Mechanism of Cadmium-Mediated Inhibition of Msh2-Msh6 Function in DNA Mismatch Repair

Abstract

The observation that Cadmium (Cd²⁺) inhibits Msh2-Msh6, which is responsible for identifying base pair mismatches and other discrepancies in DNA, has led to the proposal that selective targeting of this protein and consequent suppression of DNA repair or apoptosis promote the carcinogenic effects of the heavy metal toxin. It has been suggested that Cd²⁺ binding to specific sites on Msh2-Msh6 blocks its DNA binding and ATPase activities. To investigate the mechanism of inhibition, we measured Cd²⁺ binding to Msh2-Msh6, directly and by monitoring changes in protein structure and enzymatic activity. Global fitting of the data to a multiligand binding model revealed that binding of about 100 Cd²⁺ ions per Msh2-Msh6 results in its inactivation. This finding indicates that the inhibitory effect of Cd²⁺ occurs via a nonspecific mechanism. Cd²⁺ and Msh2-Msh6 interactions involve cysteine sulfhydryl groups, and the high Cd²⁺:Msh2-Msh6 ratio implicates other ligands such as histidine, aspartate, glutamate, and the peptide back bone as well. Our study also shows that cadmium inactivates several unrelated enzymes similarly, consistent with a nonspecific mechanism of inhibition. Targeting of a variety of proteins, including Msh2-Msh6, in this generic manner would explain the marked broad-spectrum impact of Cd²⁺ on biological processes. We propose that the presence of multiple nonspecific Cd²⁺ binding sites on proteins and their propensity to change conformation on interaction with Cd²⁺ are critical determinants of the susceptibility of corresponding biological systems to cadmium toxicity.

Humans can be exposed to toxic levels of cadmium (Cd²⁺) from cigarette smoke and industrial pollution associated with the manufacture and use of materials such as nickel–cadmium batteries, plastics, pigments, metal alloys, and coatings. Cd²⁺ poses a severe and intractable health hazard because it can accumulate in and damage several tissues, including the liver, kidneys and lungs, pancreas, testis, and bone, and because it is a potent carcinogen (the biological half-life of Cd²⁺ is 15–30 years in humans). Cd²⁺ has been reported to alter cell growth, differentiation, and apoptosis by disrupting a variety of cellular processes, including signal transduction, gene regulation, responses to oxidative stress, and DNA processing and repair. For these reasons, the targets and mechanisms by which this heavy metal toxin effects cellular changes are subject to intense investigation (1–3).

The mutagenic effects of Cd²⁺ have been attributed, at least in part, to inhibition of DNA mismatch repair (MMR)¹ (4). This evolutionarily conserved DNA system is responsible for correcting base pair mismatches and insertion/deletion loops (IDL) formed during DNA replication and inducing cell cycle arrest and apoptosis in response to extensive DNA damage. Thus, inhibition of MMR increases genomic instability and increases susceptibility to cancer (5, 6). Recent studies of the effects of Cd²⁺ on MMR in eukaryotes suggest that the metal ion targets Msh2-Msh6, an essential protein in the system, and it has been proposed that Cd²⁺ binding to a specific site on Msh2-Msh6 blocks its activities (7, 8). The DNA repair reaction sequence involves (a) recognition of an error in a DNA strand, (b) incision of the strand in the vicinity of the error, (c) excision of the strand past the error, followed by (d) strand resynthesis. Msh2 and Msh6, eukaryotic homologues of MutS protein, form a heterodimer that primarily recognizes base pair mismatches and short IDLs (1–2 bases). Eukaryotic homologues of MutL, another conserved MMR protein (e.g., Saccharomyces cerevisiae Mlh1-Pms1 or Homo sapiens Mlh1-Pms2), have an endonuclease activity that preferentially acts on a DNA strand containing a nick and may create sites for initiation of excision in the vicinity of the mismatch (9). ExoI exonuclease catalyzes strand excision, and RPA single-stranded DNA binding protein, DNA polymerase δ, PCNA sliding clamp, and RFC clamp loader are responsible for DNA resynthesis. Msh2-Msh6 also recognizes mismatches containing damaged bases such as O6-methylguanine, and in the case of the apoptosis reaction sequence, debate continues as to whether the recognition event signals the cell cycle arrest and apoptosis machinery directly or perhaps indirectly through toxic DNA intermediates created by futile repair attempts (5, 6).

The mechanistic basis for suppression of DNA MMR by Cd²⁺ has yet to be resolved. The MMR pathway clearly has several potential protein targets for Cd²⁺-mediated inactivation, and alteration of DNA structure and dynamics by Cd²⁺ binding may be a contributing factor as well (10). Thus far, the focus of investigation has been on Msh2-Msh6, the best characterized of the core MMR proteins. The common model for prokaryotic MutS and eukaryotic Msh2-Msh6 actions in MMR is as follows: (a) MutS/Msh2-Msh6, likely in an ADP-bound state (11, 12), slides on DNA, interrogating base pair structure and/or dynamics to find mismatches (13); (b) conserved phenylalanine and glutamate residues help MutS/Msh2-Msh6 make contacts with base pairs and recognize a mismatch (14–17); (c) mismatch binding triggers replacement of ADP with adenosine 5′-triphosphate (ATP) (18) and stabilization of MutS/Msh2-Msh6 in an ATP-bound state (11, 12, 19); and, (d) ATP-bound MutS/Msh2-Msh6 activates MutL/Mlh1-Pms1 and initiates the next steps in MMR (18, 20, 21).

Recent studies of S. cerevisiae Msh2-Msh6, which is related closely to the human Msh2-Msh6 protein, indicate that its mismatched DNA binding and ATPase activities are suppressed in the presence of Cd²⁺, with one report suggesting that the ATPase activity is impacted more severely than mismatch binding (7, 8). The mechanism by which Cd²⁺ inhibits Msh2-Msh6 function is not known, although it has been speculated that selective binding of Cd²⁺ to a site on Msh2-Msh6, such as a hypothetical Zn²⁺ finger domain, could disrupt its interactions with ATP and DNA, either directly or through allostery.

