A chemoproteomic platform to quantitatively map targets of lipid-derived electrophiles

Abstract

Cells produce electrophilic products with the potential to modify and affect the function of proteins. Chemoproteomic methods have provided a means to qualitatively inventory proteins targeted by endogenous electrophiles; however, ascertaining the potency and specificity of these reactions to identify the sites in the proteome that are most sensitive to electrophilic modification requires more quantitative methods. Here we describe a competitive activity-based profiling method for quantifying the reactivity of electrophilic compounds against >1,000 cysteines in parallel in the human proteome. Using this approach, we identified a select set of proteins that constitute 'hot spots' for modification by various lipid-derived electrophiles, including the oxidative stress product 4-hydroxy-2-nonenal (HNE). We show that one of these proteins, ZAK kinase, is labeled by HNE on a conserved, active site-proximal cysteine and that the resulting enzyme inhibition creates a negative feedback mechanism that can suppress the activation of JNK pathways normally induced by oxidative stress.

Figure 1

Competitive isoTOP-ABPP for quantitative mapping of cysteine-lipid-derived electrophile (LDE) reactions in proteomes. (a) Competitive isoTOP-ABPP involves treatment of proteomes with DMSO or LDE, proteome labeling with an IA-alkyne (IA) probe, CuAAC-based incorporation of isotopically-labeled, TEV protease-cleavable biotin tags, enrichment with streptavidin, and sequential on-bead protease digestions to afford probe-labeled peptides for MS analysis. (b) Structures of the IA-probe and the competitive blockade of IA-cysteine reactions by an LDE. (c) Structures of three LDEs, HNE, 15d-PGJ2 and 2-HD, used in competitive isoTOP-ABPP experiments with their sites of reactivity marked with asterisks.

Figure 2. Quantitative profiling of LDE-cysteine reactions in proteomes

(a) Distribution of competitive isoTOP-ABPP ratios (R values) quantified from reactions with the human MDA-MB-231 proteome treated with 100 μM HNE (left, 5 technical replicates from 4 biological replicates), 15d-PGJ2 (middle, 6 technical replicates from 5 biological replicates), or 2-HD (right, 4 technical replicates from 3 biological replicates). A cut-off of five-fold or greater blockade of IA-probe labeling (R values > 5) is shown by a dashed line to mark cysteines that exhibit high sensitivity to LDEs, and proteins with cysteines showing the strongest competitive reactivity with LDEs are labeled by names. (b) Heat map of cysteines with R values > 5 illustrating examples of cysteines that display selectivity for reacting with one of the three tested LDEs. (c) Representative MS1 profiles for peptides containing cysteines that show selective competition with HNE (left) or 15d-PGJ2 (right). (d) Representative MS1 profiles for multiple cysteine-containing peptides from the same protein, only one of which shows sensitivity to LDE competition.

Figure 3

Determining the potency of HNE-cysteine reactions in proteomes and in cells. (a) Box-and-whisker (maximum 1.5 IQR) plots showing the distribution of R values for ~1100 cysteines quantified from competitive isoTOP-ABPP experiments with the MDA-MB-231 proteome treated with 5, 10, 50, 100, and 500 μM HNE. The IA-labeling of cysteines from ZAK, EEF2, RTN4, and FN3KRP exhibit exceptional sensitivity to HNE competition compared to the rest of the cysteines in the proteome. The data reported here is representative of an experiment run in duplicates. (b) Representative MS1 profiles for HNE-sensitive cysteines in ZAK and RTN4 showing concentration-dependent blockade of IA-labeling by HNE. (c) Distribution of R values quantified from competitive isoTOP-ABPP experiments with proteomes from MDA-MB-231 cells treated in situ with DMSO or HNE (100 μM, 60 min), confirming that cysteines in ZAK, EEF2, RTN4 and FN3KRP are also highly sensitive to HNE competition in living cells (2 biological replicates). (d) Comparison of R values obtained from in vitro versus in situ competitive isoTOP-ABPP experiments. Red and black diamonds mark cysteines that show similar or different in vitro versus in situ R values, respectively.

Figure 4

Functional characterization of HNE modification of ZAK kinase. (a) Crystal structure of human MAP3K9 (left, PDB: 3DTC) and multiple sequence alignment of ZAK with other 19 human MAP3Ks showing the HNE-sensitive cysteine C22 of ZAK is located next to the kinase’s ATP binding loop (“P-loop”; note that C22 corresponds to I150 in MAP3K9) and is unique to ZAK relative to other MAP3K enzymes. (b) Selective IA-labeling of wild-type (WT), but not C22A-ZAK, and concentration-dependent competition of IA-labeling of WT-ZAK by HNE as measured by gel-based ABPP using an IA-rhodamine probe. ZAK were expressed as FLAG-tagged proteins in HEK293T cells by stable transfection and immunoprecipitated prior to IA-probe labelling and analysis. (c) An HNE-alkyne probe (HNEyne) selectively labels WT-, but not C22A-ZAK in proteomes and in living cells as determined by gel-based ABPP. For b and c, the experiments were repeated three times with consistent results. (d) Catalytic activity of immunoprecipitated WT-, but not C22A-ZAK is inhibited by HNE as measured using a Myelin Basic Protein (MBP) substrate assay. A K45M-ZAK mutant, in which a conserved active-site lysine was mutated, showed no detectable activity and thus served as a catalytically dead control enzyme. All three ZAK variants (WT, C22A, and K45M were expressed at similar levels in transfected HEK293T cells; Supplementary Fig. 4). Data are presented as mean values ± s.e.m.; N = 3 experiments/group. **, P < 0.01, ***, P < 0.001, t-test, two sided. (e) SILAC-ABPP of acylphosphate-ATP probe reactivity for ZAK-transfected HEK293T cell proteomes pre-treated with DMSO (heavy cell proteome) or HNE (light cell proteome; 100 μM, 30 min) showing that the probe interaction with ZAK, but not other proteins, is substantially impaired by HNE treatment.

Figure 5

HNE modification of ZAK suppresses JNK pathway activation in cells. (a) WT-ZAK-transfected HEK-293T cells show higher basal JNK activation compared to mock-, C22A-ZAK-, or K45M-ZAK-transfected. (b,c) Western blots (b) and normalized phosphorylated JNK levels (c) showing that H₂O₂ treatment (1 mM, 30 min) increases JNK activation in WT- and C22A-ZAK cells and this increase is blocked or amplified in WT- and C22A-ZAK cells, respectively by pre-treatment with HNE (100 μM, 30 min). (d) A model diagramming ZAK-dependent and ZAK-independent pathways for HNE modulation of JNK activation. Dashed line designates the potential for oxidative stress to generate HNE and initiate a negative feedback loop to limit JNK activation. (e,f) Western blots (e) and normalized phosphorylated JNK levels (f) showing dramatic, concentration-dependent activation of JNK by HNE (50 or 100 μM, 60 min) in C22A-ZAK cells, but not in WT-ZAK cells. Note that mock- and K45M-ZAK transfected cells also show modest, but significant elevations in JNK activity following HNE treatment, which is consistent with previous studies indicating that HNE can activate JNK by multiple pathways. For a, c and f, data are presented as mean values ± s.e.m.; N >= 4 experiments/group (biological replicates). *, P < 0.05; **, P < 0.01; ^##, P < 0.01; t-test, two-sided.