A robust and versatile mass spectrometry platform for comprehensive assessment of the thiol redox metabolome

Abstract

Several diseases are associated with perturbations in redox signaling and aberrant hydrogen sulfide metabolism, and numerous analytical methods exist for the measurement of the sulfur-containing species affected. However, uncertainty remains about their concentrations and speciation in cells/biofluids, perhaps in part due to differences in sample processing and detection principles. Using ultrahigh-performance liquid chromatography in combination with electrospray-ionization tandem mass spectrometry we here outline a specific and sensitive platform for the simultaneous measurement of 12 analytes, including total and free thiols, their disulfides and sulfide in complex biological matrices such as blood, saliva and urine. Total assay run time is < 10 min, enabling high-throughput analysis. Enhanced sensitivity and avoidance of artifactual thiol oxidation is achieved by taking advantage of the rapid reaction of sulfhydryl groups with N-ethylmaleimide. We optimized the analytical procedure for detection and separation conditions, linearity and precision including three stable isotope labelled standards. Its versatility for future more comprehensive coverage of the thiol redox metabolome was demonstrated by implementing additional analytes such as methanethiol, N-acetylcysteine, and coenzyme A. Apparent plasma sulfide concentrations were found to vary substantially with sample pretreatment and nature of the alkylating agent. In addition to protein binding in the form of mixed disulfides (S-thiolation) a significant fraction of aminothiols and sulfide appears to be also non-covalently associated with proteins. Methodological accuracy was tested by comparing the plasma redox status of 10 healthy human volunteers to a well-established protocol optimized for reduced/oxidized glutathione. In a proof-of-principle study a deeper analysis of the thiol redox metabolome including free reduced/oxidized as well as bound thiols and sulfide was performed. Additional determination of acid-labile sulfide/thiols was demonstrated in human blood cells, urine and saliva. Using this simplified mass spectrometry-based workflow the thiol redox metabolome can be determined in samples from clinical and translational studies, providing a novel prognostic/diagnostic platform for patient stratification, drug monitoring, and identification of new therapeutic approaches in redox diseases.

Keywords: Glutathione; Hydrogen sulfide; Oxidative stress; Persulfides; Reactive species interactome; Redox status; Thiol-maleimide michael addition.

Copyright © 2018 The Authors. Published by Elsevier B.V. All rights reserved.

Figures

Fig. 1

Development and optimization of the analytical procedure for detection of the thiol redox metabolome. (A) Metabolic pathways defining the thiol redox metabolome. (B) Reaction of NEM with sulfide and aminothiols (using cysteine as an example); the two positions where the sulfhydryl group can add to the double bond (Michael addition reaction with sulfur adding either to C3 or C4 of the maleimide ring) leading to the formation of two distinct diastereomers are indicated (with sulfide four different diastereomers can be formed). (C) Chromatographic separation and selective detection by tandem mass spectrometry of authentic stock solutions of all analytes using full registration for the entire run. (D) Chromatograms of stable isotope labelled internal standards. (E) Representative chromatogram of the same analytes at their natural abundance in human plasma using specific time windows for selected groups of compounds. (F) Linearity of detector response for main analytes (n = 3).

Fig. 2

Effects of pH on the measurement of aminothiols and sulfide using NEM. The effect of pH was investigated in a simple aqueous buffer (A) and in plasma. (B). (A) Standard curves were prepared at four different pH values: 6, 7.4 (physiological), 8 and 9, using NEM in ammonium phosphate buffer of the appropriate pH; the insets represent the percentual changes of peak area/internal standard peak area as compared to the same values obtained at pH 7.4 (black bars) for a 1 µM standard concentration. The largest impact of pH is seen for Cys with a tenfold decrease in peak areas from pH 6 to pH 9, whereas sulfide shows the least difference between the different pHs. Data are from 2 independent measurements. For GSH and sulfide the difference between groups was not significant. 1-way ANOVA p < 0.01; Dunnet's v. pH 7.4 (black bar) * p < 0.05. (B) NEM in ammonium phosphate buffer at each pH was added to fresh plasma samples. The percentual changes of peak area/internal standard area is presented relative to those at pH 7.4 (black bars), which was considered as a control group. Contrary to the analyses carried out in buffer (shown in panel A), the largest differences in measured peak areas were seen with sulfide, with large increases above pH 7.4 as well as a slight increase at pH 6. Data are from 6 independent samples. For GSH, and HCys the differences among the groups were not significant. 1-way-ANOVA p < 0.001; Dunnet's v. pH 7.4 (black bar) * p < 0.05.