To understand the mechanism of Msh2-Msh6 inactivation by Cd²⁺, which could in turn facilitate efforts to mitigate its toxicity, we undertook a quantitative analysis of the impact of Cd²⁺ on S. cerevisiae Msh2-Msh6. The specific aim was to identify Cd²⁺ binding site(s) on Msh2-Msh6 and determine how the interaction triggers changes in its DNA binding and ATP binding/hydrolysis activities. Surprisingly, our data revealed that Msh2-Msh6 has a very high capacity for binding Cd²⁺ ions in a nonspecific manner, and these interactions lead to alteration of protein structure and loss of function. Cd²⁺ blocks the activities of several other unrelated proteins in a similar fashion, consistent with the nonspecific nature of the inhibition mechanism. On the basis of these findings, we propose that interactions between Cd²⁺ and multiple binding sites on a protein that lead to changes in its conformation constitute a generic mechanism of Cd²⁺-mediated inhibition of function.

Experimental Procedures

Proteins, DNA, Nucleotides, and Other Reagents. S. cerevisiae Msh2-Msh6 (22), Thermus aquaticus MutS (11), S. cerevisiae RFC (22), and Escherichia coli phosphate binding protein (PBP) (23) were overexpressed and purified from E. coli, and PBP was labeled with N-[2(1-maleimidyl)ethyl]-7-(diethylamino)coumarin-3-carboxamide (MDCC) as described previously. E. coli γ complex and overexpression clones for E. coli MutS and MutL were gifts from Michael O'Donnell (The Rockefeller University). Synthetic DNAs (37 bp) were purchased from Integrated DNA Technologies, purified by denaturing polyacrylamide gel electrophoresis, and annealed to prepare duplex DNA substrates, as described (12). Radioactive nucleotides were purchased from Perkin-Elmer Life Sciences, and nonradioactive nucleotides and CdCl₂ were purchased from Sigma-Aldrich. Fluorescein-5-maleimide, SYBR Green, and Measure-iT cadmium assay kit were purchased from Invitrogen. PEI-Cellulose F thin-layer chromatography (TLC) plates were purchased from EM Science.

DNA Binding Assays

DNA binding to Msh2-Msh6 was measured by gel mobility shift assays. In 10 μL reactions containing buffer A (20 mM Tris-HCl, pH 7.5, 100 mM NaCl, and 5 mM MgCl₂), Msh2-Msh6 (0.2 μM) was incubated with DNA (0.16 μM) and CdCl₂ (0–400 μM) for 20 min at 25 °C, followed by electrophoresis at 4 °C on a native 4.5% TBE-acrylamide gel. The gels were stained with SYBR Green according to the manufacturer's protocol, and the DNA was visualized and quantified on a PhosphorImager (GE Healthcare).

ATPase Assays

The ATPase activity was measured under steady-state conditions at 30 °C with Msh2-Msh6 (0.03–2 μM) and 500 μM α³²P-ATP in buffer A with varying amounts of CdCl₂ (0–500 μM). PEI cellulose TLC (0.6 M potassium phosphate, pH 3.4) was used to resolve α³²P-ADP from α³²P-ATP, and k_cat was calculated from initial ADP production rates for each Msh2-Msh6 and CdCl₂ concentration (experiments with DTT were performed in the presence of 5 mM freshly dissolved DTT). Steady-state ATPase activities of other proteins (0.5 μM) were measured similarly; T. aquaticus MutS (40 °C; 50 mM Hepes-NaOH, pH 7.8, 100 mM KCl, and 5 mM MgCl₂); S. cerevisiae RFC [25 °C; 30 mM Hepes-NaOH, pH 7.5, 10 mM Mg(OAc)₂, and 5% glycerol]; E. coli γ complex (25 °C; 20 mM Tris-HCl, pH 7.5, 50 mM NaCl, and 8 mM MgCl₂); and, E. coli MutS and MutL (25 °C; 20 mM Tris-HCl, pH 8.0, 100 mM NaCl, and 5 mM MgCl₂).

The ATPase activity was measured under presteadystate conditions using a stopped-flow instrument (KinTek Corp., Austin TX) and MDCC-labeled PBP to detect phosphate (Pi) release (11). Forty microliters of Msh2-Msh6 (2 μM) and MDCC-PBP (20 μM) was mixed with 40 μL of ATP (1 mM) and CdCl₂ (0–1 mM) in buffer A at 30 °C, and the change in MDCC-PBP fluorescence on binding Pi was measured by excitation at 425 nm and emission at >450 nm (final concentrations: 1 μM Msh2-Msh6, 10 μM MDCC-PBP, 500 μM ATP, and 0–500 μM CdCl₂). The kinetic traces were fit to a burst equation ([Pi] = (A₀ e^−kt + Vt + F₀)/m_Pi) where [Pi] is the Pi concentration, A₀ is the amplitude, k is the observed burst rate constant, V is the velocity of the linear phase, F₀ is the initial fluorescence intensity, and m_Pi is the slope of the Pi standard curve measured under the same conditions.

Protein Structure Assays

The change in Msh2-Msh6 (0.05–2 μM) intrinsic fluorescence on titration with CdCl₂ (0–500 μM) was measured using a FluoroMax-3 spectrofluorometer (Horiba Jobin Yvon). Msh2-Msh6 in buffer A was mixed with increasing CdCl₂ for 1 min at 25 °C, followed by excitation at 290 nm and emission measured at 340 nm. Control experiments with buffer replacing CdCl₂ were performed concurrently to correct for photobleaching effects. The Cd²⁺ effect on cysteine labeling was measured by incubating Msh2-Msh6 (0.5 μM) with fluorescein-5-maleimide (2.5 μM) and 0–300 μM CdCl₂ in 20 μL of 50 mM potassium phosphate, pH 7.2, and 100 mM NaCl for 5 min at 25 °C. The reactions were quenched by adding SDS and DTT and analyzed by SDS-PAGE, and the fluorescein-labeled protein was quantified on a PhosphorImager.

Cadmium Binding Assays

Cadmium binding was assayed by incubating Msh2-Msh6 (1 μM) or MutS (1 μM) in 10 μL of buffer A with CdCl₂ (0–200 μM) for 10minat 25 °C and measuring the free Cd²⁺ remaining in solution by the fluorescence-based Measure-iT cadmium assay (Invitrogen), according to the manufacturer's protocol (lower detection limit, 1 pmol or 0.1 μM Cd ⁺ in 10 μL of solution). Free Cd²⁺ concentrations were determined from linear standard plots generated concurrently with each experiment, using both freshly prepared and the assay kit CdCl₂ stock solutions and corresponding bound Cd²⁺ concentrations plotted vs total CdCl₂.

