Enzymology of reactive intermediate protection: kinetic analysis and temperature dependence of the mesophilic membrane protein catalyst MGST1

Glutathione transferases (GSTs) are a class of phase II detoxifying enzymes catalysing the conjugation of glutathione (GSH) to endogenous and exogenous electrophilic molecules, with microsomal glutathione transferase 1 (MGST1) being one of its key members. MGST1 forms a homotrimer displaying third‐of‐the‐sites‐reactivity and up to 30‐fold activation through modification of its Cys‐49 residue. It has been shown that the steady‐state behaviour of the enzyme at 5 °C can be accounted for by its pre‐steady‐state behaviour if the presence of a natively activated subpopulation (~ 10%) is assumed. Low temperature was used as the ligand‐free enzyme is unstable at higher temperatures. Here, we overcame enzyme lability through stop‐flow limited turnover analysis, whereby kinetic parameters at 30 °C were obtained. The acquired data are more physiologically relevant and enable confirmation of the previously established enzyme mechanism (at 5 °C), yielding parameters relevant for in vivo modelling. Interestingly, the kinetic parameter defining toxicant metabolism, kcat/KM, is strongly dependent on substrate reactivity (Hammett value 4.2), underscoring that glutathione transferases function as efficient and responsive interception catalysts. The temperature behaviour of the enzyme was also analysed. Both the KM and KD values decreased with increasing temperature, while the chemical step k3 displayed modest temperature dependence (Q10: 1.1–1.2), mirrored in that of the nonenzymatic reaction (Q10: 1.1–1.7). Unusually high Q10 values for GSH thiolate anion formation (k2: 3.9), kcat (2.7–5.6) and kcat/KM (3.4–5.9) support that large structural transitions govern GSH binding and deprotonation, which limits steady‐state catalysis.


Introduction
Chemically reactive molecules, that are formed during metabolism (of foreign or endogenous compounds), pose a threat, as they can covalently interact with cellular nucleophiles (abundant in proteins, lipids and DNA), causing toxicity and tumour development [1,2]. In order to protect from these reactive (mostly electrophilic) compounds [3], enzymes have evolved to intercept reactive intermediates [4,5]. Given the high concentration of nucleophilic targets in a cell, this is a formidable catalytic challenge. It is thus of interest to understand the mechanistic enzymology underlying protection.
The major enzyme class protecting against hydrophobic electrophiles is glutathione transferases (GSTs) [6,7]. These enzymes catalyse the conjugation of glutathione (GSH) to reactive electrophilic centres via the stabilization of the nucleophilic cysteine thiolate in GSH [8]. Testament to the importance of interception, glutathione transferases have evolved distinct soluble and membrane-bound forms [9], comprising cytosolic, mitochondrial and microsomal enzyme classes [10]. Each class contains various isoforms displaying broad and overlapping substrate specificities, with over 20 individual enzymes in humans alone. As GSTs are often overexpressed in tumours [11] the ability to remove reactive molecules can also lead to anticancer drug resistance [12][13][14]. Together, the substrate promiscuity, enzyme variety and pathophysiological importance prompted extensive research on the fundamental mechanistic aspects and functions of these enzymes [15].
One important member of this family is membranebound microsomal GST1 (MGST1) that constitutes 3% and 1% of the endoplasmic reticulum protein in rat and human liver, respectively, as well as 5% of the outer mitochondrial membrane in rat liver [9,16]. In conjunction with other GSTs, this enzyme is involved in cellular protection through the detoxification of xenobiotics and, uniquely, also the reduction in membrane-embedded lipid hydroperoxides [17,18].
Previously, the steady-state and pre-steady-state kinetic behaviour of MGST1 was characterized at low temperature, since, at higher temperature, the enzyme loses stability [19]. Here, using stop-flow limited turnover technology, we have overcome this stability limitation and elucidated the global kinetic mechanism of MGST1 at 30°C. The data yield insight into the mechanistic basis of reactive intermediate protection at a physiological relevant temperature and temperature behaviour.

