Stochastic palmitoylation of accessible cysteines in membrane proteins revealed by native mass spectrometry

Abstract

Palmitoylation affects membrane partitioning, trafficking and activities of membrane proteins. However, how specificity of palmitoylation and multiple palmitoylations in membrane proteins are determined is not well understood. Here, we profile palmitoylation states of three human claudins, human CD20 and cysteine-engineered prokaryotic KcsA and bacteriorhodopsin by native mass spectrometry. Cysteine scanning of claudin-3, KcsA, and bacteriorhodopsin shows that palmitoylation is independent of a sequence motif. Palmitoylations are observed for cysteines exposed on the protein surface and situated up to 8 Å into the inner leaflet of the membrane. Palmitoylation on multiple sites in claudin-3 and CD20 occurs stochastically, giving rise to a distribution of palmitoylated membrane-protein isoforms. Non-native sites in claudin-3 indicate that membrane-protein function imposed evolutionary restraints on native palmitoylation sites. These results suggest a generic, stochastic membrane-protein palmitoylation process that is determined by the accessibility of palmitoyl-acyl transferases to cysteines on membrane-embedded proteins, and not by a preferred substrate-sequence motif.

Fig. 1

Native mass spectra of micelle-released poly-palmitoylated claudins. a Spectrum of Cld3 in complex with C-CPE, showing a mass distribution corresponding to Cld3 harboring zero to five attached palmitates. Gray boxes contain Cld3 peaks with the corresponding charge state annotated. Peaks labeled with an asterisk represent C-CPE peaks. All panels in this figure show representative data from biological triplicates. b Close up of the 12 + Cld3 charge state with the number of palmitates bound to Cld3 annotated for each peak. c Deconvoluted native ESI-MS spectra of poly-palmitoylated Cld4 and Cld6. The Cld6 peaks annotated with a dot correspond to a Cld6 mass without its initiator methionine removed. d Deconvoluted spectra of native Cld3 and of Cld3 with putative palmitoylated cysteines mutated to alanine one by one, showing that mutation of Cys 103, 106, 182, or 184 to Ala result in the loss of a single palmitate, whereas mutation of Cys 181 does not

Fig. 2

Palmitoylation status of cysteine-scanned Cld3. a Palmitoylation of single cysteines introduced in Cld3 with all putative palmitoylated Cys 103, 106, 181, 182, and 184 mutated to Ala (ΔCys). The average palmitoylation state is obtained by using integrated peak areas of the deconvoluted spectra of intact micelle-released Cld3. No palmitoylation could be observed for Leu 180 Cys because of low expression of this mutant. A palmitoylation state of 0.2 was used as threshold value to classify the introduced Cys as palmitoylation site. Data are presented as mean values of biological triplicates and independent data points are shown as open circles. b Introduced cysteine residues mapped on the Cld3 homology model. In all panels, the Cα atom of each cysteine is represented as a sphere colored in a red–white–blue gradient normalized according to its palmitoylation level, in which full palmitoylation is shown as blue, 50% palmitoylation as white no palmitoylation as red. c Cysteine residues introduced in helix 4 are not palmitoylated when located further than 8 Å into the membrane. The membrane depth is the measured distance of the Cα atom of the introduced cysteine to the average position of all phosphor atoms of the inner leaflet after superimposition of the Cld3 model to mCLD19 in MemprotMD. Palmitoylation data are presented as mean values of biological triplicates and error bars represent standard deviations. d Left panel, structural side view of Cld3. Right panel, cytoplasmic view of Cld3 with palmitoylated Cα atoms in shown as spheres. The majority of palmitoylated residues are exposed

Fig. 3

Quantitative analysis of Cld3 and CD20 palmitoylation. Cartoon representation of the Cld3 model (a) and schematic representation of CD20 (b). Probability of cysteines to be palmitoylated in Cld3 is p ₁₀₃ = 0.69, p ₁₀₆ = 0.69, p ₁₈₂ = 0.86, p ₁₈₄ = 0.64 and in CD20 is p ₁₁₁ = 0.74, p ₂₂₀ = 0.82. Cβ atoms of palmitoylated cysteines shown as spheres and colored according to the probability to be palmitoylated in a white–blue gradient, in which full palmitoylation is shown as blue and no palmitoylation as white. b, d Comparison of the observed and calculated protein fractions for each palmitoylation state (P(n)) of wild-type Cld3 (b) and CD20 (d) after fitting the cooperative stochastic model to palmitoyl distributions of all Cld3 or CD20 constructs

Fig. 4

Cysteine palmitoylation is restricted by palmitoyl-acyl transferases (PAT) accessibility. PATs are restricted to lateral diffusion for its catalytic DHHC domain to interact with substrates. The membrane orientation of Cld3 prevents PATs to access palmitoylation sites located on helix 3, whereas palmitoylation sites on helix 1, 2, and 4 are accessible