompson, W. A.; Kriedemann, P. E. Biochim. Biophys. Acta 1989, 975, 384. (18) Markwell, J. P.; Thornber, J. P.; Skrdla, M. P. Biochim. Biophys. Acta 1980, 591, 391. (19) Beno, M. A.; Jennings, G.; Engbretson, M.; Knapp, G. S.; Kurtz, C.; Zabransky, B.; Linton, J.; Seifert, S.; Wiley, C.; Montano, P. A. Nucl. Instrum. Methods Phys. Res., Sect. A 2001, 467, 690. (20) Fischetti, R.; Stepanov, S.; Rosenbaum, G.; Barrea, R.; Black, E.; Gore, D.; Heurich, R.; Kondrashkina, E.; Kropf, A. J.; Wang, S.; Zhang, K.; Irving, T. C.; Bunker, G. B. J. Synchrotron Radiat. 2004, 11, 399.

J. Phys. Chem. B, Vol. 111, No. 16, 2007 4219 (21) Seifert, S.; Winans, R. E.; Tiede, D. M.; Thiyagarajan, P. J. Appl. Crystallogr. 2000, 33, 782. (22) Glatter, O.; Kratky, O. Small-angle X-ray Scattering; Academic Press: London, 1982. (23) Guinier, A.; Fournet, G. Small-angle Scattering of X-rays; Wiley: New York, 1955. (24) Svergun, D. I. J. Appl. Crystallogr. 1992, 25, 495. (25) Pedersen, J. S. J. Appl. Crystallogr. 2000, 33, 637. (26) Pedersen, J. S. AdV. Colloid Interface Sci. 1997, 70, 171. (27) Heller, W. T. J. Appl. Crystallogr. 2006, 39, 671. (28) Paradies, H. H. J. Phys. Chem. 1980, 84, 599. (29) Taylor, J. R. An Introduction to Error Analysis: The Study of Uncertainties in Physical Measurements, 2nd ed.; University Science Book: Sausalito, CA, 1997. (30) Teixeira, J. J. Appl. Crystallogr. 1988, 21, 781. (31) Chen, S. H.; Teixeira, J. Phys. ReV. Lett. 1986, 57, 2583. (32) Mullet, J. E.; Burke, J. J.; Arntzen, C. J. Plant Physiol. 1980, 65, 814. (33) Jolley, C.; Ben-Shem, A.; Nelson, N.; Fromme, P. J. Biol. Chem. 2005, 280, 33627. (34) Baba, K.; Itoh, S.; Hastings, G.; Hoshina, S. Photosynth. Res. 1996, 47, 121. (35) Bassi, R.; Simpson, D. Eur. J. Biochem. 1987, 163, 221. (36) Burke, J. J.; Ditto, C. L.; Arntzen, C. J. Arch. Biochem. Biophys. 1978, 187, 252. (37) Klughammer, C.; Schreiber, U. Measuring P700 absorbance changes in the near infrared spectral region with a dual waVelength pulse modulation system; Kluwer Academic Press Publishers: Dordrecht, The Netherlands, 1998; Vol. 5. (38) Fischer, H.; Polikarpov, I.; Craievich, A. F. Protein Sci. 2004, 13, 2825. (39) Guo, X. H.; Zhao, N. M.; Chen, S. H.; Teixeira, J. Biopolymers 1990, 29, 335. (40) Santos, S. F.; Zanette, D.; Fischer, H.; Itri, R. J. Colloid Interface Sci. 2003, 262, 400. (41) Gelamo, E. L.; Itri, R.; Tabak, M. J. Biol. Chem. 2004, 279, 33298. (42) Gelamo, E. L.; Itri, R.; Alonso, A.; da Silva, J. V.; Tabak, M. J. Colloid Interface Sci. 2004, 277, 471. (43) Bruce, B. D.; Malkin, R. J. Biol. Chem. 1988, 263, 7302. (44) Ben-Shem, A.; Nelson, N.; Frolow, F. Acta Crystallogr., Sect. D 2003, 59, 1824. (45) Fromme, P.; Witt, H. T. Biochim. Biophys. Acta 1998, 1365, 175. (46) Enright, M. B.; Leitner, D. M. Phys. ReV. E 2005, 71. (47) Simidjiev, I.; Stoylova, S.; Amenitsch, H.; Javorfi, T.; Mustardy, L.; Laggner, P.; Holzenburg, A.; Garab, G. Proc. Natl. Acad. Sci. U.S.A. 2000, 97, 1473. (48) Gross, E. L.; Grenier, J. Arch. Biochem. Biophys. 1978, 187, 387. (49) Lee, J. W. Ph.D. Dissertation, Cornell University, Ithaca, NY, 1993. (50) Ikegami, I.; Itoh, S.; Iwaki, M. Plant Cell Physiol. 1995, 36, 857. (51) Misquitta, L. V.; Misquitta, Y.; Cherezov, V.; Slattery, O.; Mohan, J. M.; Hart, D.; Zhalnina, M.; Cramer, W. A.; Caffrey, M. Structure 2004, 12, 2113. (52) Misquitta, Y.; Caffrey, M. Biophys. J. 2003, 85, 3084.