Results
Validation of the MGST1 batch used-specific activity and NEM activation pattern at 30°C The specific activities of unactivated and NEMactivated MGST1, for the enzyme batches used in this study, are in good agreement with previously published data by Andersson et al. [20] (Table 1 and Table S1). The data show similarity between the batches and ensure that the pre-steady-state and steady-state kinetic parameters generated in this study could be compared with kinetic data determined previously.
Global kinetic model for MGST1 and pre-steadystate parameters at 30°C The global kinetic mechanism for MGST1 [19] and the resulting equation are shown in Scheme 1. A full description of the kinetic mechanism of MGST1 involves elucidation of equilibrium and microscopic steps consisting of GSH binding, GSH thiolate formation, electrophilic substrate binding, the chemical conjugation step and product release. It is assumed that substrate binding to the various enzyme forms is identical (as well as thiolate anion formation). The following sections outline the pre-steady-state techniques that were employed to obtain the relevant parameters at 30°C, which are summarized in Table 2.
Determination of the GSH binding constant (K G ) and GSH thiolate anion formation (k 2 ) through limited turnover experiments When MGST1 is mixed with GSH the enzyme binds the substrate whereafter the GSH thiolate forms, which can be followed by recording the change in absorbance at 239 nm [8]. As the GSH-free enzyme is unstable at 30°C, we used GSH-bound enzyme, to which we added (in the stop-flow) a small amount of electrophilic substrate. As the substrate is rapidly conjugated this leaves the enzyme devoid of GSH thiolate leading to an initial burst (lowering absorbance) in the signal (Burst, Fig. 1A). When the small excess (2-fold, used to ensure complete consumption of enzymebound GSH thiolate) of electrophilic substrate is fully consumed (during steady state, Fig. 1A), the enzyme is free to rebind GSH from solution and form a thiolate (Thiolate formation, Fig. 1A and Figs S1 and S2). To determine k 2 and K G , the thiolate stabilization signal in these traces was fit to a single exponential to obtain k obs at varied GSH concentration and analysed by Eqn (2) (Fig. 1B, Table 2). The apparent off rate k −2 (strictly speaking GSH thiolate re-protonation) is very slow and therefore cannot be determined with precision in this setup (theoretically as the Y-axis intercept in Fig. 1B and Fig. S3). The K G for GSH is well within the physiological GSH concentration (1-10 mM) and close to the K M values (vide infra). GSH thiolate anion formation is very slow, below 1 s −1 , but as discussed below, there is no evolutionary pressure to favour rapid turnover.
The activated enzyme fraction stabilizes the GSH thiolate much more rapidly and has a strong impact on the steady-state turnover. By contrast, the ensuing thiolate stabilization, that follows the depletion of the electrophile, quantitatively depends on, and defines k 2 , for the dominating unactivated form (Fig. 1A). This experimental setup thus enabled us to determine k 2 and K G at 30°C (Table 2) obviating enzyme loss due to instability, at a temperature that is closer to physiological. Determination of the apparent GSH off rate k −2 (i.e. re-protonation of the GSH thiolate anion) The crystal structure and mass spectrometry analysis of MGST1 revealed that three GSH molecules are bound per trimer [21,22]. Binding experiments showed one tightly bound GSH [23] and spectroscopic analysis has shown that this tightly bound GSH harbours the catalytically competent thiolate anion [24]. Two protonated GSH molecules are bound with lower affinity to the remaining sites [25]. When we add a strong GSH competitive inhibitor we observe the loss of GSH thiolate signal as a very slow process (that differs in the unactivated and activated enzyme forms as shown previously [24]). We interpret the signal as resulting from the rapid replacement of the loosely bound GSH molecules, then as the GSH thiolate becomes protonated and loses affinity, it is replaced by the high excess of inhibitor. We can regard GSH thiolate re-protonation as the rate-limiting step governing GSH thiolate release.
