Pil Seok
Chae
a,
Philip D.
Laible
*b and
Samuel H.
Gellman
*a
aDepartment of Chemistry, University of Wisconsin-Madison, 1101 University Avenue, Madison, WI 53706, USA. E-mail: gellman@wisc.edu; Fax: +1 608-265-4534; Tel: +1 608-262-3303
bBiosciences Division Argonne National Laborotory, 9700 South Cass Avenue, Argonne, IL 60439, USA. E-mail: laible@anl.gov; Fax: +1 630-252-3387
First published on 14th October 2009
Integral membrane proteins (IMPs) are crucial biological components, mediating the transfer of material and information between cells and their environment. Many IMPs have proven to be difficult to isolate and study. High-resolution structural information on this class of proteins is limited, largely because of difficulties in generating soluble forms of such proteins that retain native folding and activity, and difficulties in generating high-quality crystals from such preparations. Isolated IMPs typically do not dissolve in aqueous solution, a property that arises from the large patches of hydrophobic surface necessary for favorable interactions with the core of a lipid bilayer. Detergents are generally required for IMP solubilization: hydrophobic segments of detergent molecules cluster around and shield from water the hydrophobic protein surfaces. The critical role played by detergents in membrane protein manipulation, and the fact that many IMPs are recalcitrant to solubilization and/or crystallization with currently available detergents, suggest that it should be valuable to explore new types of amphiphiles for these purposes. This review constitutes a progress report on our long-term effort to develop a new class of organic molecules, collectively designated “tripod amphiphiles,” that are intended as alternatives to conventional detergents for membrane protein manipulation. One long-range goal of this research is to identify new types of amphiphiles that facilitate IMP crystallization. This review should help introduce an important biochemical need to organic chemists, and perhaps inspire new approaches to the problem.
Over the past two decades, a few research groups have examined unusual synthetic amphiphiles for the ability to stabilize IMPs in aqueous solution. Examples include peptitergents,13 amphiphilic polymers (amphipols),14–16 hemifluorinated amphiphiles,17,18lipopeptidedetergents (LPDs),19 and deoxycholate-based facial amphiphiles.20,21 Peptitergents are peptides designed to be lipophilic on one side and hydrophilic on the other upon folding to an α-helical conformation; this design has its roots in earlier fundamental exploration of amphiphilic secondary structures.22 Amphipols are co-polymers that have been shown to stabilize a number of IMPs after they have been extracted from the membrane with conventional detergents.23 Some hemifluorinated amphiphiles were observed to be mild relative to non-fluorinated homologues for stabilization of a few of IMPs, particularly delicate assemblies composed of multiple subunits, such as cytochrome b6f complexes.17,18 LPDs, featuring an amphiphilic α-helix, have been reported to keep several IMPs (e.g., bacteriorhodopsin, PagP, Lac permease-cytochromeb562 fusion protein) soluble in water while maintaining native structure.19 However, few of these novel amphiphiles are commercially available, and there is no report of their use for IMP crystallization so far. Thus, there is a continuing need for synthetically accessible amphiphiles that can serve as tools for manipulating IMPs.
Organic chemists can readily envision new types of low-molecular-weight amphiphiles and plan synthetic routes to generate such molecules, but there have been relatively few efforts of this type to date. In our view, this situation arises for several reasons. (1) Very few organic chemists are aware of the problem. (2) Specialized biochemical skills are required to work with IMPs, and few, if any, organic chemistry groups have the necessary skills. Therefore, pursuing the development of new amphiphiles for use with IMPs requires long-term collaborations. (3) Assessing structural and functional integrity of an IMP in an amphiphile-solubilized state is often challenging. (4) Each integral membrane protein has its own peculiarities as a subject of experimental analysis, and expertise developed with one particular IMP does not necessarily transfer directly to the study of other IMPs. Most laboratories that explore IMP structure and function are focused on just one or a few examples. This situation makes it challenging to assess the extent to which new agents for membrane protein manipulation might have broad utility.24,25
Fig. 1 Chemical structures of representative classical detergents (LDAO, DDM) and selected non-classical amphiphiles (CHAPS, tween detergents, deoxycholate-based amphiphile, amphipol (A8-35; n ≈ 80), and LPD). |
Fig. 2 Schematic representation of conventional detergents (I) and tripod amphiphiles (TPAs; II and III). |
Tripod amphiphile architecture,33–35 illustrated in structures II and III of Fig. 2, differs fundamentally from that of classical detergents in that the tripods contain a branch point within the hydrophobic portion (II) or within both the hydrophobic and hydrophilic portions (III). A branch point, particularly a quarternary center, provides a subtle restriction on conformational mobility relative to linear molecular fragments,36–38 such as the n-alkyl tails common among conventional detergents. (Of course, a steroidal structure, as in CHAPS, has several branch points, but in this case conformational restriction is more severe, resulting from the cyclic constraints.) Our motivation for including branch points was to try to decrease the flexibility that characterizes most conventional detergents, which we believed might contribute to the difficulty of maintaining membrane proteins in an active state in solution and, perhaps, to the crystallization of membrane protein–detergent complexes.
