RedoxiFluor: A microplate technique to quantify target-specific protein thiol redox state in relative percentage and molar terms.

Unravelling how reactive oxygen species regulate fundamental biological processes is hampered by the lack of an accessible microplate technique to quantify target-specific protein thiol redox state in percentages and moles. To meet this unmet need, we present RedoxiFluor. RedoxiFluor uses two spectrally distinct thiol-reactive fluorescent conjugated reporters, a capture antibody, detector antibody and a standard curve to quantify target-specific protein thiol redox state in relative percentage and molar terms. RedoxiFluor can operate in global mode to assess the redox state of the bulk thiol proteome and can simultaneously assess the redox state of multiple targets in array mode. Extensive proof-of-principle experiments robustly validate the assay principle and the value of each RedoxiFluor mode in diverse biological contexts. In particular, array mode RedoxiFluor shows that the response of redox-regulated phosphatases to lipopolysaccharide (LPS) differs in human monocytes. Specifically, LPS increased PP2A-, SHP1-, PTP1B-, and CD45-specific reversible thiol oxidation without changing the redox state of calcineurin, PTEN, and SHP2. The relative percentage and molar terms are interpretationally useful and define the most complete and extensive microplate redox analysis achieved to date. RedoxiFluor is a new antibody technology with the power to quantify relative target-specific protein thiol redox state in percentages and moles relative to the bulk thiol proteome and selected other targets in a widely accessible, simple and easily implementable microplate format. Crown

Introduction

The ability of antibodies to avidly and selectively bind target proteins underpins fundamental measurement (e.g., immunoblot), manipulation (e.g., immuno-depletion), and therapeutic (e.g., vaccines) techniques [1]. Despite being a classic example of functionally essential structural disulphide (RSSR) riven proteins, antibodies have, compared to redox proteomics [[2], [3], [4]], seldom been used to assess redox signalling or develop biomarkers by measuring protein thiol redox state [5]. Instead, antibodies have mainly been used to measure antioxidant enzyme content [6], oxidative damage adducts [7,8], and protein-conjugated spin traps [9]. These uses are, however, of limited value for measuring reactive oxygen species (ROS) sensitive protein thiol defined redox signalling [10]. ROS sensitive electron exchange—shifting the balance between the reduced and oxidised (reversibly or irreversibly) chemotypes of the protein—can flip redox switches by changing protein activity, location, interactome, phase, and lifetime [[11], [12], [13], [14], [15], [16], [17], [18], [19]]. Given redox proteomics exists [20,21], it is reasonable to ask: why use antibodies to measure protein thiol redox state? Key reasons for using antibodies are five-fold:

Standalone. As several recent examples attest [[22], [23], [24], [25], [26]], antibody techniques can demonstrate a change in protein thiol redox state, which can be leveraged to discover a redox switch with systematic site-directed mutagenesis or tie the change to a functionally relevant biological output (e.g., enzyme activity). The standalone importance of immunological techniques is particularly useful when the target itself and/or many of its thiols are undetectable or difficult to detect using redox proteomics [5].

Synergy. Antibody techniques complement redox proteomics by providing an independent feedforward and feedback approach. For example, one might design a microplate protein array to select “hits” for follow-up redox proteomics or do the reverse to confirm a redox proteomic result. The latter being particularly useful for eliminating any false discovery rate concerns and for providing a simple and rapid means for others to measure a validated redox switch. Antibodies are important to redox proteomics in their own right because they can be used to enrich difficult to assess targets [27].

Biomarkers. Protein thiol biomarkers defined by a state- or process-specific target-specific redox change are of value to applied (i.e., systemic disease reporter) and basic (e.g., molecular sentinel of insulin signalling) research. When the task of pinpointing the individual thiols responsible for the redox state change is complete, it would be useful, provided the change manifests holistically (i.e., at the whole protein level), to measure the biomarker in a microplate. Developing microplate-based biomarker technologies is essential in the clinic, for example, to help appraise the therapeutic value of next-generation antioxidants [28].

Function. Antibody techniques can integrate a redox state change to a functional parameter. For example, by measuring protein thiol redox state and content—a redox-sensitive functional property [22,29,30].

Accessibility. The accessibility of antibody techniques is important because opening up the universal utility of measuring protein thiol redox state to anyone could accelerate discovery [31].

In theory, antibodies can be used in four ways to measure protein thiol redox state. First, antibodies designed to recognise the generic chemotype (e.g., sulfenic acids, RSOH, mixed disulfides with glutathione, RSSG, or S-nitrosylation, RSNO) or structurally distinct oxidised form of the protein (e.g., PTP1B) can be used in non-reducing immunoblotting [11]. Relatedly, RSSR-induced target dependent electrophoretic mobility shifts, as observed for typical 2-cys peroxiredoxin isoforms (i.e., the dimer assay [32,33]), in non-reducing immunoblotting can be used. Second, antibodies can detect electrophoretic mobility shifts induced by conjugating the target with thiol-reactive polyethylene glycol (PEG) payloads [[34], [35], [36], [37]]. Third, immunoprecipitation (IP) can enrich a target for downstream analysis: anti-chemotype blotting or streptavidin blotting of thiol-biotin conjugates [38]. Fourth, and most recently developed, ELISA techniques—the antibody-linked oxi-state assay (ALISA)—can measure target-specific protein thiol redox state in a microplate [31]. In essence, antibodies serve as a redox detector (direct:1&2) or capture handle to measure a redox reporter (indirect:3&4). In practice, only a small number of targets have proven to be amenable to direct strategies. That said, when direct strategies do work, they are invaluable as the dimer assay attests. Despite their efficacy and utility, no microplate assay can measure target-specific protein thiol redox state in percentages and moles relative to the global thiol proteome and selected other targets.

Achieving four key benefits would improve microplate techniques. Before proceeding, ALISA uses an immobilised capture antibody to bind a fluorescent maleimide (F-MAL) decorated target protein [31]. Target-specific protein thiol redox state is calculated by dividing the F-MAL signal by a total protein reporter. First, it would be useful to report the global thiol proteome redox state (i.e., global mode) to contextualise the observed target-specific findings [39]. Second, it would be invaluable to report target-specific redox state in percentages because it is difficult to interpret fold changes [21]. When an ELISA format is used (i.e., ELISA mode), percentage analysis would enable one to estimate target-specific redox state in moles—vital for advancing towards quantitative redox biology [[40], [41], [42]]. Third, improving strategies for using one antibody, when no matched pair is available, with protein A or G plates (i.e., protein A mode) is important. Presently, ALISA requires covalent immobilisation and an amine-reactive fluorescent N-hydroxysuccinimide (F-NHS), so one must elute the target to deconvolute the F-NHS signals. Finally, it would be instructive and useful to measure the redox state of several targets in parallel (i.e., array mode). Here, we present a new and improved microplate technique termed RedoxiFluor, which, as proof-of-principle experiments demonstrate, achieves four key benefits without sacrificing any of the advantages of the microplate format.