The re-protonation rates were experimentally determined by rapidly mixing GSH-bound MGST1 with the strong inhibitor GSO À 3 ( Fig. 2A) or GSH analogue GOH (Fig. S4A), following the disappearance of the thiolate signal at 239 nm. The curves were fitted with a double exponential representing unactivated (k −2,slow ) and activated (k −2,fast ) enzyme fractions ( Fig. 2 and Table 2). Essentially, no major difference in the rates could be observed using either inhibitor GSO À 3 or GOH (compare Tables S2 and S3). This experiment was also performed with NEM activated enzyme and the curves were fit to a single exponential to obtain k −2,NEM = 0.98 AE 0.03 s −1 (Table S4). The rates obtained for the activated enzyme fraction and fully NEM-modified enzyme, k −2,fast and k −2,NEM are very similar, supporting that the inherently activated enzyme fraction has similar properties as the NEM-modified enzyme. To further validate the presence of two enzyme fractions (i.e. unactivated and activated enzyme), re-protonation curves were also fit with a single exponential (Fig. 2B). In comparison, single exponential fits showed a much higher deviation, thus clearly showing the presence of the two distinct fractions. In conclusion, GSH stabilized in the Literature values were taken from the graphs in Fig. 5 of Andersson et al. [20].
Scheme 1. Global kinetic model of MGST1 [19] corresponding to one active site and derived steady-state rate Eqn (1). E represents MGST1 without GSH in its active site; C, electrophile; P, product; K C and K G , dissociation constants for the electrophile and GSH; k 2 , rate constant of thiolate anion (GS − ) formation; k −2 , reverse process (re-protonation of the thiolate); k 3 , chemical rate constant. (1) and (2) mark observed steps in limited turnover experiments (Fig. 1A). k −2,slow and k −2,fast are the re-protonation rates for the unactivated and activated fraction of the unmodified enzyme. Re-protonation rates depicted here were determined using GSO À 3 as inhibitor (see Table S3 for re-protonation rates determined using GOH). thiolate anion form is re-protonated very slowly and thus catalytically competent.
Determination of the chemical conjugation rate constant (k 3 ) and the electrophile dissociation constant (K C ) The concentration dependence ( Fig. 3 and Fig. S5) of the burst of product formation upon rapidly mixing GSH-bound MGST1 with its electrophilic substrate yielded the chemical step rate constant (k 3 ) and apparent affinity of the electrophilic substrate (K C ; Table 2). The burst can be observed when the chemical step rate exceeds subsequent GSH thiolate reformation. While the apparent affinity for the electrophilic substrate is modest (0.2-0.6 mM) chemical reactivity fully translates into increased catalysis. With a k 3 of 810 s −1 the burst of the most reactive substrate, CDNB, approaches the dead time of the stop-flow, whereas the less reactive substrate CNAP, with a k 3 of 17 s −1 , yields a 48-fold lower conjugation rate (Table 2). Based on CNAP and CDNB reactivity σ − values this difference translates into a Hammet ρ value of 4.2, showing that chemical reactivity directly translates into catalysis, and that no slow steps impede chemical conjugation. This mechanistic behaviour is significant in terms of the physiological role (and evolved properties) of the enzyme (discussed below).
