Isotope-coded, iodoacetamide-based reagent to determine individual cysteine pK_a values by MALDI-TOF mass spectrometry

Abstract

Cysteine reactivity in enzymes is imparted to a large extent by the stabilization of the deprotonated form of the reduced cysteine (i.e. the thiolate) within the active site. While this is likely to be an important chemical attribute of many thiol-based enzymes including cysteine-dependent peroxidases (peroxiredoxins) and proteases, only relatively few pK_a values have been determined experimentally. Presented here is a new technique for determining the pK_a value of cysteine residues through quantitative mass spectrometry following chemical modification with an iodoacetamide-based reagent over a range of pH buffers. This isotope-coded reagent, N-phenyl iodoacetamide (iodoacetanilide), is readily prepared in deuterated (d₅) and protiated (d₀) versions and is more reactive toward free cysteine than is iodoacetamide. Using this approach, the pK_a values for the two cysteine residues in Escherichia coli thioredoxin were determined to be 6.5 and > 10, in good agreement with previous reports using chemical modification approaches. This technique allows the pK_a of specific cysteine residues to be determined in a clear, fast, and simple manner and, because cysteine residues on separate tryptic peptides are measured separately, is not complicated by the presence of multiple cysteines within the protein of interest.

Reactive cysteine residues in proteins are typically stabilized to at least some degree in their deprotonated, thiolate form (i.e, with the pK_a of the thiol lowered from an unperturbed value of about 8.7), and are important players in a variety of enzymes catalyzing redox and hydrolytic reactions through an initial nucleophilic attack on substrate [1; 2]. Unfortunately, pK_a values for catalytically important cysteine residues have been determined in only a small number of cases relative to the large number of representatives for which knowledge of this value would be of biological and chemical significance.

Methods for determining pK_a values of cysteinyl residues in proteins typically involve the measurement of a change associated with the varying thiol/thiolate content of the protein of interest exposed to buffers over a range of pH values. Changes associated with alteration of the protonation state of a cysteine include absorbance at 240 nm, where the thiolate exhibits a modestly higher absorbance than the thiol [3; 4], NMR chemical shift changes associated with protonation/deprotonation [5; 6; 7], or changes in rates of reaction of the cysteine of interest with electrophilic chemical modifying agents wherein the more nucleophilic thiolate anion is reactive but the protonated thiol is not [8; 9; 10; 11]. In the latter case, this approach involves monitoring the progress of the chemical modification over time at each pH value; iodoacetate or its uncharged counterpart, iodoacetamide, are typically used in such studies, and progress of the reaction is measured by various methods including functional assays [11; 12; 13] or HPLC to separate unmodified from singly or doubly modified protein [14]. Particularly well studied is the thioredoxin family of proteins, where the more N-terminal of the cysteinyl residues in the CXXC motif exhibits a pK_a value as low as 3.3 – 3.5 for the oxidizing protein DsbA [10; 14] or as high as 6.7 – 7.5 for the highly reducing thioredoxins [5; 7; 15; 16].

Considering methods available for pK_a analyses, we sought to develop an approach to cysteinyl pK_a determination based on iodoacetamide modification that takes advantage of peptide analyses by quantitative mass spectrometry. Our test protein for these studies was thioredoxin from E. coli for which chemical modification data with iodoacetamide has been previously reported [11; 14]. With the approach introduced here, proteolytic digestion and matrix assisted laser desorption ionization – time of flight (MALDI-TOF) mass spectrometry (MS) of proteins is used to monitor the extent of reaction of specific cysteinyl residues at various time points, an approach that requires very little protein and accurately maps the modification(s) to one or more specific peptides. The iodoacetamide derivative used herein can be readily synthesized in two “isotope-coded” forms, incorporating no deuterium (d₀) or five deuterium atoms (d₅) into the N-phenyl iodoacetamide (iodoacetanilide) reagent to distinguish the products, similar to the original “isotope coded affinity tag” (ICAT)² technique introduced by Aebersold and colleagues [17]. In order to provide the data needed to evaluate pK_a by chemical modification, a fully or partially modified d₅-iodoacetanilide-labeled version of the protein is added in constant amounts to each quenched test sample taken over a time course of d₀-iodoacetanilide reaction, tryptic digests are analyzed by MALDI-TOF MS for their ratio of light to heavy modified peptide to monitor reaction progress, and the modification reaction is evaluated over a range of pH values. With this approach, multiple cysteinyl sites on the same or separate peptides in a digest can be simultaneously monitored, obviating the need to study mutant proteins where one or more target cysteine residues have been replaced by serine or alanine. This mass spectrometry-based approach thus provides a readily synthesized tool and new analytical technique for conducting pK_a studies of cysteine-containing proteins of interest.