Results

Exposure to Cadmium Changes Msh2-Msh6 Conformation and Suppresses Its Mismatch Binding and ATPase Activities

The effect of cadmium on S. cerevisiae Msh2-Msh6 was assessed by measuring the mismatched DNA binding, ATP binding, and hydrolysis activities of the protein, as well as monitoring its conformation via intrinsic fluorescence in the presence of CdCl₂. Gel mobility shift analysis (GMSA) revealed loss of Msh2-Msh6 binding to G:T mismatch-containing DNA [37 nucleotide (nt)] on titration with CdCl₂ (Figure 1A), and the Msh2-Msh6 · DNA complex was essentially undetectable at about 200 μM CdCl₂; a similar inhibition profile was observed for Msh2-Msh6 binding to fully matched DNA (data not shown). The saturating concentration of CdCl₂ is very high relative to that of Msh2-Msh6 (0.2 μM). However, Cd²⁺ ions may also bind DNA in the reaction via contacts with guanine-N7, the A:T base pair, and backbone ${\text{PO}}_4^{2-}$ groups, as dissociation constants for these interactions have been reported in the low micromolar range (10,24). Thus, it is possible that the GMSA reports Cd²⁺ binding to both protein and DNA, as altering the properties of either macromolecule could disrupt Msh2-Msh6·DNA complex.

Figure 1.

Cadmium changes Msh2-Msh6 conformation and inhibits its mismatch binding and ATPase activities. (A) The gel mobility shift analysis shows a decrease in the Msh2-Msh6·G:T DNA complex with increasing CdCl₂ in the reaction. (B) The steady-state Msh2-Msh6 ATPase activity decreases with increasing CdCl₂ (k_cat = 0.17 to 0.02 s⁻¹), and the data fit to a hyperbola yield K_1/2 = 9 μM. (C) The presteady state Pi release assays show a burst of Msh2-Msh6-catalyzed ATP hydrolysis at 1.3 s⁻¹ rate and 1.1 μM amplitude. Increasing CdCl₂ inhibits both the burst and the steady-state ATPase rates. (D) Intrinsic Msh2-Msh6 fluorescence decreases with increasing CdCl₂, with K_1/2 = 6 μM.

The ATPase activity of Msh2-Msh6 was measured in the absence of DNA, under steady-state conditions, and it was found to decrease from a k_cat value of 0.17 s⁻¹ to about 0.02 s⁻¹ with increasing CdCl₂ (Figure 1B). A plot of the k_cat values vs CdCl₂ concentration fit to a hyperbola yields an apparent inhibition constant K_1/2 of 9.2 ± 2 μM. We also measured Msh2-Msh6 ATPase activity under presteady-state conditions to determine which steps in the reaction are affected by Cd²⁺ (ATP binding to Msh2-Msh6 appears not to be disrupted significantly by Cd²⁺, according to data from nitrocellulose membrane filtration experiments; Supporting Information, Figure S1A). A series of stopped-flow experiments was performed in which the change in MDCC-labeled PBP fluorescence on binding Pi was used as a reporter of ATP hydrolysis and Pi release by Msh2-Msh6 (Figure 1C). As reported previously, in the absence of DNA, Msh2-Msh6 catalyzes a burst of ATP hydrolysis and Pi release (1.3 s⁻¹) in the first catalytic turnover, followed by the slower, linear steady-state phase (0.3 s⁻¹) (12). The amplitude of the burst phase decreases with increasing CdCl₂ in the reaction, as does the ATPase rate (Figure 1C). Rapid-quench experiments, which directly measure α³²P-ATP hydrolysis, yielded similar results (data not shown; ref 12). These data indicate that Cd²⁺ suppresses the ATP hydrolysis step in the reaction. Curiously, a substantially higher concentration of CdCl₂ is required to inhibit Msh2-Msh6 ATPase activity in presteady state experiments (Figure 1C) as compared with steady-state experiments (Figure 1B), which differ only in the concentration of Msh2-Msh6 in the reactions, i.e., 1 vs 0.03 μM, respectively; the reason for this discrepancy is clarified in the next section. In any case, the results all confirm that cadmium has a severe inhibitory effect on both the DNA binding and the ATPase activities of Msh2-Msh6.

The active sites for DNA binding and ATP binding/hydrolysis are positioned at opposite ends of Msh2-Msh6, across a distance of approximately 50 Å, according to the structure of the human protein (14). If Cd²⁺ binding to Msh2-Msh6 underlies the inhibitory effect, it is likely that the interaction induces changes in protein conformation that affect both active sites. To test this hypothesis, we assessed Msh2-Msh6 conformation in the presence of CdCl₂ by measuring its intrinsic fluorescence (Figure 1D). Titration of Msh2-Msh6 (0.05 μM) with CdCl₂ causes a decrease in protein fluorescence, and the data fit to a hyperbola yield a K_1/2 value of 5.5 ± 1 μM, which is comparable to the apparent inhibition constant determined from steady-state ATPase assays (Figure 1B, K_1/2 = 9.2 μM). Together, these data support the hypothesis that Cd²⁺ binding triggers conformational changes in Msh2-Msh6 that cause loss of both its ATPase and DNA binding activities.

Stoichiometry of Cadmium Binding to Msh2-Msh6 That Results in Loss of Function

According to the steady-state ATPase and protein fluorescence data, cadmium inhibits Msh2-Msh6 activities with an apparent inhibition constant in the 5–10 μM range. To determine the affinity and stoichiometry of the interaction between Cd²⁺ and Msh2-Msh6, we titrated the protein with increasing amounts of CdCl₂ and measured the concentration of free Cd²⁺ remaining in solution using a fluorescence-based assay. Surprisingly, cadmium binding saturates at a ratio of about 170–180 Cd²⁺ ions per Msh2-Msh6 (Figure 2A, 1 μM Msh2-Msh6 in the reaction). The inset in Figure 2A shows the isotherm at low CdCl₂ concentrations, and there is no detectable difference between the first Cd²⁺ ion and the subsequent Cd²⁺ ions binding to Msh2-Msh6. Thus, the interactions between Cd²⁺ and Msh2-Msh6 appear nonspecific, with no particular binding site displaying significantly higher affinity for Cd²⁺ than any other site on the protein. In a parallel experiment, we also measured Cd²⁺ binding to T. aquaticus MutS, a prokaryotic homologue of Msh2-Msh6, to compare and contrast Cd²⁺ effects on these evolutionarily related proteins. MutS also binds a large number of Cd²⁺ nonspecifically, although in this case binding saturates at a ratio of about 50 Cd²⁺ ions per MutS dimer (Figure 2A). One distinction between these two proteins is that the Msh2-Msh6 dimer has 27 cysteines while MutS dimer has two cysteines. Free thiol groups are known as good ligands of Cd²⁺ ions, and the difference in cysteine count between Msh2-Msh6 and MutS may be linked to differences in their interactions with and response to cadmium. This hypothesis is considered further in a later section.

