Solubilization of a membrane protein by combinatorial supercharging

Abstract

Hydrophobic and aggregation-prone, membrane proteins often prove too insoluble for conventional in vitro biochemical studies. To engineer soluble variants of human caveolin-1, a phage-displayed library of caveolin variants targeted the hydrophobic intramembrane domain with substitutions to charged residues. Anti-selections for insolubility removed hydrophobic variants, and positive selections for binding to the known caveolin ligand HIV gp41 isolated functional, folded variants. Assays with several caveolin binding partners demonstrated the successful folding and functionality by a solubilized, full-length caveolin variant selected from the library. This caveolin variant allowed assay of the direct interaction between caveolin and cavin. Clustered along one face of a putative helix, the solubilizing mutations suggest a structural model for the intramembrane domain of caveolin. The approach provides a potentially general method for solubilization and engineering of membrane-associated proteins by phage display.

Figures

Figure 1

Selection, design, and expression of caveolin selectants. a) Solubility selections using a phage-displayed library of caveolin intra-membrane domain (IMD) variants. The anti-selection eliminates aggregation-prone, hydrophobic variants binding to hydrophobic interaction chromatography resin. Then, a positive selection step isolated variants binding to gp41. The increased salt concentration and washes during the selections increases the stringency of the selections. b) SDS-PAGE and c) Western blot of E. coli crude lysates expressing caveolin variants, using an antibody that recognizes the N-terminus of caveolin. Purified cav(1-104) in lane 1 is a positive control. The negative control in lane 2 is the lysate fraction of E. coli carrying the empty expression vector. An additional control in lane 3 shows expression of a 17 kDa caveolin variant (cavΔN-term) which lacks the antibody recognition site. The red arrow indicates the expected size of full-length caveolin fused to a His6 tag (≈24.9 kDa). d) SDS-PAGE and e) Western blot of E. coli lysates expressing full-length caveolin variants with a N-terminal MBP fusion having the expected size indicated by the red arrow (≈64.5 kDa). An antibody specific for MBP detects the presence of the overexpressed proteins. The pellet fraction (P) represents the insoluble proteins after cell lysis and centrifugation, and S indicates the soluble fraction. The control in lanes 1 and 2 is lysate from E. coli cells expressing only MBP.

Figure 2

Binding by the solubilized full-length caveolin variant (cav11-MBP) to cavin and three other binding partners. a) In this ELISA, cav11-MBP binds to known binding partners gp41 ectodomain, protein kinase A (PKA), or truncated caveolin, cav(1-104), and cavin. Binding to the blocking agent (non-fat milk), was subtracted to provide the net HRP activity. Cav11-MBP binds to all four targets. b) As negative controls, MBP fails to the blocking agent used in panel a (non-fat milk), and no signal is observed for control wells with MBP omitted. c) In addition, cav11-MBP (4 μM) binds well to the target proteins, when BSA is used as a blocking agent; the negative control, MBP (4 μM) fails to bind to the target proteins. Error bars indicate standard deviation (n=3). d) The role of cavin in caveolae formation. Membrane bending and formation of caveolae requires the interaction of cavin with caveolin. Results reported here demonstrate that caveolin can bind directly to cavin. Dissociation of cavin destabilizes the caveolae, and caveolin can then be targeted for lysosomal degradation. Schematic adapted from (34).

Figure 3

Predicted membrane topology of the caveolin IMD. a) Helical wheel representations of the IMDs of wild-type (wt) caveolin and cav11. The aliphatic and aromatic faces of the helix are shown, and N and C superscripts indicate the N- and C-termini. Residues in blue were targeted for mutagenesis in the library. Orange residues designate positions with charged residues substituted for the soluble variant. b) Proposed membrane topology of the wild-type caveolin IMD α-helix.