Results

Overview of RedoxiFluor

RedoxiFluor uses two thiol-reactive fluorescent reporters and a capture antibody functionalised solid support to measure target-specific protein thiol redox state in percentages and moles in a microplate (Fig. 1). Reduced thiols are labelled with a fluorescent maleimide (F-MAL1) reporter via a Michael addition reaction dependent thioether bond. Reversibly oxidised thiols (RSOX) are reduced to the maleimide reactive sulfhydryl state (RSH/RS−) using 1-4-dithiothreitol (DTT) mediated sulphur exchange reactions and labelled with a spectrally distinct fluorescent maleimide reporter (F-MAL2). Optionally, a specific chemotype (e.g., sulfenic acids, RSOH) can be labelled using selective reductants or a direct reactivity strategy [43,44]. Direct reactivity strategies can be used to label DTT irreducible sulfinic and sulfonic acids [45]. A capture antibody functionalised solid support is used to bind the target protein thiol from a biological sample (e.g., cell lysate). The thiol redox state encoded F-MAL reporters enable one to quantify target-specific protein thiol redox state in percentages (i.e., protein A mode). A biotin-conjugated detector antibody and recombinant protein standard curve (i.e., ELISA mode) enable one to calculate target-specific protein thiol redox state in percentages and moles. The percentage and molar terms are relative because the labelling is, for the purposes of safeguarding antibody binding, performed under native conditions. Accordingly, even if, 100% or near thiol labelling is achieved, solvent hidden groups will remain and so it cannot be absolute, but this applies to most techniques. A bonus of native labelling: it might stabilise certain reversibly oxidised thiols and keep some redox active transition metal ions protein bound [46]. Global mode RedoxiFluor (i.e., untargeted F-MAL1/2 analysis) can contextualise target-specific protein thiol redox state relative to the thiol proteome. Array mode protein A based RedoxiFluor can quantify the redox state of several proteins in percentages. In redox proteomics, dual fluorescent reporters have been extensively used to separate proteins in one or two dimensions by SDS-PAGE before mass spectrometry based peptide fingerprinting [[47], [48], [49], [50]]. In contrast, RedoxiFluor uses antibodies to separate the target and verify its identity so as to measure target-specific redox state in a microplate. RedoxiFluor, therefore, rests on solid and valid redox (fluorescent DIGE) and immunological (antibody and ELISA) principles.

Fig. 1

RedoxiFluor overview. Design. RedoxiFluor uses two spectrally distinct thiol-reactive fluorescent reporters to encode target-specific redox state. For example, reduced and reversibly oxidised thiols are labelled with green and red thiol-reactive fluorophores, respectively (see main text). A target-specific capture antibody immobilised in a microplate is used to selectively bind the target from a sample. Target-specific redox state can be quantified in relative percentage and molar terms in a microplate. Modes. RedoxiFluor can operate in protein A (percentages), ELISA (percentages and moles), array, and global (context) mode. Key redox and microplate related benefits of RedoxiFluor are shown. Key RedoxiFluor uses are illustrated. (For interpretation of the references to colour in this figure legend, the reader is referred to the Web version of this article.)

RedoxiFluor is a valid immunological technique for measuring target-specific redox state in relative percentage and molar terms

To measure target-specific redox state in percentages and moles, it is necessary to convert F-MAL signals to percentages in a mathematically valid and robust way. Mixing equimolar F-MAL1/2 standards to construct artificial redox states from 90 to 10% reversibly oxidised in l-cysteine buffer to exact any thiol-dependent turn-on fluorescence confirmed percentage analysis is possible (Supplementary Fig. 1). Experiments with fully F-MAL1/2 labelled recombinant bovine serum albumin (BSA) show that RedoxiFluor can discern between different redox states from 100, 75, 50, 25, to 0% reversibly oxidised (Supplementary Fig. 1). No effect of fluorophore labelling order is observed: the same result was obtained if F-MAL1 or F-MAL2 was used first (Supplementary Fig. 1). The labelling order experiment shows that the different fluorescent reporter payloads tested didn't sterically impair the thiol labelling capacity of the maleimide warhead.

Logically, to convert any F-MAL value to absolute percentages or moles, as opposed to relative quantification, F-MAL labelling must be complete (i.e., 100%). It must also be completely faithful: off-target labelling must be eliminated. As the extensive thiol-reactive literature attests (reviewed in Ref. [51]), neither logical requirement can be satisfied in the strictest possible terms. Although maleimide reacts quickly with thiols, off-target labelling of amines (e.g., lysine residues) becomes appreciable as the pH rises. Accordingly, one must endeavour to label as many thiols as possible while minimising off-target binding, which is practically achieved by titrating the amount of F-MAL and the reaction time [51]. We stress that this varies on a target-by-target basis and is a general drawback of using the current staple of thiol-reactive warheads [52,53]. Nevertheless, under the experimental reaction conditions, minimal F-MAL1 labelling (i.e., amine dependent) of pre-alkylated (i.e., N-ethylmaleimide, NEM, conjugated) BSA occurred (Supplementary Fig. 2). Consistent with previous work [54], we estimate F-MAL labelling, as determined by adding F-MAL2 after F-MAL1 labelling of reduced BSA, proceeded to near completion for BSA (95%) (Supplementary Fig. 2). Consistent with specificity, F-MAL minimally labelled a rabbit immunoglobin (IG) control, which is expected to have few free thiol groups. In contrast, substantial F-MAL labelling of the pre-reduced IG occurred (Supplementary Fig. 2).

From the above and the wider literature [51], unless both criteria are absolutely satisfied for the target absolute percentage and molar quantification cannot be achieved. Incidentally, it would also depend on strict reduced chemotype specificity and the ability of common thiol-reactive warheads to react with RSOH and protein persulfides (RSSH) is problematic [43]. RedoxiFluor like all assays can be updated as the chemical tool kit evolves (e.g., by using reduced sulfhydryl selective labelling agents). Nevertheless, relative percentage and molar quantification is valuable [21]. For RedoxiFluor to achieve relative percentage and molar quantification, the antibody must bind the target protein and the off-target binding (i.e., random) of proteins to the solid support must be minimal. To determine whether both criteria could be satisfied, we selected the catalytic subunit (i.e., PPP2CA, UniProt ID: P67775) of the serine/threonine protein phosphatase PP2A because it is a strategically important redox regulated protein [55]. After repeating the same labelling controls in complex samples derived from Xenopus laevis (X. laevis), their labelling proceeded to 98% (and practically it is difficult to know if the remaining 2% from F-MAL2 corresponds to off-target labelling of amines when there are no available thiol groups), and confirming the suitability of gel filtration chromatography (i.e., no reagent spill-over, Supplementary Fig. 2), we tested target and random binding using 50% F-MAL1/2 labelled X. laevis sample standards in protein A mode wherein a PP2A capture antibody is passively bound to a protein A derivatised plate. Consistent with target-specific binding, no discernible F-MAL signals were observed when standards were incubated with rabbit IG control or blank (i.e., protein A only) wells and immunodepleting PP2A from the standards abolished the signal (Supplementary Fig. 3).