Figure 2.

High stoichiometry of Cd²⁺ binding is required for inhibition of Msh2-Msh6 function. (A) Titration of Msh2-Msh6 and T. aquaticus MutS with CdCl₂ results in saturation of Cd²⁺ binding at ∼180 ions per Msh2-Msh6 and ∼50 ions per MutS dimer. The inset shows the binding isotherms at low CdCl₂ concentrations. (B) The ATPase activity of 0.03 (red), 0.5 (green), 1.0 (blue), and 2.0 μM (brown) Msh2-Msh6 and (C) the intrinsic fluorescenceof0.05 (red), 0.5 (green), and 2.0 μM (blue, brown, and cyan) Msh2-Msh6 in the presence of increasing CdCl₂. Experimental data and corresponding fits to amultisite binding model are shown. The best global fit of the data yielded an apparent dissociation constant K_D = 0.5 ± 0.14 μM for the Cd²⁺ binding sites, an equilibrium constant K_C = 4.0 ± 1.1 for protein conformation change, and the number of Cd²⁺ binding sites S= 107.9 ± 3.1. Simulations using these parameters are shown as solid lines of the same color as the corresponding data. (D) Protein fluorescence data for 2 μM Msh2-Msh6 are shown as circles, and a corresponding simulation using the best fit parameters is shown as a black line. Other lines show simulations performed with the same parameters except with the number of binding sites, S, set to 1(blue), 10(red), 50(cyan), and 150 (green).

The high Cd²⁺:Msh2-Msh6 binding ratio offers an explanation for our observation that far greater amounts of CdCl₂ appear necessary to block the ATPase activity of 1 vs 0.03 μM Msh2-Msh6 under otherwise identical reaction conditions (Figure 1). To explicitly determine the stoichiometry of Cd²⁺ ion binding required to inhibit Msh2-Msh6 function, we performed a series of experiments with varying protein concentrations and measured the effects of Cd²⁺ on both Msh2-Msh6 conformation and ATPase activity. The resulting titration curves showed an increasingly pronounced sigmoidal shape with higher Msh2-Msh6 concentrations (Figure 2B, C, 0.5, 1.0, and 2.0 μM). This protein concentration-dependent sigmoidicity can be explained by the fact that the changes in fluorescence and ATPase activity occur only after a threshold of Cd²⁺ ions bound to Msh2-Msh6 is reached.

The data were fit to the Hill equation, which is used commonly to analyze sigmoidal titration curves and estimate the number of ligands binding to a macromolecule. The fits yielded Hill coefficients ranging from 1.5 to 6.8 and apparent equilibrium dissociation constants from 58 to 290 μM for the data sets with 0.5 to 2 μM protein, respectively (note: Msh2-Msh6 is a stable dimer within this concentration range). These parameters have no physical meaning, although the Hill coefficient may provide a reliable estimate of the number of ligand binding sites in the case of extreme positive cooperativity (25). There is, however, no evidence to justify an assumption of cooperativity in Cd²⁺ binding to Msh2-Msh6. Therefore, we developed a simple model in which multiple, independent binding sites of equal affinity on Msh2-Msh6 are occupied by cadmium, after which the protein undergoes conformational change that alters its intrinsic fluorescence and renders it inactive.

As shown in reaction scheme, Cd²⁺ ions (L) bind to Msh2-Msh6 complex (M) with S binding sites. Once the binding sites are populated, the complex transforms into inactive form (M′L_S).

$$\begin{array}{c}\mathrm{M}+\mathrm{L}\rightleftharpoons \text{ML}\\ \text{ML}+\mathrm{L}\rightleftharpoons {\text{ML}}_2\\ ⋮\\ {\text{ML}}_{S-1}+\mathrm{L}\rightleftharpoons {\text{ML}}_S\\ {\text{ML}}_S\rightleftharpoons {\mathrm{M}}^{\prime }{\mathrm{L}}_S\end{array}$$ (1)

The fraction of altered, Cd²⁺-bound complex, [M′L_S]/[M_T], is calculated as described in the Appendix. The value F of the observed property of the reaction (either fluorescence or ATPase k_cat) is calculated as

$$F=F_0+\frac{[{\mathrm{M}}^{\prime }{\mathrm{L}}_S]}{{\mathrm{M}}_T}(F_1-F_0)$$ (2)

where F₀ and F₁ are values of the properties of the native protein in the absence of Cd²⁺ and for the altered Cd²⁺-bound protein, respectively (in the case of the ATPase activity, F₁ = 0).

To test the model and obtain accurate estimates of its parameters, all of the fluorescence and ATPase data were fit simultaneously to the model using gfit, a program designed for global analysis of data from different types of experiments (Supporting Information; 26). The best fit to the data yielded stoichiometry S = 107 ± 3 for Cd²⁺ binding to Msh2-Msh6 with an apparent dissociation constant K_D = 0.5 ± 0.1 μM and a conformational change equilibrium constant K_C = 4 ± 1 (Figure 2). The search for alternative good or better fits was performed by random restart method. The fitting procedure was repeated over 300 times with randomly chosen starting parameters. Each procedure produced either a very similar set of parameters or a significantly worse fit. Notably, the analysis provided a relatively narrow confidence interval for Cd²⁺ stoichiometry. Accordingly, simulations using different numbers of Cd²⁺ binding sites per Msh2-Msh6 (S = 1…150) show that the shapes of the titration curves are affected strongly by the value of S (Figure 2D, 2 μM protein). Thus, both the intrinsic fluorescence and the ATPase responses of Msh2-Msh6 to cadmium exposure indicate that changes in conformation and loss of function occur as a result of Cd²⁺ ions binding the protein at a high ratio of about 100:1.