Determination of the steady-state kinetic constants for electrophilic second substrates (at constant 0.5 mM GSH and varying CDNB and CNAP concentration) The steady-state kinetic constants k cat , K M and k cat / K M were determined by analysing the linear steady- [GSH] curve (B) determines thiolate anion formation k 2 and GSH binding affinity K G . (A) Limited turnover curve at 239 nm is shown for a GSH concentration of 0.5 mM, as a representation to indicate the characteristics of these curves-initial burst (1) followed by steady-state and GSH rebinding/thiolate formation (2)-are highlighted (see Scheme 1 for comparison). Thiolate anion formation is fitted with a single exponential (red) to obtain k obs . The same curve is also shown as log-scale (A inset). See Figs S1 and S2 for thiolate anion formation fits to all GSH concentrations used. (B) k obs is subsequently plotted against the GSH concentration and fitted with Eqn (2) to determine k 2 and K G (see Table 2). At least five traces were measured per GSH concentration and averaged before analysis. Note that in (B), standard errors are smaller than the symbols. state phase that followed the initial burst ( Fig. 1A) upon mixing GSH-bound MGST1 with the electrophilic substrate. The rate concentration dependence is shown in Fig. 4 and Fig. S6, and the resulting kinetic parameters are given in Table 3 (referred to as 'Experiment').
In addition to the experimental steady-state constants, k cat , K M and k cat /K M were also calculated from the microscopic and equilibrium constants (Table 2), using the steady-state equations for MGST1 (Scheme S1), as well as through modelling using KIN-TEK EXPLORER software (KinTec Corporation, Snow Shoe, PA, USA) ( Table 3 and Table S5). For the former, it was assumed that all enzyme is unactivated, whereas for the latter, it was assumed that 10% of the enzyme is activated.
The measurements were performed in the presence of 0.5 mM GSH, which yields stronger signals because of lower background, and not 5 mM GSH that is required for saturation, which explains the shift from literature values for k cat and K M (Fig. 4, Table 3). This becomes particularly evident for k cat that is approximately a factor of 2 lower in comparison to saturating conditions. Taking this into account the values obtained agree well with those obtained earlier [26].  (Fig. 1), is plotted against the second substrate concentration and fit to determine k 3 and K C ( Table 2). At least five traces were measured per electrophile concentration and averaged before analysis. Note that standard errors are smaller than the symbols.  Table 2). For this, it was taken into account that 11 μM (corresponding to the enzyme concentration) of the substrate is conjugated in the burst phase. At least five traces were measured per electrophile concentration and averaged before analysis. The resulting k cat and K M are given in Table 3. Note that standard errors are smaller than the symbols. Calculation of the steady-state kinetic constants for GSH (at varying GSH and constant CDNB and CNAP concentration) Analogue to the estimation of the steady-state kinetic constants for varying electrophile and constant GSH concentration, k cat , K M and k cat /K M values were obtained through calculation (using equation in Scheme S1), as well as through KINTEK EXPLORER assisted modelling and compared with literature values (Table 4 and Table S6). For modelling, 10% activated enzyme was assumed, while all enzyme was assumed as unactivated for the calculation. Clearly, taking activation into account the agreement to steady-state experimental data is improved, especially for CDNB as discussed further below.

Temperature dependence of kinetic parameterscomparison between 5 and 30°C
Pre-steady-state and steady-state kinetic parameters at 5°C were previously determined. The data obtained in this study allows us to determine temperature effects on these parameters. Ratios were calculated, as well as rate enhancement for each 10°C (Q 10 ), see Table 5.
The temperature effect on the chemical step is modest both for the enzyme catalysed and nonenzymatic reaction, whereas dissociation constants and K M values are lowered at 30°C, indicating higher affinity/efficiency. Most striking is the high-temperature enhancement of thiolate anion formation (k 2 ) reflected also in k cat and k cat /K M . High Q 10 values indicate large structural transitions governing GSH binding coupled with thiolate anion formation.

Discussion
Does the global kinetic model previously derived using data from 5°C also hold at 30°C?