Before proceeding to determine whether relative PP2A-specific redox state percentage analysis was possible in protein A mode, we tested other immunological techniques. Click-PEG wherein clickable PEG payloads are used to detect reversibly oxidised thiols as mass shifted bands by immunoblot [[34], [35], [36]] cannot report PP2A redox state (Fig. 2A). PEG decorated PP2A was undetectable [5]. ALISA works [31]. However, it is limited to fold changes (Fig. 2B). In contrast, RedoxiFluor can reliably and accurately report PP2A-specific redox state in percentages, as demonstrated by the tight correspondence between the known redox state of the experimental standard and the measured redox state (Fig. 2C and Table 1). Likewise, ELISA mode RedoxiFluor can accurately (e.g., the mean observed redox difference from the standard was 1.3%) and reproducibly (e.g., the mean CV value between samples was 4.1%) (Fig. 2D and Table 1) measure the redox state of PP2A in 10–90% reversible oxidised sample standards. Specificity is evidenced by the positive biotin-conjugated detector antibody dependent signal in sample standards and lack of any discernible signal in the PP2A immunodepleted, IG controls, and blanks (Supplementary Fig. 3). The recombinant protein standard curve showed the ability to detect PP2A in the picogram range (i.e., from 8,000–125 pg/ml, Supplementary Fig. 3). Importantly, F-MAL1 and F-MAL2 did not impact capture antibody binding compared to unlabelled sample controls as evidenced by mean pM values of 99.4, 101.9, and 96.7, respectively. Taking the 20 vs. 40% redox state as examples (compare to ALISA, Fig. 2B), picomoles of reduced PP2A were significantly greater in the 20 compared to the 40% reversibly oxidised redox state; as confirmed by a corresponding decrease in picomoles of reversibly oxidised PP2A (Fig. 2E–F). Overall, RedoxiFluor can measure target-specific redox state in relative percentages (i.e., protein A and ELISA mode) and moles (i.e., ELISA mode) in a microplate.

Fig. 2

RedoxiFluor can quantify target-specific protein thiol redox state in relative percentage and molar terms. A. Click-PEG cannot detect PP2A redox state, as evidenced by the loss of signal in the “PEGylated” lanes (3–5) compared to lysates (lane 1) and the PEG-free clickable maleimide handle only control (lane 2). B. ALISA detected a significant difference (unpaired t-test, P = 0.0192, n = 3) between the 20 and 40% redox states. C. Protein A mode RedoxiFluor can accurately and reproducibly discern between different PP2A redox states ranging from 10 to 90% reversibly oxidised (n = 3 per standard, see methods). D. ELISA mode RedoxiFluor can accurately and reproducibly discern between different PP2A redox states ranging from 10 to 90% reversibly oxidised (n = 3 per standard, see methods). A separate PP2A ELISA mode standard experiment quantifying significant differences (unpaired t-tests, P = < 0.0001 in panels E–G, n = 6) between the 20 (n = 6) and 40% (n = 6) reversibly oxidised states in percentages (E) and picomoles of reduced (F) and reversibly oxidised (G) protein. All standards and samples were derived from Xenopus laevis lysates (see methods). Data are presented as the mean (M) and standard deviation (SD).

Table 1

Measured PP2A-specific RedoxiFluor standard values in protein A and ELISA mode. Values reported are the mean (M), standard deviation (SD), mean observed difference from the standard (i.e., 2% if a mean of 48 was registered for the 50% standard), and coefficient of variation (CV) between standards. All values are reported in percentages. All CV values within standards (i.e., values from triplicate readings of the same sample) were less than 5%. All standards were derived from Xenopus laevis lysates (see methods).

Mode	Protein A				ELISA
Standard	M	SD	Difference	CV1	M	SD	Difference	CV
90	90.2	1.1	0.2	1.7	89.1	0.48	0.9	0.5
80	81.1	1.1	1.1	1.2	81.2	2.9	1.2	3.6
70	71.5	0.9	1.5	0.7	70.7	1.4	0.7	2
60	60.6	1.1	0.6	1.8	62.1	1.2	2.1	1.9
50	48	1.2	2	2.6	52.6	0.3	2.6	0.6
40	41.1	2.1	1.1	5.1	39.2	1.6	0.8	4
30	30.3	0.52	0.3	3	32	1.5	2	4.8
20	19.9	2.9	0.1	2.9	20.4	2	0.4	9.6
10	10.3	0.8	0.3	7.3	11.3	1.1	1.3	9.7
Mean	n/a	1.3	0.8	2.9	n/a	1.4	1.3	4.1

Proof-of-principle experiments demonstrate the value and potential uses of the different RedoxiFluor modes in diverse biological contexts

After validating the ability of the assay to report target-specific redox state in relative percentage and molar terms, we performed proof-of-principle experiments to demonstrate how one might use the different RedoxiFluor modes in diverse biological contexts by using an appreciated (immunology) and underappreciated (fertilisation) redox models. Experiments are intended to provide a starting point, as defined by a change in target-specific redox state, for more involved work since our focus is to critically examine the uses of a new technique, which could then, combined with other methods (e.g., activity assays, redox proteomics, site-directed mutagenesis, etc.) be used to discover redox switches or biomarkers.

First, we considered a topical and translationally important immunology model: the response of human monocyte cells to the pro-inflammatory stimulus lipopolysaccharide (LPS). To showcase global mode, we determined whether LPS (100 ng/ml for 30 min) altered proteome-wide redox state in THP-1 cells (Fig. 3). Consistent with previous research showing that the bulk thiol pool is often refractory to a redox stimulus [47], no LPS-induced change in proteome-wide redox state occurred. The highly reduced state of the proteome (15% reversible oxidation) is consistent with previous literature [39,56] and shows that our lysis protocol limits ex vivo oxidation [8]. If one does elect to convert the global percentage value to moles via normalising to protein content, consider that it can be over or under estimated by the relative abundance of the thiol-free (i.e., no cysteine residue) proteome. The proof-of-principle experiment showcases the value of global mode RedoxiFluor as a standalone assay.