Nonspecific Cadmium Binding Affects Several Proteins in a Similar Manner

The above findings argue against an inhibition mechanism whereby site-specific Cd²⁺ binding to Msh2-Msh6 directly impacts its ATPase or DNA binding activities. The lack of specificity evidenced by the high stoichiometry of Cd²⁺ ions necessary for Msh2-Msh6 inhibition implies that the heavy metal toxin may inactivate other proteins, unrelated to Msh2-Msh6, in a similar manner. We tested this hypothesis by assaying S. cerevisiae RFC complex, a clamp loader that catalyzes assembly of PCNA clamps onto DNA during replication and various DNA processing reactions, including DNA MMR. The data in Figure 3A show that CdCl₂ effects near identical suppression of both Msh2-Msh6 and RFC ATPase activities (0.03 μM proteins). A survey of yet other ATPases, assayed at higher protein concentrations of 0.5 μM, shows that correspondingly higher concentrations of CdCl₂ are required to inactivate S. cerevisiae RFC, E. coli MutS, E. coli MutL, and the E. coli γ complex clamp loader (Figure 3B). Interestingly, the ATPase activity of T. aquaticus MutS, which is a homologue of Msh2-Msh6 and also binds a large number of Cd²⁺ ions nonspecifically (Figure 2A), is comparatively resistant to inactivation under the same conditions. A possible reason for this difference is explored in the next section.

Figure 3.

Nonspecific cadmium binding inhibits different proteins in a similar manner. (A) S. cerevisiae Msh2-Msh6 and RFC clamp loader (0.03 μM) exhibit anidentical decrease in ATPase activity with increasing CdCl₂. (B) A variety of other proteins (0.5 μM) exhibit the same ATPase inhibition profile with increasing CdCl₂, except for T. aquaticus MutS.

Inhibitory Effects of Nonspecific Cadmium Binding Involve Sulfhydryl Groups on Msh2-Msh6

To characterize the Cd²⁺ interactions that lead to inhibition of Msh2-Msh6 function, we probed further the observed difference between Cd²⁺ binding to T. aquaticus MutS vs Msh2-Msh6 (Figure 2A). Numerous Cd²⁺ ions bind nonspecifically to both proteins, although the binding stoichiometry is lower for T. aquaticus MutS. Correspondingly, T. aquaticus MutS is less susceptible to inhibition by Cd²⁺, as its ATPase activity declines by only about 20% in the presence of 400 μM CdCl₂ vs greater than 90% decline for Msh2-Msh6 (Figure 4A, 0.5 μM proteins). We hypothesized that that the difference in the number of cysteines on T. aquaticus MutS relative to Msh2-Msh6 may contribute to differences in their Cd²⁺ binding stoichiometry and the impact of Cd²⁺ on their activities (2 vs 27 cysteines, respectively). Evidence supporting participation of multiple cysteine thiols on Msh2-Msh6 in Cd²⁺ binding was provided by a labeling experiment in which the protein was reacted with fluorescein-5-maleimide. As shown in Figure 4B, addition of increasing concentrations of CdCl₂ to the reaction blocks thiol labeling on both Msh2 and Msh6 subunits, consistent with their involvement in coordinating Cd²⁺ ions (labeling of 0.5 μM Msh2-Msh6 is reduced to <30%, in the presence of 300 μM CdCl₂). Furthermore, the ATPase activity of Msh2-Msh6 is resistant to high concentrations of CdCl₂ in the presence of DTT (Figure 4A). DTT likely competes effectively against thiol groups and other ligands on Msh2-Msh6, thereby protecting the protein against Cd²⁺. These results, together with the observation that all of the ATPases tested above have several cysteine residues with the exception T. aquaticus MutS, suggest that participation of thiol groups in nonspecific interactions between proteins and Cd²⁺ may play a key role in mediating its inhibitory effect on the proteins (S. cerevisiae RFC, 28 cysteines/pentamer; S. cerevisiae Msh2-Msh6, 27/dimer; E. coli γ complex, 27/pentamer; E. coli MutL, 14/dimer; E. coli MutS, 12/dimer; and T. aquaticus MutS, 2/dimer).

Figure 4.

Thiol groups are involved in cadmium-mediated inhibition of Msh2-Msh6. (A) Cd²⁺ has a relatively minor effect on T. aquaticus MutS (□) as compared with S. cerevisiae Msh2-Msh6 ATPase (●). The addition of reduced DTT to the reaction protects Msh2-Msh6 from Cd²⁺-mediated inhibition (▴). (B) Cd²⁺ binding blocks reaction between cysteines on Msh2-Msh6 and fluorescein-5-maleimide. SDS-PAGE analysis shows reduction in fluorescein-labeled Msh2-Msh6 with increasing CdCl₂ to <30% labeled protein at 300 μM CdCl₂; Coommassie staining of the same gel shows equivalent amounts of protein in each reaction.

Discussion

Multiple Targets of Cadmium-Mediated Toxicity

The potent toxic effects of cadmium, including disruption of DNA metabolic reactions leading to mutations and chromosomal aberrations, make it a group I carcinogen, classified as a hazard to humans by the International Agency for Research on Cancer (1,2). Several targets of Cd²⁺ have been identified, and various mechanisms have been proposed in the literature to account for Cd²⁺-mediated toxicity. Cd²⁺ does not participate directly in redox reactions under physiological conditions, but it induces reactive oxygen species, likely via inhibition of antioxidant enzymes, and can thus promote nucleic acid, lipid, and protein damage (27). In favor of a more direct mechanism, there is evidence for Cd²⁺ replacing Ca²⁺ or Zn²⁺ in metallo-proteins and interfering with Ca²⁺ transport (28) and Ca²⁺-dependent signaling (29) or the function of Zn²⁺-dependent enzymes and Zn²⁺ finger-containing transcription factors (30). Other examples illustrating the broad spectrum of cadmium toxicity include altered expression of transcription factors (c-fos, c-jun, and c-myc proto-oncogenes) and translation factors (initiation factor 3 and elongation factor 1-δ) (1). Cd²⁺ can also disrupt extracellular processes such as cell–cell adhesion by interfering with the function of E-cadherin (31). Importantly, Cd²⁺ is known to disrupt the activities of tumor suppressor p53 (32), which would significantly increase the risk of carcinogenesis.

It has become increasingly apparent in recent years that Cd²⁺ targets a host of DNA repair mechanisms as well, including base excision repair (BER), nucleotide excision repair (NER), and MMR (3). In BER, for example, Cd²⁺ suppresses formamidopyrimidine DNA glycosylase (33), which catalyzes removal of damaged bases, as well as 8-oxo-dGTPase (34), which hydrolyzes 8-oxo-dGTP to 8-oxo-dGMP, minimizing its misincorporation into DNA. Other reported BER targets include AP endonuclease (35), DNA polymerase β (36), and DNA ligase (37). In NER, the DNA damage recognition protein XPA is inactivated by Cd²⁺ (33), and other key NER proteins, including TFIIH and TFIIA, and proteins responsible for repair DNA synthesis, appear to be affected as well (3). More recently, Cd²⁺ was found to inhibit the MMR system (4). Thus far, only the eukaryotic mismatch recognition protein Msh2-Msh6 has been identified clearly as a target of Cd²⁺ (7, 8). Together, these studies highlight the marked destructive impact of cadmium on critical processes responsible for DNA damage response and repair and help to explain the high level of mutagenesis, genome instability, and cancer risk associated with exposure to this toxin.