If the previously proposed global kinetic model (Scheme 1) is valid also at a physiologically relevant temperature the pre-steady-state behaviour should predict the steady-state behaviour of MGST1 at 30°C. Steady-state kinetic parameters were calculated from microscopic and equilibrium constants ( Table 2; Scheme 1) and compared to literature data, hereby  showing good agreement between predicted and observed data (Tables 3 and 4). The calculated k cat values for electrophilic substrates in Table 3 were derived assuming that the enzyme is 100% unactivated, which leads to an expected underestimation. When 10% activated enzyme is simulated in KINTEK EXPLORER higher values that are very close to the experimental parameters are derived. The data reaffirm the previous suggestion that MGST1 exists in a dynamic equilibrium between unactivated and activated forms. It should be noted that literature data for k cat in Table 3 were obtained at 10-fold higher GSH (5 mM) explaining that these are 2-fold higher. Overall, the proposed mechanism shows excellent agreement between experimental steady-state k cat with those calculated based on pre-steady-state data. The same holds for literature values [20] for k cat obtained at varied GSH as expected ( Table 4).
The K M values for electrophilic substrates agree well between experiment and literature values (Table 3) but appear to be higher by a factor of ≈ 3 when compared to those calculated in KINTEK (assuming a 10% activated subpopulation). As k 2 has a strong impact on K M electrophile a slight underestimation of activated enzyme fraction could explain this discrepancy. Experimentally, it is also challenging to obtain initial rates at very low substrate concentrations, which can lead to an overestimation of K M , whereas pre-steady-state data that yield k 2 are well-defined. The literature K M values for GSH on the other hand are well-predicted and robust (Table 4).
The above factors affecting k cat and K M apply in aggregate to k cat /K M , as it is calculated as a product. The efficiency, experimental k cat /K M , is approximately two orders of magnitude higher for CDNB compared with GSH underscoring that the function of the enzyme is to intercept reactive intermediates whereas GSH binding and, specifically, forming the catalytically competent thiolate is not rate limiting at physiological conditions.
We conclude that the (albeit simplified) mechanism captures the essential kinetic behaviour of MGST1 as follows. MGST1 is in rapid equilibrium with its substrates, which appears instrumental given that protection from reactive electrophiles entails intercepting covalent modification of cellular nucleophiles (e.g. DNA). However, after GSH has bound, slow GSH thiolate formation ensues. Slow GSH thiolate formation would appear counterproductive but put into perspective of the known turnover in vivo [5] is actually of no consequence. In fact, individual GSTs (including all enzyme forms present) can be calculated to perform one catalytic event every second day [5]. The massive concentration of GSTs that has evolved is necessary to intercept reactive intermediates whereas each Table 5. Temperature dependence of kinetic parameters and rate enhancement for each 10°C (Q 10 ) were calculated for the following kinetic constants: k 2 -rate constant of thiolate anion formation; k −2 -reverse process (re-protonation of the thiolate); k 3 -chemical rate constant; k (nonenzymatic) -nonenzymatic reaction between electrophile and thiolate anion; K D -equilibrium binding constant. 5°C values were taken from [19,24,27].  individual enzyme seldom catalyses turnover [5]. Should all GSTs perform catalysis at their full potential the substrate GSH would rapidly be depleted, leading to cell death. In kinetic terms, k cat is not under evolutionary pressure whereas k cat /K M electrophile is. The electrophile being in rapid equilibrium and k cat / K M linearly dependent on substrate reactivity is clearly the most efficient solution to protect from toxicity caused by reactive intermediates. Still, it is puzzling why a third of the site reactivity did evolve in MGST1 and MGST2 [25,28] as this lowers the efficiency correspondingly.