MATERIALS AND METHODS

Materials

Trichloracetic acid (TCA), dihydroxybenzoic acid, dibasic sodium phosphate, boric acid, ammonium sulfate, sodium citrate, 2-mercaptoethanol, aniline, iodoacetic acid and iodoacetamide were from Sigma-Aldrich. Ammonium bicarbonate, cysteine, and EDTA were from Research Organics. Dithiothreitol was from Anatrace, Inc. (Maumee, OH). Acetonitrile and dimethylsulfoxide were from Fisher. 5,5’-dithiobis(2-nitrobenzoic acid) (DTNB) and sinapic acid were from Fluka. Trifluoroacetic acid and tris(2-carboxyethyl)phosphine (TCEP) were from Pierce Chemical Co. NADPH was from Roche and ether was from Mallinckrodt. Perdeuterated aniline was from Cambridge Isotope Laboratories (Andover, MA).

Iodoacetanilide (also denoted d₀-iodoacetanilide) was prepared from aniline and iodoacetic acid using dicyclohexylcarbodiimide (DCC) coupling (Scheme 1) as previously described [18]. Substitution of per-deuterated aniline for aniline in this procedure yields d₅-iodoacetanilide. E. coli thioredoxin reductase was expressed and purified as described previously [19; 20]. E. coli thioredoxin was expressed from E. coli strain CHW170 [21] which was a gift from Dr. Charles H. Williams at the University of Michigan (this is a wild-type E. coli strain transformed with an expression plasmid with wild-type thioredoxin under control of a temperature-sensitive λ-promoter). Thioredoxin was purified as previously described [22] with the exception that all chromatography was conducted by FPLC at 4 °C, a 75 ml Q-Sepharose HP column was used instead of a 200 ml column, and a 250 ml Suparose 12 prep grade FPLC column was used instead of a Sephadex G50 size exclusion column. Pure protein was concentrated to 10 mg/ml in 20 mM potassium phosphate buffer, 1 mM EDTA, pH 7.6 and stored in aliquots at −80 °C.

Scheme 1.

Comparison of rates of reaction of iodoacetamide and iodoacetanilide with free cysteine and with reduced thioredoxin

For analysis of reaction rates of each reagent with free cysteine, reaction mixtures included 50 µM cysteine in 5 ml of buffer containing 10 mM sodium phosphate, 10 mM boric acid, 10 mM sodium citrate, 1 mM EDTA, and 100 mM ammonium sulfate (BPACE buffer) at pH 7 to which iodoacetamide (0.25 to 2 mM) or iodoacetanilide (0.25 to 1 mM) (from 200 mM stocks prepared in dimethylsulfoxide) was added and incubated at room temperature. An aliquot (0.3 ml) was removed at each time point and the content of free thiols was determined by the addition of 0.2 ml of a solution containing 750 µM 5,5'-dithiobis(2-nitrobenzoic acid) (DTNB, from a 50 mM stock in dimethylsulfoxide) and spectral monitoring (on an Agilent HP8453 diode array spectrophotometer) of the appearance of the colored product, 2-nitro-5-thiobenzoate (ε₄₁₂ = 14,150 M⁻¹ cm⁻¹) [23]. The reaction time point was taken as the moment the DTNB was added to the cuvette and all values were corrected for A₄₁₂ observed when DTNB was added to buffer alone. The reaction rate monitored by loss of thiol content was determined by direct fit to a single exponential equation. Pseudo-first order rate constants thus obtained were then replotted versus reagent concentration to obtain the second order rate constants for modification of cysteine at pH 7.