Fig. 3

No difference in the redox state of the bulk thiol proteome in unstimulated (control) and LPS-stimulated human monocytes. Significant differences (paired t-tests, P = < 0.0001 in panels A–B, n = 6) in percent reduced compared to percent oxidised protein in unstimulated (control, panel A) and LPS-stimulated (panel B) human monocytes. C. No significant difference (unpaired t-test, P = 0.9694, n = 6) in percent reversibly oxidised protein in LPS-stimulated human monocytes compared to unstimulated human monocytes. D. A representative SDS-PAGE gel image showing that both proteomes are highly reduced and similar band patterns in the reduced (red channel, F-MAL1) and reversibly oxidised (green channel, F-MAL2) channels in control and LPS samples. Data are presented as the mean (M) and standard deviation (SD). (For interpretation of the references to colour in this figure legend, the reader is referred to the Web version of this article.)

Second, we interrogated the redox state of interleukin-1 receptor-associated kinase 1 (IRAK1, UniProt ID: P51617) a protein kinase downstream of LPS-induced toll-like receptor activation in the innate immune response. Protein A mode RedoxiFluor revealed that LPS increased IRAK1-specific reversible thiol oxidation by 33% from 41 to 74% (Fig. 4A–C). Note that IG controls were challenging in THP-1 cells because they express Fc binding proteins. ELISA mode confirmed the LPS-induced increase (+48%) in IRAK1-specific reversible thiol oxidation (Fig. 4D–F). We attribute the greater increase in ELISA compared to protein A mode to baseline variability. While IRAK1 protein content remained at ∼1.8 pM (Fig. 4G), reversibly oxidised IRAK1 increased by 0.9 pM in LPS-stimulated cells compared to unstimulated controls (Fig. 4H, Supplementary Fig. 4). The percentage and molar terms are relative and just like, in redox proteomics, could be underestimated if over-oxidation was appreciable. This result demonstrates the boon and bane of RedoxiFluor. It shows that a global change across all labelled target-specific thiols can happen, but it also demonstrates that, albeit similar to any other immunological technique, without proteomics or systematic site-directed mutagenesis screens one cannot know, which of the 17 thiols in IRAK1 caused the observed change. Interpretationally, the change in IRAK1-specific redox state can reflect a structural difference and/or a change in the rate of reversible thiol oxidation formation and or removal. This proof-of-principle experiment showcases the value of ELISA mode RedoxiFluor.

Fig. 4

LPS increases IRAK1-specific reversible thiol oxidation. A. In protein A mode, precent reduced is significantly (paired t-test, P = 0.0419, n = 6) greater than percent oxidised IRAK1 in unstimulated (control) human monocytes. B. In protein A mode, precent oxidised is significantly (paired t-test, P = < 0.0001, n = 6) greater than percent reduced IRAK1 in LPS-stimulated human monocytes. C. Protein A mode RedoxiFluor revealed a significant (unpaired t-test, P = < 0.0001, n = 6) LPS-induced increase in percent oxidised IRAK1 in LPS stimulated compared to unstimulated cells. D. In ELISA mode, precent reduced is significantly (paired t-test, < 0.0001, n = 6) greater than percent oxidised IRAK1 in unstimulated monocytes. E. In ELISA mode, precent oxidised is significantly (paired t-test, P = < 0.0001, n = 6) greater than percent reduced IRAK1 in LPS-stimulated human monocytes. F. ELISA mode RedoxiFluor revealed a significant (unpaired t-test, P = < 0.0001, n = 6) LPS-induced increase in oxidised IRAK1 in LPS-stimulated monocytes compared to unstimulated monocytes. G. No significant difference (unpaired t-test, P = 0.9159, n = 6) in picomoles of IRAK1 in unstimulated monocytes compared to LPS-stimulated monocytes. H. A significant (unpaired t-test, P = < 0.0001, n = 6) LPS-induced increase in picomoles of oxidised IRAK1 in LPS-stimulated human monocytes compared to unstimulated human monocytes. Data are presented as the mean (M) and standard deviation (SD).

Third, we designed a protein A microplate to simultaneously measure the redox state of several redox regulated phosphatases (i.e., array mode) in unstimulated and LPS-stimulated monocytes (Supplementary Fig. 5 and Supplementary Table 1). In unstimulated cells, protein phosphatase redox state ranged from 75.6% (i.e., PTP1B) to 88.2% (i.e., SHP2) reduced, with most targets clustering (i.e., within 3%) around 85%—the global protein thiol redox state. PTP1B and PP2A were, however, displaced from 85% by 10 and 6%, respectively. Array mode RedoxiFluor revealed no significant difference in SHP2, calcineurin, and PTEN redox state—they remained highly reduced (i.e., ∼90-85%) in unstimulated and LPS-stimulated cells (Fig. 5, Supplementary Figs. 6–7). Consistent with both the specificity of LPS-induced redox changes and heterogeneity within an enzyme class (i.e., phosphatases), array mode RedoxiFluor revealed a significant LPS-induced increase in PTP1B (+4.8%), SHP1 (+10.4%), CD45 (+10.1%), and PP2A (+8.4%) specific reversible thiol oxidation (Fig. 5, Supplementary Figs. 6–7). Within the responsive phosphatases, the magnitude of the LPS-induced change ranged from 4.8% to 10.4% (Fig. 5, Supplementary Figs. 6–7). Aside from PTP1B (∼75%–70%), LPS decreased their redox state from between 88 and 80% to 78-70% reduced. Regarding state-specific redox signatures (Supplementary Fig. 5), the unrefined new global redox state metric—the summed redox state of the 86 thiols distributed across seven phosphatases—was insensitive to LPS. However, the rationally refined four-parameter metric was sensitive to LPS (i.e., a weighted mean LPS redox sentinel metric). To validate one of the “hits”, we selected PP2A for follow-up ELISA mode RedoxiFluor. ELISA mode RedoxiFluor confirmed the LPS-induced increase in PP2A specific reversible thiol oxidation (+6% from 19 to 25%, Supplementary Fig. 8). After confirming PP2A content remained at ∼44 pM, ELISA mode revealed that the relative percentage change corresponded to a relative 3.6 pM increase in the amount of reversibly oxidised PP2A from 8.4 to 11 pM (Supplementary Fig. 8). This proof-of-principle experiment showcases the value of array mode RedoxiFluor.

Fig. 5

LPS increases PP2A-, PTP1B-, SHP1-, and CD45-specific reversible thiol oxidation. No significant difference (all unpaired t-tests and n = 6) in PTEN (P = 0.1871), SHP2 (P = 0.3054), and calcineurin (P = 0.2780) specific reversible thiol oxidation (i.e., percent oxidised protein) in unstimulated (control) and LPS-stimulated human monocytes as determined by array mode RedoxiFluor. A significant (all unpaired-tests and n = 6) LPS-induced increases in PP2A (P = 0.0181), SHP1 (P < 0.0001), PTP1B (P = 0.0483), and CD45 (P = 0.0018) specific reversible thiol oxidation occurred in LPS-stimulated human monocytes compared to unstimulated controls. Data are presented as the mean (M) and standard deviation (SD).