Fig. 3

Optimization of sample preparation procedure: anticoagulation, and stabilization with NEM. For all experiments depicted, NEM (or other thiol alkylating agent) was added directly to whole blood (panels A-D) after which the sample was centrifuged at 800 ×g for 10 min at room temperature. In panel E NEM was added to plasma directly after separation from blood cells by centrifugation of whole blood. (A) The choice of anticoagulant affects the concentrations of thiol measured; because of its metal chelating properties EDTA is the most suitable anticoagulant for assessment of the thiol redox metabolome. 2-way ANOVA p < 0.0019, Dunnet's vs. EDTA * p < 0.01 (B) Delay in addition of NEM to whole blood leads to progressive decreases in GSH and CysGly concentrations as compared to time = 0. (C) The apparent concentrations of GSH and sulfide increase with increasing concentrations of NEM added to whole blood, as compared to whole blood treated with 1 mM NEM as a control (centrifugation 800 ×g x 10 min). (D) The concentration of sulfide detected increases with increasing concentrations of alkylating agents added to whole blood. (E) The NEM concentration-dependent increases of GSH levels (as compared to samples treated with 1 mM NEM) were not observed when NEM was added to plasma (although also absolute GSH concentrations were considerably lower), suggesting that these increases were largely due to leakage of GSH from blood cells; however, apparent plasma sulfide concentrations still increased with increasing NEM concentrations, suggesting removal of sulfide from bound forms in plasma at elevated concentrations of the alkylating agent. For all panels data were obtained by analysis of two independent biological samples taken form different human individuals (mean ± SD); measurements were carried out at least in duplicate.

Fig. 4

Optimization of sample preparation procedure: effects of centrifugation speed and duration, effects of freeze/thawing and sample stability. (A,B) NEM addition to whole blood (shown in A) increases stability of thiols in comparison to NEM addition to plasma (shown in B), as demonstrated by comparing overall absolute concentrations of the same thiols in A vs. B. However, when low speed/long duration centrifugation (800 ×g, 10 min) was chosen to separate plasma form blood cells this procedure artificially increases plasma GSH concentrations (shown in A, pink bar vs. blue bar), probably due to leakage of NEM-adducts form cellular blood components (RBCs); see main text. (C) Freezing of NEM stabilized plasma leads to an increase in GSH, Cys and CysGly concentrations compared to their levels in fresh samples; concentrations of GSH dramatically decreased, and GSSG increased when samples were frozen without stabilization by NEM (n = 3; 2-way RM ANOVA p < 0.001; Dunnet's vs. fresh plasma p < 0.01). (D) In samples stabilized with NEM freeze/thawing or maintenance of samples in thermostatted autosampler at 5 °C did not affect concentrations of the redox thiol metabolome (n = 3, differences among groups are not significant). For all panels 2 or 3 independent biological samples from different individuals were analyzed; measurements were carried out in triplicates. (For interpretation of the references to color in this figure legend, the reader is referred to the web version of this article).

Fig. 5

Determination of the accuracy of the method (Method A) by comparison with an independent method optimized for GSH/GSSG detection (Method B) in plasma samples from 10 healthy human volunteers. In Method A detection of the thiol redox metabolome was carried out after ultrafiltration, separation on the Aqua UPLC column and detection by triple-quadrupole mass spectrometry. In Method B, thiols and disulfides were measured following deproteinization with sulfosalicylic acid, separation on a C18-UPLC column, and detection by Q-ToF mass spectrometry. (A) Concentrations of GSH are grossly underestimated by Method B, while Cys, HCys and HCysSS are overestimated by Method B compared to Method A (n = 10; 2-way RM ANOVA Method A vs. B p < 0.0001; Sidak's multiple comparison test p < 0.01; # T-Test-p < 0.01). (B) Bland-Altman plots of the absolute differences in GSH and GSSG concentrations as assessed by Method A - Method B in relation to the mean values of the analytes. (C) Effects of freeze/thawing were absent when Method B (with SSA deproteinization) was used (n = 3). Both methods revealed very similar values when the same sample clearing/deproteinization procedures were applied. For all panels 3–10 independent biological samples were analysed; measurements were carried out in triplicates.

Fig. 6

Assessement offree,totalandacid-labilethiols in human blood cells, urine and saliva. (A) Workflow for determination of free (1), total (2) and acid-labile thiols (3) in biological samples such as blood plasma (Table 2), blood cells (panel B) and other biofluids (panels C,D). (B-D) For the determination of free, total and acid-labile thiols in blood cells (depicted in B), urine (shown in C) and saliva (shown in D), samples were divided into 3 aliquots; one aliquot was used for determination of free thiols, one for total thiols following DTT reduction, and one for acid-labile thiols after addition of hydrochloric acid (HCl) and subsequent neutralization by sodium hydroxide (NaOH). All measurements were carried out in triplicate. Data in B are means from individually analysed cell pellets of the same 10 volunteers of whom circulating plasma thiol concentrations were analysed for inclusion in Table 2; data in panels C and D were from pooled saliva and urine samples of 5 healthy volunteers.

Scheme 1

Mechanism of the reaction between N-ethylmaleimide and thiols (RSH) in aqueous solution. R- alkyl chain with additional functional groups (as in aminothiols), hydrogen (as for sulfide) or sulfur chains of various length (as in polysulfides).