For evaluation of the reactivity of each of these reagents with a test protein, E. coli thioredoxin was first reduced by incubation with dithiothreitol (20 mM final concentration) at 25 °C for 10–20 min. Dithiothreitol was removed and the protein was exchanged into BPACE buffer at pH 7.0 by passing it through a PD-10 size exclusion column (GE-Healthcare). Protein concentration in the pooled fractions was determined using an extinction coefficient at 280 nm of 13,700 M⁻¹cm⁻¹ [24]. The solution was brought to a final concentration of 8 µM thioredoxin in 7.5 ml, and the reaction was started by the addition of 80 µM iodoacetamide or iodoacetanilide. A portion of the reaction mixture (0.48 ml) was removed at each time point and the content of free thiols was determined as described above for cysteine, except that 5 µl of a 30 mM stock of DTNB in dimethylsulfoxide was added. Apparent first order rate constants were again obtained by fitting the time course data to a single exponential equation.

Quantitative analysis by mass spectrometry of labeled cysteine-containing peptides

E. coli thioredoxin (250 µM) in 25 mM potassium phosphate buffer at pH 7 was incubated with 1 µM thioredoxin reductase and 500 µM NADPH at room temperature for 10–15 min to fully reduce the thioredoxin, then diluted 50-fold into BPACE buffer at pH 7.0. To this solution was added 50 µM of either d₀-iodoacetanilide or d₅-iodoacetanilide and the reaction was allowed to continue at 23 °C for varying times before being quenched with 2-mercaptoethanol. For each labeling agent, two samples were prepared; one reaction mixture was incubated for 30 min to obtain exclusively singly-labeled thioredoxin, and a separate reaction mixture for each was incubated for 25–40 h to obtain the doubly labeled protein. The amount and nature of labeled protein in each reaction mixture was determined by MALDI-TOF MS analysis of the intact proteins. For each set (singly and doubly labeled), the labeled proteins were combined in known ratios of d₅-labeled (1.25 nmol) and d₀-labeled (0 – 12.5 nmol) proteins, assuming that all of the protein in each sample was found in the desired labeled state. To concentrate and exchange the samples into a buffer suitable for trypsin digestion and MALDI-TOF MS analysis, each 0.5 ml sample was precipitated by the addition of 50 µl 100% (w/v) trichloroacetic acid for 20 min at −20° C followed by centrifugation at 14,000 × g for 15 min at 4 °C. The protein pellet was washed with 0.5 mL of 1:1 ethanol:ether, recentrifuged and resuspended in 50 µl of 36 mM ammmonium bicarbonate, 1 mM CaCl, and 10% acetonitrile for digestion with 1.25 µg trypsin (16 h at 37 °C). Each sample was mixed in a 1:1 ratio with matrix [0.01 mg sinapic acid (Fluka) in 1 ml of 50% acetonitrile, 0.3% trifluoroacetic acid] and spotted onto a MALDI target plate. The tryptic peptide of interest from thioredoxin contains both cysteinyl residues (ADGAILVDFWAEWCGPCK, 1979.89 Da unmodified), gaining 134.06 Da for each modification by iodoacetanilide. Following sample preparation and application to the target plate, the ratio of the heavy (d₅) and light (d₀) modified peptides in each sample was determined experimentally by measuring the relative intensities of the monoisotopic peaks for the singly-labeled peptides (at 2114 and 2119 Da for the d₀- and d₅-labeled peptides, respectively) and the doubly-labeled peptides (with alkylation of both cysteinyl residues in thioredoxin, at 2248 and 2258 Da for the d₀- and d₅-labeled Peptides, respectively) using a Bruker Daltonics MALDI-TOF mass spectrometer.

