A comparative cross-linking strategy to probe conformational changes in protein complexes

Abstract

Chemical cross-linking, together with mass spectrometry (MS), is a powerful combination for probing subunit interactions within static protein assemblies. To probe conformational changes in response to stimuli, we have developed a comparative cross-linking strategy, using lysine-specific deuterated and nondeuterated bis(sulfosuccinimidyl)suberate cross-linking reagents (BS3). Here we describe the experimental procedures as well as the data analysis, validation and interpretation. The protocol involves first assigning cross-linked peptides in the complex without ligand binding, or with post-translational modifications (PTMs) at natural abundance, using a standard procedure with labeled cross-linkers, proteolysis and assignment of cross-linked peptides after liquid chromatography-tandem MS (LC-MS/MS) and database searching. An aliquot of the protein complex is then exposed to a stimulus: either ligand binding or incubation with a phosphatase or kinase to bring about changes in PTMs. Two solutions--one containing the apo/untreated complex and the other containing the enzymatically modified/ligand-bound complex--are then cross-linked independently. Typically, nondeuterated BS3-d0 is used for the untreated complex and deuterated BS3-d4 is used for the experiment. The two aliquots are then incubated at equal concentrations, digested and processed as before. The ratios of labeled and unlabeled cross-linked peptides provide a direct readout of the effect of the stimulus. We exemplify our method by quantifying changes in subunit interactions induced by dephosphorylation of an ATP synthase. The protocol can also be used to determine the conformational changes in protein complexes induced by various stimuli including ligand/drug binding, oligomerization and other PTMs. Application of the established protocol takes ~9 d, including protein complex purification.

Figure 1. Comparative cross-linking workflow

(i) The first step is to assign cross-links in the protein complex without perturbation or ligand binding. The protein complex is cross-linked using a 1:1 mixture of BS3-d0/d4. (ii) For comparative cross-linking, the same protein complex present in different conformations is cross-linked with non-deuterated (d0) and deuterated (d4) BS3, respectively. Both samples are subsequently pooled in equal quantities. Gel electrophoresis confirms cross-linking of the protein subunits by appearance of bands at higher molecular weight. (iii) The proteins are digested with trypsin and the peptide mixture containing cross-linked and non-cross-linked peptides is separated by nanoLC. (iv) Peptides are directly eluted into a mass spectrometer and peak pairs with a specific mass difference (Δm = 4Da) are indicative of light (d0) and heavy (d4) cross-linked peptides. (v) MS/MS and database searching identifies the cross-linked peptides. (vi) Ratios of the peptides corresponding to the different conformations of the protein complex (d0:d4) are determined by generation of extracted ion chromatograms (XICs) for both light (d0) and heavy (d4) cross-linked peptides.

Figure 2. Schematic of Protocol and approximate timing

Comparative cross-linking protocol and timing of each step starting with the isolation of the complex and considering the alternative strategies for cross-linking and separation of cross-linked peptides is given. The time taken for data analysis is also considered and possible pause points throughout the protocol are highlighted in red.

Figure 3. Comparative cross-linking to probe conformational changes

Protein interactions affected by PTMs, oligomerisation, activation/deactivation and ligand binding are shown. Cross-linking intensities are obtained from peak pairs in mass spectra representing intensities before and after stimulating the protein complex. Changes in protein-interaction intensities are shown as dotted lines: no change (black), increased cross-linking intensities (green) and reduced cross-linking intensities (red).

Figure 4. Data analysis workflow

BS3-d0 and BS3-d4 cross-linked peptides co-elute during LC-MS/MS analysis (i). MS and MS/MS spectra are submitted to database searching (ii) yielding a list of potential cross-links (PXLs). MS spectra are inspected for specific peak pairs corresponding to BS3-d0 and BS3-d4 cross-linked peptides (iii). Spectra without peak pairs are rejected from further analysis (false positives, FP) (iv). MS/MS spectral quality of precursors with specific peak pair is checked (v) and spectra of poor quality are rejected (false positives, FP) (vi). The false discovery rate is calculated from the number of false positives (FP) and the total number of potential cross-linked peptides (PXLs) (vii). For comparative cross-linking XICs are generated for both the light (d0) and heavy (d4) cross-linked peptides (viii). d0/d4 cross-linking ratios are calculated (ix). Protein-protein interactions that change upon stimulation of the protein complex show d0:d4 fractions that are < or > 1:1. (x). If the majority of d0:d4 ratios deviate from 1:1, normalization of the dataset can be performed (xi).

Figure 5. Gel electrophoresis of a cross-linked protein complex

20 μl of the cATPase (approx. 20 μM) were cross-linked with 1 mM BS3. The gel shows the un-cross-linked and the cross-linked protein complex. Protein bands of covalently linked protein subunits are visible at higher molecular weight.

Figure 6. Comparative cross-linking of the cATPase

The cATPase was treated with a phosphatase and cross-linked with BS3-d4, while the untreated cATPase was cross-linked with BS3-d0. Both forms were pooled in a 1:1 ratio and proteins were separated by gel electrophoresis. After tryptic digestion and LC-MS/MS analysis, cross-links were identified by a database search (a). Protein interactions were compared by generating XICs and calculation of the ratios of their peak areas. Cross-links that showed equal intensities in MS spectra and in XICs represent protein interactions that are not affected by dephosphorylation (b). Cross-links showing reduced intensities in the dephosphorylated cATPase are identified by lower d4 intensities in MS spectra and smaller d4 XIC peak areas (c).

Figure 7. Summary of comparative cross-linking of phosphorylated and dephosphorylated cATPase

Interactions within the untreated (phosphorylated, phos) and dephosphorylated (dephos) cATPase are represented with dotted lines. Changes in cross-linking intensities are colour-coded according to the legend. Phosphosites are shown space filled (yellow), cross-linked residues (red) and the nucleotide binding site (red, yellow circle). Interactions between the α/β head and I, II, γ, δ, and the extended helix of ε are reduced after dephosphorylation. Interactions between ε and the membrane ring (III) are not affected. Interactions on top of the α/β head did not change or showed only small changes. Interactions at the base were dramatically reduced inducing nucleotide release.