Intrinsically disordered region affecting GSH binding, activation and product release GSH binding is in rapid equilibrium whereas thiolate anion formation is very slow also at 30°C. The GSH binding site in MGST1 is covered by a loop between helix 1 and 2 as shown by electron crystallography data [21]. The electron density is not well-defined and we interpret this as reflecting either intrinsic disorder and/or different structural organization over GSH thiolate in one site of the homotrimer and protonated GSH in the other two. Recently the structure of MGST2 was solved demonstrating partial GSH occupancy [29] consistent with one catalytically competent GSH-bound per trimer [28] supporting the latter mechanism also in MGST1. Dynamics of the loop in MGST1 is necessary to permit entry of GSH and release of the product. Once the product is formed it rapidly leaves its active site, with data showing [25] that the product can, in addition, stimulate a neighbouring active site to stabilize GSH thiolate. This behaviour ensures the highest possible concentration of catalytically competent MGST1. There is of course no rationale for evolving slow GSH thiolate formation per se and it does not occur in cytosolic glutathione transferases [8]. These enzymes bind GSH in an open active site. We suggest that in MGST1, comparatively rapid structural transitions take place to bind three GSH molecules in low affinity mode with loops partially disordered. Then, slow structural transitions in one site are coupled with GSH thiolate formation and proton release resulting in tight binding. The other two protonated GSH molecules remain loosely bound and free to interchange rapidly with solvent. As the catalytically competent GSH thiolate is conjugated the GSH sulfur loses its negative charge, promoting the transition to a low affinity product binding mode and rapid product release [the binding affinity for the product CDNB conjugate is an order of magnitude lower compared with GSH (thiolate) [25]]. Thus, one rationale for a third of the site reactivity could be to avoid trapping the enzyme in a product-bound state. Given the broad substrate acceptance [30] and the observed tight binding of certain GSH conjugates [31]. This is a viable suggestion. In support, the MAPEG family member, unspecific MGST2 also displays a third of the site reactivity [28] whereas the specific, leukotriene C 4 synthase does not (all three site reactivity [32]). MGST1 is also a phospholipid hydroperoxide GSH peroxidase. During catalysis, a GSH sulphenic acid (ester) intermediate is postulated [33]. Applying the same reasoning as above, it might be of advantage to avoid trapping this uncharged intermediate. As a third of the site reactivity nevertheless does lower the overall conjugation capacity, the high expression of MGST1 in liver endoplasmic reticulum, mitochondrial outer membranes [34] and other cell membranes [35], (making it the single most abundant GST in this tissue) appears of consequence to fulfil protection from chemically reactive intermediates.

Temperature dependence (ratios of kinetic parameters at 5 and 30°C)
Having data for kinetic behaviour at 5 and 30°C enables a discussion on MGST1 temperature behaviour in comparison to other enzymes [36]. MGST1 is a mesophilic enzyme and the rate enhancement for each 10°C (Q 10 ) is expected to be around 2 [36]. The Q 10 values for MGST1 k 2 and k cat are, however, closer to 4. Thus, thiolate anion formation, a process we view as governed by structural transitions and subunit communication, displays a comparatively high Q 10 value. In support, large structural transitions that govern enzyme reactions are typically known to display high Q 10 values [36]. The values for k cat CDNB and CNAP are also high as expected since k cat is limited by GSH thiolate anion formation with these substrates. The high Q 10 value would imply that the enzyme could be even more active at higher temperatures still. We have indeed observed that MGST1 is highly active at higher temperatures (e.g. 50°C unpublished data). Thus, either enzyme structural transitions governing activity are increased per se or there is also a temperaturedependent shift to the activated form. Since the ratio of the unactivated and intrinsically activated enzymes was not determined with great precision this remains to be determined.
The chemical conjugation step k 3 is only increased marginally with temperature in the enzyme active site (Q 10 , 1.2-1) in line with the nonenzymatic chemical rates ( Table 5). The important kinetic parameter that governs the enzyme behaviour in vivo is k cat /K M , and since k cat increases markedly with temperature whereas K M decreases, the temperature rate behaviour of MGST1, i.e. high Q 10 , appears optimal in this respect.