Generation of a “heavy” (d₅-labeled) thioredoxin standard

D₅-iodoacetanilide-labeled thioredoxin was generated by incubating 2.8 mM d₅-iodoacetanilide with 0.29 mM thioredoxin and 0.42 mM TCEP in 25 mM potassium phosphate buffer, with 1 mM EDTA, at pH 7.6 for 4 h at room temperature in the presence or absence of 4 M guanidine-HCl. The labeling reactions were quenched by the addition of 100 mM 2-mercaptoethanol and the protein samples were exchanged into 25 mM phosphate buffer containing 1 mM EDTA, pH 7.0, using multiple cycles of centrifugation and redilution using a Microcon-30 ultrafiltration device (Millipore). After establishing the content of singly and doubly labeled protein in each sample by MALDI-TOF MS (with the doubly labeled product predominating in the sample to which the guanidine had been added), the samples were combined to ensure the presence of approximately equal amounts of both singly and doubly labeled protein and aliquots were frozen at −80 °C until needed.

Determination of the pK_a value for reduced thioredoxin

E. coli thioredoxin was reduced using the physiological reductants (NADPH and thioredoxin reductase) as described above. The sample was then brought to a final concentration of 5.0 µM thioredoxin in 9 ml of BPACE buffer adjusted to the desired pH with either ammonium hydroxide or 20% sulfuric acid. The reaction was started by the addition of 50 µM d₀-iodoacetanilide and the samples were incubated at 23 °C in a temperature-controlled water bath. The final pH of each reaction mixture was measured. At each time point, 0.5 ml of the solution (2.5 nmol of thioredoxin) was removed and quenched with a final concentration of 250 mM 2-mercaptoethanol and 1.25 nmol of the d₅-labeled thioredoxin standard was added. As described above for the peptide quantitation, samples were TCA precipitated and exchanged into 36 mM ammonium bicarbonate, 1 mM CaCl, and 10% acetonitrile and incubated with 1.25 µg of trypsin for 16 h at 37 °C.

The extent of the thioredoxin reaction with d₀-iodoacetanilide at each time point was measured by MALDI-TOF MS analysis. Samples were mixed with sinapic acid and spotted onto a MALDI target plate as described above. Data were collected in positive ion mode using the Autorun feature and each sample was analyzed three times. To determine the relative ratio of the singly-labeled thioredoxin peptide, the intensity of the monoisotopic peak of the light derivative (2114 Da) was divided by the intensity of the monoisotopic peak of the heavy derivative (2119 Da). The same procedure was used for the doubly-labeled peaks at 2247 Da and 2257 Da for the light and heavy derivatives, respectively.

Rates of reaction were determined between pH 4 and pH 10.6 for thioredoxin reacting with one or two molecules of iodoacetanilide according to the consecutive, irreversible kinetic model of A → B → C, where A is unlabeled peptide, B is singly-labeled peptide, C is doubly-labeled peptide, and k₁ and k₂ are the rate constants for each Step, A → B and B → C, respectively [25]. Data for the generation of “B” over time t were plotted and directly fit to equation 1 using Kaleidagraph 4.0 (Synergy Software).

$${[\mathrm{B}]}_t=\frac{{[\mathrm{A}]}_0{k}_1}{k_2-k_1}[\text{exp}(-k_{1}t)-\text{exp}(-k_{2}t)]$$ Equation 1

The rate of appearance of species “C” can be described by equation 2.

$${[\mathrm{C}]}_t={[\mathrm{A}]}_0\left\{1+\frac{1}{k_1-k_2}[k_2\text{exp}(-k_{1}t)-k_1\text{exp}(-k_{2}t)]\right\}$$ Equation 2

Because k₁>> k₂ at all pH values, the rate equation for the less reactive cysteine (B → C) was simplified to a single exponential equation (equation 3).

$${[\mathrm{C}]}_t={[\mathrm{A}]}_0(1-\text{exp}(-k_{2}t))$$ Equation 3

The pK_a values for the fast- and slow-reacting cysteine residues were then calculated by direct fits of each observed rate constant to equation 4 [26], where y is the observed rate constant (k₁ or k₂) at a given x (pH), k_HA is the limiting rate constant for the reaction at low pH (the protonated, thiol form), and k_A⁻ is the limiting rate constant for the reaction at high pH (the deprotonated, thiolate form); k_HA and k_A⁻ were allowed to vary during the fits.

