Chemoselective umpolung of thiols to episulfoniums for cysteine bioconjugation

Abstract

Cysteine conjugation is an important tool in protein research and relies on fast, mild and chemoselective reactions. Cysteinyl thiols can either be modified with prefunctionalized electrophiles, or converted into electrophiles themselves for functionalization with selected nucleophiles in an independent step. Here we report a bioconjugation strategy that uses a vinyl thianthrenium salt to transform cysteine into a highly reactive electrophilic episulfonium intermediate in situ, to enable conjugation with a diverse set of bioorthogonal nucleophiles in a single step. The reactivity profile can connect several nucleophiles to biomolecules through a short and stable ethylene linker, ideal for introduction of infrared labels, post-translational modifications or NMR probes. In the absence of reactive exogenous nucleophiles, nucleophilic amino acids can react with the episulfonium intermediate for native peptide stapling and protein-protein ligation. Ready synthetic access to isotopologues of vinyl thianthrenium salts enables applications in quantitative proteomics. Such diverse applications demonstrate the utility of vinyl-thianthrenium-based bioconjugation as a fast, selective and broadly applicable tool for chemical biology.

Fig. 1. Biofunctionalization with vinyl thianthrenium reagents.

a, The vinyl thianthrenium reagents VTT and VTFT utilized in this work. b, Strategy for the in situ activation of Cys based on site-selective linchpin formation in the presence of other amino acids and exogenous nucleophiles. c, Proposed mechanism for the umpolung of thiols based on the KIE determined via a competition experiment between VTFT and VTFT-d₃ and the second-order rate constant for the functionalization of GSH with VTFT. d, Bioconjugation of sfGFP S147C with VTFT and sodium azide. The following publicly available protein structure was used: sfGFP (PDB:2B3P). His, histidine; Lys, Lysine; Tyr, tyrosine; Nu, nucleophile.

Fig. 2. Selectivity of bioconjugation with vinyl thianthrenium salts.

a, Three-dimensional representation of the average ¹H and ¹³C NMR shift perturbations of each amino acid after functionalization of ¹³C,¹⁵N-labelled ubiquitin T12C and the NOESY correlation plot of scaled cross-peak heights before and after functionalization of ubiquitin T12C. The linear regression of the experimental data is indicated as a solid line. b, Mean number of modified residues in E. coli lysate resulting from reaction with VTT/VTFT and sodium azide at different pH values. Data are derived from two technical replicates. TCEP was used for 1 h pre-reduction at 37 °C. c, Reactivity of DHAR1 mutants with VTT and N-methylmaleimide (NMM) and enzyme assay subsequent to the modification. The data are expressed as mean ± s.e. of nine independent experiments. The statistical analysis was performed via a two-sided analysis of variance, followed by post hoc Tukey test with P < 0.05 (see Supplementary Tables 11 and 12 for exact P values). The following publicly available protein structures were used: ubiquitin (PDB:1D3Z) and DHAR1 (PDB:5EL8). Source data

Fig. 3. Scope of vinyl thianthrenium salts as bioconjugation reagents.

a, Scope of proteins used for conjugation reactions with NaN₃. b, Scope of functionalities introduced with VTT/VTFT. ^a40 equiv. VTFT, yield not determined; ^byield not determined; ^c10 mM Nu; ^d3 mM Nu, 35 equiv. VTT; ^e40 mM sodium iodide as nucleophile with subsequent addition of 40 mM S-centred nucleophiles after 3 min; ^fubiquitin T9C, HEPES (8.0, 50 mM), 40 mM sodium iodide as nucleophile with subsequent addition of 20 mM S-centred nucleophile after 2 min, yield, 74%. The following publicly available protein structures were used: sfGFP (PDB:2B3P), DHAR2 (PDB:5LOL), ubiquitin (PDB:1UBQ), Trxh1 (PDB:1XFL), DHAR1 (PDB:5EL8), BSA (PDB:3V03). MDAR, monodehydroascorbate reductase; PrxIIB, type 2 peroxiredoxin; Trxh1, thioredoxin-h1; DTT, dithiothreitol.

Fig. 4. Utilization of vinyl thianthreniums for stapling and macrocyclization of peptides.

a, Stapling of nucleophilic amino acids in native peptides with VTT. ^aMeCN:H₂O 9:1, 2.2 equiv. NEt₃. ^bDMF:H₂O 1:1, 5.0 equiv. NEt₃, ^cMeCN:H₂O 9:1, 3.0 equiv. NEt₃. b, Macrocyclization of reduced disulfide bonds with VTFT. ^dNaP_i (pH 6.5, 50 mM). AA, amino acid.

Fig. 5. Applications of VTT in proteomics.

a, Application of VTT isotopologues in quantitative proteomics. Reaction with control and heat-shock lysates from E. coli cultures incubated at 37 °C or 43 °C, respectively, prior to lysis; mean of MS¹ intensities from Cys-containing peptides functionalized with isotopologues of VTT and their respective protein origins. Data are derived from two technical replicates; volcano-plot with auxiliary lines resulting from two-sided linear models for microarray data (LIMMA)-based differential expression analysis with P < 0.05 derived from MS¹ intensities of identified peptides based on two technical replicates. All P values are associated with empirical Bayes moderated t-statistics without adjustments for multiple comparisons. The horizontal auxiliary line indicates the limit of significance based on P < 0.05. The vertical auxiliary lines indicate the values at which the protein populations have at least halved or doubled. b, Identified and simulated PPI between poly(A) polymerase I and lysine-sensitive aspartokinase 3 via VTT induced cross-linking in an E. coli lysate. MS¹, ion spectra of intact peptides. Source data