Mechanisms of membrane protein crystallization in 'bicelles'

Abstract

Despite remarkable progress, mainly due to the development of LCP and 'bicelle' crystallization, lack of structural information remains a bottleneck in membrane protein (MP) research. A major reason is the absence of complete understanding of the mechanism of crystallization. Here we present small-angle scattering studies of the evolution of the "bicelle" crystallization matrix in the course of MP crystal growth. Initially, the matrix corresponds to liquid-like bicelle state. However, after adding the precipitant, the crystallization matrix transforms to jelly-like state. The data suggest that this final phase is composed of interconnected ribbon-like bilayers, where crystals grow. A small amount of multilamellar phase appears, and its volume increases concomitantly with the volume of growing crystals. We suggest that the lamellar phase surrounds the crystals and is critical for crystal growth, which is also common for LCP crystallization. The study discloses mechanisms of "bicelle" MP crystallization and will support rational design of crystallization.

Figure 1

Sequence of the experimental steps. (A) Schematic representation of the experimental stages of crystallization of bacteriorhodopsin in bicelles within capillaries. Step 1—preparation of the crystallization system on ice (mixture of bicelles and purple membranes); step 2—loading of the bicellar/purple membrane mixture into a capillary; step 3—addition of a precipitant and monitoring of subsequent changes of the crystallization matrix; step 4—monitoring of the formation of the BR crystals. During steps 2–4, monitoring of the structure of the crystallization system and the growth of the crystals were performed using real-time SAXS. (B)—Photographs of the crystallization system in different steps of the experiments (aligned vertically with A) corresponding to steps 1 (day 0), 2 (day 4), 3 (day 8 & 17); the crystallization system presents a transparent/semi-transparent homogeneous phase. In step 4 (day 24, 25 & 65), small crystals appeared, and then they grew to a maximum size.

Figure 2

Evolution of the crystallization matrix during the crystallization process. The SAXS curves for the pure DMPC/CHAPSO mixture without PM (left) and the crystallization system DMPC/CHAPSO/PM (right). Experimental data for crystallization system with and without PM are shown as light purple and orange hollow circles, correspondingly. SAXS curves for PM are shown as dark purple squares. The approximations by form-factors of the bicelles and the ribbons are shown as blue lines. The graphical representations of the structural organizations of the crystallization matrix are shown adjacent to the corresponding SAXS curves.

Figure 3

Transformation of the SAXS curves for the crystallization system during the different steps of the crystallization process. (A) The SAXS curves related to different steps of the crystallization process. The red arrows in the small angle region indicate the interference peaks from the lipid/detergent smectic phase (curve designations are given in the legend; steps numbering is described in Fig. 1). (B) The peaks extracted from the SAXS curves (part A) by subtraction of the baseline. The curves are scaled to separate them vertically for better visualization. The black arrows “Lα₁” and “Lα₂” (shown in both A,B) indicate the lamellar peaks of the first and second order; the other black arrow “L_cryst 64 Å” indicates the peak from the local lamellar phase bonded with the protein surface, the spacing of this L_cryst is 64 Å. The arrow “68 Å” indicates the precursor of the L_cryst phase.

Figure 4

SAXS peaks from the lipid/detergent crystallization phase prior to or at the moment of crystal formation. The peaks’ positions correspond to distances 500–700 Å (these values were calculated from the positions of 1st order peaks). The graphs are presented after baseline subtraction. The observed peaks disappeared several days after they appeared. Each curve is measured in a different capillary. The curves are scaled to separate them vertically for better visualization. The dashed rectangle highlights two consecutive curves of the same sample: these curves show that, with time, peak intensity increases and shifts to smaller angles. The arrows indicate the position of the 2nd order peaks.

Figure 5

SAXS data from the crystallization matrix after crystal formation (BM29, ESRF). (A) The SAXS curve for the crystallization matrix with crystals (blue curve) and the corresponding baseline (red curve). (B) The SAXS curve from the crystallization matrix after baseline subtraction (orange). Intensity is multiplied by q⁴ for better observation of the wide-angle peaks. The Gaussian approximations of the peaks are represented in black. The Miller indexes (for BR crystals) and the reflex numbers (for the lipidic multilayers Lα or L_cryst) are marked above the corresponding peaks (for more details, see Table S4).

Figure 6

Behavior of the peak for the local lamellar phase L_cryst. (A) 2D pattern for the sample containing the BR crystals (corresponds to 1D curve step 4.3 in Fig. 3). The reflections corresponding to the local lamellar phase L_cryst are shown by white arrows. The scattering ring belongs to the multilamellar phase Lα with a spacing of about 84 Å. (B) Behavior of the peak for the local lamellar phase L_cryst on the 1D curves. All the SAXS curves belong to different time points of the same sample. The crystallization conditions are the same as for the sample in Fig. 3.

Figure 7

The scheme demonstrating the evolution of the crystallization matrix and sequential appearance/disappearance of various structural elements: a mixture of bicelles and PMs, ribbons, the lamellar phase Lα, the “phase 500–700 Å”, the local lamellar phase L_cryst and BR crystals (see more extended description in the main text). Following Katsaras et al., we present the “phase 500–700 Å” as chiral nematic; however, the true nature of this phase is still unclear. The axis correspond to time, complexity (i.e. number/quantity of the new appeared structural element), and concentration, respectively. Concentration is given in arbitrary units (structural elements have different concentration ranges; here the concentration is presented on the same scale for clarity; dependencies of concentrations vs. time are shown qualitatively).

Figure 8

Models used for the approximation of the SAS data. (A) Model 1: circular cylinder with a core–shell scattering length density profile (in our calculations, we used ε = 1 for the bicelles). (B) Model 2: elliptical cylinder with a core–shell scattering length density profile.