Finally, we explored how the fundamental biological process of fertilisation [57] impacted PTEN redox state in X. laevis. To do so, we measured PTEN redox state in unfertilised eggs and 1-cell zygotes (i.e., 30 min post-fertilisation). Protein A mode RedoxiFluor revealed a significant fertilisation-induced increase (+20%) in PTEN-specific reversible thiol oxidation (Fig. 6, Supplementary Fig. 9). In line with specificity, we observed no background F-MAL signal in IG compared to blank wells (Supplementary Fig. 9). When matched paired antibodies are unavailable, it can be useful to confirm microscale findings in macroscale mode to visually verify PTEN. Macroscale RedoxiFluor is also valuable because it reports a redox modifiable functional property: the protein interactome [16]. Before proceeding, it is important to note that the interactome was accessible to IP only—no PTEN interactants were identified when we concentrated microplate samples and performed SDS-PAGE, as evidenced by the presence one F-MAL positive band at 55 kDa (i.e., PTEN). It is unlikely, therefore, that interactants survive the harsher washing procedures of microplate methods to make any appreciable contribution, even when summed, to the signal (Supplementary Fig. 9). To study intermolecular RSSR, one should covalently bind the capture antibody and complete the F-MAL labelling following elution (i.e., to preserve the bond and eliminate antibody thiols). After verifying successful PTEN “pull-down” by immunoblot (Supplementary Fig. 9), we measured the aggregated redox state of the PTEN interactome in a microplate (i.e., measuring the redox state of eluent aliquots). The summed redox state of the six coeluting proteins (i.e., the PTEN interactome) is impervious to fertilisation (Fig. 6, Supplementary Fig. 9). However, band 2 only coelutes in zygotes and is mostly oxidised. Consistent with the microscale protein A finding, PTEN-specific reversible thiol oxidation, as measured by analysing the eluent from the excised 55 kDa band in a plate reader, was increased (+14%) in 1-cell zygotes compared to unfertilised eggs (Fig. 6, Supplementary Fig. 9). PTEN-specific redox state is displaced from the bulk thiol proteome in X. laevis, but this result and the related redox proteomic data will be published separately in due course. This proof-of-principle experiment shows that RedoxiFluor can be adapted (i.e., used in macroscale) and applied to diverse biological contexts.

Fig. 6

Fertilisation increases PTEN-specific reversible thiol oxidation. A. Protein A mode RedoxiFluor revealed a significant (unpaired t-test, P = 0.0002, n = 6) fertilisation-induced increase in PTEN-specific reversible thiol oxidation in 1-cell zygotes compared to unfertilised eggs in X. laevis. B. Macroscale RedoxiFluor found no significant (unpaired t-test, P = 0.2592, n = 3) difference in PTEN interactome redox state in 1-cell zygotes compared to unfertilised eggs in X. laevis. C. A representative SDS-PAGE gel image showing the PTEN (highlighted) and the redox state (reduced = green channel; oxidised = red channel) of its interactome (arrows 1–6 correspond to coeluting proteins with molecular weights of ∼100, 75, 60, 37, 25 and 10 kDa, respectively) in unfertilised eggs (E) and zygotes (Z). D. Quantifying the redox state of the PTEN-specific band manually excised and eluted from (C) in unfertilised eggs and 1-cell zygotes in X. laevis confirmed the significant (unpaired t-test, P = 0.0020, n = 3) fertilisation-induced increase in PTEN-specific reversible thiol oxidation. Data are presented as the mean (M) and standard deviation (SD). (For interpretation of the references to colour in this figure legend, the reader is referred to the Web version of this article.)

Discussion

Advantages

We have validated a new immunological technique for measuring protein thiol redox state and thereby take the field a step closer to realising the full power and benefits of antibodies as capture handles for reading thiol-redox state encoded reporters. The inherent accessibility and elegant simplicity of antibody techniques, as exemplified by a microplate assay, open up the universal utility of measuring protein thiol redox state to anyone. In a field beset by flawed assays like the notorious TBARS technique for lipid peroxidation [8], it is important to note that RedoxiFluor draws on valid and established principles, from the labelling techniques and reagents long-used in thiol research (e.g., F-MALs have a rich history in redox proteomics [47]) to the firm foundations of the ELISA assay. Delivering a simple and readily interpretable measure of protein thiol redox state in an easily implementable microplate format could facilitate wider use. As the proof-of-principle experiments demonstrate, RedoxiFluor achieves four key benefits: global mode (i.e., context), relative percentages, relative moles, and array mode. Together, each benefit defines the most complete microplate redox analysis achieved to date. Achieving relative percentage and molar analysis is of particular importance because of their interpretational value and relevance to advancing quantitative redox biology [21]. Moreover, all of the benefits of the microplate format are intact (e.g., high-throughput, rapid, and sensitive automated analysis). Importantly, RedoxiFluor is inherently evolvable and can be updated as new reagents emerge. For example, recent turn-on fluorescent reaction dependent RSOH probes could readily be incorporated [58]. The advantages of combining RedoxiFluor and redox proteomics are considered below. We hope that RedoxiFluor proves useful for advancing knowledge of protein thiol redox biology.

Limitations

It is essential to bear several limitations in mind when using RedoxiFluor irrespective of whether they are general caveat of the field or methodological pillar itself. In considering general caveats of the field, all the well-articulated caveats associated with the current staple of thiol-reactive warheads apply. Principally, their inability to completely label the proteome, chemically limited reactivity with protonated thiols, off-target labelling, and reactivity with RSOH and RSSH [43,51]. Relevant to RedoxiFluor, native labelling conditions mean the steric penalty of the payloads must be evaluated (i.e., examine difference between the two reporters) and the microenvironment plays more of a role—a more solvent exposed thiol could be labelled in one state but not the other—in native compared to denaturing labelling [59,60]. For this reason, a structural dimension is associated with interpreting the change. Potential structural sensitivity is useful and potentially mechanistically insightful when combined with antibody enrichment native labelling redox proteomics (see below). Even when immunodepleting or genetically deleting the target abolishes the signal, it could be argued that there is a potential F-MAL signal from any target-specific protein interactants able to withstand multiple exposures to detergents and washes. In practice, this is insignificant as the PTEN microscale experiments in X. laevis confirm. If it were found to be appreciable for another protein, then, at worst, the microscale results reflect the target-specific interactome and only percentage analysis is possible. As the macroscale PTEN examples attests, IP retains protein-protein interactions, which means RedoxiFluor can be used to study the redox-sensitive protein interactome [[61], [62], [63]]. Like all immunological techniques, RedoxiFluor cannot disclose the identity of the oxidised thiols—only redox proteomics or systematic site-directed mutagenesis screens can. Likewise, it also requires a suitable antibody, so their availability can be rate-limiting. Availability concerns are, however, mitigated by significant recent developments in phage directed antibody production and nanobodies [1,64], as well as, genome engineering strategies to express tagged proteins [65]. Finally, and similar to redox proteomics and immunological techniques, the temporal resolution of RedoxiFluor is limited to the number of time-points collected. Fleeting target-specific redox state changes may be missed because they cannot currently be captured in real-time [66].

