Progress in Biophysics and Molecular Biology 88 (2005) 339���357 Review Membrane protein crystallization in amphiphile phases: practical and theoretical considerations Peter Nollert deCODE BioStructures, 7869 NE Day Rd. W, Bainbridge Island, WA 98110, USA Available online 7 October 2004 Abstract Integral membrane proteins are amphiphilic molecules. In order to enable chromatographic purification and crystallization, a complementary amphiphilic microenvironment must be created and maintained. Various types of amphiphilic phases have been employed in crystallizations and intricate amphiphilic microenvironmental structures have resulted from these and are found inside membrane protein crystals. In this review the process of crystallization is put into the context of amphiphile phase transitions. Finally, practical factors are considered and a pragmatic way is suggested to pursue membrane protein crystallization trials. r 2004 Elsevier Ltd. All rights reserved. Keywords: Membrane protein crystallization Amphiphile Crystal packing Lipid Detergent Contents 1. Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 340 2. Amphiphilic microenvironments conducive to crystallization . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 341 3. The amphiphilic microenvironment inside membrane protein crystals . . . . . . . . . . . . . . . . . . . . . . . . . . . . 346 4. Amphiphile phase transitions during crystallization. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 349 ARTICLE IN PRESS www.elsevier.com/locate/pbiomolbio 0079-6107/$ - see front matter r 2004 Elsevier Ltd. All rights reserved. doi:10.1016/j.pbiomolbio.2004.07.006 E-mail address: pnollert@decode.com(P. Nollert).

5. Practical considerations . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 351 6. Conclusions. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 353 Acknowledgements. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 354 References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 354 1. Introduction Integral membrane proteins are amphiphilic molecules, they ������love both������, water as well as oil. This duality is rooted in the physical nature of their surfaces: they all possess two fundamentally different types of surfaces, a hydrophobic perimeter and two hydrophilic caps (Fig. 1). While soluble proteins interact with water molecules and ions in an aqueous medium, membrane proteins do this only with a fraction of their surface. The hydrophobic perimeter faces alkyl chains of lipids and hydrophobic surfaces of other integral membrane proteins. Indeed, both of these media need to be arranged within certain dimensions into a suitable microenvironment in order to maintain the native conformation of the protein molecule. However, having individual particles is required to subject these proteins to chromatographic purification procedures and to arrange theminto well-ordered three-dimensional arrays, crystals suitable for X-ray diffraction experiments. These particles have to be free to translate and rotate in space to bind, aggregate, forma crystal nucleus and associate with the faces of a growing crystal. The restrictions set by ARTICLE IN PRESS Fig. 1. Schematic depiction of a soluble protein and a transmembrane protein in their native environment, an aqueous solution and a membrane bilayer, respectively. Both models are based on high-resolution X-ray crystallographic experimental structures showing protein, water and lipids. Water molecules are red and blue, the surface of the protein is blue where there is negative charge and red where there is positive charge, hydrophobic core and lipids are colored gray. (A) Soluble protein particles are dissolved in a homogenous medium with a dielectric constant e close to that of distilled water. On their surface they interact with water molecules and with ions. (B) Integral protein particles are dissolved a low dielectric amphipilic medium, the lipid bilayer membrane, contacting the hydrophobic core, while their hydrophilic caps are exposed to an aqueous medium similar to that shown in A for soluble proteins. P. Nollert / Progress in Biophysics and Molecular Biology 88 (2005) 339���357 340

these two requirements, (i) maintaining an amphiphilic microenvironment and, (ii) allowing essentially free diffusion of small units are severe and are at the heart of the challenge to grow crystals of integral membrane proteins. Nonetheless, many membrane protein crystallizations have been reported (for a current update on published membrane protein structures visit http://www.mpibp���frankfurt.mpg.de/michel/ public/memprotstruct.html or http://blanco.biomol.uci.edu/Membrane_Proteins_xtal.html) and it is the goal of this review to point out practical as well as theoretical considerations these crystallizations are based on. The focus is on highlighting the delicate interplay between small molecule amphiphiles, detergents and lipids, and integral membrane proteins before, during and after crystallization. The critical role of amphiphiles in crystallization cannot be fully addressed in the context of this review and the reader is therefore directed to reviews and monographs by Garavito and Ferguson-Miller (2001), Garavito and Picot (1990), Wiener (2001), Michel (1983, 1991), Hunte et al. (2003) and Iwata (2003). For a fundamental introduction to the physical chemistry of detergents consult Tanford (1980), Rosen (1978) and Wennerstroemand Lindman (1979). 2. Amphiphilic microenvironments conducive to crystallization The first reports of successful membrane protein crystallizations (Michel and Oesterhelt, 1980 Henderson and Shotton, 1980 Garavito and Rosenbusch, 1980) created a paradigm for membrane protein crystallization: transfer membrane proteins from their native environment into particulate detergent micelles in order to purify and to crystallize them in the same way as soluble proteins. It was reasoned that homogenous, lipid-free protein detergent micelles of uniform size would be most suitable for crystallization experiments. At the time the choice of the detergent for crystallization purposes was based on three factors: (i) stabilization of the native conformation of the membrane protein in monodisperse form, (ii) enabling protein���protein contacts in the packed crystal and, (iii) preventing detrimental phase separations during crystal growth. This line of thinking was expanded in the recent years, particularly with respect to lipids being recognized as beneficial and sometimes crucial crystallization components. Most detergents belong into one of the following categories: ionic, non-ionic or zwitterionic. Their characteristic behavior depends on their shape, stereochemistry of the head group and tail. According to the ���intrinsic curvature hypothesis��� (Gruner, 1985) they formsupramolecular structures in water due to the hydrophobic effect (Tanford, 1980) and their shape (Fig. 2A). At sufficiently high concentrations, i.e. above the critical micellar concentration (CMC), detergents form micelles. These form roughly spherical objects in which detergent molecules are primarily packed with their alkyl chains towards the center and their head groups towards the surface (Rosen, 1978 Wennerstroemand Lindman, 1979). Detergent molecules in these micelles are flexible and exhibit a high degree of mobility, allowing for dramatic fluctuations in overall micellar shape including deformations, fusion and fission (Tieleman et al., 2000). Amphiphiles generally display a rich phase behavior commonly described in phase diagrams (Fig. 2B). Detergents typically have consolute boundaries, separating a single-phase micellar region from a ARTICLE IN PRESS P. Nollert / Progress in Biophysics and Molecular Biology 88 (2005) 339���357 341