$$y=\frac{(k_{{\mathrm{A}}^{-}}\times {10}^x)+(k_{\text{HA}}\times {10}^{\text{pKa}})}{{10}^x+{10}^{\text{pKa}}}$$ Equation 4

Sequencing of the singly labeled thioredoxin peptide

A sample containing singly-labeled thioredoxin was generated by incubating pre-reduced thioredoxin with iodoacetanilide for 20 min in BPACE buffer at pH 6.0. As above, the sample was exchanged into 36 mM ammonium bicarbonate, 1 mM CaCl, and 10% acetonitrile and digested with 1.25 µg trypsin for 16 h at 37° C. The sample was spotted onto a MALDI target using a matrix made up of saturated dihydroxybenzoic acid in 50:50 acetonitrile:water with 0.1% trifluoroacetic acid, and the sequence of the peptide corresponding to a single N-phenyl iodoacetamide label (2113.8 Da) was determined by the fragmentation pattern obtained from a 4700 Proteomic Analyzer MALDI-TOF-TOF MS (Applied Biosystems).

RESULTS

Synthesis and testing of isotope-coded iodoacetamide derivatives

For development of an isotope-coded thiol reagent based on iodoacetamide that would allow for quantitative assessment of reaction progress by MALDI-TOF MS, we chose to derivatize iodoacetic acid with aniline using a simple, one step synthetic procedure where either normal aniline or perdeuterated aniline was coupled to iodoacetamide to yield N-phenyl iodoacetamide (d₀- or d₅-iodoacetanilide, Scheme 1) [18]. Thus, upon reaction with a thiol group, the products of the modification with these compounds differ by five atomic mass units, a difference that is sufficient to readily discriminate between the multiple isotopic peaks for the peptides labeled with d₀- iodoacetanilide and those for the peptides labeled with d₅-iodoacetanilide [27].

To assess the reactivity of iodoacetanilide with thiol groups compared with the commonly-used iodoacetamide reagent, various concentrations of the alkylating agents (0.25 to 2 mM) were added to solutions of 50 µM cysteine at pH 7, and thiol contents were assessed at each time point using DTNB assays. Pseudo-first order rate constants were then replotted versus reagent concentration to obtain the second order rate constants for reaction (Figure 1). By this measure, d₀-iodoacetanilide is 3-fold more reactive toward cysteine (at 110 M⁻¹ min⁻¹) than is iodoacetamide (36 M⁻¹ min⁻¹), indicating that the addition of the phenyl group to the amido nitrogen does not adversely affect the reactivity of this reagent toward thiol groups. As the iodoacetanilide was to be used with thioredoxin in the subsequent experiments and accessibility of sites of reaction could be more restricted in the protein than for cysteine, we also assessed the rate of reaction of 80 µM of each of the reagents toward 8 µM thioredoxin at pH 7. Under these conditions, a ~7-fold higher rate of reaction was observed for d₀-iodoacetanilide and thioredoxin (1.7 min⁻¹) compared with iodoacetamide and thioredoxin (0.24 min⁻¹; data not shown), verifying that iodoacetanilide is both reactive and not adversely affected by any restriction in its access to the most reactive thiol group in thioredoxin.

Figure 1. Rate constants for free cysteine reacting with iodoacetanilide and iodoacetamide.

Free cysteine (50 µM) was incubated with either iodoacetamide (circles) or iodoacetanilide (squares) at pH 7.0 and 23 °C. For each timepoint at a given reagent concentration, a portion of the reaction was removed and the free thiol content was determined by measuring the A₄₁₂ upon the addition of 300 µM DTNB. Values for the pseudo-first order rate constants (k_obs) were determined by direct fit to a single exponential equation at each reagent concentration. Shown are the k_obs values plotted versus reagent concentration; the second order rate constants were determined from the slope of the lines fit to the data in this plot.