Applications, recommendations and future directions

Three main uses for RedoxiFluor are envisaged (see Fig. 1). First, as a standalone protein thiol measurement technique to develop biomarkers (single or panels) or study redox regulation. Second, as a feedforward technique to select targets for follow-up redox proteomic analysis. Third, as a feedback technique to confirm redox proteomic changes. Combining RedoxiFluor and redox proteomics delivers added value via integrating a holistic and specific thiol metric with the added functional dimensions—lifetime and interactome on the microscale and macroscale, respectively. Specifically, RedoxiFluor can report the individual holistic weighted mean redox state of all the target thiols, which is significant if some are undetected despite enrichment. Meanwhile, redox proteomics can report on the faithfulness and extent of the labelling for the detectable thiols and simultaneously report individual redox state changes in percentages and identify any interacting proteins, and their redox state (which likely differs from the bulk pool) of any interactants in macroscale. Of special consideration, with respect to biomarkers, is the possibility of developing data-led (i.e., redox proteomic screen informed) state- or process-specific redox signatures using array mode RedoxiFluor. Relevant to clinical screening, pre-antibody decorated plates could be manufactured so as to readily and quickly assess multiparametric redox outputs in a microplate.

Irrespective of the intended use, general recommendations apply. We recommend that users validate the F-MAL labelling protocol for their samples of interest, as well as the antibody selectivity and plate orthogonality by repeating the control experiments (or similar) detailed in the present manuscript. Comparing target content in labelled and unlabelled standards in ELISA mode is recommended for confirming that there is no defect in antibody recognition. Interpretationally, unless certain criteria are met (see results) the relative nature of the observed findings should be clearly stated. Specific recommendations mostly relate to standalone mode: no biological meaning can be ascribed to a change in protein thiol redox state without additional evidence (e.g., functional assays). Moreover, a redox switch cannot be elucidated without performing the site-directed mutagenesis experiments. If using standalone mode to report a molecular process (e.g., a molecular sentinel of insulin dependent redox signalling [67]), then the target should be one that has been independently identified to respond to the stimulus. Ideally, the native labelling would have also been validated by targeted redox proteomics. If using RedoxiFluor in tandem with redox proteomics, for the target-specific steps the labelling protocols should be similar, in at least some samples, so as to directly compare the results; which would allow by exclusion reasoning certain undetected but redox sensitive thiols to be deduced.

In considering future directions, we wish to highlight two promising avenues. First, it may be possible with the synthesis of some new reagents to directly integrate RedoxiFluor with redox proteomics. That is, measuring the target-specific global redox state (sum of all labelled thiols) before reading their individual redox state from the same samples. Given the difficulties of detecting F-MAL decorated peptides [47], doing so would necessitate creating chemically cleavable compounds with different masses on the thiol warheads to discriminate the reduced and reversibly oxidised thiols via mass spectrometry. Second, and with respect to sensitivity, signal amplification comes from the number of thiols labelled. It is rate-limited, even if completed, by the amount of target captured, which places an upper-bound on the theoretically possible sensitivity. It would be useful, especially in the clinic (i.e., plasma or liquid biopsies) to amplify the redox state. We are actively exploring each possibility.

Conclusion

RedoxiFluor is a new antibody technology with the power to quantify relative target-specific protein thiol redox state in percentages and moles relative to the bulk thiol proteome (i.e., context) and selected other targets (i.e., array mode) in a widely accessible, simple and easily implementable microplate format.

Methods

RedoxiFluor protocol: A detailed step-by-step protocol is included in the supplementary material.

Samples: Xenopus laevis (X. laevis)

Following ethical approval (#ETH2021-0222), unfertilised eggs and 1-cell zygotes collected 30 min post-fertilisation were harvested from three different adult females housed at 18 °C in groups of 20 (i.e., each X. laevis sample represents the weighted mean of 20 eggs/zygotes). Note that it isn't possible to comment on the sex of the samples at this developmental stage. Samples were immediately lysed in IP buffer (25 mM Tris-HCl, 150 mM NaCl, 1 mM EDTA, 1% NP-40, 5% glycerol, pH 7.1) supplemented with a protease inhibitor tablet (Sigma Aldrich, UK, #11697498001) and 1 mM F-MAL (F-MAL1: Fluorescin-5-maleimide, ThermoFisher, UK, #62245; or F-MAL2: AlexaFluor™647-C2-maleimide, ThermoFisher, UK; #A20347).

Samples: THP-1 human monocytes

THP-1 human monocytes (ECACC, #88081201) were cultured (95% air, 5% CO2 at 37 °C) in RPMI 1640 medium supplemented with 10% fetal bovine serum, 2000 mg/L glucose, 2 mM l-glutamine and 1% penicillin/streptomycin at a density of 5 x 105 cells. Unstimulated or LPS stimulated (100 ng/ml for 30 min, Sigma Aldrich, UK, #L6529) cells were rapidly lysed in IP buffer supplemented with 1 mM F-MAL and processed for RedoxiFluor (see below).

Antibodies

The antibodies used for RedoxiFluor were as follows: PP2A (Abcam, UK, #ab226790), PP2A matched pair ELISA (Abcam, UK, #ab218174), IRAK1 matched pair ELISA (Abcam, UK, #ab210071), SHP1 (Abcam, UK, #ab227503), SHP2 (Abcam, UK, #ab30283), CD45 (Abcam, UK, #ab208022), PTEN (ThermoFisher, UK, #PA5-20418), calcineurin (ThermFisher, UK, #PA5-17446), and PTP1B (Abcam, UK, #ab244207).