The brief overview of known Cd²⁺ targets highlights the fact that this metal toxin can interfere with the function of a wide variety of proteins, and ongoing efforts in several laboratories are aimed at identifying the corresponding mechanisms of inhibition. Cd²⁺ may bind specific site(s) on a protein and disrupt its function, as appears to be the case for Cd²⁺ binding to the Zn²⁺ finger motif in XPA (38). However, most of the proteins identified as Cd²⁺ targets are not canonical metal-binding proteins, and the exact mechanisms by which Cd²⁺ alters their structure and/or function remain unknown. This is also true in the case of Msh2-Msh6, where it has been speculated that Cd²⁺ may directly bind a critical residue or compete with Mg²⁺ in the ATPase site or bind a hypothetical Zn²⁺ finger motif, thereby disrupting its function in DNA repair (7,8).

Model Mechanism for Nonspecific Cadmium Binding-Mediated Disruption of Protein Structure and Function

In an effort to determine the mechanism by which cadmium inhibits MMR, we quantified Cd²⁺ binding to S. cerevisiae Msh2-Msh6 and assessed its effects on protein conformation and activity. In parallel, we also examined a prokaryotic homologue, T. aquaticus MutS, with the goal of identifying possible evolutionarily conserved features of the inhibition mechanism. Initial DNA binding and ATPase data indicated that inhibition requires high concentrations of Cd²⁺ relative to Msh2-Msh6 (Figure 1). Direct measurement of Cd²⁺ binding showed that the interaction saturates at 170–180 Cd²⁺ ions per Msh2-Msh6 (Figure 2A). A high ratio of Cd²⁺: Msh2-Msh6 was also found necessary for alteration of Msh2-Msh6 structure and activity. Isotherms of intrinsic Msh2-Msh6 fluorescence and ATPase activity from titrations with CdCl₂ showed increasing sigmoidal character at higher protein concentrations, consistent with Cd²⁺ ions binding to multiple sites on Msh2-Msh6 (Figure 2B, C). Global analysis of all of the data yielded good fits to a minimal model, in which binding of multiple Cd²⁺ ions to Msh2-Msh6 is followed by change in protein conformation associated with loss of function. The results show that a strikingly high stoichiometry of about 100 Cd²⁺ ions (S = 107) is required to bind Msh2-Msh6 with low micromolar affinity to induce loss of function (apparent K_D = 0.5 μM).

Nonspecific Interactions between Proteins and Cadmium

These findings raise interesting questions regarding the mechanism of cadmium-mediated inhibition; for example, on the nature of the Cd²⁺ binding sites on Msh2-Msh6 and, more importantly, the likelihood that an inhibition mechanism that involves Cd²⁺ binding Msh2-Msh6 at a ratio of about 100:1 is specific to this particular protein.

Several studies have shown that free sulfhydryl groups on cysteine residues are good ligands of Cd²⁺ ions, since cadmium, a soft acid, interacts strongly with soft donor atoms like sulfur (39). Other known biological ligands of Cd²⁺ ions include the methylsulfanyl group on methionine, amino, and carboxyl groups, as well as carbonyl groups in the peptide bond (40). We find that Cd²⁺blocks the reaction between fluorescein-5-maleimide and cysteines on Msh2-Msh6 (Figure 4), consistent with a recent report of thiol involvement in the Cd²⁺-Msh2-Msh6 interaction; the authors found that addition of free cysteine to the reaction protects Msh2-Msh6 from Cd²⁺, as does histidine to slightly lesser extent (8). In addition to 27 cysteines, S. cerevisiae Msh2-Msh6 has 57 methionine and 47 histidine residues. Thus, a variety of Cd²⁺–Msh2-Msh6 complexes are possible, with coordination numbers ranging from 2 to 8. Ligands in solution may also be involved, as illustrated by association of an excessive number of adenosine 5′-3-O-(thio)-triphosphate (ATPγS) molecules with Msh2-Msh6 in the presence of CdCl₂ (Supporting Information, Figure S1B). At 200 μM CdCl₂, the ratio is 13 ATPγS per Msh2-Msh6, which is remarkable given that the dimer has only two specific binding sites for nucleotides (12). This phenomenon can be explained by the participation of ATPγS molecules in multiple Cd²⁺ coordination complexes formed on Msh2-Msh6. Given the number of ions bound and the broad variety of ligands involved, binding of Cd²⁺ to Msh2-Msh6 has to be defined as a nonspecific interaction. Similar nonspecific interactions with Cd²⁺ have been reported for the metal-binding protein metallothionein (MT), which is best known for chelating seven Cd²⁺ ions with low nanomolar affinity via multiple sulfur atoms; however, the binding does not saturate even at 20 mol equiv of Cd²⁺ ions per MT (41).

Despite their nonspecific nature, the Cd²⁺–Msh2-Msh6 interactions are characterized by fairly high affinity (apparent K_D = 0.5 μM), suggesting that other proteins could be susceptible to this type of Cd²⁺ inhibition mechanism. Consistent with this hypothesis, higher concentrations of other proteins required correspondingly higher Cd²⁺ concentrations for inactivation (Figure 3). T. aquaticus MutS, which has only one cysteine per monomer, presented an interesting case as it binds multiple Cd²⁺ ions, but its ATPase activity is relatively resistant to Cd²⁺-mediated inactivation (Figures 2A and 4A). It is possible that nonspecific Cd²⁺ coordination complexes that include one or more sulfur atoms (as terminal or bridging ligands) play a special role in Cd²⁺-mediated alteration of protein structure and function. Another possible reason for the differential effect of cadmium on T. aquaticus MutS is that it is a thermostable protein and may thus be more resistant to conformational changes on nonspecific Cd²⁺ binding, as compared with Msh2-Msh6 and the other proteins tested in this study. This hypothesis is consistent with the observation that at very high CdCl₂ concentrations, Cd²⁺ binding-mediated changes in Msh2-Msh6 conformation eventually lead to protein aggregation over time (detected as increasing turbidity at 630 nm), whereas in the case of MutS, no such structural perturbation can be detected even at 2 mM CdCl₂ (Supporting Information, Figure S2).