Evaluation of the applicability of d₀- and d₅-iodoacetanilide modification to quantification of peptides by MALDI-TOF MS

To confirm that these reagents could be used to provide quantitative data for MALDI-TOF MS analyses, four separate labeled Trx samples were generated with either d₀-iodoacetanilide or d₅-iodoacetanilide, and incubated for either short (30 min) or long (25 – 40 h) times to obtain predominantly singly or doubly labeled protein, respectively. The differentially labeled pairs of protein samples were then mixed at various ratios, from 10-fold lower to 10-fold higher d₀-labeled protein compared with the constant amount of d₅-labeled protein. Samples were subsequently precipitated, digested with trypsin, and analyzed by MALDI-TOF mass spectrometry. The relative signal intensities for pairs of peptide ions labeled with the d₀- or d₅-iodoacetanilide were determined and compared with the expected ratios. As shown in Fig. 2, expected ratios from 1:5 to 5:1 of the d₀- versus d₅-labeled peptides gave experimental intensity ratios that were linear when compared with the expected ratios. While the fit was linear in both cases, the slope of the line was closer to 1 for singly labeled (1.09) than for doubly labeled Trx peptide (0.76). The variation from 1 for the slope of the doubly-labeled samples can be explained by the fact that the reaction did not proceed to as full an extent for the d₀ reagent as for the d₅-iodoacetanilide. This was because, despite the length of the reaction, it was difficult to consistently obtain a fully labeled Trx due to the slow reaction of Cys35; in this case, MALDI-TOF MS of the intact proteins showed a small, but detectable peak of singly-labeled Trx in the doubly-labeled sample generated with the d₀-iodoacetanilide. While not ideal, this was not a significant problem since neither this nor later experiments required an absolute concentration for the standard protein, but rather the consistent use of the same standard. The good correlation between expected and measured ratios does, in any case, indicate the suitability of this technique for measuring reaction progress, as long as a single sample of the d₅-modified Trx is used as the quantification standard for all of the d₀-modified samples to be analyzed at a given pH, and the amount of d₀-modified peptide is within 5-fold of the d₅-modified peptide. For our purposes, a large batch of d₅-acetanilide-labeled Trx was prepared which contained both singly- and doubly-labeled protein, and this standard was aliquoted for use in subsequent experiments.

Figure 2. Quantitative comparison of measured peptide ratios versus expected peptide ratios for d₀- and d₅-iodoacetanilide-labeled thioredoxin (Trx).

Pre-reduced E. coli Trx at pH 7 was incubated with either d₀-iodoacetanilide (light) or d₅-iodoacetanilide (heavy) in separate experiments for either 30 min or for 25–40 h to obtain predominantly singly or doubly labeled protein, respectively. Each pair of heavy and light protein samples was mixed together in known ratios of the heavy (1.25 nmol) and light (0 – 12.5 nmol) proteins, precipitated with TCA and exchanged into 10% acetonitrile-containing ammonium bicarbonate buffer as described in Materials and Methods, and digested with trypsin. The ratios of the heavy and light modified peptides were then determined for singly (open circles) and doubly labeled (closed circles) Trx samples by measuring the relative peptide intensities using a Bruker Daltonics MALDI-TOF MS. Plotted are the observed ratios versus the expected ratios (from 0.2 to 5) based on protein concentrations and the fit to a straight line.

In separate experiments, comparisons were made of the tryptic cleavage products of modified thioredoxin to determine whether or not this modification adversely affected cleavage by trypsin. For all modifications described in this paper (doubly-labeled, singly-labeled, d₀ or d₅) under the digestion conditions used, no peptides were observed that corresponded to a missed cleavage site on either side of the labeled peptide.

Evaluation of pK_a values for reduced Trx based on pH titration of the rates for iodoacetanilide modification