Sample processing: thiol labelling procedure

Samples were incubated with 1 mM F-MAL1 or F-MAL2 (depending on the labelling order) for 30 min on ice and centrifuged at 14,000 g for 5 min at 4 °C. Soluble supernatants were passed through a 6 kDa spin column (Bio-Rad, UK, #7326222) to remove excess F-MAL1. Flow throughs were treated with 5 mM DTT (ThermoFisher, UK, #RO861) for 30 min on ice. After removing excess DTT with a spin column, samples were treated with 1 mM F-MAL1 or 2 for 30 min on ice. Unreacted F-MAL1 or 2 was removed with a spin column. To prepare assay calibrants, samples were lysed in 5 mM DTT for 30 min and centrifuged at 14,000 g for 5 min at 4 °C. After removing excess DTT with a spin column, samples were incubated with 1 mM F-MAL1 or 2 for 30 min. Unreacted F-MAL1 or 2 was removed with a spin column. Fully labelled F-MAL1/2 standards were mixed as appropriate to produce the 10–90% reversibly oxidised redox states (e.g., for a 10 μl final volume, 9 μl of F-MAL1 was mixed with 1 μl of F-MAL2 to prepare the 90% reversibly oxidised standard). To prepare BSA standards, 50 μg BSA was reduced with 1 mM DTT for 15 min at room temperature (RT). Excess DTT was removed with a spin column and flow-throughs were treated with 0.5 mM F-MA1 or 2. After removing excess F-MAL1/2, the desired redox states were prepared by mixing the standards as appropriate. Samples were protected from ambient light throughout. For the experiments to determine that all of the F-MAL or DTT was removed from the spin column, the exact amount of reagent in lysis buffer was added to the column and the fluorescence or absorbance (320 nm for oxidised DTT) compared to the unfiltered control and blanks was measured in a plate reader.

Isolated fluorophore and recombinant BSA experiments

For the isolated fluorophore experiments described in Supplementary Figs. 1 and 1 mM of each fluorophore was incubated in l-cysteine buffer (25 mM Tris, pH 7.2, 2 mM l-cysteine) for 30 min at RT in the dark to capture any thiol-dependent turn-on fluorescence. To construct the 90 to 10% standard curve, appropriate amounts of fluorophore were mixed to a final volume of 10 μl. Aliquots (1 μl) were analysed in triplicate in a plate reader (see RedoxiFluor analysis). BSA standards (prepared as described above) were analysed in the same way. For the gel experiments, BSA standards (1 μg) were resolved by SDS-PAGE and analysed as described below. For the experiments to determine the extent of BSA labelling, reduced BSA was incubated with 5 mM NEM or 1 mM F-MAL2 for 30 min and the excess was removed with a spin column. F-MAL2 (1 mM) was then added for 30 min, removed with a spin column, and the fluorescence was measured on a plate reader (i.e., the degree of labelling must reflect off-target or incomplete labelling). Similar experiments were performed to optimise sample labelling.

Protein A mode RedoxiFluor

To assess a single target (i.e., PP2A, PTEN, or IRAK1), 0.1 μg of capture antibody in binding buffer (50%: 0.05% Tween-20 in phosphate buffered saline [PBST]; 50% Superblock [ThermoFisher, UK, #37580) was added to each well of a black protein A derivatised microplate (ThermoFisher, UK, #15155) for 1 h at RT at 350 rpm on a plate shaker. Unbound capture antibody was removed by washing (3 × 2 min PBST washes at 400 rpm), before assay calibrants (i.e., 10–90% reversibly oxidised standards), controls (i.e., immunodepleted, see below), or samples were added in duplicate and incubated in the dark for 2 h at RT. For X. laevis, 5 μl of sample/assay calibrant was diluted in 95 μl PBS. For THP-1 cells, 10 μl of sample was diluted in 90 μl PBS. After removing unbound sample, wells were washed (3 × 2 min PBST washes at 400 rpm), rinsed in PBS to remove excess Tween-20, and incubated with denaturing buffer (4% SDS) for 15 min at RT with vigorous shaking (500–700 rpm). After measuring F-MAL/2 fluorescence in a plate reader, the target-specific protein thiol redox state was calculated (see below).

ELISA mode RedoxiFluor

To measure PP2A- and IRAK1-specific protein thiol redox state in ELISA mode matched paired antibodies were used. Black MaxiSorp immuno microplates (ThermoFisher, UK, #437111) were incubated with 50 μl of 2 μg/ml capture antibody overnight at 4 °C in binding buffer (35 mM NaHCO3, 15 mM Na2CO3, pH 9.6) on a plate shaker at 350 rpm. Unbound capture antibody was removed by washing (3 × 2 min PBST washes at 400 rpm) before wells were blocked (50% PBST, 50% Superblock) for 2 h at RT at 350 rpm and washed (3 × 2 min PBST washes at 400 rpm). The recombinant protein standards, assay calibrants (i.e., 10–90% reversibly oxidised), assay controls (i.e., immunodepleted sample, and samples (diluted as above to a final volume of 50 μl) were added in duplicate and incubated for 2 h at RT at 350 rpm in the dark. Excess sample were removed, wells were washed (3 × 2 min PBST washes at 400 rpm), and 0.5 μg/ml biotin-conjugated detector antibody was added for 1 h at RT at 350 rpm in the dark. After a wash step, 0.05 μg/ml of HRP-conjugated streptavidin (Abcam, UK, #ab210901) was added for 1 h at RT at 350 rpm in the dark. After a final wash step, wells were incubated with QuantaBlu™ (ThermoFisher, UK, #15169) prepared according to the manufacturer's guidelines for 10 min at RT at 400 rpm in the dark. The QuantaBlu signal was measured at 325 (excitation) and 425 (emission) nm for 100 ms on a plate reader. To stop the HRP reaction and unmask the F-MAL1/2 signal wells were incubated in denaturing buffer for 15 min at RT at 500–700 rpm. After measuring F-MAL/2 fluorescence in a plate reader, the target-specific protein thiol redox state was calculated (see below).

Assay controls

For the PP2A and PTEN microplate experiments in X. laevis described, sample aliquots (diluted as above) were also added to rabbit isotype control wells in protein A and ELISA mode to assess nonspecific binding. To check specificity in the case of PP2A, samples were incubated with PP2A capture antibody functionalised protein A magnetic beads overnight at 4 °C with gentle rotation (see below). The PP2A immunodepleted sample was added to a second capture antibody functionalised magnetic bead for 1 at RT before being added to the protein A or ELISA microplate. Sample-specific (i.e., THP-1 for IRAK1) calibrated standards (e.g., 10–90% reversibly oxidised) were used in every RedoxiFluor experiment.

Array mode RedoxiFluor

To assess the redox state of multiple targets in array mode, 0.1 μg of SHP1 (row B), SHP2 (row C), PTP1B (row D), PP2A (row E), PTEN (row F), CD45 (row G), and calcineurin (row G) capture antibodies were added in binding buffer (50%: PBST; 50% Superblock) to a black protein A derivatised microplate for 1 h at RT at 350 rpm on a plate shaker. Row A was reserved as a blank well. Unstimulated (lanes 1–3 & 7–9) and LPS stimulated (lanes 4–6 & 10–12) samples (diluted 1:10) were added for 2 h at 350 rpm in the dark. Thereafter, array mode experiments were identical to the protein A mode RedoxiFluor.