Stoichiometry and affinity constants defining Cd²⁺ interactions have been reported for specific metal-binding proteins, such as MTs, as well as Cd²⁺-binding peptides and small molecules, but not for a majority of the proteins reported as targets of Cd²⁺ toxicity. It is therefore difficult to determine from past in vitro studies whether these proteins are inhibited by a nonspecific mechanism similar toMsh2-Msh6. Moreover, in many cases, the data are generated from reactions containing DTT, which may or may not be in reduced form, further complicating interpretation (7). According to most studies, micromolar concentrations of Cd²⁺ are sufficient to inactivate the protein in question, although the reported concentration range can vary widely, even for closely related proteins. For example, we can estimate a K_1/2 ∼ 5 μM from a study of Cd²⁺-mediated inhibition of human 8-oxoguanine-DNA glycosylase (estimated hOGG1 concentration = 1 nM) (42), and a K_1/2 > 100 μM for mouse 8-oxoguanine-DNA glycosylase (reported mOGG1 concentration = 0.5–2 nM) (43). In case of in vitro studies using cell-free extracts, estimates of stoichiometry or binding constants are further complicated by the presence of Cd²⁺ binding molecules other than the protein in question. For example, Cd²⁺ apparently inhibits the cleavage activity of purified N-methylpurine-DNA glycosylase (reported MPG concentration = 0.35 μM) with a K_1/2 ∼ 100 μM, whereas in a Hela cell-free extract, which likely has lower amounts of MPG, the K_1/2 is ∼ 250 μM (44). Notably, all of these proteins have multiple cysteine, methionine, and histidine residues, and in light of our results with Msh2-Msh6 and other ATPases, the reported range of Cd²⁺ concentrations is consistent with a nonspecific binding and inhibition mechanism. In the case of in vivo studies, the addition of low micromolar concentrations of Cd²⁺ to the growth medium can lower the activity of the protein under investigation. However, the concentration of cadmium inside cells can vary widely depending on uptake by transporters and interaction with other ligands (45). Thus, it is difficult to ascertain inhibition constants based on in vivo data, and the nonspecific Cd²⁺ binding mechanism proposed here could well be applicable, and may even be predominant.

Cadmium can clearly have a disastrous impact on cellular processes by affecting many different proteins. It has been proposed that MMR is a specific target of cadmium toxicity, since exposure to CdCl₂ leads to high mutability in S. cerevisiae, while DNA recombination levels and cell viability remain relatively stable (4). Our investigation indicates that Msh2-Msh6 is not affected specifically by Cd²⁺, but is instead subject to a generic mechanism of inhibition. Results from both in vivo and in vitro analysis are likely influenced by variation in the stoichiometry and affinity of nonspecific Cd²⁺ binding sites on different proteins, the ability of proteins to adopt “inactive” conformations on Cd²⁺ binding, the concentration of proteins in the cell or reaction, the status of redundant or compensatory protective mechanisms against DNA damage and cytotoxicity, and the status of protective mechanisms against metal ion toxicity. To unequivocally identify a specific target of cadmium toxicity in vitro, which is distinguished by a selective Cd²⁺ binding and inhibition mechanism, it is important to assess the Cd²⁺ binding stoichiometry and affinity of the protein, its amino acid composition, its propensity for conformational changes, as well as the influence of other ligands in the reaction, such as DTT. To identify a specific target of cadmium toxicity in vivo, it is important to assess the extent to which Cd²⁺ accumulates inside cells and its impact on the cellular redox status, as well as compare its effects on unrelated proteins under the same experimental conditions, particularly if they correlate with protein concentrations. These are all pertinent indicators of the type of mechanism underlying the impact of cadmium on a biological system and, therefore, inform development of strategies to mitigate its toxicity.

Generic Antidotes for a Generic Poison

The primary mechanism of cadmium detoxification in various organisms is chelation by thiol-containing molecules such as glutathione (GSH) and metalloproteins (MT) (46–48). Cd²⁺ interacts predominantly with the sulfhydryl and amino groups on GSH and can make weak contacts with the glutamyl and glycyl carboxylic acid groups. Dissociation constants for Cd²⁺-GSH interaction are in the low micromolar range(47,49), which implies that GSH would be ineffective as a competitor against proteins that bind Cd²⁺ with similar affinity. However, the GSH concentration can range from 1 to 10 mM inside cells, and the ratio of reduced GSH to oxidized GSSG in the cytosol is maintained at 100:1. Thus, high amounts of GSH are available to mop up Cd²⁺ and protect nonspecific Cd²⁺-binding proteins from its toxic effects. Moreover, exposure to Cd²⁺ results in strong stimulation of cysteine and GSH biosynthesis, consistent with the hypothesis that GSH is a key part of the defense mechanism against Cd²⁺ toxicity (46). MTs are ubiquitous transition metal-binding proteins characterized by high cysteine content and recognized for their role in homeostasis of essential metals and protection against toxic metals (48). Mammalian MTs are reported to bind up to seven Zn²⁺ or Cd²⁺ ions in metal–thiolate clusters with dissociation constants in the low nanomolar range (50). Such efficient interaction with Cd²⁺ would allow MTs to compete effectively with nonspecific Cd ⁺-binding proteins and thus block its toxic effects, at least until excessive cadmium exposure and accumulation overwhelms these defense mechanisms.

In summary, our results confirm that Cd²⁺ binding inactivates Msh2-Msh6, which in turn would disrupt MMR and lead to a high mutator phenotype and increased risk of carcinogenesis. We have discovered that the mechanism of inhibition does not involve binding of Cd²⁺ to a particular site on Msh2-Msh6 but rather binding of a large number of Cd²⁺ ions to multiple sites, which leads to changes in protein conformation and loss of function. Not surprisingly, this inhibition mechanism is not unique to Msh2-Msh6, and Cd²⁺ inactivates several unrelated proteins in a similar manner. These findings highlight the critical role that GSH and MT play in shielding cellular processes against cadmium toxicity, by competing effectively against nonspecific Cd²⁺-binding proteins and thus sparing them from catastrophic disruption of structure and function.

Appendix

The model of Msh2-Msh6 inhibition by cadmium uses the following assumptions:

• – The Msh2-Msh6 complex has S binding sites for Cd²⁺ with equal affinity K_A.

