Simultaneous Native Mass Spectrometry Analysis of Single and Double Mutants To Probe Lipid Binding to Membrane Proteins

Abstract

Lipids are critical modulators of membrane protein structure and function. However, it is challenging to investigate the thermodynamics of protein-lipid interactions because lipids can simultaneously bind membrane proteins at different sites with different specificities. Here, we developed a native mass spectrometry (MS) approach using single and double mutants to measure the relative energetic contributions of specific residues on Aquaporin Z (AqpZ) toward cardiolipin (CL) binding. We first mutated potential lipid-binding residues on AqpZ, and mixed mutant and wild-type proteins together with CL. By using native MS to simultaneously resolve lipid binding to the mutant and wild-type proteins in a single spectrum, we directly determined the relative affinities of CL binding, thereby revealing the relative Gibbs free energy change for lipid binding caused by the mutation. Comparing different mutants revealed that W14 contributes to the tightest CL binding site, with R224 contributing to a lower affinity site. Using double mutant cycling, we investigated the synergy between W14 and R224 sites on CL binding. Overall, this novel native MS approach provides unique insights into the binding of lipids to specific sites on membrane proteins.

Figure 1.

Schematic of single mutant analysis experiment and its data processing workflow. (A) Wild-type (WT) and W14A mutant proteins are mixed with CL at an approximate ratio of 1:1:100 of WT:W14A:CL. Wild type and W14A are annotated as tetrameric squares colored in green and blue, respectively, and CL is represented in a grey graphic. (B) Raw and (C) deconvolved mass spectra for a representative replicate at 25 °C for the wild type and mutant proteins with a series of lipids bound where deconvolved peaks are colored in green and blue, respectively. The number of lipids bound is annotated. Each peak area was extracted, and (D) the bar chart shows the average and standard deviation from the three replicate measurements. Calculated from Equation 3, (E) the differences in the Gibbs free energy change (ΔΔG) plot for binding up to four lipids reveal that (F) the W14A mutation is unfavorable for CL binding. The W14 site on one monomer is annotated in blue while the entire protein complex is grey.

Figure 2.

Schematic of the double mutant cycle experiment and its data processing workflow. (A) Wild-type and mutant proteins (X, Y, and XY) are mixed with CL at an approximate ratio of 1:1:1:1:200, where the proteins are represented in tetrameric squares colored in green, yellow, blue, and purple, respectively, and CL is represented in a grey graphic. (B) Deconvolved mass spectra independently resolve all four proteins, indicated in their respective colors. The numbers of bound lipids are annotated. (C) Extraction and pairwise comparison of the relative peak areas for each bound lipid enables the construction of the double mutant cycle.

Figure 3.

(A) Side-view and (B) top-view of AqpZ with mutant sites W14, R224, R75, and K4 labeled, indicated with an arrow, and colored in blue, yellow, cyan, and red, respectively. Four chains of the protein are shown in grey. PDB code: 2abm.

Figure 4.

The difference in the Gibbs free energy change (ΔΔG plots) for CL binding up to four lipids at 25 °C for each mutant (A) W14A, (B) R224A, (C) R75A, and (D) K4A compared to the wild-type protein. Error bars indicate the 95% confidence intervals. Green color bars represent unfavorable mutations, where CL binding is more favorable to the wild type. Negative blue color bars depict more favorable CL binding to the mutant. Grey bars represent statistically insignificant interactions.

Figure 5.

Double mutant cycle for the (A) first and (B) second CL bindings constructed with AqpZ WT, R224A, W14A, and R224A_W14A (DM) at 25 °C. The mean $\Delta \Delta G$ in kJ/mol with ± 95% confidence intervals are stated. Positive $\Delta \Delta G$ values are depicted in darker red for higher values. Data from lipids 2 to 3 and 3 to 4 can be found in Figure S8 C–D, respectively.