Fig. 3 Modular synthesis of tripod amphiphiles: (a) CH2(CN)2, AcOH, NH4OAc, benzene, reflux; (b) PhMgBr, CuCN, THF, 0 °C; (c) KOH, ethylene glycol, reflux; (d) EDC·HCl, HOBt, 3-(dimethylamino)-1-propylamine, DMF, then m-CPBA, CHCl3; (e) EDC·HCl, HOBt, serinol, DMF; (f) AgOTf, 2,3,4,6-tetra-O-benzoyl-α-D-glucopyranosyl bromide, DCM, then NaOMe, MeOH; (e) EDC·HCl, HOBt, ethanolamine, DMF; (h) 1,2-trans-peracetylated maltose, BF3·Et2O, DCM, then NaOMe, MeOH. |
The ease with which tripod amphiphile structure may be varied is important because it is unlikely that any single version will be a “magic bullet” that succeeds with all or even a large fraction of membrane proteins. The relationship between conventional detergent HLB and efficacy in extraction of IMPs from their native membranes illustrates this point.42–46Detergents within a narrow range of HLB values (12 to 15) display high efficiency and selectivity for the extraction of human adenosine A3 receptor in a functional state.47 Comparable optimal HLB ranges have been found for other IMPs from prokaryotic and eukaryotic sources. For instance, detergents with HLB ≈ 13 successfully extracted D-alanine carboxypeptidase from Bacillus subtilis cells.45 A similar value was found for protein extraction from mitochondrial membrane from bovine heart.48 It is notable that optimum HLB values vary somewhat from one IMP to the next. For instance, mitochondrial porin solubilization from bovine heart was more effective with detergents that have a smaller HLB, while total membrane protein solubilization from the same source was better achieved with the detergents of higher HLB.48
Fig. 4 Chemical structures of tripod amphiphile 4 and isomeric amphiphiles 5 and 6, which have more classical architectures. |
We were initially surprised by the success of N-oxides 1 and 4 for bR solubilization, because treatment of purple membrane with LDAO (Fig. 1), a widely used conventional N-oxide detergent, causes rapid bR denaturation, as indicated by a loss of the characteristic purple color.52 This contrast enabled us to probe the functional importance of amphiphile architecture. We prepared 5 and 6, isomers of tripod amphiphile 4, in which the lipophilic portion is incrementally transformed into a conventional detergent “tail”. Both 5 and 6 behaved similarly to LDAO, causing rapid bR denaturation.33 This early result provided strong support for our hypothesis that the tripod amphiphile architecture represents an advantageous complement to conventional detergent architecture.
Comparisons among a small set of N-oxide tripod amphiphiles revealed that 1 was somewhat superior to 4 in terms of behavior with bR and Rho. Solubilized forms of each protein could be readily purified (including removal of endogenous lipids), and the resulting preparations were stable for several weeks, as assessed by optical absorbance.34 Stability of solubilized membrane proteins on this time scale is essential for crystallization efforts. Two proteins have been crystallized from the 1-solubilized state, bR53 and a form of the potassium channel from Streptomyces lividans;33 however, structure-determination has not been carried out in either case.