To determine the pK_a for the cysteinyl residues in Trx by our adapted chemical modification approach, reduced Trx was incubated with d₀-iodoacetanilide in BPACE buffer at various pH values and aliquots taken from the reaction mixtures at various time points were quenched with 2-mercaptoethanol. As alluded to above, a standard amount of Trx labeled with d₅-iodoacetanilide containing both the doubly-labeled and singly-labeled products was added to each sample prior to buffer exchange and trypsin digestion. The ratio of light peptide (d₀-labeled) to heavy peptide (d₅-labeled) was calculated from the intensity values of the first monoisotopic peak from each (Fig. 3a). As shown in Fig.3b, one cysteine reacts much more quickly than the other at all pH values (especially below pH 10), an observation which agrees with previous studies showing that Cys32 reacts much more quickly than Cys35 [11]. The irreversible reaction fits to an A → B → C model where A is unlabeled peptide, B is singly labeled peptide, and C is doubly labeled peptide, and the first step is fast and the second step slow. As described in Materials and Methods, the difference in rates allows the appearance of C to be fit to a single exponential equation (equation 3) while the appearance of B over time was fit to equation 1 (Fig. 3b). The experimentally determined rate constants for each step at various pH values from 4 to 10.6 were then plotted versus pH in order to determine the pK_a value for each cysteine (Fig. 4; equation 4). Using this approach, the calculated pK_a for the most reactive cysteine was 6.49 ± 0.18. For the less reactive cysteine, the exact pK_a value was too high to be determined given the very limited data above pH 10, but the value obtained from the fit was 10.13 ± 0.05 (Fig. 4).

Figure 3. MALDI-TOF MS data and fits to monitor the rates of reaction of iodoacetanilide with thioredoxin.

The extent of iodoacetanilide reaction with Trx was determined after trypsin digestion by analyzing peak intensities of the cysteine-containing peptide using a Bruker Daltonics MALDI-TOF MS (panel A, sample data for the appearance of the singly d₀-labeled peptide over time at pH 8.0). The ratio of peptide labeled with the d₀-iodoacetanilide to that of the d₅-labeled standard, which increased with time of reaction, was determined by dividing the intensity of the monoisotopic peak for the light peptide (~2114 Da) by the corresponding intensity of the monoisotopic peak for the heavy peptide (~2119 Da). As shown in panel B, the ratios for the singly labeled (2114 or 2119 Da, open circles) and doubly labeled (2247 or 2257 Da, closed circles) peptides were plotted as a function of time. Fits of the data to a kinetic model of A → B → C, where A is unlabeled peptide, B is singly-labeled peptide (open circles), and C is doubly-labeled peptide (closed circles) as shown by the lines were carried out using equation 1 and equation 3 as described in Materials and Methods.

Figure 4. Apparent pK_a determination for the fast- and slow-reacting cysteinyl residues of E. coli thioredoxin.

The pseudo-first order rate constants for the singly-labeled peptides (k₁, open circles) and doubly-labeled peptides (k₂, closed circles) of Trx were determined over the pH range of 4 to 10.6 as described in Materials and Methods and plotted as a function of pH. Data were fit to equation 4 as described, allowing the pK_a value for each as well as the rate constants for the upper and lower plateaus to vary. Final results gave a pK_a of 6.49 ± 0.18 and upper plateau of 0.537 ± 0.039 for the fast reacting cysteine (identified by subsequent MS-MS analysis as Cys32) and pK_a of 10.13 ± 0.05 and upper plateau of 0.196 ± 0.007 for the slow reacting cysteine (Cys35). Both rates dropped to zero at low pH, within experimental error (0.001 ± 0.048 and 0.0011 ± 0.0008, respectively).

Tandem mass spectrometry for identification of the fast-reacting cysteine residue in Trx

Previous studies have indicated that Cys32 reacts much more quickly than Cys35 [11] and therefore, based on data from the literature, the fast reaction with an observed pK_a of ~6.5 can be attributed to the alkylation of Cys32 while the slow rate is attributed to the alkylation of Cys35. To confirm that this is indeed the case in our experiments, the 2113.8 Da peptide from Trx that contained only a single iodoacetanilide label was sequenced using tandem MS analysis with a MALDI-TOF-TOF instrument. The fragmentation spectra included peptide fragments that indicated the presence of the label at Cys32 of Trx, whereas none were observed that supported a label at Cys35 (Fig. 5).

Figure 5. MALDI-TOF-TOF spectrum of the singly, d₀-iodoacetanilide-labeled peptide of thioredoxin.

