LCP-Tm: an assay to measure and understand stability of membrane proteins in a membrane environment

Abstract

Structural and functional studies of membrane proteins are limited by their poor stability outside the native membrane environment. The development of novel methods to efficiently stabilize membrane proteins immediately after purification is important for biophysical studies, and is likely to be critical for studying the more challenging human targets. Lipidic cubic phase (LCP) provides a suitable stabilizing matrix for studying membrane proteins by spectroscopic and other biophysical techniques, including obtaining highly ordered membrane protein crystals for structural studies. We have developed a robust and accurate assay, LCP-Tm, for measuring the thermal stability of membrane proteins embedded in an LCP matrix. In its two implementations, protein denaturation is followed either by a change in the intrinsic protein fluorescence on ligand release, or by an increase in the fluorescence of a thiol-binding reporter dye that measures exposure of cysteines buried in the native structure. Application of the LCP-Tm assay to an engineered human beta2-adrenergic receptor and bacteriorhodopsin revealed a number of factors that increased protein stability in LCP. This assay has the potential to guide protein engineering efforts and identify stabilizing conditions that may improve the chances of obtaining high-resolution structures of intrinsically unstable membrane proteins.

Figure 1

Schematic diagram of the LCP-T_m protocol showing a sequence of temperature treatment steps. After each heating/cooling step samples are centrifuged at 20°C and 5600 × g for 10 min to return the LCP sample to a transparent state. Absorbance and fluorescence spectra are taken for each sample at RT. The whole protocol contains 13 heating/cooling treatment steps, takes ∼300 min and allows for processing of up to six to eight samples simultaneously (A, absorbance; F, fluorescence).

Figure 2

Heat-induced water shedding from the LCP sample. (A) Photographs of a cuvette filled with LCP: (i) after loading; (ii) after heating to 80°C and cooling to RT; and (iii) after 10 min centrifugation at 5,600 × g and 20°C. The first layer of zoom windows shows the domain-like mesoscopic structure of LCP on a scale of ∼100 μm. The second layer of zoom windows depicts the LCP microstructure at a 100 Å scale, showing a cubic lattice formed by a single lipid bilayer. In the initial sample (i) the LCP domains are tightly packed resulting in a homogeneous and transparent appearance. Heating (ii) induces shrinkage of the cubic lattice shedding water into the interdomain space. The water droplets of a few microns in size scatter light impeding spectroscopic measurements (see Fig. S1). Cooling alone does not restore transparency due to a prominent hysteresis in the LCP swelling behavior. A mechanical force such as centrifugation is required to achieve a transparent sample (iii). (B) MO/water temperature-composition phase diagram (re-drawn from Briggs et al. (21)). The phase diagram is metastable at temperatures below 17°C (36). Ia3d and Pn3m represent two bicontinuous cubic phases with different symmetries. Initial samples are prepared at RT and 40% w/w hydration in the cubic-Pn3m phase (yellow dot). Heating brings the samples along the yellow dashed line into a region where the cubic-Pn3m phase and water coexist. The higher the temperature, the more water separates from the LCP.

Figure 3

Thermal denaturation curves for β₂AR-T4L obtained by the LCP-T_m protocol. (A) Comparison between the intrinsic protein fluorescence and the CPM probe fluorescence for monitoring the unfolding of β₂AR-T4L/timolol in the MO cubic phase. (B) Effect of ligands on denaturation of β₂AR-T4L in the MO cubic phase. Individual points represent averaged data obtained from at least three samples. Continuous curves represent fits by the Boltzmann sigmoidal function.

Figure 4

Apparent melting temperatures obtained with the LCP-T_m assay using intrinsic protein fluorescence. (A) Effect of ligands. (B) Effect of LCP host lipids. (C) Effect of lipid additives. (D) Effect of pH. The solid bars represent data for β₂AR-T4L and the open bars for β₂AR samples. Experiments in B, C, and D were carried out with receptors bound to timolol. The T_m data represent averaged values obtained by curve fitting from at least three samples. The error bars show standard deviation for T_m data.

Figure 5

Comparisons of stability of bR and β₂AR-T4L/timolol in detergent (open symbols) and LCP (solid symbols) environments at RT. Fractions of folded proteins were estimated using absorbance and fluorescence as described in the Materials and Methods. The original raw data are shown in Figs. S11–S18 in the Supporting Material. bR was prepared in 1.2% w/v OG, 25 mM Na/K phosphate pH 5.6, and β₂AR-T4L was prepared in 0.05% w/v DDM, 0.01% w/v CHS, 20 mM Hepes pH 7.5, 150 mM NaCl, 0.5 mM timolol.