Attempted solubilization of the LHI–RC superassembly from R. capsulatus membranes with N-oxide tripod amphiphile 1 led to extensive disruption.35 However, tripod amphiphile 2, which has a branched diglucoside head group, could extract the intact LHI–RC superassembly from the native membrane. Tripod amphiphile 3 is very similar to 2, but there is no branch point in the hydrophilic portion of 3. This subtle structural difference leads to a substantial functional difference: considerable degradation of LHI was observed when solubilization was undertaken with 3. The contrast between 2 and 3 suggests that a branch point in the hydrophilic portion can complement a branch point in the hydrophobic portion in terms of membrane protein stabilization in aqueous solution. Interestingly, this functional distinction does not arise from HLB, which is quite similar for the two tripod amphiphiles, nor is the distinction reflected in critical micelle concentrations (CMC), which are almost identical for 2 and 3 (3.6 and 4.0 mM). It should be pointed out that 3, lacking a branch point in the hydrophilic portion, is significantly less soluble than analogue 2.
We prepared a series of detergents (7a–e) bearing the branched diglucoside head group found in 2 connected to a linear alkyl tail. Among this series, the version with a 12-carbon tail (7c) was most effective at extracting the LHI–RC superassembly from R. capsulatus membranes. However, the extraction efficiency of this detergent was much lower than that observed with tripod amphiphile 2. This comparison seems to complement the comparison of N-oxides 4–6 discussed above in suggesting that the tripod architecture displays distinctive advantages in the context of membrane protein solubilization and stabilization. Interestingly, some conventional detergents feature branching in the hydrophilic portion, as exemplified by the tween series (Fig. 1). Among non-classical amphiphiles, amphipols and LPDs (Fig. 1) can be viewed as containing branch points. We could not compare 7c with the analogue bearing a maltose (i.e., non-branched) head group because the latter was insoluble. Thus, even in the context of a conventional hydrophobic tail, hydrophilic group branching seems to confer favorable behavior. Collectively, these observations raise the possibility that new amphiphile architectures, featuring different placements of internal branch points and manifesting useful properties for IMP manipulation, remain to be discovered by imaginative chemists.
A recent comparison of >120 commercially available detergents in terms of LHI–RC superassembly solubilization from R. capsulatus membranes indicates that DDM is one of the most effective among conventional detergents.58 This demonstration of the value of DDM as a tool for membrane protein manipulation complements results obtained from studies with other IMPs, including Rho,59 diacyl glycerol kinase,60lactose permease,61 and human apelin receptor.62 In this context, it is noteworthy that we found DDM to be less effective than 2 for long-term stabilization of the R. capsulatus LHI–RC superassembly in aqueous solution;35 this comparison highlights the promise of tripod amphiphiles for membrane protein research.
A molecular design that is conducive to facile variation enables exploration of structure–property relationships. Comparisons among tripod amphiphiles having closely related hydrophilic portions, as reported in our publications,34,35 have shown that performance is quite sensitive to these variations. More recently we have explored the impact of variations in the hydrophobic portions and observed profound effects on function. The two TPAs shown in Fig. 5, related structurally to 2, proved to be inferior to 2 for solubilization of the LHI–RC superassembly from R. capsulatus (unpublished results). These findings show that variations in the TPA hydrophobic group can have a significant influence on performance.
Fig. 5 Tripod amphiphiles with variations in the hydrophobic portion relative to 2. These amphiphiles are inferior to 2 for R. capsulatus LHI–RC superassembly solubilization. |
The many integral membrane proteins that remain difficult to isolate and purify, or that are refractory to crystallization, constitute a strong impetus for the invention of new synthetic amphiphiles. This work requires the creativity and skill set of the organic chemist, which must be deployed in concert with the techniques and insights of membrane protein biochemists. In addition to solubilization, stabilization and ultimately crystallization of membrane proteins, novel amphiphiles can be applied to solution NMR spectroscopy63–65 and might support exciting new methods for study of membrane proteins, such as mass spectrometry.66 It is hoped that this review will inspire more chemists to apply their imaginations to the development of new types of amphiphilic agents for applications in membrane protein science.
This journal is © The Royal Society of Chemistry 2010 |