Global thiol proteome redox state

To measure the redox state of the thiol proteome, samples (1 μl diluted in 199 μl PBS) were measured in triplicate in a plate reader.

Macroscale: RedoxiFluor

Aliquots (10 μg) of PTEN capture antibody or rabbit isotype control (ThermoFisher, UK, #31235) were incubated with 50 μl protein A derivatised magnetic beads for 1 h at RT. Beads were magnetised to remove excess capture antibody and incubated with undiluted X. laevis samples (∼500 μg protein) overnight at 4 °C. Beads were magnetised to remove unbound sample, washed in PBST (0.025% Tween-20), and resuspended in denaturing buffer (4% SDS, 200 mM Tris, 20% glycerol, pH 7.1). To prevent any potential heat-induced fluorescence artifacts, PTEN was eluted from the capture antibody by chemically breaking the antibody apart: the disulfide bonds required to structure the antibody for target binding were reduced with 10 mM DTT for 15 min at RT. This procedure proved effective as magnetic beads were negative for F-MAL1-2 following elution. Note that glycine elution is incompatible with RedoxiFluor because the low pH destroys F-MAL fluorescence. Eluents were isolated by magnetising the beads. To assess PTEN interactome redox state, triplicate eluent aliquots (2 μl) were measured in a plate reader. To identify coeluting proteins, eluents were resolved by SDS-PAGE using a precast 4–15% gradient gel (Bio-Rad, UK, #4561085) and F-MAL1/2 signals were captured on a gel scanner for 1 s using the appropriate filters. To assess the PTEN-specific redox state, eluents were resolved by SDS-PAGE and the specific-PTEN band (as confirmed by immunoblot, see below) was manually excised and passively eluted in distilled water (dH2O) for 24 h at RT with vigorous agitation. Passive elution was successful because the bands were F-MAL1/2 negative after 24 h. For PTEN-specific redox state analysis, no gel scanner imaging was performed to eliminate potential phototoxicity interfering with microplate analysis. The passively eluted PTEN redox state was measured using a plate reader.

Microscale-PTEN interactome experiment

To determine for a protein with known interactants at the macroscale whether any protein interactants were observable at the microscale, PTEN bound from 100% F-MAL2 labelled X. laevis samples in protein A mode (i.e., incubated with a PTEN capture antibody as above) was eluted in denaturing buffer. Sets of eluents were pooled (n = 10), concentrated in a 3 kDa Amicon column, resolved by SDS-PAGE and analysed for the presence of F-MAL1/2 positive bands other than the expected 50 kDa PTEN specific band.

RedoxiFluor analysis

Irrespective of the RedoxiFluor mode, target-specific protein thiol redox state was assessed by measuring F-MAL1 (494–518 nm) and F-MAL2 (651–671 nm) with a 5 nm bandwidth for 100 ms on a plate reader. After blank subtraction, F-MAL1 and F-MAL2 signals (v) were corrected using the following equation: v = (v/E)/q or a derivative thereof were E and q represent the extinction coefficient and quantum yield, respectively. Corrected values were totalled, and precent reduced or reversibly oxidised protein was calculated (i.e., reduced = (reduced/total)*100). In ELISA mode RedoxiFluor, pg/ml protein content values computed from the recombinant protein standard curve were converted to picomoles by dividing them by the molecular weight of the target protein (i.e., picomoles = picogram/target molecular weight). The amount of reduced and reversibly oxidised target protein in picomoles could then be calculated (i.e., pM reduced = total pM*reduced%).

Immunoblot

After resolving samples by SDS-PAGE, they were transferred to a 0.45 μM 100% methanol activated PVDF membrane for 1 h. Membranes were blocked for 1 h in 5% non-fat dry milk (NFDM) and incubated with a primary PTEN antibody overnight (1:1000) at 4 °C with gentle agitation. To avoid interference from the eluted capture antibody light and heavy chains, the PTEN primary antibody aliquot was incubate with 100 μM DyLight755 fluorescent conjugated N-hydroxysuccinimide ester (F-NHS, ThermoFisher, UK, #62278) for 30 min at RT. Unreacted F-NHS was quenched with 1 mM Tris pH 7 for 15 min at RT. Fluorescence was assessed on a gel scanner.

ALISA

PTEN primary antibody aliquots (0.1 μg) were incubated with an epoxy group functionalised plate (PolyAn, Germany, #00695251) in binding buffer (150 mM Na2HPO4, 50 mM NaCl, pH 8.5) overnight at 4 °C at 350 rpm on a plate shaker [31]. Unreacted epoxy groups were quenched (100 mM Tris, pH 9) for 2 h at RT, washed in PBST, and incubated with samples labelled with a single F-MAL for 2 h at RT with gentle agitation. After removing unbound sample, wells were washed in PBST (3x), resuspended in PBS, and incubated with 350 μMF-NHS for 30 min. Excess F-NHS was removed, wells were washed in PBST (3x), and resuspended in denaturing buffer for 1 h. Eluent redox state was assessed in new microplate. After blank subtraction, the target-specific redox state was calculated as: F-MAL/F-NHS.

Click-PEG

Catalyst-free inverse electron demand Diels Alder chemistry Click-PEG wherein trans-cyclooctene (TCO) labelled reversibly oxidised thiols are conjugated to 6-methyltetrazine derivatised PEG 5 kDa was performed as previously described [24,36]. Click reacted samples were analysed by PP2A immunoblot.

Statistics

Data-set normality was assessed using Shapiro-Wilk and Kolmogorov-Smirnov testing. For the within sample redox state (i.e., reduced vs. reversibly oxidised), data were analysed using a paired t-test or nonparametric equivalent. For between samples redox state (i.e., %reversibly oxidised), data were analysed with an unpaired t-test or non-parametric equivalent. In all cases, alpha was set to P > 0.05 and tests were performed on GraphPad Prism Version 9 (https://www.graphpad.com). No data exclusion criteria were applied (i.e., all values including outliers were considered). The figure legends report exact P values and statistical tests. Data are presented as the mean (M) and standard deviation (SD).

Author contributions

Conceptualization, J.N.C. and M.G.; Methodology, J.N.C.; Investigation, ALL; Writing—Original Draft, J.N.C; Writing—Review and Editing, ALL.; Funding Acquisition, J.N.C & M.G.; Resources, J.N.C. and M.G.; Supervision, J.N.C. & M.G.

Declaration of competing interest

J.N.C. has filed a UK patent application relating to RedoxiFluor.