The protein adduct was digested with trypsin in order to generate an 18-residue cysteine-containing peptide. The digest was spotted onto the MALDI target using DHB as the matrix and the labeled peptide (m/z 2113.85) was selected for fragmentation. The spectrum shown is the sum of 10 separate measurements. Cleavage of the amide bond results in N-terminal fragments designated as “b” and C-terminal fragments designated as “y”. The masses of both sets of ions clearly identify the d₀-iodoacetanilide linked covalently to Cys32 (b14 − b13 = y5 − y4 = 236.06 m/z). Peptides in the table marked in bold are found in the spectrum and asterisks indicate peaks in the spectrum that demonstrate labeling on Cys32. No peaks were observed that were consistent with labeling on Cys35.

DISCUSSION

We present here a new and sensitive MS-based method to determine the pK_a for cysteine residues in proteins using a combination of the well established technique of following chemical modification rates over a range of pH as a measure of thiol pK_a with a simple isotopic labeling of an iodoacetamide derivative to enable readout of the data by MS. The reagents required for this method are readily synthesized in reasonable quantities from commercially available precursors. In the original report of the synthesis of d₅-iodoacetanilide, the deuterated label was incorporated into proteins for neutron scattering studies [18]. In later modifications, ¹³C-labeled iodoacetanilide was generated for quantification of peptides and proteins using MS as a potential alternative to ICAT technology [27; 28]. With the present focus on pK_a analysis, these reagents allow for the use of minimal amounts of protein, given that MALDI-TOF MS is capable of detecting protein amounts considerably lower than those used in this study. While altered chromatographic separation of peptides possessing deuterium labels can adversely affect MS-based quantification and has led to a preference for ¹³C or ¹⁵N labels over ²H in isotope-coded tags [29; 30], the present method requires no separation of peptides before MALDI-TOF analysis and is unaffected by this limitation. Were analysis of a complex mixture of peptides to be carried out that required chromatographic separation prior to MS analysis, ¹³C-iodoacetanilide would be the better choice as the modifying agent [28].

With previous approaches using iodoacetamide reactivity to measure the pK_a of cysteine residues, subsequent assays for modification were unable to distinguish between or separately evaluate multiple cysteines in the same protein. By using tryptic digestion and MALDI-TOF MS as introduced here, the extent of reactivity with each cysteine residue can be monitored directly without the need for mutagenesis to distinguish cysteinyl residues. This method can be easily adapted to other less well characterized proteins that contain cysteine residues, including those for which the physiological substrate is unknown. Because each cysteine residue is monitored independently, this technique can easily accommodate situations where the various cysteines in the protein react at very different rates and exhibit quite different pK_a values. In most cases, the protease, or even multiple proteases, used for the digestion can be optimized to ensure that each cysteine in the protein is located on a separate peptide of appropriate size. Although this is not readily accomplished in the case of E. coli Trx, the two rates are easily distinguished. Cys32 has been shown to be solvent accessible while Cys35 is buried and protected from solvent [31]. This rate difference is further enhanced at pH values below 8 due to the differences in the pK_a values between Cys32 and Cys35. In addition, as we have done here, the identity of the fast and slow reacting cysteinyl residues (if they are on the same peptide) can easily be confirmed by using MS-MS analysis to sequence the modified peptides. While we did not observe any missed cleavage sites in Trx even after modification, the presence of missed cleavage sites for other modified peptides will have to be determined experimentally for any other proteins and proteases used and, if present, their ratios will need to be accounted for during data analysis.

To confirm that our method gives results comparable with other chemical modification approaches, we tested it with E. coli thioredoxin. This protein contains 2 cysteine residues that are critical for its activity, and the pK_a values for these residues have been determined by chemical modification with iodoacetamide in several instances, allowing us to easily compare our results to those previous determinations. The pK_a values presented here of 6.49 ± 0.18 for Cys32 and greater than 10 for Cys35 closely agree with previously published pK_a values using iodoacetamide and iodoacetic acid (6.7 or 7.1 for Cys32 and greater than 9 for Cys35) [11; 14]. Thus, we now have a MS-based technique for evaluating the pK_a values of specific cysteine residues that will be widely applicable and is not complicated by the presence of multiple cysteines within the protein of interest.