Conclusion
By using a limited turnover experimental setup, we were able to characterize the pre-steady-state kinetic behaviour of MGST1 in detail at 30°C. In addition to consolidating the understanding of the global mechanism of MGST1, the kinetic model underscores how k cat /K M for reactive intermediates is under selective evolutionary pressure to ensure cellular protection by interception. We confirm that the enzyme is present in two forms, one that binds and releases GSH more rapidly (activated form; constitutes~10% of enzyme) and one that is more stable and slower in this regard (unactivated enzyme). The purified enzyme does behave as the enzyme in its natural environment since the specific activity in liver microsomes towards CDNB is in agreement with a 10% activated enzyme population. Our view of MGST1 as an efficient interception enzyme ties in very well with that of Brian Ketterer et al. [37] who analysed and put into perspective the behaviour of cytosolic GSTs. They concluded that the high amount and specific activities of GSTs (although sometimes modest), were quite sufficient for protecting against (conjugating) a range of reactive intermediates. In their analysis, reactive intermediates reached GSTs in the aqueous phase (membrane-bound GSTs were just being discovered). As electrophilic reactive intermediates are predominantly hydrophobic, a considerable fraction resides in membranes and membrane-bound GSTs can dominate metabolism [38] simply based on the partitioning of the substrate. Clearly, the ubiquitous and abundant cellular distribution, as well as kinetic behaviour of GSTs, is paramount to the interception of reactive intermediates.
Our kinetic analysis of MGST1 shows that k cat /K M , the kinetic parameter of physiological and toxicological significance for reactive intermediates, is directly governed by chemical reactivity (as is also the case for soluble GSTs [39]). . All other chemicals were of reagent grade and obtained from common commercial sources. The desthio analogue of glutathione (GOH), was a kind gift from Michael J. Goodman, Vanderbilt University, and synthesized as described previously [40].

Enzyme preparation
MGST1 was purified from male Sprague-Dawley rat livers as described previously [41] with the exception that 0.2% Triton X-100 was used during the final purification step. Before use, the buffer was exchanged to 0.1 M potassium phosphate buffer pH 6.5 containing 0.1% Triton X-100, 1 mM GSH, 0.1 mM EDTA, 20% glycerol using an Econo-PAC 10DG desalting column (Bio-Rad Laboratories AB, Solna, Sweden). Two enzyme batches were used throughout this study. For clarity, only batch 1 is shown in the main paper, whereas the data for batch 2 is given in the Supporting Information.

Enzyme activation
A peculiar feature of MGST1 is its ability to become activated by reactive molecules, proteolysis and even heat treatment [42]. Indeed, a fraction of the purified enzyme is inherently activated. As this fraction appears to vary, we have analysed two enzyme batches in parallel (all data for the second batch is summarized in the Supporting Information). Analysis of two batches of enzyme in this work also supports confidence for the extensive comparisons to data from earlier work (comprising many different enzyme batches).
Activation of MGST1 was performed by adding NEM to a final concentration of 5 mM at 4°C [17]. When the maximal activity was reached (approx. within 30 min), the reaction was stopped by adding an equimolar concentration of GSH, thus giving a final free thiol concentration of 1 mM. Subsequently, the buffer was exchanged to 0.1 M potassium phosphate buffer pH 6.5 containing 0.1% Triton X-100, 1 mM GSH, 0.1 mM EDTA, 20% glycerol using an Econo-PAC 10DG desalting column (Bio-Rad). The activated enzyme was kept on ice at all times and was prepared prior to the experiment on the same day.
Measurements were performed at 30°C in 0.1 M potassium phosphate buffer pH 6.5, containing 0.1% Triton X-100. Enzymatic activities were calculated after correction for the nonenzymatic reaction and were in general agreement with the values reported previously [20]. All measurements were taken in triplicate and slopes were fitted using the CARY WINUV software package (Agilent Technologies). These measurements were performed in order to validate the activity of the enzyme preparations and to estimate the degree of the activated enzyme. Specific activity was also measured in the stop-flow setup as described under 'Burst followed by steady' state in the following section.