• – Binding of Cd²⁺ to each site can occur independently.

• – After all S sites are occupied, unfolding can occur with the equilibrium constant K_C.

• – Unfolding occurs in one step, which renders the Msh2-Msh6 complex enzymatically inactive and changes its intrinsic tryptophan fluorescence.

• – Properties of folded species are not affected by the number of bound Cd²⁺.

The fraction of unfolded Msh2-Msh6 complex [M′L_S]/M_T is calculated using concentrations of all intermediate complexes. An intermediate with i Cd²⁺ bound is formed in the reaction:

$${\text{ML}}_{i-1}+\mathrm{L}\rightleftharpoons {\text{ML}}_i$$ (S1)

Because multiple empty sites may be present in the left-hand-side complex and the ligand can bind to any one of them, the apparent equilibrium constant for complex formation is directly proportional to S−i + 1, the number of unoccupied sites. The reverse reaction can occur by dissociation of any of i ligands from the right-hand-side complex. Therefore, at equilibrium:

$$K_{\mathrm{A}}\frac{S-i+1}{i}=\frac{[{\text{ML}}_i]}{[{\text{ML}}_{i-1}][\mathrm{L}]}$$ (S2)

Concentrations of intermediate complexes:

$$\begin{array}{c}[\text{ML}]=[\mathrm{L}]K_{\mathrm{A}}S[\mathrm{M}]\\ [{\text{ML}}_2]=[\mathrm{L}]K_{\mathrm{A}}\frac{S-1}{2}[\text{ML}]=[\mathrm{M}]{([\mathrm{L}]K_{\mathrm{A}})}^{2}S\frac{S-1}{2}\\ [{\text{ML}}_3]=[\mathrm{L}]K_{\mathrm{A}}\frac{S-2}{3}[{\text{ML}}_2]=[\mathrm{M}]{([\mathrm{L}]K_{\mathrm{A}})}^{3}S\frac{S-1}{2}\frac{S-2}{3}\\ ⋮\\ [{\text{ML}}_i]=[\mathrm{M}]{([\mathrm{L}]K_{\mathrm{A}})}^i\left(\begin{array}{c}S\\ i\end{array}\right)\\ ⋮\\ [{\text{ML}}_S]=[\mathrm{M}]{([\mathrm{L}]K_{\mathrm{A}})}^S\\ [{\mathrm{M}}^{\prime }{\mathrm{L}}_S]=[{\text{ML}}_S]K_{\mathrm{C}}=[\mathrm{M}]{([\mathrm{L}]K_{\mathrm{A}})}^S{K}_{\mathrm{C}}\end{array}$$ (S3)

where binomial coefficient $\left(\begin{array}{c}S\\ i\end{array}\right)=\frac{S!}{(S-i)!i!}$

Total concentration of the complex:

$$\begin{array}{c}{M}_T=[\mathrm{M}]+[{\text{ML}}_1]+\cdots +[{\text{ML}}_S][{\mathrm{M}}^{\prime }{\mathrm{L}}_S]=\\ [\mathrm{M}](1+\sum _{i=1}^S{([\mathrm{L}]K_{\mathrm{A}})}^i(\begin{array}{c}S\\ i\end{array})+K_{\mathrm{C}}{([\mathrm{L}]K_{\mathrm{A}})}^S)=\\ =[\mathrm{M}]({([\mathrm{L}]K_{\mathrm{A}}+1)}^S+K_{\mathrm{C}}{([\mathrm{L}]K_{\mathrm{A}})}^S)\end{array}$$ (S4)

Total concentration of ligand:

$$\begin{array}{c}{\mathrm{L}}_T=[\mathrm{L}]+[\text{ML}]+\cdots +i[{\text{ML}}_i]+\cdots +S[{\text{ML}}_S]+S[{\mathrm{M}}^{\prime }{\mathrm{L}}_S]=\\ =[\mathrm{L}]+[\mathrm{M}]\left(\sum _{i=1}^{S}i{([\mathrm{L}]K_{\mathrm{A}})}^i\right(\begin{array}{c}S\\ i\end{array})+K_{\mathrm{C}}S{([\mathrm{L}]K_{\mathrm{A}})}^S)=\\ =[\mathrm{L}]+[\mathrm{M}]S([\mathrm{L}]K_{\mathrm{A}}{([\mathrm{L}]K_{\mathrm{A}}+1)}^{S-1}+K_{\mathrm{C}}{([\mathrm{L}]K_{\mathrm{A}})}^S)\end{array}$$ (S5)

Combining eqs S4 and S5 produces an expression for ligand binding density, from which the variable [M] can be eliminated.

$$\frac{L_{\mathrm{T}}-[\mathrm{L}]}{M_{\mathrm{T}}}=\frac{S[\mathrm{L}]K_{\mathrm{A}}{([\mathrm{L}]K_{\mathrm{A}}+1)}^{S-1}+S{K}_{\mathrm{C}}{([\mathrm{L}]K_{\mathrm{A}})}^S}{{([\mathrm{L}]K_{\mathrm{A}}+1)}^S+K_{\mathrm{C}}{([\mathrm{L}]K_{\mathrm{A}})}^S}$$ (S6)

Equation S6 has one unknown variable, [L], and can be solved numerically. However, to avoid computational overflow errors, substitute

$$x=\frac{[\mathrm{L}]K_{\mathrm{A}}}{[\mathrm{L}]K_{\mathrm{A}}+1}$$ (S7)

and

$$\frac{L_{\mathrm{T}}}{M_{\mathrm{T}}}=S+\frac{S(x-1)}{K_{\mathrm{C}}{x}^S+1}+\frac{x}{K_{\mathrm{A}}{M}_{\mathrm{T}}(1-x)}$$ (S8)

Equation S8 was numerically solved for x using the fzero MATLAB function. The fraction of unfolded complex was calculated as

$$\frac{[{\mathrm{M}}^{\prime }{\mathrm{L}}_S]}{M_{\mathrm{T}}}=\frac{K_{\mathrm{C}}{([\mathrm{L}]K_{\mathrm{A}})}^S}{{([\mathrm{L}]K_{\mathrm{A}}+1)}^S+K_{\mathrm{C}}{([\mathrm{L}]K_{\mathrm{A}})}^S}=\frac{K_{\mathrm{C}}S{x}^S}{K_{\mathrm{C}}{x}^S+1}$$ (S9)

The model, data and results of the analysis are available at http://gfit.sourceforge.net.