Cyclic 5-membered disulfides are not selective substrates of thioredoxin reductase, but are opened nonspecifically

Abstract

The cyclic five-membered disulfide 1,2-dithiolane has been widely used in chemical biology and in redox probes. Contradictory reports have described it either as nonspecifically reduced in cells, or else as a highly specific substrate for thioredoxin reductase (TrxR). Here we show that 1,2-dithiolane probes, such as "TRFS" probes, are nonspecifically reduced by thiol reductants and redox-active proteins, and their cellular performance is barely affected by TrxR inhibition or knockout. Therefore, results of cellular imaging or inhibitor screening using 1,2-dithiolanes should not be interpreted as reflecting TrxR activity, and previous studies may need re-evaluation. To understand 1,2-dithiolanes' complex behaviour, probe localisation, environment-dependent fluorescence, reduction-independent ring-opening polymerisation, and thiol-dependent cellular uptake must all be considered; particular caution is needed when co-applying thiophilic inhibitors. We present a general approach controlling against assay misinterpretation with reducible probes, to ensure future TrxR-targeted designs are robustly evaluated for selectivity, and to better orient future research.

Fig. 1. 1,2-dithiolanes in chemical biology.

a Linear disulfide-based cellular delivery: irreversible cleavage of linear disulfides after cell entry leads to intracellular cargo release (Pr-SH: intracellular protein thiol). b From linear, to cyclic 6-membered, to increasingly ring-strained 5-membered disulfides, to ETPs. c Strained 1,2-dithiolanes in natural products. d Fast, irreversible and nonspecific thiol-disulfide interchange of 1,2-dithiolanes by exofacial thiols, followed by dynamic transmembrane exchange cascades, enhances cellular uptake and intracellular delivery. [^a:data from Matile et al.]. e Some of the 1,2-dithiolane-based “TRFS” probes and prodrugs that have been reported as selective cellular substrates of thioredoxin reductase [TRFS = “Thioredoxin Reductase Fluorogenic Substrate“^,,].

Fig. 2. 1,2-dithiolane probe design and synthesis.

a After opening or reduction of the 1,2-dithiolane in SS50-PQ, thiolate cyclisation releases the precipitating phenol PQ-OH that gives ESIPT-based fluorescence in the solid state. b Synthesis of SS50-PQ: (i) Boc₂O, NEt₃, dioxane/H₂O, r.t., 15 h (96%); then either MsCl, py, DCM followed by KSAc, acetone, 60 °C, 2 h (89%), or HSAc, PPh₃, DIAD, THF, 0 °C to r.t., 15 h (88%). (ii) KOH, MeOH, open to air, r.t., 15 h (98%). (iii) MeI, NaH, DMF, 0 °C to r.t., 0.5 h (70%). (iv) PQ-OH, triphosgene, NEt₃, DCM, 0 °C to r.t., 1 h; then 6, NEt₃, DCM, 0 °C to r.t., 1 h (49%).

Fig. 3. 1,2-dithiolane probes are activated by a range of chemical and biological reductants.

a GSH challenge. Fluorescence timecourses of SS50-PQ exposed to GSH, and the corresponding dose-response plot (t = 6 h at 37 °C). b Chemical reductant assays. Normalised signal from SS50-PQ challenged with monothiols (GSH, cysteamine (CA), cysteine (Cys), and mercaptoethyl-dimethylamine (MEDA) each at 1 mM), dithiol (1 mM dithiothreitol (DTT)), non-reductants (GSH disulfide (GSSG) and serine (Ser) each at 1 mM), or tris(2-carboxyethyl)phosphine (TCEP) (100 µM); showing all endpoint results (6 h at 37 °C) and selected kinetics. c Redox enzyme assays. Normalised signal from SS50-PQ and TRFS-green challenged with TrxR/Trx or GR/GSH/Grx network proteins (20 nM TrxR1/GR, 10 µM Trx1/Grx1/TRP14*, 100 µM GSH as indicated; 100 µM NADPH in all TrxR/GR assays). d Dose–response plots for redox effector proteins (20 nM TrxR1/TrxR2/GR, 0.03–10 µM Trx1/Grx1/TRP14*, 100 µM GSH as indicated; 100 µM NADPH in all TrxR/GR assays). (a–d: data are for single representative examples from either 2 (b) or 3 (a, c, d) independent experiments; probes at 10 µM in aqueous TE-buffer; notation TRP14* indicates pre-reduced TRP14 (see “Methods”).

Fig. 4. The phenolic carbamate design of SS50-PQ gives reliable performance across biological assays.

a Cellular fluorescence timecourses of SS50-PQ in HeLa and A549 cells (50 µM probe). b Dose-dependency of cellular fluorescence timecourses with SS50-PQ, as compared to TRFS-green (HeLa cells). (a, b: data as mean ± SD of ≥3 independent experiments). c Microscopy of SS50-PQ-treated HeLa cells shows fluorescent intracellular solid precipitates of released PQ-OH (representative images from 3 independent experiments with similar results). d Flow cytometry-based single-cell statistics of cellular fluorescence after SS50-PQ treatment (25 µM; Jurkat T-cells). (c,d: representative examples of ≥3 independent experiments with similar results). e Fluorescence imaging of embryonic zebrafish before and after SS50-PQ treatment (representative example of 3 independent experiments). (c, e: brightfield transmission image in greyscale, fluorescence superimposed in green).

Fig. 5. Cellular signal from 1,2-dithiolane probes does not report on TrxR.

a–e Cellular fluorescence timecourse studies of SS50-PQ and TRFS-green, with the TrxR-independent SS00-PQ and the TrxR-dependent RX1 as benchmarks for assay outcomes. a A549 cells starved of, or supplemented with, selenium (as Na₂SeO₃) (probes at 100 µM; full data in Supplementary Fig. 8a). A549 cells treated with Na₃PO₃S (probes at 100 µM; full data in Supplementary Fig. 8b). c TrxR1-knockout (−/−) and -wildtype (fl/fl) MEF cells (probes at 100 µM; full data in Supplementary Fig. 8c). d, e A549 cells pre-incubated for 2 h with the gold-free TrxR inhibitors TRi-1 and TRi-3 (probes at 50 µM; full data in Supplementary Fig. 10–13). (a–e: data shown as mean from 3 independent experiments; full representations in Supplementary Information). Note: plots on the same graphs are run under identical conditions except for the indicated variable (TrxR knockout/presence, etc.) so are kinetically comparable).

Fig. 6. The fluorescence of non-cargo-releasing 1,2-dithiolane Fast-TRFS activates in the absence of TrxR and of reductants, after partitioning into membranes: coherent with strain-promoted oligomerisation and environment-dependent fluorescence.

a Weak, environment-dependently fluorescent Fast-TRFS; and strong environment-dependently fluorescent compounds dithiol-TRFS (its reduction product) and Linear-TRFS (a reference compound). b The inhomogenous media of cellular assays may be closer modelled with lipid vesicle suspensions than as an all-aqueous system. c Extractive concentration of Fast-TRFS into lipid membranes may drive concentration-dependent, strain-promoted ring-opening polymerisation (ROP) to nonstrained polydisulfide PolyLinear-TRFS, whose fluorescence may mimic that of nonstrained monodisulfide Linear-TRFS. Thus, ROP may be a non-reductive mechanism for fluorescence turn-on of Fast-TRFS. d Fast-TRFS and Linear-TRFS (each 10 µM) incubated with lecithin vesicles. After 60 min, TCEP (0.1 mM) was added to benchmark complete reduction. (Representative example from 2 independent experiments; full data in Supplementary Fig. 16).