Pre-steady-state kinetics
Measurements were performed on an SX20 stopped-flow system (Applied Photophysics, Leatherhead, UK) equipped with an XBO 150 W/4 (Osram, Munich, Germany) as described previously [24]. Between 50 and 100 μL from each of the two syringes were rapidly mixed in a 10-or a 2-mm path length cell, and absorbance was recorded at various wavelengths as indicated. The 2-mm cell was employed to allow for the high-background absorbance of the GSH thiolate anion when measuring at 239 nm. All experiments were performed at 30°C in 0.1 M potassium phosphate (pH 6.5), 20% glycerol, 0.1% Triton X-100 and 0.1 mM EDTA. All concentrations given are the resulting final concentrations in the observation cell.

Limited turnover
Absorbance was measured at 239 nm for 100 s to capture the whole enzymatic process: the initial burst and steadystate formation of the CDNB-GS conjugate followed by the rebinding of the thiolate anion (to yield K G and k 2 ). A final concentration of 11 μM MGST1 (trimer) and 20 μM CDNB was used and GSH was varied between 0.25 and 5 mM final. The following extinction coefficients were used: thiolate anion (ε 239 = 5000 M −1 Ácm −1 ) and CDNB (ε 239 = 2700 M −1 Ácm −1 ).
Burst followed by steady state MGST1 (trimer, 11 μM final) was rapidly mixed with varying concentrations of CDNB (12.5-1000 μM final) and CNAP (25-1000 μM final) in the presence of 0.5 mM GSH. Absorbance was measured at 340 nm for 10 s in the case of CDNB and at 297 nm for 30 s in the case of CNAP to measure the conjugation to GSH [CDNB: ε 340 = 9.6 mM −1 Ácm −1 ; CNAP: ε 297 = 11.9 mM −1 Ácm −1 (see also Fig. 5)]. The initial burst was fitted to an exponential yielding the chemical conjugation step (k 3 ) and electrophile affinity (K C ) followed by a linear part (representing the steady state, limited largely by GSH thiolate formation). The burst amplitude was determined using only the curves from lower CDNB concentrations and subsequently fixed at this value for plotting the observed 'rate constants' (k obs ) against the electrophile concentration to determine K C and k 3 for the chemical step. At high CDNB concentrations, part of the burst is hidden in the dead time of the instrument and thus not suitable to determine the amplitude.
Determination of GSH off rate (i.e. GSH thiolate reprotonation) Inhibitors GSO À 3 and GOH were used to outcompete the GSH thiolate. GSH-bound (0.5 mM) MGST1 (trimer, 11 μM final) was rapidly mixed with either GSO À 3 or GOH, respectively, to a final concentration of 10 or 20 mM. Protonation of the thiolate anion was followed at 239 nm (GSH thiolate: ε 239 = 5 mM −1 Ácm −1 ) until the decline stabilized (~100 s). The curves were subsequently fitted to a double exponential to determine the rate constants (k −2 ) for GSH thiolate loss from the activated and unactivated enzyme fractions.

Data simulation and analysis
Microscopic rate and steady-state rate constants (including standard errors) were determined by nonlinear regression using the program package GRAPHPAD PRISM 6 (GraphPad Software, San Diego, CA, USA). Macroscopic rate constants were also calculated from the model equation (see Scheme S1) and compared to the experimental data. In addition, we used KINTEK EXPLORER software to derive steady-state rate constants for a mixed (activated and unactivated) enzyme composition [43,44]. Specifically, we assumed a 10% activated enzyme fraction. To do this, we used the global kinetic model for MGST1 (Scheme 1) in KINTEK EXPLORER. Taking the pre-steady-state constants for activated and unactivated enzymes into account, we then generated a series of kinetics with either constant GSH and varying electrophile concentration or vice versa. The steady-state rate that followed the burst phase was plotted against the second substrate concentration and fitted to Michaelis-Menten kinetics to determine k cat and K M .

Supporting information
Additional supporting information may be found online in the Supporting Information section at the end of the article.